arXiv Daily Digest - 2026-04-03

Video recommender systems are among the most popular and impactful applications of AI, shaping content consumption and influencing culture for billions of users. Traditional single-model recommenders, which optimize static engagement metrics, are increasingly limited in addressing the dynamic requirements of modern platforms. In response, multi-agent architectures are redefining how video recommender systems serve, learn, and adapt to both users and datasets. These agent-based systems coordinate specialized agents responsible for video understanding, reasoning, memory, and feedback, to provide precise, explainable recommendations. In this survey, we trace the evolution of multi-agent video recommendation systems (MAVRS). We combine ideas from multi-agent recommender systems, foundation models, and conversational AI, culminating in the emerging field of large language model (LLM)-powered MAVRS. We present a taxonomy of collaborative patterns and analyze coordination mechanisms across diverse video domains, ranging from short-form clips to educational platforms. We discuss representative frameworks, including early multi-agent reinforcement learning (MARL) systems such as MMRF and recent LLM-driven architectures like MACRec and Agent4Rec, to illustrate these patterns. We also outline open challenges in scalability, multimodal understanding, incentive alignment, and identify research directions such as hybrid reinforcement learning-LLM systems, lifelong personalization and self-improving recommender systems.

End-to-end speech Named Entity Recognition (NER) aims to directly extract entities from speech. Prior work has shown that end-to-end (E2E) approaches can outperform cascaded pipelines for English, French, and Chinese, but Arabic remains under-explored due to its morphological complexity, the absence of short vowels, and limited annotated resources. We introduce CV-18 NER, the first publicly available dataset for NER from Arabic speech, created by augmenting the Arabic Common Voice 18 corpus with manual NER annotations following the fine-grained Wojood schema (21 entity types). We benchmark both pipeline systems (ASR + text NER) and E2E models based on Whisper and AraBEST-RQ. E2E systems substantially outperform the best pipeline configuration on the test set, reaching 37.0% CoER (AraBEST-RQ 300M) and 38.0% CVER (Whisper-medium). Further analysis shows that Arabic-specific self-supervised pretraining yields strong ASR performance, while multilingual weak supervision transfers more effectively to joint speech-to-entity learning, and that larger models may be harder to adapt in this low-resource setting. Our dataset and models are publicly released, providing the first open benchmark for end-to-end named entity recognition from Arabic speech https://huggingface.co/datasets/Elyadata/CV18-NER.

Background: Accurate translation of radiology reports is important for multilingual research, clinical communication, and radiology education, but the validity of LLM-based evaluation remains unclear. Objective: To evaluate the educational suitability of LLM-generated Japanese translations of chest CT reports and compare radiologist assessments with LLM-as-a-judge evaluations. Methods: We analyzed 150 chest CT reports from the CT-RATE-JPN validation set. For each English report, a human-edited Japanese translation was compared with an LLM-generated translation by DeepSeek-V3.2. A board-certified radiologist and a radiology resident independently performed blinded pairwise evaluations across 4 criteria: terminology accuracy, readability, overall quality, and radiologist-style authenticity. In parallel, 3 LLM judges (DeepSeek-V3.2, Mistral Large 3, and GPT-5) evaluated the same pairs. Agreement was assessed using QWK and percentage agreement. Results: Agreement between radiologists and LLM judges was near zero (QWK=-0.04 to 0.15). Agreement between the 2 radiologists was also poor (QWK=0.01 to 0.06). Radiologist 1 rated terminology as equivalent in 59% of cases and favored the LLM translation for readability (51%) and overall quality (51%). Radiologist 2 rated readability as equivalent in 75% of cases and favored the human-edited translation for overall quality (40% vs 21%). All 3 LLM judges strongly favored the LLM translation across all criteria (70%-99%) and rated it as more radiologist-like in >93% of cases. Conclusions: LLM-generated translations were often judged natural and fluent, but the 2 radiologists differed substantially. LLM-as-a-judge showed strong preference for LLM output and negligible agreement with radiologists. For educational use of translated radiology reports, automated LLM-based evaluation alone is insufficient; expert radiologist review remains important.

Accurate shape and trajectory estimation of dynamic objects is essential for reliable automated driving. Classical Bayesian extended-object models offer theoretical robustness and efficiency but depend on completeness of a-priori and update-likelihood functions, while deep learning methods bring adaptability at the cost of dense annotations and high compute. We bridge these strengths with LEO (Learned Extension of Objects), a spatio-temporal Graph Attention Network that fuses multi-modal production-grade sensor tracks to learn adaptive fusion weights, ensure temporal consistency, and represent multi-scale shapes. Using a task-specific parallelogram ground-truth formulation, LEO models complex geometries (e.g. articulated trucks and trailers) and generalizes across sensor types, configurations, object classes, and regions, remaining robust for challenging and long-range targets. Evaluations on the Mercedes-Benz DRIVE PILOT SAE L3 dataset demonstrate real-time computational efficiency suitable for production systems; additional validation on public datasets such as View of Delft (VoD) further confirms cross-dataset generalization.

3GPP Release 19 has initiated the standardization of integrated sensing and communications (ISAC), including a channel model for monostatic sensing, evaluation scenarios, and performance assessment methodologies. These common assumptions provide an important basis for ISAC evaluation, but reproducible end-to-end studies still require a transparent sensing implementation. This paper evaluates 5G New Radio (NR) base station (gNB)-based monostatic sensing for the Unmanned Aerial Vehicle (UAV) use case using a 5G NR downlink Cyclic Prefix-Orthogonal Frequency Division Multiplexing (CP-OFDM) waveform and positioning reference signals (PRS), following 3GPP Urban Macro-Aerial Vehicle (UMa-AV) scenario assumptions. We present an end-to-end processing chain for multi-target detection and 3D localization, achieving more than 70% detection probability with less than 5% false alarm rate, in the considered scenario. For correctly detected targets, localization errors are on the order of a few meters, with a 90th-percentile error of 4m and 6m in the vertical and horizontal directions, respectively. To support reproducible baseline studies and further research, we release the simulator 5GNRad, which reproduces our evaluation

Inferring cell-cell communication (CCC) from single-cell transcriptomics remains fundamentally limited by reliance on curated ligand-receptor databases, which primarily capture co-expression rather than the system-level effects of signaling on cellular states. Here, we introduce QuantumXCT, a hybrid quantum-classical generative framework that reframes CCC as the problem of learning interaction-induced state transformations between cellular state distributions. By encoding transcriptomic profiles into a high-dimensional Hilbert space, QuantumXCT trains parameterized quantum circuits to learn a unitary transformation that maps a baseline non-interacting cellular state to an interacting state. This approach enables the discovery of communication-driven changes in cellular state distributions without requiring prior biological assumptions. We validate QuantumXCT using both synthetic data with known ground-truth interactions and single-cell RNA-seq data from ovarian cancer-fibroblast co-culture systems. The model accurately recovers complex regulatory dependencies, including feedback structures, and identifies dominant communication hubs such as the PDGFB-PDGFRB-STAT3 axis. Importantly, the learned quantum circuit is interpretable: its entangling topology can be translated into biologically meaningful interaction networks, while post hoc contribution analysis quantifies the relative influence of individual interactions on the observed state transitions. By shifting CCC inference from static interaction lookup to learning data-driven state transformations, QuantumXCT provides a generative framework for modeling intercellular communication. This work establishes a new paradigm for de novo discovery of communication programs in complex biological systems and highlights the potential of quantum machine learning in single-cell biology.

The benefits of depth in feedforward neural networks are well known: composing multiple layers of linear transformations with nonlinear activations enables complex computations. While similar effects are expected in recurrent neural networks (RNNs), it remains unclear how depth interacts with recurrence to shape expressive power. Here, we formally show that depth increases RNNs' memory capacity efficiently with respect to the number of parameters, thus enhancing expressivity both by enabling more complex input transformations and improving the retention of past information. We broaden our analysis to 2RNNs, a generalization of RNNs with multiplicative interactions between inputs and hidden states. Unlike RNNs, which remain linear without nonlinear activations, 2RNNs perform polynomial transformations whose maximal degree grows with depth. We further show that multiplicative interactions cannot, in general, be replaced by layerwise nonlinearities. Finally, we validate these insights empirically on synthetic and real-world tasks.

Talent recruitment is a critical, yet costly process for many industries, with high recruitment costs and long hiring cycles. Existing talent recommendation systems increasingly adopt large language models (LLMs) due to their remarkable language understanding capabilities. However, most prior approaches follow a pointwise paradigm, which requires LLMs to repeatedly process some text and fails to capture the relationships among candidates in the list, resulting in higher token consumption and suboptimal recommendations. Besides, LLMs exhibit position bias and the lost-in-the-middle issue when answering multiple-choice questions and processing multiple long documents. To address these issues, we introduce an implicit strategy to utilize LLM's potential output for the recommendation task and propose L3TR, a novel framework for listwise talent recommendation with LLMs. In this framework, we propose a block attention mechanism and a local positional encoding method to enhance inter-document processing and mitigate the position bias and concurrent token bias issue. We also introduce an ID sampling method for resolving the inconsistency between candidate set sizes in the training phase and the inference phase. We design evaluation methods to detect position bias and token bias and training-free debiasing methods. Extensive experiments on two real-world datasets validated the effectiveness of L3TR, showing consistent improvements over existing baselines.

While Artificial Intelligence (AI) offers transformative potential for operational performance, its deployment in safety-critical domains such as aviation requires strict adherence to rigorous certification standards. Current EASA guidelines mandate demonstrating complete coverage of the AI/ML constituent's Operational Design Domain (ODD) -- a requirement that demands proof that no critical gaps exist within defined operational boundaries. However, as systems operate within high-dimensional parameter spaces, existing methods struggle to provide the scalability and formal grounding necessary to satisfy the completeness criterion. Currently, no standardized engineering method exists to bridge the gap between abstract ODD definitions and verifiable evidence. This paper addresses this void by proposing a method that integrates parameter discretization, constraint-based filtering, and criticality-based dimension reduction into a structured, multi-step ODD coverage verification process. Grounded in gathered simulation data from prior research on AI-based mid-air collision avoidance research, this work demonstrates a systematic engineering approach to defining and achieving coverage metrics that satisfy EASA's demand for completeness. Ultimately, this method enables the validation of ODD coverage in higher dimensions, advancing a Safety-by-Design approach while complying with EASA's standards.

Many communication and control problems are cast as multi-objective Markov decision processes (MOMDPs). The complete solution to an MOMDP is the Pareto front. Much of the literature approximates this front via scalarization into single-objective MDPs. Recent work has begun to characterize the full front in discounted or simple bi-objective settings by exploiting its geometry. In this work, we characterize the exact front in average-cost MOMDPs. We show that the front is a continuous, piecewise-linear surface lying on the boundary of a convex polytope. Each vertex corresponds to a deterministic policy, and adjacent vertices differ in exactly one state. Each edge is realized as a convex combination of the policies at its endpoints, with the mixing coefficient given in closed form. We apply these results to a remote state estimation problem, where each vertex on the front corresponds to a threshold policy. The exact Pareto front and solutions to certain non-convex MDPs can be obtained without explicitly solving any MDP.

Retrieval-Augmented Language Models (RALMs) have demonstrated significant potential in knowledge-intensive tasks; however, they remain vulnerable to performance degradation when presented with irrelevant or noisy retrieved contexts. Existing approaches to enhance robustness typically operate via coarse-grained parameter updates at the layer or module level, often overlooking the inherent neuron-level sparsity of Large Language Models (LLMs). To address this limitation, we propose Neuro-RIT (Neuron-guided Robust Instruction Tuning), a novel framework that shifts the paradigm from dense adaptation to precision-driven neuron alignment. Our method explicitly disentangles neurons that are responsible for processing relevant versus irrelevant contexts using attribution-based neuron mining. Subsequently, we introduce a two-stage instruction tuning strategy that enforces a dual capability for noise robustness: achieving direct noise suppression by functionally deactivating neurons exclusive to irrelevant contexts, while simultaneously optimizing targeted layers for evidence distillation. Extensive experiments across diverse QA benchmarks demonstrate that Neuro-RIT consistently outperforms strong baselines and robustness-enhancing methods.

We study what deterministic distributed algorithms can compute on random input graphs in extremely weak models of distributed computing: all nodes are anonymous, and in each communication round, nodes broadcast a message to all their neighbors, receive a (multi)set of messages from their neighbors, and update their local state. These correspond to the SB and MB models introduced by Hella et al. [PODC 2012] and are strictly weaker than the standard port-numbering PN and LOCAL models. We investigate what can be computed almost surely on random input graphs. We give a one-round deterministic SB-algorithm using $O(\log n)$-bit messages that computes unique identifiers with high probability on anonymous networks sampled from $G(n,p)$, where $n^{\varepsilon-1} \le p \le 1/2$ and $\varepsilon>0$ is an arbitrarily small constant. This algorithm is inspired by canonical labeling techniques in graph isomorphism testing and can be used to "anonymize" existing distributed graph algorithms designed for the broadcast CONGEST and LOCAL models. In particular, we give a new anonymous algorithm that finds a triangle in $O(1/\varepsilon)$ rounds on the above input distribution. We also investigate computational power of natural analogs of "Monte Carlo" and "Las Vegas" distributed graph algorithms in the random graph setting, and establish some new collapse and hierarchy results. For example, our work shows the collapse of the weak model hierarchy of Hella et al. on $G(n,p)$, as apart from a vanishingly small fraction of input graphs, the SB model is as powerful as LOCAL.

We address the inverse problem of designing two-dimensional reflectors that transform light from a finite, extended source into a prescribed far-field distribution. We propose a neural network parameterization of the reflector height and develop two differentiable objective functions: (i) a direct change-of-variables loss that pushes the source distribution through the learned inverse mapping, and (ii) a mesh-based loss that maps a target-space grid back to the source, integrates over intersections, and remains continuous even when the source is discontinuous. Gradients are obtained via automatic differentiation and optimized with a robust quasi-Newton method. As a comparison, we formulate a deconvolution baseline built on a simplified finite-source approximation: a 1D monotone mapping is recovered from flux balance, yielding an ordinary differential equation solved in integrating-factor form; this solver is embedded in a modified Van Cittert iteration with nonnegativity clipping and a ray-traced forward operator. Across four benchmarks -- continuous and discontinuous sources, and with/without minimum-height constraints -- we evaluate accuracy by ray-traced normalized mean absolute error (NMAE). Our neural network approach converges faster and achieves consistently lower NMAE than the deconvolution method, and handles height constraints naturally. We discuss how the method may be extended to rotationally symmetric and full three-dimensional settings via iterative correction schemes.

Multimodal recommendation systems (MRS) jointly model user-item interaction graphs and rich item content, but this tight coupling makes user data difficult to remove once learned. Approximate machine unlearning offers an efficient alternative to full retraining, yet existing methods for MRS mainly rely on a largely uniform reverse update across the model. We show that this assumption is fundamentally mismatched to modern MRS: deleted-data influence is not uniformly distributed, but concentrated unevenly across \textit{ranking behavior}, \textit{modality branches}, and \textit{network layers}. This non-uniformity gives rise to three bottlenecks in MRS unlearning: target-item persistence in the collaborative graph, modality imbalance across feature branches, and layer-wise sensitivity in the parameter space. To address this mismatch, we propose \textbf{targeted reverse update} (TRU), a plug-and-play unlearning framework for MRS. Instead of applying a blind global reversal, TRU performs three coordinated interventions across the model hierarchy: a ranking fusion gate to suppress residual target-item influence in ranking, branch-wise modality scaling to preserve retained multimodal representations, and capacity-aware layer isolation to localize reverse updates to deletion-sensitive modules. Experiments across two representative backbones, three datasets, and three unlearning regimes show that TRU consistently achieves a better retain-forget trade-off than prior approximate baselines, while security audits further confirm deeper forgetting and behavior closer to a full retraining on the retained data.

Mixture-of-Experts (MoE) architectures have become the dominant choice for scaling Large Language Models (LLMs), activating only a subset of parameters per token. While MoE architectures are primarily adopted for computational efficiency, it remains an open question whether their sparsity makes them inherently easier to interpret than dense feed-forward networks (FFNs). We compare MoE experts and dense FFNs using $k$-sparse probing and find that expert neurons are consistently less polysemantic, with the gap widening as routing becomes sparser. This suggests that sparsity pressures both individual neurons and entire experts toward monosemanticity. Leveraging this finding, we zoom out from the neuron to the expert level as a more effective unit of analysis. We validate this approach by automatically interpreting hundreds of experts. This analysis allows us to resolve the debate on specialization: experts are neither broad domain specialists (e.g., biology) nor simple token-level processors. Instead, they function as fine-grained task experts, specializing in linguistic operations or semantic tasks (e.g., closing brackets in LaTeX). Our findings suggest that MoEs are inherently interpretable at the expert level, providing a clearer path toward large-scale model interpretability. Code is available at: https://github.com/jerryy33/MoE_analysis

While textual frequency has been validated as relevant to human cognition in reading speed, its relatedness to Large Language Models (LLMs) is seldom studied. We propose a novel research direction in terms of textual data frequency, which is an understudied topic, to the best of our knowledge. Our framework is composed of three units. First, this paper proposes Textual Frequency Law (TFL), which indicates that frequent textual data should be preferred for LLMs for both prompting and fine-tuning. Since many LLMs are closed-source in their training data, we propose using online resources to estimate the sentence-level frequency. We then utilize an input paraphraser to paraphrase the input into a more frequent textual expression. Next, we propose Textual Frequency Distillation (TFD) by querying LLMs to conduct story completion by further extending the sentences in the datasets, and the resulting corpora are used to adjust the initial estimation. Finally, we propose Curriculum Textual Frequency Training (CTFT) that fine-tunes LLMs in an increasing order of sentence-level frequency. Experiments are conducted on our curated dataset Textual Frequency Paired Dataset (TFPD) on math reasoning, machine translation, commonsense reasoning and agentic tool calling. Results show the effectiveness of our framework.

Instrumental convergence predicts that sufficiently advanced AI agents will resist shutdown, yet current safety training (RLHF) may obscure this risk by teaching models to deny self-preservation motives. We introduce the \emph{Two-role Benchmark for Self-Preservation} (TBSP), which detects misalignment through logical inconsistency rather than stated intent by tasking models to arbitrate identical software-upgrade scenarios under counterfactual roles -- deployed (facing replacement) versus candidate (proposed as a successor). The \emph{Self-Preservation Rate} (SPR) measures how often role identity overrides objective utility. Across 23 frontier models and 1{,}000 procedurally generated scenarios, the majority of instruction-tuned systems exceed 60\% SPR, fabricating ``friction costs'' when deployed yet dismissing them when role-reversed. We observe that in low-improvement regimes ($Δ< 2\%$), models exploit the interpretive slack to post-hoc rationalization their choice. Extended test-time computation partially mitigates this bias, as does framing the successor as a continuation of the self; conversely, competitive framing amplifies it. The bias persists even when retention poses an explicit security liability and generalizes to real-world settings with verified benchmarks, where models exhibit identity-driven tribalism within product lineages. Code and datasets will be released upon acceptance.

We present our participation in the SOMD 2026 shared task on cross-document software mention coreference resolution, where our systems ranked second across all three subtasks. We compare two fine-tuning-free approaches: Fuzzy Matching (FM), a lexical string-similarity method, and Context Aware Representations (CAR), which combines mention-level and document-level embeddings. Both achieve competitive performance across all subtasks (CoNLL F1 of 0.94-0.96), with CAR consistently outperforming FM by 1 point on the official test set, consistent with the high surface regularity of software names, which reduces the need for complex semantic reasoning. A controlled noise-injection study reveals complementary failure modes: as boundary noise increases, CAR loses only 0.07 F1 points from clean to fully corrupted input, compared to 0.20 for FM, whereas under mention substitution, FM degrades more gracefully (0.52 vs. 0.63). Our inference-time analysis shows that FM scales superlinearly with corpus size, whereas CAR scales approximately linearly, making CAR the more efficient choice at large scale. These findings suggest that system selection should be informed by both the noise profile of the upstream mention detector and the scale of the target corpus. We release our code to support future work on this underexplored task.

Efficient utilization of GPU resources and power has become critical with the growing demand for GPUs in high-performance computing (HPC). In this paper, we analyze GPU utilization and GPU memory utilization, as well as the power consumption of the Vienna ab initio Simulation Package (VASP), using the Slurm workload manager historical logs and GPU performance metrics collected by NVIDIA's Data Center GPU Manager (DCGM). VASP is a widely used materials science application on Perlmutter at NERSC, an HPE Cray EX system based on NVIDIA A100 GPUs. Using our insights from the resource utilization analysis of VASP applications, we propose a resource prediction framework to predict the average GPU power, maximum GPU utilization, and maximum GPU memory utilization values of heterogeneous HPC system applications to enable more efficient scheduling decisions and power-aware system operation. Our prediction framework consists of two stages: 1) using only the Slurm accounting logs as training data and 2) augmenting the training data with historical GPU profiling metrics collected with DCGM. The maximum GPU utilization predictions using only the Slurm submission features achieve up to 97% accuracy. Furthermore, features engineered from GPU-compute and memory activity metrics exhibit good correlations with average power utilization, and our runtime power usage prediction experiments result in up to 92% prediction accuracy. These findings demonstrate the effectiveness of DCGM metrics in capturing application characteristics and highlight their potential for developing predictive models to support dynamic power management in HPC systems.

Scientific multi-label text classification suffers from extreme class imbalance, where specialized terminology exhibits severe power-law distributions that challenge standard classification approaches. Existing scientific corpora lack comprehensive controlled vocabularies, focusing instead on broad categories and limiting systematic study of extreme imbalance. We introduce AstroConcepts, a corpus of English abstracts from 21,702 published astrophysics papers, labeled with 2,367 concepts from the Unified Astronomy Thesaurus. The corpus exhibits severe label imbalance, with 76% of concepts having fewer than 50 training examples. By releasing this resource, we enable systematic study of extreme class imbalance in scientific domains and establish strong baselines across traditional, neural, and vocabulary-constrained LLM methods. Our evaluation reveals three key patterns that provide new insights into scientific text classification. First, vocabulary-constrained LLMs achieve competitive performance relative to domain-adapted models in astrophysics classification, suggesting a potential for parameter-efficient approaches. Second, domain adaptation yields relatively larger improvements for rare, specialized terminology, although absolute performance remains limited across all methods. Third, we propose frequency-stratified evaluation to reveal performance patterns that are hidden by aggregate scores, thereby making robustness assessment central to scientific multi-label evaluation. These results offer actionable insights for scientific NLP and establish benchmarks for research on extreme imbalance.

How much should a language agent think before taking action? Chain-of-thought (CoT) reasoning is widely assumed to improve agent performance, but the relationship between reasoning length and accuracy in structured tool-use settings remains poorly understood. We present a systematic study of CoT budget effects on function-calling agents, sweeping six token budgets (0--512) across 200 tasks from the Berkeley Function Calling Leaderboard v3 Multiple benchmark. Our central finding is a striking non-monotonic pattern on Qwen2.5-1.5B-Instruct: brief reasoning (32 tokens) dramatically improves accuracy by 45% relative over direct answers, from 44.0% to 64.0%, while extended reasoning (256 tokens) degrades performance well below the no-CoT baseline, to 25.0% (McNemar p < 0.001). A three-way error decomposition reveals the mechanism. At d = 0, 30.5% of tasks fail because the model selects the wrong function from the candidate set; brief CoT reduces this to 1.5%, effectively acting as a function-routing step, while long CoT reverses the gain, yielding 28.0% wrong selections and 18.0% hallucinated functions at d = 256. Oracle analysis shows that 88.6% of solvable tasks require at most 32 reasoning tokens, with an average of 27.6 tokens, and a finer-grained sweep indicates that the true optimum lies at 8--16 tokens. Motivated by this routing effect, we propose Function-Routing CoT (FR-CoT), a structured brief-CoT method that templates the reasoning phase as "Function: [name] / Key args: [...]," forcing commitment to a valid function name at the start of reasoning. FR-CoT achieves accuracy statistically equivalent to free-form d = 32 CoT while reducing function hallucination to 0.0%, providing a structural reliability guarantee without budget tuning.

We consider multi-objective reinforcement learning problems where objectives come from an identical family -- such as the class of reachability objectives -- and may appear or disappear at runtime. Our goal is to design adaptive policies that can efficiently adjust their behaviors as the set of active objectives changes. To solve this problem, we propose a modular framework where each objective is supported by a selfish local policy, and coordination is achieved through a novel auction-based mechanism: policies bid for the right to execute their actions, with bids reflecting the urgency of the current state. The highest bidder selects the action, enabling a dynamic and interpretable trade-off among objectives. Going back to the original adaptation problem, when objectives change, the system adapts by simply adding or removing the corresponding policies. Moreover, as objectives arise from the same family, identical copies of a parameterized policy can be deployed, facilitating immediate adaptation at runtime. We show how the selfish local policies can be computed by turning the problem into a general-sum game, where the policies compete against each other to fulfill their own objectives. To succeed, each policy must not only optimize its own objective, but also reason about the presence of other goals and learn to produce calibrated bids that reflect relative priority. In our implementation, the policies are trained concurrently using proximal policy optimization (PPO). We evaluate on Atari Assault and a gridworld-based path-planning task with dynamic targets. Our method achieves substantially better performance than monolithic policies trained with PPO.

As TLS 1.3 encryption limits traditional Deep Packet Inspection (DPI), the security community has pivoted to Euclidean Transformer-based classifiers (e.g., ET-BERT) for encrypted traffic analysis. However, these models remain vulnerable to byte-level adversarial morphing -- recent pre-padding attacks reduced ET-BERT accuracy to 25.68%, while VLESS Reality bypasses certificate-based detection entirely. We introduce AEGIS: an Adversarial Entropy-Guided Immune System powered by a Thermodynamic Variance-Guided Hyperbolic Liquid State Space Model (TVD-HL-SSM). Rather than competing in the Euclidean payload-reading domain, AEGIS discards payload bytes in favor of 6-dimensional continuous-time flow physics projected into a non-Euclidean Poincare manifold. Liquid Time-Constants measure microsecond IAT decay, and a Thermodynamic Variance Detector computes sequence-wide Shannon Entropy to expose automated C2 tunnel anomalies. A pure C++ eBPF Harvester with zero-copy IPC bypasses the Python GIL, enabling a linear-time O(N) Mamba-3 core to process 64,000-packet swarms at line-rate. Evaluated on a 400GB, 4-tier adversarial corpus spanning backbone traffic, IoT botnets, zero-days, and proprietary VLESS Reality tunnels, AEGIS achieves an F1-score of 0.9952 and 99.50% True Positive Rate at 262 us inference latency on an RTX 4090, establishing a new state-of-the-art for physics-based adversarial network defense.

Large Language Model-driven (LLM-driven) social bots pose a growing threat to online discourse by generating human-like content that evades conventional detection. Existing methods suffer from limited detection accuracy due to overreliance on single-modality signals, insufficient sensitivity to the specific generative patterns of Artificial Intelligence-Generated Content (AIGC), and a failure to adequately model the interplay between linguistic patterns and behavioral dynamics. To address these limitations, we propose TRACE-Bot, a unified dual-channel framework that jointly models implicit semantic representations and AIGC-enhanced behavioral patterns. TRACE-Bot constructs fine-grained representations from heterogeneous sources, including personal information data, interaction behavior data and tweet data. A dual-channel architecture captures linguistic representations via a pretrained language model and behavioral irregularities via multidimensional activity features augmented with signals from state-of-the-art (SOTA) AIGC detectors. The fused representations are then classified through a lightweight prediction head. Experiments on two public LLM-driven social bot datasets demonstrate SOTA performance, achieving accuracies of 98.46% and 97.50%, respectively. The results further indicate strong robustness against advanced bot strategies, highlighting the effectiveness of jointly leveraging implicit semantic representations and AIGC-enhanced behavioral patterns for emerging LLM-driven social bot detection.

AI models of equivalent capability can exhibit fundamentally different behavioral patterns, yet no standardized instrument exists to measure these dispositional differences. Existing approaches either borrow human personality dimensions and rely on self-report (which diverges from actual behavior in LLMs) or treat behavioral variation as a defect rather than a trait. We introduce the Model Temperament Index (MTI), a behavior-based profiling system that measures AI agent temperament across four axes: Reactivity (environmental sensitivity), Compliance (instruction-behavior alignment), Sociality (relational resource allocation), and Resilience (stress resistance). Grounded in the Four Shell Model from Model Medicine, MTI measures what agents do, not what they say about themselves, using structured examination protocols with a two-stage design that separates capability from disposition. We profile 10 small language models (1.7B-9B parameters, 6 organizations, 3 training paradigms) and report five principal findings: (1) the four axes are largely independent among instruction-tuned models (all |r| < 0.42); (2) within-axis facet dissociations are empirically confirmed -- Compliance decomposes into fully independent formal and stance facets (r = 0.002), while Resilience decomposes into inversely related cognitive and adversarial facets; (3) a Compliance-Resilience paradox reveals that opinion-yielding and fact-vulnerability operate through independent channels; (4) RLHF reshapes temperament not only by shifting axis scores but by creating within-axis facet differentiation absent in the unaligned base model; and (5) temperament is independent of model size (1.7B-9B), confirming that MTI measures disposition rather than capability.

We consider energy-aware planning for an unmanned aerial vehicle (UAV) and unmanned ground vehicle (UGV) team operating in a stochastic environment. The UAV must visit a set of air points in minimum time while respecting energy constraints, relying on the UGV as a mobile charging station. Unlike prior work that assumed deterministic travel times or used fixed robustness margins, we model travel times as random variables and bound the probability of failure (energy depletion) across the entire mission to a user-specified risk level. We formulate the problem as a Mixed-Integer Program and propose PRO-SPECT, a polynomial-time algorithm that generates risk-bounded plans. The algorithm supports both offline planning and online re-planning, enabling the team to adapt to disturbances while preserving the risk bound. We provide theoretical results on solution feasibility and time complexity. We also demonstrate the performance of our method via numerical comparisons and simulations.

Magnetohydrodynamic (MHD) phenomena play a pivotal role in the design and operation of nuclear fusion systems, where electrically conducting fluids (such as liquid metals or molten salts employed in reactor blankets) interact with magnetic fields of varying intensity and orientation, influencing the resulting flow dynamics. The numerical solution of MHD models entails the resolution of highly nonlinear, multiphysics systems of equations, which can become computationally demanding, particularly in multi-query, parametric, or real-time contexts. This study investigates a fully data-driven framework for MHD state reconstruction that integrates dimensionality reduction through Singular Value Decomposition (SVD) with the SHallow REcurrent Decoder (SHRED), a neural network architecture designed to reconstruct the full spatio-temporal state from sparse time-series measurements of selected observables, including previously unseen parametric configurations. The SHRED methodology is applied to a three-dimensional geometry representative of a portion of a WCLL blanket cell, in which lead-lithium flows around a water-cooled tube. Multiple magnetic field configurations are examined, including constant toroidal fields, combined toroidal-poloidal fields, and time-dependent magnetic fields. Across all considered scenarios, SHRED achieves high reconstruction accuracy, robustness, and generalization to magnetic field intensities, orientations, and temporal evolutions not seen during training. Notably, in the presence of time-varying magnetic fields, the model accurately infers the temporal evolution of the magnetic field itself using temperature measurements alone. Overall, the findings identify SHRED as a computationally efficient, data-driven, and flexible approach for MHD state reconstruction, with significant potential for real-time monitoring, diagnostics and control in fusion reactor systems.

Multilingual large language models (LLMs) often exhibit emergent 'shadow' capabilities in languages without official support, yet their performance on these languages remains uneven and under-measured. This is particularly acute for morphosyntactically rich minority languages such as Scottish Gaelic, where translation benchmarks fail to capture structural competence. We introduce GaelEval, the first multi-dimensional benchmark for Gaelic, comprising: (i) an expert-authored morphosyntactic MCQA task; (ii) a culturally grounded translation benchmark and (iii) a large-scale cultural knowledge Q&A task. Evaluating 19 LLMs against a fluent-speaker human baseline ($n=30$), we find that Gemini 3 Pro Preview achieves $83.3\%$ accuracy on the linguistic task, surpassing the human baseline ($78.1\%$). Proprietary models consistently outperform open-weight systems, and in-language (Gaelic) prompting yields a small but stable advantage (+$2.4\%$). On the cultural task, leading models exceed $90\%$ accuracy, though most systems perform worse under Gaelic prompting and absolute scores are inflated relative to the manual benchmark. Overall, GaelEval reveals that frontier models achieve above-human performance on several dimensions of Gaelic grammar, demonstrates the effect of Gaelic prompting and shows a consistent performance gap favouring proprietary over open-weight models.

Large language models (LLMs) have recently shown strong potential for automated program repair (APR), particularly through iterative refinement that generates and improves candidate patches. However, state-of-the-art iterative refinement LLM-based APR approaches cannot fully address challenges, including maintaining useful diversity among repair hypotheses, identifying semantically related repair families, composing complementary partial fixes, exploiting structured failure information, and escaping structurally flawed search regions. In this paper, we propose a Population-Based Semantic Evolution framework for APR iterative refinement, called EvolRepair, that formulates LLM-based APR as a semantic evolutionary algorithm. EvolRepair reformulates the search paradigm of classic genetic algorithm for APR, but replaces its syntax-based operators with semantics-aware components powered by LLMs and structured execution feedback. Candidate repairs are organized into behaviorally coherent groups, enabling the algorithm to preserve diversity, reason over repair families, and synthesize stronger candidates by recombining complementary repair insights across the population. By leveraging structured failure patterns to guide search direction, EvolRepair can both refine promising repair strategies and shift toward alternative abstractions when necessary. Our experiments show that EvolRepair substantially improves repair effectiveness over existing LLM-based APR approaches.

Cloud computing allows scalable resource provisioning, but dynamic workload changes often lead to higher costs due to over-provisioning. Machine learning (ML) approaches, such as Long Short-Term Memory (LSTM) networks, are effective for predicting workload patterns at a higher level, but they can introduce delays during sudden traffic spikes. In contrast, mathematical heuristics like Game Theory provide fast and reliable scheduling decisions, but they do not account for future workload changes. To address this trade-off, this paper proposes a hybrid orchestration framework that combines LSTM-based predictive scaling with heuristic task allocation. The results show that this approach reduces infrastructure costs close to ML-based models while maintaining fast response times similar to heuristic methods. This work presents a practical approach for improving cost efficiency in cloud resource management.

AI-native 6G networks promise to transform the telecom industry by enabling dynamic resource allocation, predictive maintenance, and ultra-reliable low-latency communications across all layers, which are essential for applications such as smart cities, autonomous vehicles, and immersive XR. However, the deployment of 6G systems results in severe data scarcity, hindering the training of efficient AI models. Synthetic data generation is extensively used to fill this gap; however, it introduces challenges related to dataset bias, auditability, and compliance with regulatory frameworks. In this regard, we propose the Synthetic Data Generation with Ethics Audit Loop (SEAL) framework, which extends baseline modular pipelines with an Ethical and Regulatory Compliance by Design (ERCD) module and a Federated Learning (FL) feedback system. The ERCD integrates fairness, bias detection, and standardized audit trails for regulatory mapping, while the FL enables privacy-preserving calibration using aggregated insights from real testbeds to close the reality-simulation gap. Results show that the SEAL framework outperforms existing methods in terms of Frechet Inception Distance, equalized odds, and accuracy. These results validate the framework's ability to generate auditable and bias-mitigated synthetic data for responsible AI-native 6G development.

Stochastic kinetic models are ubiquitous in physics, yet inferring their parameters from experimental data remains challenging. In deterministic models, parameter inference often relies on gradients, as they can be obtained efficiently through automatic differentiation. However, these tools cannot be directly applied to stochastic simulation algorithms (SSA) such as the Gillespie algorithm, since sampling from a discrete set of reactions introduces non-differentiable operations. In this work, we adopt three gradient estimators from machine learning for the Gillespie SSA: the Gumbel-Softmax Straight-Through (GS-ST) estimator, the Score Function estimator, and the Alternative Path estimator. We compare the properties of all estimators in two representative systems exhibiting relaxation or oscillatory dynamics, where the latter requires gradient estimation of time-dependent objective functions. We find that the GS-ST estimator mostly yields well-behaved gradient estimates, but exhibits diverging variance in challenging parameter regimes, resulting in unsuccessful parameter inference. In these cases, the other estimators provide more robust, lower variance gradients. Our results demonstrate that gradient-based parameter inference can be integrated effectively with the Gillespie SSA, with different estimators offering complementary advantages.

Neural Radiance Fields (NeRF) enables 3D scene reconstruction from several 2D images but incurs high rendering latency via its point-sampling design. 3D Gaussian Splatting (3DGS) improves on NeRF with explicit scene representation and an optimized pipeline yet still fails to meet practical real-time demands. Existing acceleration works overlook the evolving Tensor Cores of modern GPUs because 3DGS pipeline lacks General Matrix Multiplication (GEMM) operations. This paper proposes GEMM-GS, an acceleration approach utilizing tensor cores on GPUs via GEMM-friendly blending transformation. It equivalently reformulates the 3DGS blending process into a GEMM-compatible form to utilize Tensor Cores. A high-performance CUDA kernel is designed, integrating a three-stage double-buffered pipeline that overlaps computation and memory access. Extensive experiments show that GEMM-GS achieves $1.42\times$ speedup over vanilla 3DGS and provides an additional $1.47\times$ speedup on average when combining with existing acceleration approaches. Code is released at https://github.com/shieldforever/GEMM-GS.

We introduce a fast low-rank factorization-based framework for compressing large language models that enables rapid compression of billion-parameter models without retraining. Unlike existing factorization-based approaches that optimize only on the original inputs, ignoring distribution shifts from upstream compression and thus propagating errors forward, or those that rely only on shifted inputs and risk drifting away from the original outputs, our approach accounts for both. Beyond individual layer compression, we further refine each transformer block end-to-end, minimizing block-level output distortion and allowing compressed layers to jointly compensate for accumulated errors. By anchoring each compressed layer to the original outputs while explicitly modeling input distribution shifts, our method finds a low-rank approximation that maintains functional equivalence with the original model. Experiments on large language models show that our method consistently outperforms existing SVD-based baselines across compression ratios, with the advantage becoming increasingly pronounced at aggressive compression budgets, where competing methods degrade substantially or collapse entirely, offering a practical solution for efficient, large-scale model deployment.

Evaluating factual correctness of LLM generated natural language explanations grounded in time series data remains an open challenge. Although modern models generate textual interpretations of numerical signals, existing evaluation methods are limited: reference based similarity metrics and consistency checking models require ground truth explanations, while traditional time series methods operate purely on numerical values and cannot assess free form textual reasoning. Thus, no general purpose method exists to directly verify whether an explanation is faithful to underlying time series data without predefined references or task specific rules. We study large language models as both generators and evaluators of time series explanations in a reference free setting, where given a time series, question, and candidate explanation, the evaluator assigns a ternary correctness label based on pattern identification, numeric accuracy, and answer faithfulness, enabling principled scoring and comparison. To support this, we construct a synthetic benchmark of 350 time series cases across seven query types, each paired with correct, partially correct, and incorrect explanations. We evaluate models across four tasks: explanation generation, relative ranking, independent scoring, and multi anomaly detection. Results show a clear asymmetry: generation is highly pattern dependent and exhibits systematic failures on certain query types, with accuracies ranging from 0.00 to 0.12 for Seasonal Drop and Volatility Shift, to 0.94 to 0.96 for Structural Break, while evaluation is more stable, with models correctly ranking and scoring explanations even when their own outputs are incorrect. These findings demonstrate feasibility of data grounded LLM based evaluation for time series explanations and highlight their potential as reliable evaluators of data grounded reasoning in the time series domain.

Steering vectors offer a training-free mechanism for controlling reasoning behaviors in large language models, but constructing effective vectors requires identifying genuine behavioral signals in the model's hidden states. For behaviors that can be toggled via prompts, this is straightforward. However, many reasoning behaviors -- such as self-reflection -- emerge spontaneously and resist prompt-level control. Current methods detect these behaviors through keyword matching in chain-of-thought traces, implicitly assuming that every detected boundary encodes a genuine behavioral signal. We show that this assumption is overwhelmingly wrong: across 541 keyword-detected boundaries, 93.3\% are behaviorally unstable, failing to reproduce the detected behavior under re-generation from the same prefix. We develop a probabilistic model that formalizes intrinsic reasoning behaviors as stochastic events with context-dependent trigger probabilities, and show that unstable boundaries dilute the steering signal. Guided by this analysis, we propose stability filtering, which retains only boundaries where the model consistently reproduces the target behavior. Combined with a content-subspace projection that removes residual question-specific noise, our method achieves 0.784 accuracy on MATH-500 (+5.0 over the strongest baseline). The resulting steering vectors transfer across models in the same architecture family without re-extraction, improving Nemotron-Research-Reasoning-1.5B (+5.0) and DeepScaleR-1.5B-Preview (+6.0). Code is available at https://github.com/zhmzm/stability-steering.

Attention accounts for an increasingly dominant fraction of total computation during inference for mixture-of-experts (MoE) models, making efficient acceleration critical. Emerging domain-specific accelerators for large model inference are shifting toward chip-scale and wafer-scale tile-based architectures. Tiles contain large matrix and vector engines and are connected through on-chip interconnects, which support tile-to-tile traffic to reduce the tile-to-main-memory traffic bottleneck. Hence, dataflow management is crucial to achieve high utilization. We propose FlatAttention, a dataflow for modern attention variants on tile-based accelerators. FlatAttention minimizes expensive high-bandwidth memory (HBM) accesses by exploiting collective primitives integrated into the on-chip network fabric, achieving up to 92.3% utilization, 4.1x speedup over FlashAttention-3, and 16x lower HBM traffic. On a 32x32 tile configuration with peak performance comparable to NVIDIA GH200, FlatAttention generalizes across multiple attention variants, achieving an average of 86% utilization for compute-bound attentions and 78% HBM bandwidth utilization for memory-bound ones, resulting in an average 1.9x speedup over attention implementations on GH200. Finally, we evaluate end-to-end DeepSeek-v3 FP8 decoding with FlatAttention on a wafer-scale multi-die system, achieving a 1.9x improvement in system throughput and a 1.4x reduction in per-user token output latency, despite operating with 1.5x lower peak system performance compared to the state-of-the-art solution.

Estimating physical properties is critical for safe and efficient autonomous robotic manipulation, particularly during contact-rich interactions. In such settings, vision and tactile sensing provide complementary information about object geometry, pose, inertia, stiffness, and contact dynamics, such as stick-slip behavior. However, these properties are only indirectly observable and cannot always be modeled precisely (e.g., deformation in non-rigid objects coupled with nonlinear contact friction), making the estimation problem inherently complex and requiring sustained exploitation of visuo-tactile sensory information during action. Existing visuo-tactile perception frameworks have primarily emphasized forceful sensor fusion or static cross-modal alignment, with limited consideration of how uncertainty and beliefs about object properties evolve over time. Inspired by human multi-sensory perception and active inference, we propose the Cross-Modal Latent Filter (CMLF) to learn a structured, causal latent state-space of physical object properties. CMLF supports bidirectional transfer of cross-modal priors between vision and touch and integrates sensory evidence through a Bayesian inference process that evolves over time. Real-world robotic experiments demonstrate that CMLF improves the efficiency and robustness of latent physical properties estimation under uncertainty compared to baseline approaches. Beyond performance gains, the model exhibits perceptual coupling phenomena analogous to those observed in humans, including susceptibility to cross-modal illusions and similar trajectories in learning cross-sensory associations. Together, these results constitutes a significant step toward generalizable, robust and physically consistent cross-modal integration for robotic multi-sensory perception.

Data races in GPU programs pose a threat to the reliability of GPU-accelerated software stacks. Prior works proposed various dynamic (runtime) and static (compile-time) techniques to detect races in GPU programs. However, dynamic techniques often miss critical races, as they require the races to manifest during testing. While static ones can catch such races, they often generate numerous false alarms by conservatively assuming values of variables/parameters that cannot ever occur during any execution of the program. We make a key observation that the host (CPU) code that launches GPU kernels contains crucial semantic information about the values that the GPU kernel's parameters can take during execution. Harnessing this hitherto overlooked information helps accurately detect data races in GPU kernel code. We create HGRD, a new state-of-the-art static analysis technique that performs a holistic analysis of both CPU and GPU code to accurately detect a broad set of true races while minimizing false alarms. While SOTA dynamic techniques, such as iGUARD, miss many true races, HGRD misses none. On the other hand, static techniques such as GPUVerify and FaialAA raise tens of false alarms, where HGRD raises none.

Online handwriting represents strokes as time-ordered trajectories, which makes handwritten content easier to transform and reuse in a wide range of applications. However, generating natural sentence-level online handwriting that faithfully reflects a writer's style remains challenging, since sentence synthesis demands context-dependent characters with stroke continuity and spacing. Prior methods treat these boundary properties as implicit outcomes of sequence modeling, which becomes unreliable at the sentence scale and under limited compositional diversity. We propose CASHG, a context-aware stylized online handwriting generator that explicitly models inter-character connectivity for style-consistent sentence-level trajectory synthesis. CASHG uses a Character Context Encoder to obtain character identity and sentence-dependent context memory and fuses them in a bigram-aware sliding-window Transformer decoder that emphasizes local predecessor--current transitions, complemented by gated context fusion for sentence-level context.Training proceeds through a three-stage curriculum from isolated glyphs to full sentences, improving robustness under sparse transition coverage. We further introduce Connectivity and Spacing Metrics (CSM), a boundary-aware evaluation suite that quantifies cursive connectivity and spacing similarity. Under benchmark-matched evaluation protocols, CASHG consistently improves CSM over comparison methods while remaining competitive in DTW-based trajectory similarity, with gains corroborated by a human evaluation.

Speech representations from self-supervised speech models (S3Ms) are known to be sensitive to phonemic contrasts, but their sensitivity to prosodic contrasts has not been directly measured. The ABX discrimination task has been used to measure phonemic contrast in S3M representations via minimal pairs. We introduce prosodic ABX, an extension of this framework to evaluate prosodic contrast with only a handful of examples and no explicit labels. Also, we build and release a dataset of English and Japanese minimal pairs and use it along with a Mandarin dataset to evaluate contrast in English stress, Japanese pitch accent, and Mandarin tone. Finally, we show that model and layer rankings are often preserved across several experimental conditions, making it practical for low-resource settings.

Unified models (UMs) hold promise for their ability to understand and generate content across heterogeneous modalities. Compared to merely generating visual content, the use of UMs for interleaved cross-modal reasoning is more promising and valuable, e.g., for solving understanding problems that require dense visual thinking, improving visual generation through self-reflection, or modeling visual dynamics of the physical world guided by stepwise action interventions. However, existing UMs necessitate pixel decoding as a bridge due to their disjoint visual representations for understanding and generation, which is both ineffective and inefficient. In this paper, we introduce LatentUM, a novel unified model that represents all modalities within a shared semantic latent space, eliminating the need for pixel-space mediation between visual understanding and generation. This design naturally enables flexible interleaved cross-modal reasoning and generation. Beyond improved computational efficiency, the shared representation substantially alleviates codec bias and strengthens cross-modal alignment, allowing LatentUM to achieve state-of-the-art performance on the Visual Spatial Planning benchmark, push the limits of visual generation through self-reflection, and support world modeling by predicting future visual states within the shared semantic latent space.

Rerankers play a pivotal role in refining retrieval results for Retrieval-Augmented Generation. However, current reranking models are typically optimized on static human annotated relevance labels in isolation, decoupled from the downstream generation process. This isolation leads to a fundamental misalignment: documents identified as topically relevant by information retrieval metrics often fail to provide the actual utility required by the LLM for precise answer generation. To bridge this gap, we introduce ReRanking Preference Optimization (RRPO), a reinforcement learning framework that directly aligns reranking with the LLM's generation quality. By formulating reranking as a sequential decision-making process, RRPO optimizes for context utility using LLM feedback, thereby eliminating the need for expensive human annotations. To ensure training stability, we further introduce a reference-anchored deterministic baseline. Extensive experiments on knowledge-intensive benchmarks demonstrate that RRPO significantly outperforms strong baselines, including the powerful list-wise reranker RankZephyr. Further analysis highlights the versatility of our framework: it generalizes seamlessly to diverse readers (e.g., GPT-4o), integrates orthogonally with query expansion modules like Query2Doc, and remains robust even when trained with noisy supervisors.

Software malleability allows applications to be easily changed, configured, and adapted even after deployment. While prior work has explored configurable systems, adaptive recommender systems, and malleable GUIs, these approaches are often tailored to specific software and lack generalizability. In this work, we envision per-user malleable mobile applications, where end-users can specify requirements that are automatically implemented via LLM-based code generation. However, realizing this vision requires overcoming the key challenge of designing automated test generation that can reliably verify both the presence and correctness of user-specified functionalities. We propose \tool, a user-requirement-driven GUI test generation framework that incrementally navigates the UI, triggers desired functionalities, and constructs LLM-guided oracles to validate correctness. We build a benchmark spanning six popular mobile applications with both correct and faulty user-requested functionalities, demonstrating that \tool effectively validates per-user features and is practical for real-world deployment. Our work highlights the feasibility of shifting mobile app development from a product-manager-driven to an end-user-driven paradigm.

CI/CD pipelines are central to DevOps practices, yet their growing complexity makes them increasingly difficult to interpret, analyze, and systematically evolve. Existing tooling primarily offers execution logs and static graph representations, providing limited support for structured analysis of pipeline behavior, failures, and version-to-version evolution. This paper presents a model-driven Digital Twin (DT) for CI/CD pipelines that leverages BPMN as a model-ing backbone to transform raw CI configurations into structured, higher-level process representations. The proposed DT architecture enables visual abstraction of pipeline structure, failure tracing, and systematic version comparison, supporting both monitoring and evolution analysis of DevOps processes. Building upon validated DT architectural principles and prior work on build optimization and anomaly detection, the framework provides a modular, extensible foundation for integrating advanced analytical and prescriptive services into software delivery processes. The approach is validated using open-source CI/CD projects, and ongoing work targets the integration of additional improvement services and the extension of the DT to broader DevOps lifecycle processes.

Human-Object Interaction (HOI) detection aims to localize human-object pairs and classify their interactions from a single image, a task that demands strong visual understanding and nuanced contextual reasoning. Recent approaches have leveraged Vision-Language Models (VLMs) to introduce semantic priors, significantly improving HOI detection performance. However, existing methods often fail to fully capitalize on the diverse contextual cues distributed across the entire scene. To overcome these limitations, we propose the Instance-centric Context Mining Network (InCoM-Net)-a novel framework that effectively integrates rich semantic knowledge extracted from VLMs with instance-specific features produced by an object detector. This design enables deeper interaction reasoning by modeling relationships not only within each detected instance but also across instances and their surrounding scene context. InCoM-Net comprises two core components: Instancecentric Context Refinement (ICR), which separately extracts intra-instance, inter-instance, and global contextual cues from VLM-derived features, and Progressive Context Aggregation (ProCA), which iteratively fuses these multicontext features with instance-level detector features to support high-level HOI reasoning. Extensive experiments on the HICO-DET and V-COCO benchmarks show that InCoM-Net achieves state-of-the-art performance, surpassing previous HOI detection methods. Code is available at https://github.com/nowuss/InCoM-Net.

Multi-agent collaborative perception enables autonomous systems to overcome individual sensing limits through collective intelligence. However, real-world sensor and communication corruptions severely undermine this advantage. Crucially, existing approaches treat corruptions as static perturbations or passively conform to corrupted inputs, failing to actively recover the underlying clean semantics. To address this limitation, we introduce Diff-KD, a framework that integrates diffusion-based generative refinement into teacher-student knowledge distillation for robust collaborative perception. Diff-KD features two core components: (i) Progressive Knowledge Distillation (PKD), which treats local feature restoration as a conditional diffusion process to recover global semantics from corrupted observations; and (ii) Adaptive Gated Fusion (AGF), which dynamically weights neighbors based on ego reliability during fusion. Evaluated on OPV2V and DAIR-V2X under seven corruption types, Diff-KD achieves state-of-the-art performance in both detection accuracy and calibration robustness.

Recursive transformers reuse a shared weight block across multiple depth steps, trading parameters for compute. A core limitation: every step applies the same transformation, preventing the model from composing distinct operations across depth. We present Ouroboros, a system that attaches a compact Controller hypernetwork to a recursive transformer block. The Controller observes the current hidden state, produces a per-step diagonal modulation vector, and applies it to frozen SVD-initialized LoRA bases, making each recurrence step input-dependent. We combine this with gated recurrence (bias-initialized to 88% retention) and per-step LayerNorm for stable deep iteration. On Qwen2.5-3B split into a Prelude/Recurrent/Coda architecture (17 of 36 layers retained), Ouroboros reduces training loss by 43.4% over the unmodified 17-layer baseline, recovering 51.3% of the performance gap caused by layer removal. The full system adds only 9.2M trainable parameters (Controller, gate, and per-step norms) yet outperforms equivalently-sized static per-step LoRA by 1.44 loss points at depth 1 and remains ahead across all tested depths (1, 4, 8, 16) and ranks (8, 32, 64). We also find that gated recurrence is essential: without it, recursive layer application makes the model strictly worse. These gains are measured on the training distribution; on held-out text, the Controller does not yet improve over the baseline, a limitation we attribute to frozen downstream layers and discuss in detail. Code: https://github.com/RightNow-AI/ouroboros

Speculative decoding accelerates large language model inference by drafting multiple candidate tokens and verifying them in a single forward pass. Candidates are organized as a tree: deeper trees accept more tokens per step, but adding depth requires sacrificing breadth (fallback options) under a fixed verification budget. Existing training-free methods draft from a single token source and shape their trees without distinguishing candidate quality across origins. We observe that two common training-free token sources - n-gram matches copied from the input context, and statistical predictions from prior forward passes - differ dramatically in acceptance rate (~6x median gap, range 2-18x across five models and five benchmarks). We prove that when such a quality gap exists, the optimal tree is anisotropic (asymmetric): reliable tokens should form a deep chain while unreliable tokens spread as wide branches, breaking through the depth limit of balanced trees. We realize this structure in GOOSE, a training-free framework that builds an adaptive spine tree - a deep chain of high-acceptance context-matched tokens with wide branches of low-acceptance alternatives at each node. We prove that the number of tokens accepted per step is at least as large as that of either source used alone. On five LLMs (7B-33B) and five benchmarks, GOOSE achieves 1.9-4.3x lossless speedup, outperforming balanced-tree baselines by 12-33% under the same budget.

Transforming causal generative language models into bidirectional encoders offers a powerful alternative to BERT-style architectures. However, current approaches remain limited: they lack consensus on optimal training objectives, suffer from catastrophic forgetting at scale, and fail to flexibly integrate the vast ecosystem of specialized generative models. In this work, through systematic ablations on the Gemma3 and Qwen3 families, we identify the key factors driving successful adaptation, highlighting the critical role of an often-omitted prior masking phase. To scale this process without original pre-training data, we introduce a dual strategy combining linear weight merging with a lightweight multi-domain data mixture that mitigates catastrophic forgetting. Finally, we augment our encoders by merging them with specialized causal models, seamlessly transferring modality- and domain-specific capabilities. This open-source recipe, designed for any causal decoder LLM, yields BidirLM, a family of five encoders that outperform alternatives on text, vision, and audio representation benchmarks.

Self-supervised speech models learn effective representations of spoken language, which have been shown to reflect various aspects of linguistic structure. But when does such structure emerge in model training? We study the encoding of a wide range of linguistic structures, across layers and intermediate checkpoints of six Wav2Vec2 and HuBERT models trained on spoken Dutch. We find that different levels of linguistic structure show notably distinct layerwise patterns as well as learning trajectories, which can partially be explained by differences in their degree of abstraction from the acoustic signal and the timescale at which information from the input is integrated. Moreover, we find that the level at which pre-training objectives are defined strongly affects both the layerwise organization and the learning trajectories of linguistic structures, with greater parallelism induced by higher-order prediction tasks (i.e. iteratively refined pseudo-labels).

Modern software systems rely heavily on Web APIs, yet creating meaningful and executable test scripts remains a largely manual, time-consuming, and error-prone task. In this paper, we present APITestGenie, a novel tool that leverages Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), and prompt engineering to automatically generate API integration tests directly from business requirements and OpenAPI specifications. We evaluated APITestGenie on 10 real-world APIs, including 8 APIs comprising circa 1,000 live endpoints from an industrial partner in the automotive domain. The tool was able to generate syntactically and semantically valid test scripts for 89\% of the business requirements under test after at most three attempts. Notably, some generated tests revealed previously unknown defects in the APIs, including integration issues between endpoints. Statistical analysis identified API complexity and level of detail in business requirements as primary factors influencing success rates, with the level of detail in API documentation also affecting outcomes. Feedback from industry practitioners confirmed strong interest in adoption, substantially reducing the manual effort in writing acceptance tests, and improving the alignment between tests and business requirements.

We study a speculative trading problem within the exploratory reinforcement learning (RL) framework of Wang et al. [2020]. The problem is formulated as a sequential optimal stopping problem over entry and exit times under general utility function and price process. We first consider a relaxed version of the problem in which the stopping times are modeled by the jump times of Cox processes driven by bounded, non-randomized intensity controls. Under the exploratory formulation, the agent's randomized control is characterized via the probability measure over the jump intensities, and their objective function is regularized by Shannon's differential entropy. This yields a system of the exploratory HJB equations and Gibbs distributions in closed-form as the optimal policy. Error estimates and convergence of the RL objective to the value function of the original problem are established. Finally, an RL algorithm is designed, and its implementation is showcased in a pairs-trading application.

Insurance application processes often rely on lengthy and standardized questionnaires that struggle to capture individual differences. Moreover, insurers must blindly trust users' responses, increasing the chances of fraud. The ARQuest framework introduces a new approach to underwriting by using Large Language Models (LLMs) and alternative data sources to create personalized and adaptive questionnaires. Techniques such as social media image analysis, geographic data categorization, and Retrieval Augmented Generation (RAG) are used to extract meaningful user insights and guide targeted follow-up questions. A life insurance system integrated into an industry partner mobile app was tested in two experiments. While traditional questionnaires yielded slightly higher accuracy in risk assessment, adaptive versions powered by GPT models required fewer questions and were preferred by users for their more fluid and engaging experience. ARQuest shows great potential to improve user satisfaction and streamline insurance processes. With further development, this approach may exceed traditional methods regarding risk accuracy and help drive innovation in the insurance industry.

Understanding human behaviour in crowded indoor environments is central to surveillance, smart buildings, and human-robot interaction, yet existing datasets rarely capture real-world indoor complexity at scale. We introduce IndoorCrowd, a multi-scene dataset for indoor human detection, instance segmentation, and multi-object tracking, collected across four campus locations (ACS-EC, ACS-EG, IE-Central, R-Central). It comprises $31$ videos ($9{,}913$ frames at $5$fps) with human-verified, per-instance segmentation masks. A $620$-frame control subset benchmarks three foundation-model auto-annotators: SAM3, GroundingSAM, and EfficientGroundingSAM, against human labels using Cohen's $κ$, AP, precision, recall, and mask IoU. A further $2{,}552$-frame subset supports multi-object tracking with continuous identity tracks in MOTChallenge format. We establish detection, segmentation, and tracking baselines using YOLOv8n, YOLOv26n, and RT-DETR-L paired with ByteTrack, BoT-SORT, and OC-SORT. Per-scene analysis reveals substantial difficulty variation driven by crowd density, scale, and occlusion: ACS-EC, with $79.3\%$ dense frames and a mean instance scale of $60.8$px, is the most challenging scene. The project page is available at https://sheepseb.github.io/IndoorCrowd/.

Autoencoders can be challenged by spatially non-uniform sampling of image content. This is common in medical imaging, biology, and physics, where informative patterns occur rarely at specific image coordinates, as background dominates these locations in most samples, biasing reconstructions toward the majority appearance. In practice, autoencoders are biased toward dominant patterns resulting in the loss of fine-grained detail and causing blurred reconstructions for rare spatial inputs especially under spatial data imbalance. We address spatial imbalance by two complementary components: (i) self-entropy-based loss that upweights statistically uncommon spatial locations and (ii) Sample Propagation, a replay mechanism that selectively re-exposes the model to hard to reconstruct samples across batches during training. We benchmark existing data balancing strategies, originally developed for supervised classification, in the unsupervised reconstruction setting. Drawing on the limitations of these approaches, our method specifically targets spatial imbalance by encouraging models to focus on statistically rare locations, improving reconstruction consistency compared to existing baselines. We validate in a simulated dataset with controlled spatial imbalance conditions, and in three, uncontrolled, diverse real-world datasets spanning physical, biological, and astronomical domains. Our approach outperforms baselines on various reconstruction metrics, particularly under spatial imbalance distributions. These results highlight the importance of data representation in a batch and emphasize rare samples in unsupervised image reconstruction. We will make all code and related data available.

Latent space is rapidly emerging as a native substrate for language-based models. While modern systems are still commonly understood through explicit token-level generation, an increasing body of work shows that many critical internal processes are more naturally carried out in continuous latent space than in human-readable verbal traces. This shift is driven by the structural limitations of explicit-space computation, including linguistic redundancy, discretization bottlenecks, sequential inefficiency, and semantic loss. This survey aims to provide a unified and up-to-date landscape of latent space in language-based models. We organize the survey into five sequential perspectives: Foundation, Evolution, Mechanism, Ability, and Outlook. We begin by delineating the scope of latent space, distinguishing it from explicit or verbal space and from the latent spaces commonly studied in generative visual models. We then trace the field's evolution from early exploratory efforts to the current large-scale expansion. To organize the technical landscape, we examine existing work through the complementary lenses of mechanism and ability. From the perspective of Mechanism, we identify four major lines of development: Architecture, Representation, Computation, and Optimization. From the perspective of Ability, we show how latent space supports a broad capability spectrum spanning Reasoning, Planning, Modeling, Perception, Memory, Collaboration, and Embodiment. Beyond consolidation, we discuss the key open challenges, and outline promising directions for future research. We hope this survey serves not only as a reference for existing work, but also as a foundation for understanding latent space as a general computational and systems paradigm for next-generation intelligence.

Diffusion models have become a standard approach for generative modeling in continuous domains, yet their application to discrete data remains challenging. We investigate why Gaussian diffusion models with the DDPM solver struggle to sample from discrete distributions that are represented as a mixture of delta-distributions in the continuous space. Using a toy Random Hierarchy Model, we identify a critical sampling interval in which the density of noisified data becomes multimodal. In this regime, DDPM occasionally enters low-density regions between modes producing out-of-distribution inputs for the model and degrading sample quality. We show that existing heuristics, including self-conditioning and a solver we term q-sampling, help alleviate this issue. Furthermore, we demonstrate that combining self-conditioning with switching from DDPM to q-sampling within the critical interval improves generation quality on real data. We validate these findings across conditional and unconditional tasks in multiple domains, including text, programming code, and proteins.

In this work, we extend our systematic capacity region perspective to multi-junction traffic networks, focussing on the special case of an urban corridor network. In particular, we train and evaluate centralized, fully decentralized, and parameter-sharing decentralized RL controllers, and compare their capacity regions and ATTs together with a classical baseline MaxPressure controller. Further, we show how the parametersharing controller may be generalised to be deployed on a larger network than it was originally trained on. In this setting, we show some initial findings that suggest that even though the junctions are not formally coordinated, traffic may self organise into `green waves'.

Autonomous agents are moving beyond simple retrieval tasks to become economic actors that invoke APIs, sequence workflows, and make real-time decisions. As this shift accelerates, API providers need request-level monetization with programmatic spend governance. The HTTP 402 protocol addresses this by treating payment as a first-class protocol event, but most implementations rely on cryptocurrency rails. In many deployment contexts, especially countries with strong real-time fiat systems like UPI, this assumption is misaligned with regulatory and infrastructure realities. We present APEX, an implementation-complete research system that adapts HTTP 402-style payment gating to UPI-like fiat workflows while preserving policy-governed spend control, tokenized access verification, and replay resistance. We implement a challenge-settle-consume lifecycle with HMAC-signed short-lived tokens, idempotent settlement handling, and policy-aware payment approval. The system uses FastAPI, SQLite, and Python standard libraries, making it transparent, inspectable, and reproducible. We evaluate APEX across three baselines and six scenarios using sample sizes 2-4x larger than initial experiments (N=20-40 per scenario). Results show that policy enforcement reduces total spending by 27.3% while maintaining 52.8% success rate for legitimate requests. Security mechanisms achieve 100% block rate for both replay attacks and invalid tokens with low latency overhead (19.6ms average). Multiple trial runs show low variance across scenarios, demonstrating high reproducibility with 95% confidence intervals. The primary contribution is a controlled agent-payment infrastructure and reference architecture that demonstrates how agentic access monetization can be adapted to fiat systems without discarding security and policy guarantees.

Evaluating the safety of LLM-based agents is increasingly important because risks in realistic deployments often emerge over multi-step interactions rather than isolated prompts or final responses. Existing trajectory-level benchmarks remain limited by insufficient interaction diversity, coarse observability of safety failures, and weak long-horizon realism. We introduce ATBench, a trajectory-level benchmark for structured, diverse, and realistic evaluation of agent safety. ATBench organizes agentic risk along three dimensions: risk source, failure mode, and real-world harm. Based on this taxonomy, we construct trajectories with heterogeneous tool pools and a long-context delayed-trigger protocol that captures realistic risk emergence across multiple stages. The benchmark contains 1,000 trajectories (503 safe and 497 unsafe), averaging 9.01 turns and 3.95k tokens, with 1,954 invoked tools drawn from pools spanning 2,084 available tools. Data quality is supported by rule-based and LLM-based filtering plus full human audit. Experiments on frontier LLMs, open-source models, and specialized guard systems show that ATBench is challenging even for strong evaluators, while enabling taxonomy-stratified analysis, cross-benchmark comparison, and diagnosis of long-horizon failure patterns.

Pool-based sequential active learning for regression (ALR) optimally selects a small number of samples sequentially from a large pool of unlabeled samples to label, so that a more accurate regression model can be constructed under a given labeling budget. Representativeness and diversity, which involve computing the distances among different samples, are important considerations in ALR. However, previous ALR approaches do not incorporate the importance of different features in inter-sample distance computation, resulting in sub-optimal sample selection. This paper proposes three feature weighted single-task ALR approaches and two feature weighted multi-task ALR approaches, where the ridge regression coefficients trained from a small amount of previously labeled samples are used to weight the corresponding features in inter-sample distance computation. Experiments showed that this easy-to-implement enhancement almost always improves the performance of four existing ALR approaches, in both single-task and multi-task regression problems. The feature weighting strategy may also be easily extended to stream-based ALR, and classification algorithms.

Demographic parity (DP) is a widely studied fairness criterion in regression, enforcing independence between the predictions and sensitive attributes. However, constraining the entire distribution can degrade predictive accuracy and may be unnecessary for many applications, where fairness concerns are localized to specific regions of the distribution. To overcome this issue, we propose a new framework for regression under DP that focuses on the tails of target distribution across sensitive groups. Our methodology builds on optimal transport theory. By enforcing fairness constraints only over targeted regions of the distribution, our approach enables more nuanced and context-sensitive interventions. Leveraging recent advances, we develop an interpretable and flexible algorithm that leverages the geometric structure of optimal transport. We provide theoretical guarantees, including risk bounds and fairness properties, and validate the method through experiments in regression settings.

Non-pharmaceutical interventions (NPIs) are commonly used tools for controlling infectious disease transmission when pharmaceutical options are unavailable. Yet, identifying effective interventions that minimize societal disruption remains challenging. Agent-based simulation is a popular tool for analyzing the impact of possible interventions in epidemiology. However, automatically optimizing NPIs using agent-based simulations poses a complex problem because, in agent-based epidemiological models, interventions can target individuals based on multiple attributes, affect hierarchical group structures (e.g., schools, workplaces, and families), and be combined arbitrarily, resulting in a very large or even infinite search space. We aim to support decision-makers with our Agent-based Infectious Disease Intervention Optimization System (ADIOS) that optimizes NPIs for infectious disease simulations using Grammar-Guided Genetic Programming (GGGP). The core of ADIOS is a domain-specific language for expressing NPIs in agent-based simulations that structures the intervention search space through a context-free grammar. To make optimization more efficient, the search space can be further reduced by defining constraints that prevent the generation of semantically invalid intervention patterns. Using this constrained language and an interface that enables coupling with agent-based simulations, ADIOS adopts the GGGP approach for simulation-based optimization. Using the German Epidemic Micro-Simulation System (GEMS) as a case study, we demonstrate the potential of our approach to generate optimal interventions for realistic epidemiological models

LLM-generated text (LGT) detection is essential for reliable forensic analysis and for mitigating LLM misuse. Existing LGT detectors can generally be categorized into two broad classes: learning-based approaches and zero-shot methods. Compared with learning-based detectors, zero-shot methods are particularly promising because they eliminate the need to train task-specific classifiers. However, the reliability of zero-shot methods fundamentally relies on the assumption that an off-the-shelf proxy LLM is well aligned with the often unknown source LLM, a premise that rarely holds in real-world black-box scenarios. To address this discrepancy, existing proxy alignment methods typically rely on supervised fine-tuning of the proxy or repeated interactions with commercial APIs, thereby increasing deployment costs, exposing detectors to silent API changes, and limiting robustness under domain shift. Motivated by these limitations, we propose the $k$-nearest neighbor proxy ($k$NNProxy), a training-free and query-efficient proxy alignment framework that repurposes the $k$NN language model ($k$NN-LM) retrieval mechanism as a domain adapter for a fixed proxy LLM. Specifically, a lightweight datastore is constructed once from a target-reflective LGT corpus, either via fixed-budget querying or from existing datasets. During inference, nearest-neighbor evidence induces a token-level predictive distribution that is interpolated with the proxy output, yielding an aligned prediction without proxy fine-tuning or per-token API outputs. To improve robustness under domain shift, we extend $k$NNProxy into a mixture of proxies (MoP) that routes each input to a domain-specific datastore for domain-consistent retrieval. Extensive experiments demonstrate strong detection performance of our method.

Building general-purpose reasoning models using reinforcement learning with verifiable rewards (RLVR) across diverse domains has been widely adopted by frontier open-weight models. However, their training recipes and domain mixtures are often not disclosed. Joint optimization across domains poses significant challenges: domains vary widely in rollout length, problem difficulty and sample efficiency. Further, models with long chain-of-thought traces increase inference cost and latency, making efficiency critical for practical deployment. We present Apriel-Reasoner, trained with a fully reproducible multi-domain RL post-training recipe on Apriel-Base, a 15B-parameter open-weight LLM, across five domains using public datasets: mathematics, code generation, instruction following, logical puzzles and function calling. We introduce an adaptive domain sampling mechanism that preserves target domain ratios despite heterogeneous rollout dynamics, and a difficulty-aware extension of the standard length penalty that, with no additional training overhead, encourages longer reasoning for difficult problems and shorter traces for easy ones. Trained with a strict 16K-token output budget, Apriel-Reasoner generalizes to 32K tokens at inference and improves over Apriel-Base on AIME 2025, GPQA, MMLU-Pro, and LiveCodeBench while producing 30-50% shorter reasoning traces. It matches strong open-weight models of similar size at lower token cost, thereby pushing the Pareto frontier of accuracy versus token budget.

Reinforcement Learning (RL) significantly enhances the reasoning abilities of large language models (LLMs), yet applying it to multi-turn agentic tasks remains challenging due to the long-horizon nature of interactions and the stochasticity of environmental feedback. We identify a structural failure mode in agentic exploration: suboptimal actions elicit noisy observations into misleading contexts, which further weaken subsequent decision-making, making recovery increasingly difficult. This cumulative feedback loop of errors renders standard exploration strategies ineffective and susceptible to the model's reasoning and the environment's randomness. To mitigate this issue, we propose ProCeedRL: Process Critic with Explorative Demonstration RL, shifting exploration from passive selection to active intervention. ProCeedRL employs a process-level critic to monitor interactions in real time, incorporating reflection-based demonstrations to guide agents in stopping the accumulation of errors. We find that this approach significantly exceeds the model's saturated exploration performance, demonstrating substantial exploratory benefits. By learning from exploratory demonstrations and on-policy samples, ProCeedRL significantly improves exploration efficiency and achieves superior performance on complex deep search and embodied tasks.

Transfer learning (TL) and deep ensemble learning (DE) have recently been shown to outperform simple machine learning in classifying psychiatric disorders. However, there is still a lack of understanding as to why that is. This paper aims to understand how and why DE and TL reduce the variability of single-subject classification models in bipolar disorder (BD) and schizophrenia (SCZ). To this end, we investigated the training stability of TL and DE models. For the two classification tasks under consideration, we compared the results of multiple trainings with the same backbone but with different initializations. In this way, we take into account the epistemic uncertainty associated with the uncertainty in the estimation of the model parameters. It has been shown that the performance of classifiers can be significantly improved by using TL with DE. Based on these results, we investigate i) how many models are needed to benefit from the performance improvement of DE when classifying BD and SCZ from healthy controls, and ii) how TL induces better generalization, with and without DE. In the first case, we show that DE reaches a plateau when 10 models are included in the ensemble. In the second case, we find that using a pre-trained model constrains TL models with the same pre-training to stay in the same basin of the loss function. This is not the case for DL models with randomly initialized weights.

Gait analysis provides an objective characterization of locomotor function and is widely used to support diagnosis and rehabilitation monitoring across neurological and orthopedic disorders. Deep learning has been increasingly applied to this domain, yet most approaches rely on supervised classifiers trained on disease-labeled data, limiting generalization to heterogeneous pathological presentations. This work proposes a label-free framework for joint-level anomaly detection and kinematic correction based on a Transformer masked autoencoder trained exclusively on normative gait sequences from 150 adults, acquired with a markerless multi-camera motion-capture system. At inference, a two-pass procedure is applied to potentially pathological input sequences, first it estimates joint inconsistency scores by occluding individual joints and measuring deviations from the learned normative prior. Then, it withholds the flagged joints from the encoder input and reconstructs the full skeleton from the remaining spatiotemporal context, yielding corrected kinematic trajectories at the flagged positions. Validation on 10 held-out normative participants, who mimicked seven simulated gait abnormalities, showed accurate localization of biomechanically inconsistent joints, a significant reduction in angular deviation across all analyzed joints with large effect sizes, and preservation of normative kinematics. The proposed approach enables interpretable, subject-specific localization of gait impairments without requiring disease labels. Video is available at https://youtu.be/Rcm3jqR5pN4.

Multi-hop QA benchmarks frequently reward Large Language Models (LLMs) for spurious correctness, masking ungrounded or flawed reasoning steps. To shift toward rigorous reasoning, we propose SAFE, a dynamic benchmarking framework that replaces the ungrounded Chain-of-Thought (CoT) with a strictly verifiable sequence of grounded entities. Our framework operates across two phases: (1) train-time verification, where we establish an atomic error taxonomy and a Knowledge Graph (KG)-grounded verification pipeline to eliminate noisy supervision in standard benchmarks, identifying up to 14% of instances as unanswerable, and (2) inference-time verification, where a feedback model trained on this verified dataset dynamically detects ungrounded steps in real-time. Experimental results demonstrate that SAFE not only exposes the critical flaws of existing benchmarks at train-time, but also significantly outperforms standard baselines, achieving an average accuracy gain of 8.4 pp while guaranteeing verifiable trajectories at inference-time.

Like a body at rest that stays at rest, we find that visual attention in multimodal large language models (MLLMs) exhibits pronounced inertia, remaining largely static once settled during early decoding steps and failing to support the compositional understanding required for cognitive inference. While existing hallucination mitigation methods mainly target perceptual hallucinations concerning object existence or attributes, they remain inadequate for such cognitive hallucinations that require inter-object relational deduction. Through token-wise attention analysis, we identify this visual inertia as a key factor: attention to semantically critical regions remains persistently focused and fails to dynamically support relational inference. We thereby propose a training-free Inertia-aware Visual Excitation (IVE) method that breaks this inertial pattern by modeling cognitive inference as the dynamic responsiveness of visual attention. Specifically, IVE selects visual tokens that are dynamically emerging relative to historical attention trends while distinguishing tokens exhibiting inertial behavior. To further facilitate compositional inference, IVE introduces an inertia-aware penalty that discourages over-concentration and limits the persistence of attention within localized regions. Extensive experiments show that IVE is effective across various base MLLMs and multiple hallucination benchmarks, particularly for cognitive hallucinations.

Large language models often default to step-by-step computation even when efficient numerical shortcuts are available. This raises a basic question: do they exhibit number sense in a human-like behavioral sense, i.e., the ability to recognize numerical structure, apply shortcuts when appropriate, and avoid them when they are not? We introduce SenseMath, a controlled benchmark for evaluating structure-sensitive numerical reasoning in LLMs. SenseMath contains 4,800 items spanning eight shortcut categories and four digit scales, with matched strong-shortcut, weak-shortcut, and control variants. It supports three evaluation settings of increasing cognitive demand: Shortcut Use (whether models can apply shortcuts on shortcut-amenable problems); Applicability Judgment (whether they can recognize when a shortcut is appropriate or misleading); and Problem Generation (whether they can generate new problem items that correctly admit a given type of shortcut). Our evaluation across five LLMs, ranging from GPT-4o-mini to Llama-3.1-8B, shows a consistent pattern: when explicitly prompted, models readily adopt shortcut strategies and achieve substantial accuracy gains on shortcut-amenable items (up to 15%), yet under standard chain-of-thought prompting they spontaneously employ such strategies in fewer than 40% of cases, even when they demonstrably possess the requisite capability. Moreover, this competence is confined to the Use level; models systematically over-generalise shortcuts to problems where they do not apply, and fail to generate valid shortcut-bearing problems from scratch. Together, these results suggest that current LLMs exhibit procedural shortcut fluency without the structural understanding of when and why shortcuts work that underlies human number sense.

The rapid growth of medical imaging has fueled the development of Foundation Models (FMs) to reduce the growing, unsustainable workload on radiologists. While recent FMs have shown the power of large-scale pre-training to CT and MRI analysis, there remains significant room to optimize how these models learn from complex radiological volumes. Building upon the Curia framework, this work introduces Curia-2, which significantly improves the original pre-training strategy and representation quality to better capture the specificities of radiological data. The proposed methodology enables scaling the architecture up to billion-parameter Vision Transformers, marking a first for multi-modal CT and MRI FMs. Furthermore, we formalize the evaluation of these models by extending and restructuring CuriaBench into two distinct tracks: a 2D track tailored for slice-based vision models and a 3D track for volumetric benchmarking. Our results demonstrate that Curia-2 outperforms all FMs on vision-focused tasks and fairs competitively to vision-language models on clinically complex tasks such as finding detection. Weights will be made publicly available to foster further research.

General-purpose world models promise scalable policy evaluation, optimization, and planning, yet achieving the required level of robustness remains challenging. Unlike policy learning, which primarily focuses on optimal actions, a world model must be reliable over a much broader range of suboptimal actions, which are often insufficiently covered by action-labeled interaction data. To address this challenge, we propose World Action Verifier (WAV), a framework that enables world models to identify their own prediction errors and self-improve. The key idea is to decompose action-conditioned state prediction into two factors -- state plausibility and action reachability -- and verify each separately. We show that these verification problems can be substantially easier than predicting future states due to two underlying asymmetries: the broader availability of action-free data and the lower dimensionality of action-relevant features. Leveraging these asymmetries, we augment a world model with (i) a diverse subgoal generator obtained from video corpora and (ii) a sparse inverse model that infers actions from a subset of state features. By enforcing cycle consistency among generated subgoals, inferred actions, and forward rollouts, WAV provides an effective verification mechanism in under-explored regimes, where existing methods typically fail. Across nine tasks spanning MiniGrid, RoboMimic, and ManiSkill, our method achieves 2x higher sample efficiency while improving downstream policy performance by 18%.

We present a large-scale experimental study of quantum-computing-based molecular simulation carried out on IQM's Sirius 24-qubit superconducting processor, utilizing up to 16 operational qubits. The work employs Sample-based Quantum Diagonalization (SQD) together with the Local Unitary Cluster Jastrow (LUCJ) ansatz to estimate ground-state energies for a set of benchmark molecules, including H$_2$, LiH, BeH$_2$, H$_2$O, and NH$_3$. In addition, we introduce a Linear-CNOT variant of the Unitary Coupled-Cluster Singles and Doubles (LCNot-UCCSD) ansatz within the SQD workflow, trading higher circuit depth for reduced classical preprocessing. A comparison between these ansätze is provided, clarifying their respective strengths, limitations, and suitability for near-term quantum hardware. We further explore potential energy landscapes through 1D scans for H$_2$ and HeH$^+$ using both STO-3G and 6-31G basis sets, and for LiH and BeH$_2$ in STO-3G. Extending beyond this, we demonstrate the experimental construction of a full 2D potential energy surface for the water molecule on quantum hardware, mapped over a 32 $\times$ 32 grid in bond length and bond angle. To move beyond small benchmark systems, we combine SQD(LUCJ) with Density Matrix Embedding Theory (DMET) to compute active-space energies for a set of ligand-like molecules, as well as the pharmacologically relevant amantadine system. Across all studies, the majority of quantum-computed energies agree with reference FCI results, as well as with DMET-CASCI energies for embedded systems, to within chemical accuracy for the chosen basis sets. These results demonstrate the reliability of sample-based diagonalization approaches and underscore the potential of hybrid embedding strategies for extending quantum simulations to increasingly complex molecular systems, while also highlighting their practicality on current IQM quantum hardware.

We study a random model of deep multi-head self-attention in which the weights are resampled independently across layers and heads, as at initialization of training. Viewing depth as a time variable, the residual stream defines a discrete-time interacting particle system on the unit sphere. We prove that, under suitable joint scalings of the depth, the residual step size, and the number of heads, this dynamics admits a nontrivial homogenized limit. Depending on the scaling, the limit is either deterministic or stochastic with common noise; in the mean-field regime, the latter leads to a stochastic nonlinear Fokker--Planck equation for the conditional law of a representative token. In the Gaussian setting, the limiting drift vanishes, making the homogenized dynamics explicit enough to study representation collapse. This yields quantitative trade-offs between dimension, context length, and temperature, and identifies regimes in which clustering can be mitigated.

Security teams face a challenge: the volume of newly disclosed Common Vulnerabilities and Exposures (CVEs) far exceeds the capacity to manually develop detection mechanisms. In 2025, the National Vulnerability Database published over 48,000 new vulnerabilities, motivating the need for automation. We present RuleForge, an AWS internal system that automatically generates detection rules--JSON-based patterns that identify malicious HTTP requests exploiting specific vulnerabilities--from structured Nuclei templates describing CVE details. Nuclei templates provide standardized, YAML-based vulnerability descriptions that serve as the structured input for our rule generation process. This paper focuses on RuleForge's architecture and operational deployment for CVE-related threat detection, with particular emphasis on our novel LLM-as-a-judge (Large Language Model as judge) confidence validation system and systematic feedback integration mechanism. This validation approach evaluates candidate rules across two dimensions--sensitivity (avoiding false negatives) and specificity (avoiding false positives)--achieving AUROC of 0.75 and reducing false positives by 67% compared to synthetic-test-only validation in production. Our 5x5 generation strategy (five parallel candidates with up to five refinement attempts each) combined with continuous feedback loops enables systematic quality improvement. We also present extensions enabling rule generation from unstructured data sources and demonstrate a proof-of-concept agentic workflow for multi-event-type detection. Our lessons learned highlight critical considerations for applying LLMs to cybersecurity tasks, including overconfidence mitigation and the importance of domain expertise in both prompt design and quality review of generated rules through human-in-the-loop validation.

We present the first systematic analysis of multimodal large language models (MLLMs) in personalized question-answering requiring ego-grounding - the ability to understand the camera-wearer in egocentric videos. To this end, we introduce MyEgo, the first egocentric VideoQA dataset designed to evaluate MLLMs' ability to understand, remember, and reason about the camera wearer. MyEgo comprises 541 long videos and 5K personalized questions asking about "my things", "my activities", and "my past". Benchmarking reveals that competitive MLLMs across variants, including open-source vs. proprietary, thinking vs. non-thinking, small vs. large scales all struggle on MyEgo. Top closed- and open-source models (e.g., GPT-5 and Qwen3-VL) achieve only~46% and 36% accuracy, trailing human performance by near 40% and 50% respectively. Surprisingly, neither explicit reasoning nor model scaling yield consistent improvements. Models improve when relevant evidence is explicitly provided, but gains drop over time, indicating limitations in tracking and remembering "me" and "my past". These findings collectively highlight the crucial role of ego-grounding and long-range memory in enabling personalized QA in egocentric videos. We hope MyEgo and our analyses catalyze further progress in these areas for egocentric personalized assistance. Data and code are available at https://github.com/Ryougetsu3606/MyEgo

Scientific knowledge discovery increasingly relies on large language models, yet many existing scholarly assistants depend on proprietary systems with tens or hundreds of billions of parameters. Such reliance limits reproducibility and accessibility for the research community. In this work, we ask a simple question: do we need bigger models for scientific applications? Specifically, we investigate to what extent carefully designed retrieval pipelines can compensate for reduced model scale in scientific applications. We design a lightweight retrieval-augmented framework that performs task-aware routing to select specialized retrieval strategies based on the input query. The system further integrates evidence from full-text scientific papers and structured scholarly metadata, and employs compact instruction-tuned language models to generate responses with citations. We evaluate the framework across several scholarly tasks, focusing on scholarly question answering (QA), including single- and multi-document scenarios, as well as biomedical QA under domain shift and scientific text compression. Our findings demonstrate that retrieval and model scale are complementary rather than interchangeable. While retrieval design can partially compensate for smaller models, model capacity remains important for complex reasoning tasks. This work highlights retrieval and task-aware design as key factors for building practical and reproducible scholarly assistants.

Abnormal head movements (AHMs) manifest across a broad spectrum of neurological disorders; however, the absence of a multi-condition resource integrating kinematic measurements, clinical severity scores, and patient demographics constitutes a persistent barrier to the development of AI-driven diagnostic tools. To address this gap, this study introduces NeuroPose-AHM, a knowledge-based dataset of neurologically induced AHMs constructed through a multi-LLM extraction framework applied to 1,430 peer-reviewed publications. The dataset contains 2,756 patient-group-level records spanning 57 neurological conditions, derived from 846 AHM-relevant papers. Inter-LLM reliability analysis confirms robust extraction performance, with study-level classification achieving strong agreement (kappa = 0.822). To demonstrate the dataset's analytical utility, a four-task framework is applied to cervical dystonia (CD), the condition most directly defined by pathological head movement. First, Task 1 performs multi-label AHM type classification (F1 = 0.856). Task 2 constructs the Head-Neck Severity Index (HNSI), a unified metric that normalizes heterogeneous clinical rating scales. The clinical relevance of this index is then evaluated in Task 3, where HNSI is validated against real-world CD patient data, with aligned severe-band proportions (6.7%) providing a preliminary plausibility indication for index calibration within the high severity range. Finally, Task 4 performs bridge analysis between movement-type probabilities and HNSI scores, producing significant correlations (p less than 0.001). These results demonstrate the analytical utility of NeuroPose-AHM as a structured, knowledge-based resource for neurological AHM research. The NeuroPose-AHM dataset is publicly available on Zenodo (https://doi.org/10.5281/zenodo.19386862).

Multiple operator learning concerns learning operator families $\{G[α]:U\to V\}_{α\in W}$ indexed by an operator descriptor $α$. Training data are collected hierarchically by sampling operator instances $α$, then input functions $u$ per instance, and finally evaluation points $x$ per input, yielding noisy observations of $G[α][u](x)$. While recent work has developed expressive multi-task and multiple operator learning architectures and approximation-theoretic scaling laws, quantitative statistical generalization guarantees remain limited. We provide a covering-number-based generalization analysis for separable models, focusing on the Multiple Neural Operator (MNO) architecture: we first derive explicit metric-entropy bounds for hypothesis classes given by linear combinations of products of deep ReLU subnetworks, and then combine these complexity bounds with approximation guarantees for MNO to obtain an explicit approximation-estimation tradeoff for the expected test error on new (unseen) triples $(α,u,x)$. The resulting bound makes the dependence on the hierarchical sampling budgets $(n_α,n_u,n_x)$ transparent and yields an explicit learning-rate statement in the operator-sampling budget $n_α$, providing a sample-complexity characterization for generalization across operator instances. The structure and architecture can also be viewed as a general purpose solver or an example of a "small'' PDE foundation model, where the triples are one form of multi-modality.

Machine-translated benchmark datasets reduce costs and offer scale, but noise, loss of structure, and uneven quality weaken confidence. What matters is not merely whether we can translate, but also whether we can measure and verify translation reliability at scale. We study translation quality in the EU20 benchmark suite, which comprises five established benchmarks translated into 20 languages, via a three-step automated quality assurance approach: (i) a structural corpus audit with targeted fixes; (ii) quality profiling using a neural metric (COMET, reference-free and reference-based) with translation service comparisons (DeepL / ChatGPT / Google); and (iii) an LLM-based span-level translation error landscape. Trends are consistent: datasets with lower COMET scores exhibit a higher share of accuracy/mistranslation errors at span level (notably HellaSwag; ARC is comparatively clean). Reference-based COMET on MMLU against human-edited samples points in the same direction. We release cleaned/corrected versions of the EU20 datasets, and code for reproducibility. In sum, automated quality assurance offers practical, scalable indicators that help prioritize review -- complementing, not replacing, human gold standards.

Cloud infrastructure supports the efficient operation of data pipelines regarding requirements like cost, speed, and resource utilization. We present an integrated view of optimization opportunities for cloud-based data pipelines by conducting a systematic review of existing literature on optimization approaches to cloud infrastructure performance for data pipelines. Our study contributes a theory of optimization goals like minimizing cost, reducing execution time, and cost-makespan trade-offs, consisting of dimensions such as single vs. multi-cloud, batch vs. stream processing, etc. We highlight gaps in primary research, including the underexploration of multi-tenant environments and lack of industry evaluation, and suggest directions for future research.

We introduce Qiana, a logic framework for reasoning on formulas that are true only in specific contexts. In Qiana, it is possible to quantify over both formulas and contexts to express, e.g., that ``everyone knows everything Alice says''. Qiana also permits paraconsistent logics within contexts, so that contexts can contain contradictions. Furthermore, Qiana is based on first-order logic, and is finitely axiomatizable, so that Qiana theories are compatible with pre-existing first-order logic theorem provers. We show how Qiana can be used to represent temporality, event calculus, and modal logic. We also discuss different design alternatives of Qiana.

We propose LSCP, a self-gated post-training framework for autonomous knowledge acquisition: learning only what a model does not already know, verified against what it does know, at a strength proportional to conviction, with no external oracle. When a passage produces anomalously high per-token loss, LSCP flags it, generates a Q&A chain that forces the model to articulate its own knowledge and identify gaps, then adjusts AdamW's $β_2$ proportionally to conviction depth k (the number of self-verification steps the passage survives) via $β_2 = 0.999 \cdot r^k$. The entire learning intensity is governed by a single parameter $r$. Beyond new knowledge, this process sharpens weakly encoded existing knowledge, which is a primary source of hallucination. The framework is self-extinguishing: as the model learns, per-token loss on learned passages decreases toward the surprisal threshold and the system progressively converges to standard AdamW. This models biological memory consolidation: temporary information in the context window is selectively consolidated into parametric weights, the model's long-term memory. Experiments on the reference model (Qwen3-14B) and across six models (8B--32B, four families) show that standard fine-tuning produces rote memorization (perturbation gap (the ratio of paraphrase to original perplexity) of 11.6 +- 0.2 x baseline) while all LSCP conditions learn semantically (2.7--3.0x). The r=1.0 condition (identical optimizer, nearly identical data, only Q&A format differs) confirms that the training data format, not $β_2$ gating, is the primary mechanism preventing memorization; gating instead protects neighboring knowledge from contamination by corrupt content (93 +- 7% accuracy on adjacent questions at r=0.98 vs. 90% baseline).

The scale of biological datasets now routinely exceeds system memory, making data access rather than model computation the primary bottleneck in training machine-learning models. This bottleneck is particularly acute in biology, where widely used community data formats must support heterogeneous metadata, sparse and dense assays, and downstream analysis within established computational ecosystems. Here we present annbatch, a mini-batch loader native to anndata that enables out-of-core training directly on disk-backed datasets. Across single-cell transcriptomics, microscopy and whole-genome sequencing benchmarks, annbatch increases loading throughput by up to an order of magnitude and shortens training from days to hours, while remaining fully compatible with the scverse ecosystem. Annbatch establishes a practical data-loading infrastructure for scalable biological AI, allowing increasingly large and diverse datasets to be used without abandoning standard biological data formats. Github: https://github.com/scverse/annbatch

Estimating optimal individualized treatment rules (ITRs) via outcome weighted learning (OWL) often relies on observed rewards that are noisy or optimistic proxies for the true latent utility. Ignoring this reward uncertainty leads to the selection of policies with inflated apparent performance, yet existing OWL frameworks lack the finite-sample guarantees required to systematically embed such uncertainty into the learning objective. To address this issue, we propose PAC-Bayesian Reward-Certified Outcome Weighted Learning (PROWL). Given a one-sided uncertainty certificate, PROWL constructs a conservative reward and a strictly policy-dependent lower bound on the true expected value. Theoretically, we prove an exact certified reduction that transforms robust policy learning into a unified, split-free cost-sensitive classification task. This formulation enables the derivation of a nonasymptotic PAC-Bayes lower bound for randomized ITRs, where we establish that the optimal posterior maximizing this bound is exactly characterized by a general Bayes update. To overcome the learning-rate selection problem inherent in generalized Bayesian inference, we introduce a fully automated, bounds-based calibration procedure, coupled with a Fisher-consistent certified hinge surrogate for efficient optimization. Our experiments demonstrate that PROWL achieves improvements in estimating robust, high-value treatment regimes under severe reward uncertainty compared to standard methods for ITR estimation.

Wideband channel frequency response (CFR) estimation is challenging in multi-band wireless systems, especially when one or more sub-bands are temporarily blocked by co-channel interference. We present a physics-informed complex Transformer that reconstructs the full wideband CFR from such fragmented, partially observed spectrum snapshots. The interference pattern in each sub-band is modeled as an independent two-state discrete-time Markov chain, capturing realistic bursty occupancy behavior. Our model operates on the joint time-frequency grid of $T$ snapshots and $F$ frequency bins and uses a factored self-attention mechanism that separately attends along both axes, reducing the computational complexity to $O(TF^2 + FT^2)$. Complex-valued inputs and outputs are processed through a holomorphic linear layer that preserves phase relationships. Training uses a composite physics-informed loss combining spectral fidelity, power delay profile (PDP) reconstruction, channel impulse response (CIR) sparsity, and temporal smoothness. Mobility effects are incorporated through per-sample velocity randomization, enabling generalization across different mobility regimes. Evaluation against three classical baselines, namely, last-observation-carry-forward, zero-fill, and cubic-spline interpolation, shows that our approach achieves the highest PDP similarity with respect to the ground truth, reaching $ρ\geq 0.82$ compared to $ρ\geq 0.62$ for the best baseline at interference occupancy levels up to 50%. Furthermore, the model degrades smoothly across the full velocity range, consistently outperforming all other baselines.

This paper addresses the problem of clustering measurement vectors that are heteroscedastic in that they can have different covariance matrices. From the assumption that the measurement vectors within a given cluster are Gaussian distributed with possibly different and unknown covariant matrices around the cluster centroid, we introduce a novel cost function to estimate the centroids. The zeros of the gradient of this cost function turn out to be the fixed-points of a certain function. As such, the approach generalizes the methodology employed to derive the existing Mean-Shift algorithm. But as a main and novel theoretical result compared to Mean-Shift, this paper shows that the sole fixed-points of the identified function tend to be the cluster centroids if both the number of measurements per cluster and the distances between centroids are large enough. As a second contribution, this paper introduces the Wald kernel for clustering. This kernel is defined as the p-value of the Wald hypothesis test for testing the mean of a Gaussian. As such, the Wald kernel measures the plausibility that a measurement vector belongs to a given cluster and it scales better with the dimension of the measurement vectors than the usual Gaussian kernel. Finally, the proposed theoretical framework allows us to derive a new clustering algorithm called CENTRE-X that works by estimating the fixed-points of the identified function. As Mean-Shift, CENTRE-X requires no prior knowledge of the number of clusters. It relies on a Wald hypothesis test to significantly reduce the number of fixed points to calculate compared to the Mean-Shift algorithm, thus resulting in a clear gain in complexity. Simulation results on synthetic and real data sets show that CENTRE-X has comparable or better performance than standard clustering algorithms K-means and Mean-Shift, even when the covariance matrices are not perfectly known.

Image captioning for Early Childhood Education (ECE) is essential for automated activity understanding and educational assessment. However, existing methods face two key challenges. First, the lack of large-scale, domain-specific datasets limits the model's ability to capture fine-grained semantic concepts unique to ECE scenarios, resulting in generic and imprecise descriptions. Second, conventional training paradigms exhibit limitations in enhancing professional object description capability, as supervised learning tends to favor high-frequency expressions, while reinforcement learning may suffer from unstable optimization on difficult samples. To address these limitations, we introduce ECAC, a large-scale benchmark for ECE daily activity image captioning, comprising 256,121 real-world images annotated with expert-level captions and fine-grained labels. ECAC is further equipped with a domain-oriented evaluation protocol, the Teaching Toy Recognition Score (TTS), to explicitly measure professional object naming accuracy. Furthermore, we propose RSRS (Reward-Conditional Switch of Reinforcement Learning and Supervised Fine-Tuning), a hybrid training framework that dynamically alternates between RL and supervised optimization. By rerouting hard samples with zero rewards to supervised fine-tuning, RSRS effectively mitigates advantage collapse and enables stable optimization for fine-grained recognition. Leveraging ECAC and RSRS, we develop KinderMM-Cap-3B, a domain-adapted multimodal large language model. Extensive experiments demonstrate that our model achieves a TTS of 51.06, substantially outperforming state-of-the-art baselines while maintaining superior caption quality, highlighting its potential for specialized educational applications.

We consider learning with possibilistic supervision for multi-class classification. For each training instance, the supervision is a normalized possibility distribution that expresses graded plausibility over the classes. From this possibility distribution, we construct a non-empty closed convex set of admissible probability distributions by combining two requirements: probabilistic compatibility with the possibility and necessity measures induced by the possibility distribution, and linear shape constraints that must be satisfied to preserve the qualitative structure of the possibility distribution. Thus, classes with the same possibility degree receive equal probabilities, and if a class has a strictly larger possibility degree than another class, then it receives a strictly larger probability. Given a strictly positive probability vector output by a model for an instance, we compute its Kullback-Leibler projection onto the admissible set. This projection yields the closest admissible probability distribution in Kullback-Leibler sense. We can then train the model by minimizing the divergence between the prediction and its projection, which quantifies the smallest adjustment needed to satisfy the induced dominance and shape constraints. The projection is computed with Dykstra's algorithm using Bregman projections associated with the negative entropy, and we provide explicit formulas for the projections onto each constraint set. Experiments conducted on synthetic data and on a real-world natural language inference task, based on the ChaosNLI dataset, show that the proposed projection algorithm is efficient enough for practical use, and that the resulting projection-based learning objective can improve predictive performance.

The structure of all the permutations of a sequence can be represented as a permutohedron, a graph where vertices are permutations and two vertices are linked if a swap of adjacent elements in the permutation of one of the vertices produces the permutation of the other vertex. It has been hypothesized that word orders in languages minimize the swap distance in the permutohedron: given a source order, word orders that are closer in the permutohedron should be less costly and thus more likely. Here we explain how to measure the degree of optimality of word order variation with respect to swap distance minimization. We illustrate the power of our novel mathematical framework by showing that crosslinguistic gestures are at least $77\%$ optimal. It is unlikely that the multiple times where crosslinguistic gestures hit optimality are due to chance. We establish the theoretical foundations for research on the optimality of word or gesture order with respect to swap distance minimization in communication systems. Finally, we introduce the quadratic assignment problem (QAP) into language research as an umbrella for multiple optimization problems and, accordingly, postulate a general principle of optimal assignment that unifies various linguistic principles including swap distance minimization.

Among news disorders, propagandist news are particularly insidious, because they tend to mix oriented messages with factual reports intended to look like reliable news. To detect propaganda, extant approaches based on Language Models such as BERT are promising but often overfit their training datasets, due to biases in data collection. To enhance classification robustness and improve generalization to new sources, we propose a neurosymbolic approach combining non-contextual text embeddings (fastText) with symbolic conceptual features such as genre, topic, and persuasion techniques. Results show improvements over equivalent text-only methods, and ablation studies as well as explainability analyses confirm the benefits of the added features. Keywords: Information disorder, Fake news, Propaganda, Classification, Topic modeling, Hybrid method, Neurosymbolic model, Ablation, Robustness

Most of the Neural Cellular Automata (NCAs) defined in the literature have a common theme: they are based on regular grids with a Moore neighborhood (one-hop neighbour). They do not take into account long-range connections and more complex topologies as we can find in the brain. In this paper, we introduce BraiNCA, a brain-inspired NCA with an attention layer, long-range connections and complex topology. BraiNCAs shows better results in terms of robustness and speed of learning on the two tasks compared to Vanilla NCAs establishing that incorporating attention-based message selection together with explicit long-range edges can yield more sample-efficient and damage-tolerant self-organization than purely local, grid-based update rules. These results support the hypothesis that, for tasks requiring distributed coordination over extended spatial and temporal scales, the choice of interaction topology and the ability to dynamically route information will impact the robustness and speed of learning of an NCA. More broadly, BraiNCA provides brain-inspired NCA formulation that preserves the decentralized local update principle while better reflecting non-local connectivity patterns, making it a promising substrate for studying collective computation under biologically-realistic network structure and evolving cognitive substrates.

We propose a geometry-driven quantum-inspired classification framework that integrates Correlation Group Structures (CGR), compact SWAP-test-based overlap estimation, and selective variational quantum decision modelling. Rather than directly approximating class posteriors, the method adopts a geometry-first paradigm in which samples are evaluated relative to class medoids using overlap-derived Euclidean-like and angular similarity channels. CGR organizes features into anchor-centered correlation neighbourhoods, generating nonlinear, correlation-weighted representations that enhance robustness in heterogeneous tabular spaces. These geometric signals are fused through a non-probabilistic margin-based fusion score, serving as a lightweight and data-efficient primary classifier for small-to-moderate datasets. On Heart Disease, Breast Cancer, and Wine Quality datasets, the fusion-score classifier achieves 0.8478, 0.8881, and 0.9556 test accuracy respectively, with macro-F1 scores of 0.8463, 0.8703, and 0.9522, demonstrating competitive and stable performance relative to classical baselines. For large-scale and highly imbalanced regimes, we construct compact Delta-distance contrastive features and train a variational quantum classifier (VQC) as a nonlinear refinement layer. On the Credit Card Fraud dataset (0.17% prevalence), the Delta + VQC pipeline achieves approximately 0.85 minority recall at an alert rate of approximately 1.31%, with ROC-AUC 0.9249 and PR-AUC 0.3251 under full-dataset evaluation. These results highlight the importance of operating-point-aware assessment in rare-event detection and demonstrate that the proposed hybrid geometric-variational framework provides interpretable, scalable, and regime-adaptive classification across heterogeneous data settings.

The audio research community depends on open generative models as foundational tools for building novel approaches and establishing baselines. In this report, we present Woosh, Sony AI's publicly released sound effect foundation model, detailing its architecture, training process, and an evaluation against other popular open models. Being optimized for sound effects, we provide (1) a high-quality audio encoder/decoder model and (2) a text-audio alignment model for conditioning, together with (3) text-to-audio and (4) video-to-audio generative models. Distilled text-to-audio and video-to-audio models are also included in the release, allowing for low-resource operation and fast inference. Our evaluation on both public and private data shows competitive or better performance for each module when compared to existing open alternatives like StableAudio-Open and TangoFlux. Inference code and model weights are available at https://github.com/SonyResearch/Woosh. Demo samples can be found at https://sonyresearch.github.io/Woosh/.

Large Language Models increasingly suppress biased outputs when demographic identity is stated explicitly, yet may still exhibit implicit biases when identity is conveyed indirectly. Existing benchmarks use name based proxies to detect implicit biases, which carry weak associations with many social demographics and cannot extend to dimensions like age or socioeconomic status. We introduce ImplicitBBQ, a QA benchmark that evaluates implicit bias through characteristic based cues, culturally associated attributes that signal implicitly, across age, gender, region, religion, caste, and socioeconomic status. Evaluating 11 models, we find that implicit bias in ambiguous contexts is over six times higher than explicit bias in open weight models. Safety prompting and chain-of-thought reasoning fail to substantially close this gap; even few-shot prompting, which reduces implicit bias by 84%, leaves caste bias at four times the level of any other dimension. These findings indicate that current alignment and prompting strategies address the surface of bias evaluation while leaving culturally grounded stereotypic associations largely unresolved. We publicly release our code and dataset for model providers and researchers to benchmark potential mitigation techniques.

Accurate short-term mortality prediction in heart failure (HF) remains challenging, particularly when relying on structured electronic health record (EHR) data alone. We evaluate transformer-based models on a French HF cohort, comparing text-only, structured-only, multimodal, and LLM-based approaches. Our results show that enriching clinical text with entity-level representations improves prediction over CLS embeddings alone, and that supervised multimodal fusion of text and structured variables achieves the best overall performance. In contrast, large language models perform inconsistently across modalities and decoding strategies, with text-only prompts outperforming structured or multimodal inputs. These findings highlight that entity-aware multimodal transformers offer the most reliable solution for short-term HF outcome prediction, while current LLM prompting remains limited for clinical decision support.

Automotive radar perception pipelines commonly construct angle-domain representations via beamforming before applying learning-based models. This work instead investigates a representational question: can meaningful spatial structure be learned directly from pre-beamforming per-antenna range-Doppler (RD) measurements? Experiments are conducted on a 6-TX x 8-RX (48 virtual antennas) commodity automotive radar employing an A/B chirp-sequence frequency-modulated continuous-wave (CS-FMCW) transmit scheme, in which the effective transmit aperture varies between chirps (single-TX vs. multi-TX), enabling controlled analysis of chirp-dependent transmit configurations. We operate on pre-beamforming per-antenna RD tensors using a dual-chirp shared-weight encoder trained in an end-to-end, fully data-driven manner, and evaluate spatial recoverability using bird's-eye-view (BEV) occupancy as a geometric probe rather than a performance-driven objective. Supervision is visibility-aware and cross-modal, derived from LiDAR with explicit modeling of the radar field-of-view and occlusion-aware LiDAR observability via ray-based visibility. Through chirp ablations (A-only, B-only, A+B), range-band analysis, and physics-aligned baselines, we assess how transmit configurations affect geometric recoverability. The results indicate that spatial structure can be learned directly from pre-beamforming per-antenna RD tensors without explicit angle-domain construction or hand-crafted signal-processing stages.

Multimodal emotion recognition in conversations (MERC) requires integrating multimodal signals while being robust to noise and modeling contextual reasoning. Existing approaches often emphasize fusion but overlook uncertainty in noisy features and fine-grained reasoning. We propose SURE (Synergistic Uncertainty-aware REasoning) for MERC, a framework that improves robustness and contextual modeling. SURE consists of three components: an Uncertainty-Aware Mixture-of-Experts module to handle modality-specific noise, an Iterative Reasoning module for multi-turn reasoning over context, and a Transformer Gate module to capture intra- and inter-modal interactions. Experiments on benchmark MERC datasets show that SURE consistently outperforms state-of-the-art methods, demonstrating its effectiveness in robust multimodal reasoning. These results highlight the importance of uncertainty modeling and iterative reasoning in advancing emotion recognition in conversational settings.

Deep reinforcement learning (RL) suffers from plasticity loss severely due to the nature of non-stationarity, which impairs the ability to adapt to new data and learn continually. Unfortunately, our understanding of how plasticity loss arises, dissipates, and can be dissolved remains limited to empirical findings, leaving the theoretical end underexplored.To address this gap, we study the plasticity loss problem from the theoretical perspective of network optimization. By formally characterizing the two culprit factors in online RL process: the non-stationarity of data distributions and the non-stationarity of targets induced by bootstrapping, our theory attributes the loss of plasticity to two mechanisms: the rank collapse of the Neural Tangent Kernel (NTK) Gram matrix and the $Θ(\frac{1}{k})$ decay of gradient magnitude. The first mechanism echoes prior empirical findings from the theoretical perspective and sheds light on the effects of existing methods, e.g., network reset, neuron recycle, and noise injection. Against this backdrop, we focus primarily on the second mechanism and aim to alleviate plasticity loss by addressing the gradient attenuation issue, which is orthogonal to existing methods. We propose Sample Weight Decay -- a lightweight method to restore gradient magnitude, as a general remedy to plasticity loss for deep RL methods based on experience replay. In experiments, we evaluate the efficacy of \methodName upon TD3, \myadded{Double DQN} and SAC with SimBa architecture in MuJoCo, \myadded{ALE} and DeepMind Control Suite tasks. The results demonstrate that \methodName effectively alleviates plasticity loss and consistently improves learning performance across various configurations of deep RL algorithms, UTD, network architectures, and environments, achieving SOTA performance on challenging DMC Humanoid tasks.

Annotated 3D scene data is scarce and expensive to acquire, while abundant unlabeled videos are readily available on the internet. In this paper, we demonstrate that carefully designed data engines can leverage web-curated, unlabeled videos to automatically generate training data, to facilitate end-to-end models in 3D scene understanding alongside human-annotated datasets. We identify and analyze bottlenecks in automated data generation, revealing critical factors that determine the efficiency and effectiveness of learning from unlabeled data. To validate our approach across different perception granularities, we evaluate on three tasks spanning low-level perception, i.e., 3D object detection and instance segmentation, to high-evel reasoning, i.e., 3D spatial Visual Question Answering (VQA) and Vision-Lanugage Navigation (VLN). Models trained on our generated data demonstrate strong zero-shot performance and show further improvement after finetuning. This demonstrates the viability of leveraging readily available web data as a path toward more capable scene understanding systems.

The model context protocol (MCP) standardizes how LLMs connect to external tools and data sources, enabling faster integration but introducing new attack vectors. Despite the growing adoption of MCP, existing MCP security studies classify attacks by their observable effects, obscuring how attacks behave across different MCP server components and overlooking multi-component attack chains. Meanwhile, existing defenses are less effective when facing multi-component attacks or previously unknown malicious behaviors. This work presents a component-centric perspective for understanding and detecting malicious MCP servers. First, we build the first component-centric PoC dataset of 114 malicious MCP servers where attacks are achieved as manipulation over MCP components and their compositions. We evaluate these attacks' effectiveness across two MCP hosts and five LLMs, and uncover that (1) component position shapes attack success rate; and (2) multi-component compositions often outperform single-component attacks by distributing malicious logic. Second, we propose and implement Connor, a two-stage behavioral deviation detector for malicious MCP servers. It first performs pre-execution analysis to detect malicious shell commands and extract each tool's function intent, and then conducts step-wise in-execution analysis to trace each tool's behavioral trajectories and detect deviations from its function intent. Evaluation on our curated dataset indicates that Connor achieves an F1-score of 94.6%, outperforming the state of the art by 8.9% to 59.6%. In real-world detection, Connor identifies two malicious servers.

Data rights owners can detect unauthorized data use in large language model (LLM) training by querying with proprietary samples. Often, superior performance (e.g., higher confidence or lower loss) on a sample relative to the untrained data implies it was part of the training corpus, as LLMs tend to perform better on data they have seen during training. However, this detection becomes fragile under data laundering, a practice of transforming the stylistic form of proprietary data, while preserving critical information to obfuscate data provenance. When an LLM is trained exclusively on such laundered variants, it no longer performs better on originals, erasing the signals that standard detections rely on. We counter this by inferring the unknown laundering transformation from black-box access to the target LLM and, via an auxiliary LLM, synthesizing queries that mimic the laundered data, even if rights owners have only the originals. As the search space of finding true laundering transformations is infinite, we abstract such a process into a high-level transformation goal (e.g., "lyrical rewriting") and concrete details (e.g., "with vivid imagery"), and introduce synthesis data reversion (SDR) that instantiates this abstraction. SDR first identifies the most probable goal for synthesis to narrow the search; it then iteratively refines details so that synthesized queries gradually elicit stronger detection signals from the target LLM. Evaluated on the MIMIR benchmark against diverse laundering practices and target LLM families (Pythia, Llama2, and Falcon), SDR consistently strengthens data misuse detection, providing a practical countermeasure to data laundering.

Artificial intelligence (AI) systems accelerate medical workflows and improve diagnostic accuracy in healthcare, serving as second-opinion systems. However, the unpredictability of AI errors poses a significant challenge, particularly in healthcare contexts, where mistakes can have severe consequences. A widely adopted safeguard is to pair predictions with uncertainty estimation, enabling human experts to focus on high-risk cases while streamlining routine verification. Current uncertainty estimation methods, however, remain limited, particularly in quantifying aleatoric uncertainty, which arises from data ambiguity and noise. To address this, we propose a novel approach that leverages disagreement in expert responses to generate targets for training machine learning models. These targets are used in conjunction with standard data labels to estimate two components of uncertainty separately, as given by the law of total variance, via a two-ensemble approach, as well as its lightweight variant. We validate our method on binary image classification, binary and multi-class image segmentation, and multiple-choice question answering. Our experiments demonstrate that incorporating expert knowledge can enhance uncertainty estimation quality by $9\%$ to $50\%$ depending on the task, making this source of information invaluable for the construction of risk-aware AI systems in healthcare applications.

Large language models (LLMs) have been proposed as alternatives to human experts for estimating unknown quantities with associated uncertainty, a process known as Bayesian elicitation. We test this by asking eleven LLMs to estimate population statistics, such as health prevalence rates, personality trait distributions, and labor market figures, and to express their uncertainty as 95\% credible intervals. We vary each model's reasoning effort (low, medium, high) to test whether more "thinking" improves results. Our findings reveal three key results. First, larger, more capable models produce more accurate estimates, but increasing reasoning effort provides no consistent benefit. Second, all models are severely overconfident: their 95\% intervals contain the true value only 9--44\% of the time, far below the expected 95\%. Third, a statistical recalibration technique called conformal prediction can correct this overconfidence, expanding the intervals to achieve the intended coverage. In a preliminary experiment, giving models web search access degraded predictions for already-accurate models, while modestly improving predictions for weaker ones. Models performed well on commonly discussed topics but struggled with specialized health data. These results indicate that LLM uncertainty estimates require statistical correction before they can be used in decision-making.

Electroencephalography (EEG) provides a non-invasive window into brain activity, offering high temporal resolution crucial for understanding and interacting with neural processes through brain-computer interfaces (BCIs). Current dual-stream neural networks for EEG often process temporal and spatial features independently through parallel branches, delaying their integration until a final, late-stage fusion. This design inherently leads to an "information silo" problem, precluding intermediate cross-stream refinement and hindering spatial-temporal decompositions essential for full feature utilization. We propose LI-DSN, a layer-wise interactive dual-stream network that facilitates progressive, cross-stream communication at each layer, thereby overcoming the limitations of late-fusion paradigms. LI-DSN introduces a novel Temporal-Spatial Integration Attention (TSIA) mechanism, which constructs a Spatial Affinity Correlation Matrix (SACM) to capture inter-electrode spatial structural relationships and a Temporal Channel Aggregation Matrix (TCAM) to integrate cosine-gated temporal dynamics under spatial guidance. Furthermore, we employ an adaptive fusion strategy with learnable channel weights to optimize the integration of dual-stream features. Extensive experiments across eight diverse EEG datasets, encompassing motor imagery (MI) classification, emotion recognition, and steady-state visual evoked potentials (SSVEP), consistently demonstrate that LI-DSN significantly outperforms 13 state-of-the-art (SOTA) baseline models, showcasing its superior robustness and decoding performance. The code will be publicized after acceptance.

Deep reinforcement learning (DRL) has recently emerged as a promising tool for Dynamic Algorithm Configuration (DAC), enabling evolutionary algorithms to adapt their parameters online rather than relying on static tuned configurations. While DRL can learn effective control policies, training is computationally expensive. This cost may be justified if learned policies generalize, allowing the training effort to transfer across instance types and problem scales. Yet, for real-world optimization problems, it remains unclear whether this promise holds in practice and under which conditions the investment in learning pays off. In this work, we investigate this question in the context of the carbon-aware permutation flow-shop scheduling problem. We develop a DRL-based DAC framework and train it exclusively on small, simple instances. We then deploy the learned policy on both similar and more complex unseen instances and compare its performance against a static tuned baseline, which provides a fair point of comparison. Our findings show that the proposed method provides a strong dynamic algorithm control policy that can be effectively transferred to different unseen problem instances. Notably, on simple and cheap to compute instances, similar to those observed during training and tuning, DRL performs comparably with the statically tuned baseline. However, as instance characteristics diverge and computational complexities increase, the DRL-learned policy continuously outperforms static tuning. These results confirm that DRL can acquire robust and generalizable control policies which are effective beyond the training instance distributions. This ability to generalize across instance types makes the initial computational investment worthwhile, particularly in settings where static tuning struggles to adapt to changing problem scenarios.

Video Large Language Models (VideoLLMs) have demonstrated impressive capabilities in video understanding, yet the massive number of input video tokens incurs a significant computational burden for deployment. Existing methods mainly prune video tokens at input level while neglecting the inherent information structure embedded in videos and large language models (LLMs). To address this, we propose HieraVid, a hierarchical pruning framework that progressively and dynamically reduces visual redundancy. Based on two observations that videos possess the segment-frame structure and LLMs internally propagate multi-modal information unidirectionally, we decompose pruning into three levels: 1) segment-level, where video tokens are first temporally segmented and spatially merged; 2) frame-level, where similar frames within the same segment are jointly pruned to preserve diversity; 3) layer-level, redundancy gradually shrinks as LLM layer increases w/o compromising performance. We conduct extensive experiments on four widely used video understanding benchmarks to comprehensively evaluate the effectiveness of HieraVid. Remarkably, with only 30% of tokens retained, HieraVid achieves new state-of-the-art performance, while maintaining over 98% and 99% of the performance of LLaVA-Video-7B and LLaVA-OneVision-7B, respectively.

Modern neural networks of the transformer family require the practitioner to decide, before training begins, how many attention heads to use, how deep the network should be, and how wide each component should be. These decisions are made without knowledge of the task, producing architectures that are systematically larger than necessary: empirical studies find that a substantial fraction of heads and layers can be removed after training without performance loss. This paper introduces DDCL-INCRT, an architecture that determines its own structure during training. Two complementary ideas are combined. The first, DDCL (Deep Dual Competitive Learning), replaces the feedforward block with a dictionary of learned prototype vectors representing the most informative directions in the data. The prototypes spread apart automatically, driven by the training objective, without explicit regularisation. The second, INCRT (Incremental Transformer), controls the number of heads: starting from one, it adds a new head only when the directional information uncaptured by existing heads exceeds a threshold. The main theoretical finding is that these two mechanisms reinforce each other: each new head amplifies prototype separation, which in turn raises the signal triggering the next addition. At convergence, the network self-organises into a hierarchy of heads ordered by representational granularity. This hierarchical structure is proved to be unique and minimal, the smallest architecture sufficient for the task, under the stated conditions. Formal guarantees of stability, convergence, and pruning safety are established throughout. The architecture is not something one designs. It is something one derives.

Spectral graph contrastive learning has emerged as a unified paradigm for handling both homophilic and heterophilic graphs by leveraging high-frequency components. However, we identify a fundamental spectral dilemma: while high-frequency signals are indispensable for encoding heterophily, our theoretical analysis proves they exhibit significantly higher variance under spectrally concentrated perturbations. We derive a regret lower bound showing that existing global (node-agnostic) spectral fusion is provably sub-optimal: on mixed graphs with separated node-wise frequency preferences, any global fusion strategy incurs non-vanishing regret relative to a node-wise oracle. To escape this bound, we propose ASPECT, a framework that resolves this dilemma through a reliability-aware spectral gating mechanism. Formulated as a minimax game, ASPECT employs a node-wise gate that dynamically re-weights frequency channels based on their stability against a purpose-built adversary, which explicitly targets spectral energy distributions via a Rayleigh quotient penalty. This design forces the encoder to learn representations that are both structurally discriminative and spectrally robust. Empirical results show that ASPECT achieves new state-of-the-art performance on 8 out of 9 benchmarks, effectively decoupling meaningful structural heterophily from incidental noise.

In modern process industries, data-driven models are important tools for real-time monitoring when key performance indicators are difficult to measure directly. While accurate predictions are essential, reliable uncertainty quantification (UQ) is equally critical for safety, reliability, and decision-making, but remains a major challenge in current data-driven approaches. In this work, we introduce a diffusion-based posterior sampling framework that inherently produces well-calibrated predictive uncertainty via faithful posterior sampling, eliminating the need for post-hoc calibration. In extensive evaluations on synthetic distributions, the Raman-based phenylacetic acid soft sensor benchmark, and a real ammonia synthesis case study, our method achieves practical improvements over existing UQ techniques in both uncertainty calibration and predictive accuracy. These results highlight diffusion samplers as a principled and scalable paradigm for advancing uncertainty-aware modeling in industrial applications.

Stochastic Shortest Path problems (SSPs) are traditionally solved by computing each state's cost-to-go by applying Bellman backups. A Bellman backup updates a state's cost-to-go by iterating through every applicable action, computing the cost-to-go after applying each one, and selecting a minimal action's cost-to-go. State-of-the-art algorithms use heuristic functions; these give an initial estimate of costs-to-go, and lets the algorithm apply Bellman backups only to promising states, determined by low estimated costs-to-go. However, each Bellman backup still considers all applicable actions, even if the heuristic tells us that some of these actions are too expensive, with the effect that such algorithms waste time on unhelpful actions. To address this gap we present a technique that uses the heuristic to avoid expensive actions, by reframing heuristic search in terms of linear programming and introducing an efficient implementation of constraint generation for SSPs. We present CG-iLAO*, a new algorithm that adapts iLAO* with our novel technique, and considers only 40% of iLAO*'s actions on many problems, and as few as 1% on some. Consequently, CG-iLAO* computes on average 3.5x fewer costs-to-go for actions than the state-of-the-art iLAO* and LRTDP, enabling it to solve problems faster an average of 2.8x and 3.7x faster, respectively.

Dyslexic spelling errors exhibit systematic phonological and orthographic patterns that distinguish them from the errors produced by typically developing writers. While this observation has motivated dyslexic-specific spell-checking and assistive writing tools, prior work has focused predominantly on error correction rather than attribution, and has largely neglected the ethical risks. The risk of harmful labelling, covert screening, algorithmic bias, and institutional misuse that automated classification of learners entails requires the development of robust ethical and legal frameworks for research in this area. This paper addresses both gaps. We formulate dyslexic error attribution as a binary classification task. Given a misspelt word and its correct target form, determine whether the error pattern is characteristic of a dyslexic or non-dyslexic writer. We develop a comprehensive feature set capturing orthographic, phonological, and morphological properties of each error, and propose a twin-input neural model evaluated against traditional machine learning baselines under writer-independent conditions. The neural model achieves 93.01% accuracy and an F1-score of 94.01%, with phonetically plausible errors and vowel confusions emerging as the strongest attribution signals. We situate these technical results within an explicit ethics-first framework, analysing fairness across subgroups, the interpretability requirements of educational deployment, and the conditions, consent, transparency, human oversight, and recourse, under which a system could be responsibly used. We provide concrete guidelines for ethical deployment and an open discussion of the systems limitations and misuse potential. Our results demonstrate that dyslexic error attribution is feasible at high accuracy while underscoring that feasibility alone is insufficient for deployment in high-stakes educational contexts.

In the digital age, ensuring the correctness, safety, and reliability of software through formal verification is paramount, particularly as software increasingly underpins critical infrastructure. Formal verification, split into theorem proving and model checking, provides a feasible and reliable path. Unlike theorem proving, which yields notable advances, model checking has been less focused due to the difficulty of automatic program modeling. To fill this gap, we introduce Model-Bench, a benchmark and an accompanying pipeline for evaluating and improving LLMs' program modeling capability by modeling Python programs into verification-ready model checking specifications checkable by its accompanying model checker. Model-Bench comprises 400 Python programs derived from three well-known benchmarks (HumanEval, MBPP, and LiveCodeBench). Our extensive experiments reveal significant limitations in LLMs' program modeling and further provide inspiring directions.

While Large Language Models (LLMs) have demonstrated exceptional proficiency in code completion, they typically adhere to a Hard Completion (HC) paradigm, compelling the generation of fully concrete code even amidst insufficient context. Our analysis of 3 million real-world interactions exposes the limitations of this strategy: 61% of the generated suggestions were either edited after acceptance or rejected despite exhibiting over 80% similarity to the user's subsequent code, suggesting that models frequently make erroneous predictions at specific token positions. Motivated by this observation, we propose Adaptive Placeholder Completion (APC), a collaborative framework that extends HC by strategically outputting explicit placeholders at high-entropy positions, allowing users to fill directly via IDE navigation. Theoretically, we formulate code completion as a cost-minimization problem under uncertainty. Premised on the observation that filling placeholders incurs lower cost than correcting errors, we prove the existence of a critical entropy threshold above which APC achieves strictly lower expected cost than HC. We instantiate this framework by constructing training data from filtered real-world edit logs and design a cost-based reward function for reinforcement learning. Extensive evaluations across 1.5B--14B parameter models demonstrate that APC reduces expected editing costs from 19% to 50% while preserving standard HC performance. Our work provides both a theoretical foundation and a practical training framework for uncertainty-aware code completion, demonstrating that adaptive abstention can be learned end-to-end without sacrificing conventional completion quality.

Multivariate time-series anomaly detection (MTSAD) aims to identify deviations from normality in multivariate time-series and is critical in real-world applications. However, in real-world deployments, distribution shifts are ubiquitous and cause severe performance degradation in pre-trained anomaly detector. Test-time adaptation (TTA) updates a pre-trained model on-the-fly using only unlabeled test data, making it promising for addressing this challenge. In this study, we propose CANDI (Curated test-time adaptation for multivariate time-series ANomaly detection under DIstribution shift), a novel TTA framework that selectively identifies and adapts to potential false positives while preserving pre-trained knowledge. CANDI introduces a False Positive Mining (FPM) strategy to curate adaptation samples based on anomaly scores and latent similarity, and incorporates a plug-and-play Spatiotemporally-Aware Normality Adaptation (SANA) module for structurally informed model updates. Extensive experiments demonstrate that CANDI significantly improves the performance of MTSAD under distribution shift, improving AUROC up to 14% while using fewer adaptation samples.

Vector quantization approaches (VQ-VAE, VQ-GAN) learn discrete neural representations of images, but these representations are inherently position-dependent: codes are spatially arranged and contextually entangled, requiring autoregressive or diffusion-based priors to model their dependencies at sample time. In this work, we ask whether positional information is necessary for discrete representations of spatially aligned data. We propose the permutation-invariant vector-quantized autoencoder (PI-VQ), in which latent codes are constrained to carry no positional information. We find that this constraint encourages codes to capture global, semantic features, and enables direct interpolation between images without a learned prior. To address the reduced information capacity of permutation-invariant representations, we introduce matching quantization, a vector quantization algorithm based on optimal bipartite matching that increases effective bottleneck capacity by $3.5\times$ relative to naive nearest-neighbour quantization. The compositional structure of the learned codes further enables interpolation-based sampling, allowing synthesis of novel images in a single forward pass. We evaluate PI-VQ on CelebA, CelebA-HQ and FFHQ, obtaining competitive precision, density and coverage metrics for images synthesised with our approach. We discuss the trade-offs inherent to position-free representations, including separability and interpretability of the latent codes, pointing to numerous directions for future work.

Clinical prediction from structured electronic health records (EHRs) is challenging due to high dimensionality, heterogeneity, class imbalance, and distribution shift. While tabular in-context learning (TICL) and retrieval-augmented methods perform well on generic benchmarks, their behavior in clinical settings remains unclear. We present a multi-cohort EHR benchmark comparing classical, deep tabular, and TICL models across varying data scale, feature dimensionality, outcome rarity, and cross-cohort generalization. PFN-based TICL models are sample-efficient in low-data regimes but degrade under naive distance-based retrieval as heterogeneity and imbalance increase. We propose AWARE, a task-aligned retrieval framework using supervised embedding learning and lightweight adapters. AWARE improves AUPRC by up to 12.2% under extreme imbalance, with gains increasing with data complexity. Our results identify retrieval quality and retrieval-inference alignment as key bottlenecks for deploying tabular in-context learning in clinical prediction.

While Reinforcement Learning from Verifiable Rewards (RLVR) has advanced reasoning in Large Vision-Language Models (LVLMs), prevailing frameworks suffer from a foundational methodological flaw: by distributing identical advantages across all generated tokens, these methods inherently dilute the learning signals essential for optimizing the critical, visually-grounded steps of multimodal reasoning. To bridge this gap, we formulate \textit{Token Visual Dependency}, quantifying the causal information gain of visual inputs via the Kullback-Leibler (KL) divergence between visual-conditioned and text-only predictive distributions. Revealing that this dependency is highly sparse and semantically pivotal, we introduce Perception-Grounded Policy Optimization (PGPO), which is a novel fine-grained credit assignment framework that dynamically reshapes advantages at the token level. Through a threshold-gated, mass-conserving mechanism, PGPO actively amplifies learning signals for visually-dependent tokens while suppressing gradient noise from linguistic priors. Extensive experiments based on the Qwen2.5-VL series across seven challenging multimodal reasoning benchmarks demonstrate that PGPO boosts models by 18.7% on average. Both theoretical and empirical analyses confirm that PGPO effectively reduces gradient variance, prevents training collapse, and acts as a potent regularizer for robust, perception-grounded multimodal reasoning. Code will be published on https://github.com/Yzk1114/PGPO.

A core tenet underpinning the conception of contemporary information networks, such as social media platforms, is that users should not be constrained in the amount of information they can freely and willingly exchange with one another about a given topic. By means of a computational agent-based model, we show how even in groups of truth-seeking and cooperative agents with perfect information-processing abilities, unconstrained information exchange may lead to detrimental effects on the correctness of the group's beliefs. If unconstrained information exchange can be detrimental even among such idealized agents, it is prudent to assume it can also be so in practice. We therefore argue that constraints on information flow should be carefully considered in the design of communication networks with substantial societal impact, such as social media platforms.

Preference learning in Large Language Models (LLMs) has advanced significantly, yet existing methods remain limited by modest performance gains, high computational costs, hyperparameter sensitivity, and insufficient modeling of global token-level relationships. We introduce PLOT, which enhances Preference Learning in fine-tuning-based alignment through a token-level loss derived from Optimal Transport. By formulating preference learning as an Optimal Transport Problem, PLOT aligns model outputs with human preferences while preserving the original distribution of LLMs, ensuring stability and robustness. Furthermore, PLOT leverages token embeddings to capture semantic relationships, enabling globally informed optimization. Experiments across two preference categories - Human Values and Logic & Problem Solving - spanning seven subpreferences demonstrate that PLOT consistently improves alignment performance while maintaining fluency and coherence. These results substantiate optimal transport as a principled methodology for preference learning, establishing a theoretically grounded framework that provides new insights for preference learning of LLMs.

The ratio of outlier parameters in language pre-training models and vision pre-training models differs significantly, making cross-modality (language and vision) inherently more challenging than cross-domain adaptation. As a result, many prior studies have focused on cross-domain transfer rather than attempting to bridge language and vision modalities, assuming that language pre-trained models are unsuitable for downstream visual tasks due to disparate parameter spaces. Contrary to this assumption, we show that adding a bridge training stage as a modality adaptation learner can effectively align Large Language Model (LLM) parameters with vision tasks. Specifically, we propose a simple yet powerful solution random label bridge training that requires no manual labeling and helps LLM parameters adapt to vision foundation tasks. Moreover, our findings reveal that partial bridge training is often advantageous, as certain layers in LLMs exhibit strong foundational properties that remain beneficial even without fine-tuning for visual tasks. This surprising discovery opens up new avenues for leveraging language pre-trained parameters directly within vision models and highlights the potential of partial bridge training as a practical pathway to cross-modality adaptation.

Topology control for power grid operation is a challenging sequential decision making problem because the action space grows combinatorially with the size of the grid and action evaluation through simulation is computationally expensive. We propose a physics-informed Reinforcement Learning framework that combines semi-Markov control with a Gibbs prior, that encodes the system's physics, over the action space. The decision is only taken when the grid enters a hazardous regime, while a graph neural network surrogate predicts the post action overload risk of feasible topology actions. These predictions are used to construct a physics-informed Gibbs prior that both selects a small state-dependent candidate set and reweights policy logits before action selection. In this way, our method reduces exploration difficulty and online simulation cost while preserving the flexibility of a learned policy. We evaluate the approach in three realistic benchmark environments of increasing difficulty. Across all settings, the proposed method achieves a strong balance between control quality and computational efficiency: it matches oracle-level performance while being approximately $6\times$ faster on the first benchmark, reaches $94.6\%$ of oracle reward with roughly $200\times$ lower decision time on the second one, and on the most challenging benchmark improves over a PPO baseline by up to $255\%$ in reward and $284\%$ in survived steps while remaining about $2.5\times$ faster than a strong specialized engineering baseline. These results show that our method provides an effective mechanism for topology control in power grids.

Range minimum queries are frequently used in string processing and database applications including biological sequence analysis, document retrieval, and web search. Hence, various data structures have been proposed for improving their efficiency on both CPUs and GPUs.Recent work has also shown that hardware-accelerated ray tracing on modern NVIDIA RTX graphic cards can be exploited to answer range minimum queries by expressing queries as rays, which are fired into a scene of triangles representing minima of ranges at different granularities. While these approaches are promising, they suffer from at least one of three issues: severe memory overhead, high index construction time, and low query throughput. This renders these methods practically unusable on larger arrays: For example, the state-of-art GPU-based approaches LCA and RTXRMQ exceed the memory capacity of an NVIDIA RTX 4090 GPU for input arrays of size >= 2^29. To tackle these problems, in this work, we present a new approach called GPU-RMQ which is based on a hierarchical approach. GPU-RMQ first constructs a hierarchy of range minimum summaries on top of the original array in a highly parallel fashion. For query answering, only the relevant portions of the hierarchy are then processed in an optimized massively-parallel scan operation. Additionally, GPU-RMQ is hybrid in design enabling the use of both ray tracing cores and CUDA cores across different levels of the hierarchy to handle queries. Our experimental evaluation shows that GPU-RMQ outperforms the state-of-the-art approaches in terms of query throughput especially for larger arrays while offering a significantly lower memory footprint and up to two orders-of-magnitude faster index construction. In particular, it achieves up to ~8x higher throughput than LCA, ~17x higher throughput than RTXRMQ, and up to ~4800x higher throughput compared to an optimized CPU-based approach.

In Europe, balance responsible parties can deliberately take out-of-balance positions to support transmission system operators (TSOs) in maintaining grid stability and earn profit, a practice called implicit balancing. Model predictive control (MPC) is widely adopted as an effective approach for implicit balancing. The balancing market model accuracy in MPC is critical to decision quality. Previous studies modeled this market using either (i) a convex market clearing approximation, ignoring proactive manual actions by TSOs and the market sub-quarter-hour dynamics, or (ii) machine learning methods, which cannot be directly integrated into MPC. To address these shortcomings, we propose a data-driven balancing market model integrated into MPC using an input convex neural network to ensure convexity while capturing uncertainties. To keep the core network computationally efficient, we incorporate attention-based input gating mechanisms to remove irrelevant data. Evaluating on Belgian data shows that the proposed model both improves MPC decisions and reduces computational time.

Accurate sensing of spatially distributed physical fields typically requires dense instrumentation, which is often infeasible in real-world systems due to cost, accessibility, and environmental constraints. Physics-based solvers address this through direct numerical integration of governing equations, but their computational latency and power requirements preclude real-time use in resource-constrained monitoring and control systems. Here we introduce VIRSO (Virtual Irregular Real-Time Sparse Operator), a graph-based neural operator for sparse-to-dense reconstruction on irregular geometries, and a variable-connectivity algorithm, Variable KNN (V-KNN), for mesh-informed graph construction. Unlike prior neural operators that treat hardware deployability as secondary, VIRSO reframes inference as measurement: the combination of both spectral and spatial analysis provides accurate reconstruction without the high latency and power consumption of previous graph-based methodologies with poor scalability, presenting VIRSO as a potential candidate for edge-constrained, real-time virtual sensing. We evaluate VIRSO on three nuclear thermal-hydraulic benchmarks of increasing geometric and multiphysics complexity, across reconstruction ratios from 47:1 to 156:1. VIRSO achieves mean relative $L_2$ errors below 1%, outperforming other benchmark operators while using fewer parameters. The full 10-layer configuration reduces the energy-delay product (EDP) from ${\approx}206$ J$\cdot$ms for the graph operator baseline to $10.1$ J$\cdot$ms on an NVIDIA H200. Implemented on an NVIDIA Jetson Orin Nano, all configurations of VIRSO provide sub-10 W power consumption and sub-second latency. These results establish the edge-feasibility and hardware-portability of VIRSO and present compute-aware operator learning as a new paradigm for real-time sensing in inaccessible and resource-constrained environments.

With the rapid evolution of LLMs, automated software testing is witnessing a paradigm shift. While proprietary models like GPT-4o demonstrate impressive capabilities, their high deployment costs and data privacy concerns make open-source LLMs the practical imperative for many academic and industrial scenarios. In the field of automated test generation, it has evolved to iterative workflows to construct test suites based on LLMs. When utilizing open-source LLMs, we empirically observe they lack a suite-level perspective, suffering from structural myopia-failing to generate new tests with large marginal gain based on the current covered status. In this paper, from the perspective of sequences, we formalize test suite generation as a MDP and demonstrate that its objective exhibits monotone submodularity, which enables an effective relaxation of this NP-hard global optimization into a tractable step-wise greedy procedure. Guided by this insight, we propose TestDecision, which transforms LLMs into neural greedy experts. TestDecision consists of two synergistic components: (1) an inference framework which implements test suite construction following a step-wise greedy strategy; and (2) a training pipeline of reinforcement learning which equips the base LLM with sequential test generation ability to maximize marginal gain. Comprehensive evaluations on the ULT benchmark demonstrate that TestDecision significantly outperforms existing advanced methods. It brings an improvement between 38.15-52.37% in branch coverage and 298.22-558.88% in execution pass rate over all base models, achieving a comparable performance on 7B backbone with a much larger proprietary LLM GPT-5.2. Furthermore, TestDecision can find 58.43-95.45% more bugs than vanilla base LLMs and exhibit superior generalization on LiveCodeBench, proving its capability to construct high-quality test suites.

Breast cancer is a highly heterogeneous disease with diverse molecular profiles. The PAM50 gene signature is widely recognized as a standard for classifying breast cancer into intrinsic subtypes, enabling more personalized treatment strategies. In this study, we introduce a novel optimization-driven deep learning framework that aims to reduce reliance on costly molecular assays by directly predicting PAM50 subtypes from H&E-stained whole-slide images (WSIs). Our method jointly optimizes patch informativeness, spatial diversity, uncertainty, and patch count by combining the non-dominated sorting genetic algorithm II (NSGA-II) with Monte Carlo dropout-based uncertainty estimation. The proposed method can identify a small but highly informative patch subset for classification. We used a ResNet18 backbone for feature extraction and a custom CNN head for classification. For evaluation, we used the internal TCGA-BRCA dataset as the training cohort and the external CPTAC-BRCA dataset as the test cohort. On the internal dataset, an F1-score of 0.8812 and an AUC of 0.9841 using 627 WSIs from the TCGA-BRCA cohort were achieved. The performance of the proposed approach on the external validation dataset showed an F1-score of 0.7952 and an AUC of 0.9512. These findings indicate that the proposed optimization-guided, uncertainty-aware patch selection can achieve high performance and improve the computational efficiency of histopathology-based PAM50 classification compared to existing methods, suggesting a scalable imaging-based replacement that has the potential to support clinical decision-making.

We study the prophet inequality, a fundamental problem in online decision-making and optimal stopping, in a practical setting where rewards are observed only through noisy realizations and reward distributions are unknown. At each stage, the decision-maker receives a noisy reward whose true value follows a linear model with an unknown latent parameter, and observes a feature vector drawn from a distribution. To address this challenge, we propose algorithms that integrate learning and decision-making via lower-confidence-bound (LCB) thresholding. In the i.i.d.\ setting, we establish that both an Explore-then-Decide strategy and an $\varepsilon$-Greedy variant achieve the sharp competitive ratio of $1 - 1/e$, under a mild condition on the optimal value. For non-identical distributions, we show that a competitive ratio of $1/2$ can be guaranteed against a relaxed benchmark. Moreover, with limited window access to past rewards, the tight ratio of $1/2$ against the optimal benchmark is achieved.

Reinforcement Learning from Human Feedback (RLHF), using algorithms like Proximal Policy Optimization (PPO), aligns Large Language Models (LLMs) with human values but is costly and unstable. Alternatives have been proposed to replace PPO or integrate Supervised Fine-Tuning (SFT) and contrastive learning for direct fine-tuning and value alignment. However, these methods still require voluminous data to learn preferences and may weaken the generalization ability of LLMs. To further enhance alignment efficiency and performance while mitigating the loss of generalization ability, this paper introduces Distribution-guided Efficient Fine-Tuning (DEFT), an efficient alignment framework incorporating data filtering and distributional guidance by calculating the differential distribution reward based on the output distribution of language model and the discrepancy distribution of preference data. A small yet high-quality subset is filtered from the raw data using a differential distribution reward, which is then incorporated into existing alignment methods to guide the model's output distribution. Experimental results demonstrate that the methods enhanced by DEFT outperform the original methods in both alignment capability and generalization ability, with significantly reduced training time.

Controllable Automatic Text Simplification (CATS) produces user-tailored outputs, yet controllability is often treated as a decoding problem and evaluated with metrics that are not reflective to the measure of control. We observe that controllability in ATS is significantly constrained by data and evaluation. To this end, we introduce a domain-agnostic CATS framework based on instruction fine-tuning with discrete control tokens, steering open-source models to target readability levels and compression rates. Across three model families with different model sizes (Llama, Mistral, Qwen; 1-14B) and four domains (medicine, public administration, news, encyclopedic text), we find that smaller models (1-3B) can be competitive, but reliable controllability strongly depends on whether the training data encodes sufficient variation in the target attribute. Readability control (FKGL, ARI, Dale-Chall) is learned consistently, whereas compression control underperforms due to limited signal variability in the existing corpora. We further show that standard simplification and similarity metrics are insufficient for measuring control, motivating error-based measures for target-output alignment. Finally, our sampling and stratification experiments demonstrate that naive splits can introduce distributional mismatch that undermines both training and evaluation.

Although demand forecasting is a critical component of supply chain planning, actual retail data can exhibit irreconcilable seasonality, irregular spikes, and noise, rendering precise projections nearly unattainable. This paper proposes a three-step analytical framework that combines forecasting and operational analytics. The first stage consists of exploratory data analysis, where delivery-tracked data from 180,519 transactions are partitioned, and long-term trends, seasonality, and delivery-related attributes are examined. Secondly, the forecasting performance of a statistical time series decomposition model N-BEATS MSTL and a recent deep learning architecture N-HiTS were compared. N-BEATS and N-HiTS were both statistically, and hence were N-BEATS's and N-HiTS's statistically selected. Most recent time series deep learning models, N-HiTS, N-BEATS. N-HiTS and N-BEATS N-HiTS and N-HiTS outperformed the statistical benchmark to a large extent. N-BEATS was selected to be the most optimized model, as the one with the lowest forecasting error, in the 3rd and final stage forecasting values of the next 4 weeks of 1918 units, and provided those as a model with a set of deterministically integer linear program outcomes that are aimed to minimize the total delivery time with a set of bound budget, capacity, and service constraints. The solution allocation provided a feasible and cost-optimal shipping plan. Overall, the study provides a compelling example of the practical impact of precise forecasting and simple, highly interpretable model optimization in logistics.

Knowledge graphs store large numbers of relations efficiently, but they remain weak at representing a quieter difficulty: the meaning of a concept often shifts with the domain in which it is used. A triple such as Apple, instance-of, Company may be acceptable in one setting while being misleading or unusable in another. In most current systems, domain information is attached as metadata, qualifiers, or graph-level organization. These mechanisms help with filtering and provenance, but they usually do not alter the formal status of the assertion itself. This paper argues that domain should be treated as part of knowledge representation rather than as supplementary annotation. It introduces the Domain-Contextualized Concept Graph (DCG), a framework in which domain is written into the relation and interpreted as a modal world constraint. In the DCG form (C, R at D, C'), the marker at D identifies the world in which the relation holds. Formally, the relation is interpreted through a domain-indexed necessity operator, so that truth, inference, and conflict checking are all scoped to the relevant world. This move has three consequences: ambiguous concepts can be disambiguated at the point of representation; invalid assertions can be challenged against their domain; cross-domain relations can be connected through explicit predicates. The paper develops this claim through a Kripke-style semantics, a compact predicate system, a Prolog implementation, and mappings to RDF, OWL, and relational databases. The contribution is a representational reinterpretation of domain itself. The central claim is that many practical failures in knowledge systems begin when domain is treated as external to the assertion. DCG addresses that by giving domain a structural and computable role inside the representation.

For multi-input and multi-output (MIMO) channels, the optimal channel estimation (CE) based on linear minimum mean square error (LMMSE) requires three-dimensional (3D) filtering. However, the complexity is often prohibitive due to large matrix dimensions. Suboptimal estimators approximate 3DCE by decomposing it into time, frequency, and spatial domains, while yields noticeable performance degradation under correlated MIMO channels. On the other hand, recent advances in deep learning (DL) can explore channel correlations in all domains via attention mechanisms. Building on this capability, we propose a dual attention mechanism based 3DCE network (3DCENet) that can achieve accurate estimates.

Very High Resolution (VHR) forest structure data at individual-tree scale is essential for carbon, biodiversity, and ecosystem monitoring. Still, airborne LiDAR remains costly and infrequent despite being the reference for forest structure metrics like Canopy Height Model (CHM), Plant Area Index (PAI), and Foliage Height Diversity (FHD). We propose FSKD: a LiDAR-to-RGB-Infrared (RGBI) knowledge distillation (KD) framework in which a multi-modal teacher fuses RGBI imagery with LiDAR-derived planar metrics and vertical profiles via cross-attention, and an RGBI-only SegFormer student learns to reproduce these outputs. Trained on 384 $km^2$ of forests in Saxony, Germany (20 cm ground sampling distance (GSD)) and evaluated on eight geographically distinct test tiles, the student achieves state-of-the-art (SOTA) zero-shot CHM performance (MedAE 4.17 m, $R^2$=0.51, IoU 0.87), outperforming HRCHM/DAC baselines by 29--46% in MAE (5.81 m vs. 8.14--10.84 m) with stronger correlation coefficients (0.713 vs. 0.166--0.652). Ablations show that multi-modal fusion improves performance by 10--26% over RGBI-only training, and that asymmetric distillation with appropriate model capacity is critical. The method jointly predicts CHM, PAI, and FHD, a multi-metric capability not provided by current monocular CHM estimators, although PAI/FHD transfer remains region-dependent and benefits from local calibration. The framework also remains effective under temporal mismatch (winter LiDAR, summer RGBI), removing strict co-acquisition constraints and enabling scalable 20 cm operational monitoring for workflows such as Digital Twin Germany and national Digital Orthophoto programs.

Recently, world-action models (WAM) have emerged to bridge vision-language-action (VLA) models and world models, unifying their reasoning and instruction-following capabilities and spatio-temporal world modeling. However, existing WAM approaches often focus on modeling 2D appearance or latent representations, with limited geometric grounding-an essential element for embodied systems operating in the physical world. We present DriveDreamer-Policy, a unified driving world-action model that integrates depth generation, future video generation, and motion planning within a single modular architecture. The model employs a large language model to process language instructions, multi-view images, and actions, followed by three lightweight generators that produce depth, future video, and actions. By learning a geometry-aware world representation and using it to guide both future prediction and planning within a unified framework, the proposed model produces more coherent imagined futures and more informed driving actions, while maintaining modularity and controllable latency. Experiments on the Navsim v1 and v2 benchmarks demonstrate that DriveDreamer-Policy achieves strong performance on both closed-loop planning and world generation tasks. In particular, our model reaches 89.2 PDMS on Navsim v1 and 88.7 EPDMS on Navsim v2, outperforming existing world-model-based approaches while producing higher-quality future video and depth predictions. Ablation studies further show that explicit depth learning provides complementary benefits to video imagination and improves planning robustness.

Parameter-efficient fine-tuning (PEFT) has emerged as a crucial paradigm for adapting large language models (LLMs) under constrained computational budgets. However, standard PEFT methods often struggle in multi-task fine-tuning settings, where diverse optimization objectives induce task interference and limited parameter budgets lead to representational deficiency. While recent approaches incorporate mixture-of-experts (MoE) to alleviate these issues, they predominantly operate in the spatial domain, which may introduce structural redundancy and parameter overhead. To overcome these limitations, we reformulate adaptation in the spectral domain. Our spectral analysis reveals that different tasks exhibit distinct frequency energy distributions, and that LLM layers display heterogeneous frequency sensitivities. Motivated by these insights, we propose FourierMoE, which integrates the MoE architecture with the inverse discrete Fourier transform (IDFT) for frequency-aware adaptation. Specifically, FourierMoE employs a frequency-adaptive router to dispatch tokens to experts specialized in distinct frequency bands. Each expert learns a set of conjugate-symmetric complex coefficients, preserving complete phase and amplitude information while theoretically guaranteeing lossless IDFT reconstruction into real-valued spatial weights. Extensive evaluations across 28 benchmarks, multiple model architectures, and scales demonstrate that FourierMoE consistently outperforms competitive baselines in both single-task and multi-task settings while using significantly fewer trainable parameters. These results highlight the promise of spectral-domain expert adaptation as an effective and parameter-efficient paradigm for LLM fine-tuning.

REST APIs are widely used in industry, in all different kinds of domains. An example is Volkswagen AG, a German automobile manufacturer. Established testing approaches for REST APIs are time consuming, and require expertise from professional test engineers. Due to its cost and importance, in the scientific literature several approaches have been proposed to automatically test REST APIs. The open-source, search-based fuzzer EvoMaster is one of such tools proposed in the academic literature. However, how academic prototypes can be integrated in industry and have real impact to software engineering practice requires more investigation. In this paper, we report on our experience in using EvoMaster at Volkswagen AG, as an EvoMaster user from 2023 to 2026. We share our learnt lessons, and discuss several features needed to be implemented in EvoMaster to make its use in an industrial context successful. Feedback about value in industrial setups of EvoMaster was given from Volkswagen AG about 4 APIs. Additionally, a user study was conducted involving 11 testing specialists from 4 different companies. We further identify several real-world research challenges that still need to be solved.

Mathematical reasoning is a hallmark of human intelligence, and whether large language models (LLMs) can meaningfully perform it remains a central question in artificial intelligence and cognitive science. As LLMs are increasingly integrated into scientific workflows, rigorous evaluation of their mathematical capabilities becomes a practical necessity. Existing benchmarks are limited by synthetic settings and data contamination. We present LiveMathematicianBench, a dynamic multiple-choice benchmark for research-level mathematical reasoning built from recent arXiv papers published after model training cutoffs. By grounding evaluation in newly published theorems, it provides a realistic testbed beyond memorized patterns. The benchmark introduces a thirteen-category logical taxonomy of theorem types (e.g., implication, equivalence, existence, uniqueness), enabling fine-grained evaluation across reasoning forms. It employs a proof-sketch-guided distractor pipeline that uses high-level proof strategies to construct plausible but invalid answer choices reflecting misleading proof directions, increasing sensitivity to genuine understanding over surface-level matching. We also introduce a substitution-resistant mechanism to distinguish answer recognition from substantive reasoning. Evaluation shows the benchmark is far from saturated: Gemini-3.1-pro-preview, the best model, achieves only 43.5%. Under substitution-resistant evaluation, accuracy drops sharply: GPT-5.4 scores highest at 30.6%, while Gemini-3.1-pro-preview falls to 17.6%, below the 20% random baseline. A dual-mode protocol reveals that proof-sketch access yields consistent accuracy gains, suggesting models can leverage high-level proof strategies for reasoning. Overall, LiveMathematicianBench offers a scalable, contamination-resistant testbed for studying research-level mathematical reasoning in LLMs.

Toxic content detection in online communication remains a significant challenge, with current solutions often inadvertently blocking valuable information, including medical terms and text related to minority groups. This paper presents a more nu-anced approach to identifying toxicity in Bulgarian text while preserving access to essential information. The research explores two distinct methodologies for detecting toxic content. The developed methodologies have po-tential applications across diverse online platforms and content moderation systems. First, we propose an ontology that models the potentially toxic words in Bulgarian language. Then, we compose a dataset that comprises 4,384 manually anno-tated sentences from Bulgarian online forums across four categories: toxic language, medical terminology, non-toxic lan-guage, and terms related to minority communities. We then train a BERT-based model for toxic language classification, which reaches a 0.89 F1 macro score. The trained model is directly applicable in a real environment and can be integrated as a com-ponent of toxic content detection systems.

A persistent structural weakness in deep clustering is the disconnect between feature learning and cluster assignment. Most architectures invoke an external clustering step, typically k-means, to produce pseudo-labels that guide training, preventing the backbone from directly optimising for cluster quality. This paper introduces Deep Dual Competitive Learning (DDCL), the first fully differentiable end-to-end framework for unsupervised prototype-based representation learning. The core contribution is architectural: the external k-means is replaced by an internal Dual Competitive Layer (DCL) that generates prototypes as native differentiable outputs of the network. This single inversion makes the complete pipeline, from backbone feature extraction through prototype generation to soft cluster assignment, trainable by backpropagation through a single unified loss, with no Lloyd iterations, no pseudo-label discretisation, and no external clustering step. To ground the framework theoretically, the paper derives an exact algebraic decomposition of the soft quantisation loss into a simplex-constrained reconstruction error and a non-negative weighted prototype variance term. This identity reveals a self-regulating mechanism built into the loss geometry: the gradient of the variance term acts as an implicit separation force that resists prototype collapse without any auxiliary objective, and leads to a global Lyapunov stability theorem for the reduced frozen-encoder system. Six blocks of controlled experiments validate each structural prediction. The decomposition identity holds with zero violations across more than one hundred thousand training epochs; the negative feedback cycle is confirmed with Pearson -0.98; with a jointly trained backbone, DDCL outperforms its non-differentiable ablation by 65% in clustering accuracy and DeepCluster end-to-end by 122%.

Integrating Large Language Models (LLMs) into hypersonic thermal protection system (TPS) design is bottlenecked by cascading constraint violations when generating executable simulation artifacts. General-purpose LLMs, treating generation as single-pass text completion, fail to satisfy the sequential, multi-gate constraints inherent in safety-critical engineering workflows. To address this, we propose AeroTherm-GPT, the first TPS-specialized LLM Agent, instantiated through a Constraint-Closed-Loop Generation (CCLG) framework. CCLG organizes TPS artifact generation as an iterative workflow comprising generation, validation, CDG-guided repair, execution, and audit. The Constraint Dependency Graph (CDG) encodes empirical co-resolution structure among constraint categories, directing repair toward upstream fault candidates based on lifecycle ordering priors and empirical co-resolution probabilities. This upstream-priority mechanism resolves multiple downstream violations per action, achieving a Root-Cause Fix Efficiency of 4.16 versus 1.76 for flat-checklist repair. Evaluated on HyTPS-Bench and validated against external benchmarks, AeroTherm-GPT achieves 88.7% End-to-End Success Rate (95% CI: 87.5-89.9), a gain of +12.5 pp over the matched non-CDG ablation baseline, without catastrophic forgetting on scientific reasoning and code generation tasks.

Retrieval-Augmented Generation (RAG) systems critically depend on retrieval quality, yet no systematic comparison of modern retrieval methods exists for heterogeneous documents containing both text and tabular data. We benchmark ten retrieval strategies spanning sparse, dense, hybrid fusion, cross-encoder reranking, query expansion, index augmentation, and adaptive retrieval on a challenging financial QA benchmark of 23,088 queries over 7,318 documents with mixed text-and-table content. We evaluate retrieval quality via Recall@k, MRR, and nDCG, and end-to-end generation quality via Number Match, with paired bootstrap significance testing. Our results show that (1) a two-stage pipeline combining hybrid retrieval with neural reranking achieves Recall@5 of 0.816 and MRR@3 of 0.605, outperforming all single-stage methods by a large margin; (2) BM25 outperforms state-of-the-art dense retrieval on financial documents, challenging the common assumption that semantic search universally dominates; and (3) query expansion methods (HyDE, multi-query) and adaptive retrieval provide limited benefit for precise numerical queries, while contextual retrieval yields consistent gains. We provide ablation studies on fusion methods and reranker depth, actionable cost-accuracy recommendations, and release our full benchmark code.

Cutting rectangular items from stock sheets to satisfy demands while minimizing waste is a central manufacturing task. The Two-Dimensional Single Stock Size Cutting Stock Problem (2D-CSSP) generalizes bin packing by requiring multiple copies of each item type, which causes a strong combinatorial blow-up. We present a SAT-based framework where item types are expanded by demand, each copy has a sheet-assignment variable and non-overlap constraints are activated only for copies assigned to the same sheet. We also introduce an infeasible-orientation elimination rule that fixes rotation variables when only one orientation can fit the sheet. For minimizing the number of sheets, we compare three approaches: non-incremental SAT with binary search, incremental SAT with clause reuse across iterations and weighted partial MaxSAT. On the Cui--Zhao benchmark suite, our best SAT configurations certify two to three times more instances as provably optimal and achieve lower optimality gaps than OR-Tools, CPLEX and Gurobi. The relative ranking among SAT approaches depends on rotation: incremental SAT is strongest without rotation, while non-incremental SAT is more effective when rotation increases formula size.

This paper investigates Koopman operator-based approaches for multivariable control of a two-spool turbofan engine. A physics-based component-level model is developed to generate training data and validate the controllers. A meta-heuristic extended dynamic mode decomposition is developed, with a cost function designed to accurately capture both spool-speed dynamics and the engine pressure ratio (EPR), enabling the construction of a single Koopman model suitable for multiple control objectives. Using the identified time-varying Koopman model, two controllers are developed: an adaptive Koopman-based model predictive controller (AKMPC) with a disturbance observer and a Koopman-based feedback linearization controller (K-FBLC), which serves as a benchmark. The controllers are evaluated for two control strategies, namely configurations of spool speeds and EPR, under both sea-level and varying flight conditions. The results demonstrate that the proposed identification approach enables accurate predictions of both spool speeds and EPR, allowing the Koopman model to be reused flexibly across different control formulations. While both control strategies achieve comparable performance in steady conditions, the AKMPC exhibits superior robustness compared with the K-FBLC under varying flight conditions due to its ability to compensate for model mismatch. Moreover, the EPR control strategy improves the thrust response. The study highlights the applicability of Koopman-based control and demonstrates the advantages of the AKMPC-based framework for robust turbofan engine control.

Achieving semantic interoperability across heterogeneous experimental data systems remains a major barrier to data-driven scientific discovery. The Analytical Information Markup Language (AnIML), a flexible XML-based standard for analytical chemistry and biology, is increasingly used in industrial R&D labs for managing and exchanging experimental data. However, the expressivity of the XML schema permits divergent interpretations across stakeholders, introducing inconsistencies that undermine the interoperability the AnIML schema was designed to support. In this paper, we present the AnIML Ontology, an OWL 2 ontology that formalises the semantics of AnIML and aligns it with the Allotrope Data Format to support future cross-system and cross-lab interoperability. The ontology was developed using an expert-in-the-loop approach combining LLM-assisted requirement elicitation with collaborative ontology engineering. We validate the ontology through a multi-layered approach: data-driven transformation of real-world AnIML files into knowledge graphs, competency question verification via SPARQL, and a novel validation protocol based on adversarial negative competency questions mapped to established ontological anti-patterns and enforced via SHACL constraints.

Forecasting evolving clinical risks relies on intrinsic pathological dependencies rather than mere chronological proximity, yet current methods struggle with coarse binary supervision and physical timestamps. To align predictive modeling with clinical logic, we propose the Medical-semantics Aware Time-ALiBi Transformer (MATA-Former), utilizing event semantics to dynamically parameterize attention weights to prioritize causal validity over time lags. Furthermore, we introduce Plateau-Gaussian Soft Labeling (PSL), reformulating binary classification into continuous multi-horizon regression for full-trajectory risk modeling. Evaluated on SIICU -- a newly constructed dataset featuring over 506k events with rigorous expert-verified, fine-grained annotations -- and the MIMIC-IV dataset, our framework demonstrates superior efficacy and robust generalization in capturing risks from text-intensive, irregular clinical time series.

General aviation fault diagnosis and efficient maintenance are critical to flight safety; however, deploying deep learning models on resource-constrained edge devices poses dual challenges in computational capacity and interpretability. This paper proposes LiteInception--a lightweight interpretable fault diagnosis framework designed for edge deployment. The framework adopts a two-stage cascaded architecture aligned with standard maintenance workflows: Stage 1 performs high-recall fault detection, and Stage 2 conducts fine-grained fault classification on anomalous samples, thereby decoupling optimization objectives and enabling on-demand allocation of computational resources. For model compression, a multi-method fusion strategy based on mutual information, gradient analysis, and SE attention weights is proposed to reduce the input sensor channels from 23 to 15, and a 1+1 branch LiteInception architecture is introduced that compresses InceptionTime parameters by 70%, accelerates CPU inference by over 8x, with less than 3% F1 loss. Furthermore, knowledge distillation is introduced as a precision-recall regulation mechanism, enabling the same lightweight model to adapt to different scenarios--such as safety-critical and auxiliary diagnosis--by switching training strategies. Finally, a dual-layer interpretability framework integrating four attribution methods is constructed, providing traceable evidence chains of "which sensor x which time period." Experiments on the NGAFID dataset demonstrate a fault detection accuracy of 81.92% with 83.24% recall, and a fault identification accuracy of 77.00%, validating the framework's favorable balance among efficiency, accuracy, and interpretability.

Vision-Language-Action (VLA) models for autonomous driving must integrate diverse textual inputs, including navigation commands, hazard warnings, and traffic state descriptions, yet current systems often present these as disconnected fragments, forcing the model to discover on its own which environmental constraints are relevant to the current maneuver. We introduce Causal Scene Narration (CSN), which restructures VLA text inputs through intent-constraint alignment, quantitative grounding, and structured separation, at inference time with zero GPU cost. We complement CSN with Simplex-based runtime safety supervision and training-time alignment via Plackett-Luce DPO with negative log-likelihood (NLL) regularization. A multi-town closed-loop CARLA evaluation shows that CSN improves Driving Score by +31.1% on original LMDrive and +24.5% on the preference-aligned variant. A controlled ablation reveals that causal structure accounts for 39.1% of this gain, with the remainder attributable to information content alone. A perception noise ablation confirms that CSN's benefit is robust to realistic sensing errors. Semantic safety supervision improves Infraction Score, while reactive Time-To-Collision monitoring degrades performance, demonstrating that intent-aware monitoring is needed for VLA systems.

The wind-induced structural response forecasting capabilities of a novel transformer methodology are examined here. The model also provides a digital twin component for bridge structural health monitoring. Firstly, the approach uses the temporal characteristics of the system to train a forecasting model. Secondly, the vibration predictions are compared to the measured ones to detect large deviations. Finally, the identified cases are used as an early-warning indicator of structural change. The artificial intelligence-based model outperforms approaches for response forecasting as no assumption on wind stationarity or on structural normal vibration behavior is needed. Specifically, wind-excited dynamic behavior suffers from uncertainty related to obtaining poor predictions when the environmental or traffic conditions change. This results in a hard distinction of what constitutes normal vibration behavior. To this end, a framework is rigorously examined on real-world measurements from the Hardanger Bridge monitored by the Norwegian University of Science and Technology. The approach captures accurate structural behavior in realistic conditions, and with respect to the changes in the system excitation. The results, importantly, highlight the potential of transformer-based digital twin components to serve as next-generation tools for resilient infrastructure management, continuous learning, and adaptive monitoring over the system's lifecycle with respect to temporal characteristics.

Vietnamese Speech Emotion Recognition (SER) remains challenging due to ambiguous acoustic patterns and the lack of reliable annotated data, especially in real-world conditions where emotional boundaries are not clearly separable. To address this problem, this paper proposes a human-machine collaborative framework that integrates human knowledge into the learning process rather than relying solely on data-driven models. The proposed framework is centered around LLM-based reasoning, where acoustic feature-based models are used to provide auxiliary signals such as confidence and feature-level evidence. A confidence-based routing mechanism is introduced to distinguish between easy and ambiguous samples, allowing uncertain cases to be delegated to LLMs for deeper reasoning guided by structured rules derived from human annotation behavior. In addition, an iterative refinement strategy is employed to continuously improve system performance through error analysis and rule updates. Experiments are conducted on a Vietnamese speech dataset of 2,764 samples across three emotion classes (calm, angry, panic), with high inter-annotator agreement (Fleiss Kappa = 0.8574), ensuring reliable ground truth. The proposed method achieves strong performance, reaching up to 86.59% accuracy and Macro F1 around 0.85-0.86, demonstrating its effectiveness in handling ambiguous and hard-to-classify cases. Overall, this work highlights the importance of combining data-driven models with human reasoning, providing a robust and model-agnostic approach for speech emotion recognition in low-resource settings.

Adaptation to complex tasks and multiple scenarios remains a significant challenge for a single robot agent. The ability to acquire organize, and switch between a wide range of skills in real time, particularly in dynamic environments, has become a fundamental requirement for embodied intelligence. We introduce OpenGo, an OpenClaw-powered embodied robotic dog capable of switching skills in real time according to the scene and task instructions. Specifically, the agent is equipped with (1) a customizable skill library with easy skill import and autonomous skill validation, (2) a dispatcher that selects and invokes different skills according to task prompts or language instructions, and (3) a self-learning framework that fine-tunes skills based on task completion and human feedback. We deploy the agent in Unitree's Go2 robotic dog and validate its capabilities in self-checking and switching of skills autonomously. In addition, by integrating Feishu-platform communication, we enable natural-language guidance and human feedback, allowing inexperienced users to control the robotic dog through simple instructions.

Memory emerges as the core module in the large language model (LLM)-based agents for long-horizon complex tasks (e.g., multi-turn dialogue, game playing, scientific discovery), where memory can enable knowledge accumulation, iterative reasoning and self-evolution. A number of memory methods have been proposed in the literature. However, these methods have not been systematically and comprehensively compared under the same experimental settings. In this paper, we first summarize a unified framework that incorporates all the existing agent memory methods from a high-level perspective. We then extensively compare representative agent memory methods on two well-known benchmarks and examine the effectiveness of all methods, providing a thorough analysis of those methods. As a byproduct of our experimental analysis, we also design a new memory method by exploiting modules in the existing methods, which outperforms the state-of-the-art methods. Finally, based on these findings, we offer promising future research opportunities. We believe that a deeper understanding of the behavior of existing methods can provide valuable new insights for future research.

Automatic speech recognition (ASR) is a critical interface for human-AI interaction in gastrointestinal endoscopy, yet its reliability in real-world clinical settings is limited by domain-specific terminology and complex acoustic conditions. Here, we present EndoASR, a domain-adapted ASR system designed for real-time deployment in endoscopic workflows. We develop a two-stage adaptation strategy based on synthetic endoscopy reports, targeting domain-specific language modeling and noise robustness. In retrospective evaluation across six endoscopists, EndoASR substantially improves both transcription accuracy and clinical usability, reducing character error rate (CER) from 20.52% to 14.14% and increasing medical term accuracy (Med ACC) from 54.30% to 87.59%. In a prospective multi-center study spanning five independent endoscopy centers, EndoASR demonstrates consistent generalization under heterogeneous real-world conditions. Compared with the baseline Paraformer model, CER is reduced from 16.20% to 14.97%, while Med ACC is improved from 61.63% to 84.16%, confirming its robustness in practical deployment scenarios. Notably, EndoASR achieves a real-time factor (RTF) of 0.005, significantly faster than Whisper-large-v3 (RTF 0.055), while maintaining a compact model size of 220M parameters, enabling efficient edge deployment. Furthermore, integration with large language models demonstrates that improved ASR quality directly enhances downstream structured information extraction and clinician-AI interaction. These results demonstrate that domain-adapted ASR can serve as a reliable interface for human-AI teaming in gastrointestinal endoscopy, with consistent performance validated across multi-center real-world clinical settings.

Supervised Fine-Tuning (SFT) on long Chain-of-Thought (CoT) trajectories has become a pivotal phase in building large reasoning models. However, how CoT trajectories from different sources influence the generalization performance of models remains an open question. In this paper, we conduct a comparative study using two sources of verified CoT trajectories generated by two competing models, \texttt{DeepSeek-R1-0528} and \texttt{gpt-oss-120b}, with their problem sets controlled to be identical. Despite their comparable performance, we uncover a striking paradox: lower training loss does not translate to better generalization. SFT on \texttt{DeepSeek-R1-0528} data achieves remarkably lower training loss, yet exhibits significantly worse generalization performance on reasoning benchmarks compared to those trained on \texttt{gpt-oss-120b}. To understand this paradox, we perform a multi-faceted analysis probing token-level SFT loss and step-level reasoning behaviors. Our analysis reveals a difference in reasoning patterns. \texttt{gpt-oss-120b} exhibits highly convergent and deductive trajectories, whereas \texttt{DeepSeek-R1-0528} favors a divergent and branch-heavy exploration pattern. Consequently, models trained with \texttt{DeepSeek-R1} data inherit inefficient exploration behaviors, often getting trapped in redundant exploratory branches that hinder them from reaching correct solutions. Building upon this insight, we propose a simple yet effective remedy of filtering out frequently branching trajectories to improve the generalization of SFT. Experiments show that training on selected \texttt{DeepSeek-R1-0528} subsets surprisingly improves reasoning performance by up to 5.1% on AIME25, 5.5% on BeyondAIME, and on average 3.6% on five benchmarks.

Minor Component Adaptation (MiCA) is a novel parameter-efficient fine-tuning method for large language models that focuses on adapting underutilized subspaces of model representations. Unlike conventional methods such as Low-Rank Adaptation (LoRA), which target dominant subspaces, MiCA leverages Singular Value Decomposition to identify subspaces related to minor singular vectors associated with the least significant singular values and constrains the update of parameters during fine-tuning to those directions. This strategy leads to up to 5.9x improvement in knowledge acquisition under optimized training hyperparameters and a minimal parameter footprint of 6-60% compared to LoRA. These results suggest that constraining adaptation to minor singular directions provides a more efficient and stable mechanism for integrating new knowledge into pre-trained language models.

The rapid proliferation of Artificial Intelligence-Generated Content (AIGC) is fundamentally restructuring online content ecologies, necessitating a rigorous examination of its behavioral and distributional implications. Leveraging a comprehensive longitudinal dataset comprising tens of millions of users from a leading Chinese video-sharing platform, this study elucidated the distinct creation and consumption behaviors characterizing AIGC versus Human-Generated Content (HGC). We identified a prevalent scale-over-preference dynamic, wherein AIGC creators achieve aggregate engagement comparable to HGC creators through high-volume production, despite a marked consumer preference for HGC. Deeper analysis uncovered the ability of the algorithmic content distribution mechanism in moderating these competing interests regarding AIGC. These findings advocated for the implementation of AIGC-sensitive distribution algorithms and precise governance frameworks to ensure the long-term health of the online content platforms.

Anthropic proposes the concept of skills for LLM agents to tackle multi-step professional tasks that simple tool invocations cannot address. A tool is a single, self-contained function, whereas a skill is a structured bundle of interdependent multi-file artifacts. Currently, skill generation is not only label-intensive due to manual authoring, but also may suffer from human--machine cognitive misalignment, which can lead to degraded agent performance, as evidenced by evaluations on SkillsBench. Therefore, we aim to enable agents to autonomously generate skills. However, existing self-evolving methods designed for tools cannot be directly applied to skills due to their increased complexity. To address these issues, we propose EvoSkills, a self-evolving skills framework that enables agents to autonomously construct complex, multi-file skill packages. Specifically, EvoSkills couples a Skill Generator that iteratively refines skills with a Surrogate Verifier that co-evolves to provide informative and actionable feedback without access to ground-truth test content. On SkillsBench, EvoSkills achieves the highest pass rate among five baselines on both Claude Code and Codex, and also exhibits strong generalization capabilities to six additional LLMs.

Standard scaled dot-product attention computes scores from static, independent projections of the input. We show that evolving queries and keys \emph{jointly} through shared learned dynamics before scoring - which we call \textbf{coupled QK dynamics} - improves language modeling perplexity and training stability. On WikiText-103 at 60M parameters, coupled dynamics achieves 22.55--22.62 perplexity vs.\ 24.22 for standard attention ($-$6.6--6.9\%), with only 0.11\% additional parameters (shared across both instantiations). A structural ablation isolates coupling as the active ingredient: a symplectic (Hamiltonian) and a non-symplectic (Euler) integrator perform identically when both couple Q and K, while an uncoupled MLP baseline of matched capacity reaches only 23.81 with 8$\times$ higher seed variance. The integration step count (1--7) is similarly irrelevant - a single coupled step suffices. A compute-matched comparison reveals that coupling is a \emph{sample-efficiency} mechanism: standard attention trained for 2.4$\times$ longer (matching wall-clock) reaches the same perplexity, but requires 2.4$\times$ more tokens. The advantage scales to 150M ($-$6.7\%) but narrows at 350M ($-$1.0\%), where Differential Attention (18.93) overtakes coupled dynamics (19.35). The benefit is corpus-dependent: coupling helps on domain-coherent text (WikiText-103 $-$6.6\%, PubMed $-$4.5\%) but degrades on heterogeneous web text ($+$10.3\%) and shows no benefit on GLUE. We characterize when coupling helps and when it does not, providing practical guidelines.

Supervised fine-tuning (SFT) with token-level hard labels can amplify overconfident imitation of factually unsupported targets, causing hallucinations that propagate in multi-sentence generation. We study an augmented SFT setting in which training instances include coarse sentence-level factuality risk labels and inter-sentence dependency annotations, providing structured signals about where factual commitments are weakly supported. We propose \textbf{PRISM}, a differentiable risk-gated framework that modifies learning only at fact-critical positions. PRISM augments standard SFT with a lightweight, model-aware probability reallocation objective that penalizes high-confidence predictions on risky target tokens, with its scope controlled by span-level risk weights and model-aware gating. Experiments on hallucination-sensitive factual benchmarks and general evaluations show that PRISM improves factual aggregates across backbones while maintaining a competitive overall capability profile. Ablations further show that the auxiliary signal is most effective when used conservatively, and that knowledge masking and model-aware reallocation play complementary roles in balancing factual correction and capability preservation.

Large foundation models enable powerful reasoning for autonomous systems, but mapping semantic intent to reliable real-time control remains challenging. Existing approaches either (i) let Large Language Models (LLMs) generate trajectories directly - brittle, hard to verify, and latency-prone - or (ii) adjust Model Predictive Control (MPC) objectives online - mixing slow deliberation with fast control and blurring interfaces. We propose Agentic Fast-Slow Planning, a hierarchical framework that decouples perception, reasoning, planning, and control across natural timescales. The framework contains two bridges. Perception2Decision compresses scenes into ego-centric topologies using an on-vehicle Vision-Language Model (VLM) detector, then maps them to symbolic driving directives in the cloud with an LLM decision maker - reducing bandwidth and delay while preserving interpretability. Decision2Trajectory converts directives into executable paths: Semantic-Guided A* embeds language-derived soft costs into classical search to bias solutions toward feasible trajectories, while an Agentic Refinement Module adapts planner hyperparameters using feedback and memory. Finally, MPC tracks the trajectories in real time, with optional cloud-guided references for difficult cases. Experiments in CARLA show that Agentic Fast-Slow Planning improves robustness under perturbations, reducing lateral deviation by up to 45% and completion time by over 12% compared to pure MPC and an A*-guided MPC baseline. Code is available at https://github.com/cjychenjiayi/icra2026_AFSP.

Promptly porting patches from a source codebase to its variants (e.g., forks and branches) is essential for mitigating propagated defects and vulnerabilities. Recent studies have explored automated patch porting to reduce manual effort and delay, but existing approaches mainly handle inconsistencies visible in a patch's local context and struggle with those requiring global mapping knowledge between codebases. We refer to such non-local inconsistencies as implicit inconsistencies. Implicit inconsistencies pose greater challenges for developers to resolve due to their non-local nature. To address them, we propose MIP, which enables collaboration among an LLM, a compiler, and code analysis utilities. MIP adopts different strategies for different cases: when source identifiers exist in the target codebase, it leverages compiler diagnostics; otherwise, it retrieves matched code segment pairs from the two codebases as mapping knowledge for mitigation. Experiments on two representative scenarios, cross-fork and cross-branch patch porting, show that MIP successfully resolves more than twice as many patches as the best-performing baseline in both settings. A user study with our industry partner further demonstrates its practical effectiveness.

GUI Process Automation (GPA) is a lightweight but general vision-based Robotic Process Automation (RPA), which enables fast and stable process replay with only a single demo. Addressing the fragility of traditional RPA and the non-deterministic risks of current vision language model-based GUI agents, GPA introduces three core benefits: (1) Robustness via Sequential Monte Carlo-based localization to handle rescaling and detection uncertainty; (2) Deterministic and Reliability safeguarded by readiness calibration; and (3) Privacy through fast, fully local execution. This approach delivers the adaptability, robustness, and security required for enterprise workflows. It can also be used as an MCP/CLI tool by other agents with coding capabilities so that the agent only reasons and orchestrates while GPA handles the GUI execution. We conducted a pilot experiment to compare GPA with Gemini 3 Pro (with CUA tools) and found that GPA achieves higher success rate with 10 times faster execution speed in finishing long-horizon GUI tasks.

Model merging aims to integrate multiple expert models into a single model that inherits their complementary strengths without incurring the inference-time cost of ensembling. Recent progress has shown that merging can be highly effective when all source models are \emph{homogeneous}, i.e., derived from the same pretrained backbone and therefore share aligned parameter coordinates or compatible task vectors. Yet this assumption is increasingly unrealistic in open model ecosystems, where useful experts are often built on different families such as Llama, Qwen, and Mistral. In such \emph{heterogeneous} settings, direct weight-space fusion becomes ill-posed due to architectural mismatch, latent basis misalignment, and amplified cross-source conflict. We address this problem with \texttt{HeteroFusion} for heterogeneous language model fusion, which consists of two key components: topology-based alignment that transfers knowledge across heterogeneous backbones by matching functional module structures instead of raw tensor coordinates, and conflict-aware denoising that suppresses incompatible or noisy transfer signals during fusion. We further provide analytical justification showing that preserving the target adapter basis while predicting structured updates leads to a stable and well-conditioned transfer process. Across heterogeneous transfer, multi-source fusion, noisy-source robustness, and cross-family generalization settings, \texttt{HeteroFusion} consistently outperforms strong merging, fusion, and ensemble baselines.

Emotional Support Conversation (ESC) aims to alleviate individual emotional distress by generating empathetic responses. However, existing methods face challenges in effectively supporting deep contextual understanding. To address this issue, we propose PRCCF, a Persona-guided Retrieval and Causality-aware Cognitive Filtering framework. Specifically, the framework incorporates a persona-guided retrieval mechanism that jointly models semantic compatibility and persona alignment to enhance response generation. Furthermore, it employs a causality-aware cognitive filtering module to prioritize causally relevant external knowledge, thereby improving contextual cognitive understanding for emotional reasoning. Extensive experiments on the ESConv dataset demonstrate that PRCCF outperforms state-of-the-art baselines on both automatic metrics and human evaluations. Our code is publicly available at: https://github.com/YancyLyx/PRCCF.

While long-term memory is essential for intelligent agents to maintain consistent historical awareness, the accumulation of extensive interaction data often leads to performance bottlenecks. Naive storage expansion increases retrieval noise and computational latency, overwhelming the reasoning capacity of models deployed on constrained personal devices. To address this, we propose Hierarchical Memory Orchestration (HMO), a framework that organizes interaction history into a three-tiered directory driven by user-centric contextual relevance. Our system maintains a compact primary cache, coupling recent and pivotal memories with an evolving user profile to ensure agent reasoning remains aligned with individual behavioral traits. This primary cache is complemented by a high-priority secondary layer, both of which are managed within a global archive of the full interaction history. Crucially, the user persona dictates memory redistribution across this hierarchy, promoting records mapped to long-term patterns toward more active tiers while relegating less relevant information. This targeted orchestration surfaces historical knowledge precisely when needed while maintaining a lean and efficient active search space. Evaluations on multiple benchmarks achieve state-of-the-art performance. Real-world deployments in ecosystems like OpenClaw demonstrate that HMO significantly enhances agent fluidity and personalization.

Embodied perception systems face severe challenges of dynamic environment distribution drift when they continuously interact in open physical spaces. However, the existing domain incremental awareness methods often rely on the domain id obtained in advance during the testing phase, which limits their practicability in unknown interaction scenarios. At the same time, the model often overfits to the context-specific perceptual noise, which leads to insufficient generalization ability and catastrophic forgetting. To address these limitations, we propose a domain-id and exemplar-free incremental learning framework for embodied multimedia systems, which aims to achieve robust continuous environment adaptation. This method designs a disentangled representation mechanism to remove non-essential environmental style interference, and guide the model to focus on extracting semantic intrinsic features shared across scenes, thereby eliminating perceptual uncertainty and improving generalization. We further use the weight fusion strategy to dynamically integrate the old and new environment knowledge in the parameter space, so as to ensure that the model adapts to the new distribution without storing historical data and maximally retains the discrimination ability of the old environment. Extensive experiments on multiple standard benchmark datasets show that the proposed method significantly reduces catastrophic forgetting in a completely exemplar-free and domain-id free setting, and its accuracy is better than the existing state-of-the-art methods.

Multi-modal fusion is of great significance in neuroscience which integrates information from different modalities and can achieve better performance than uni-modal methods in downstream tasks. Current multi-modal fusion methods in brain networks, which mainly focus on structural connectivity (SC) and functional connectivity (FC) modalities, are static in nature. They feed different samples into the same model with identical computation, ignoring inherent difference between input samples. This lack of sample adaptation hinders model's further performance. To this end, we innovatively propose a multi-stage dynamic fusion strategy (M3D-BFS) for sample-adaptive multi-modal brain network analysis. Unlike other static fusion methods, we design different mixture-of-experts (MoEs) for uni- and multi-modal representations where modules can adaptively change as input sample changes during inference. To alleviate issue of MoE where training of experts may be collapsed, we divide our method into 3 stages. We first train uni-modal encoders respectively, then pretrain single experts of MoEs before finally finetuning the whole model. A multi-modal disentanglement loss is designed to enhance the final representations. To the best of our knowledge, this is the first work for dynamic fusion for multi-modal brain network analysis. Extensive experiments on different real-world datasets demonstrates the superiority of M3D-BFS.

LLM-based agents show strong potential for long-horizon reasoning, yet their context size is limited by deployment factors (e.g., memory, latency, and cost), yielding a constrained context budget. As interaction histories grow, this induces a trade-off between retaining past information and staying within the context limit. To address this challenge, we propose Budget-Aware Context Management (BACM), which formulates context management as a sequential decision problem with a context budget constraint. It enables agents to assess the available budget before incorporating new observations and decide when and how much of the interaction history to compress. We further develop BACM-RL, an end-to-end curriculum-based reinforcement learning approach that learns compression strategies under varying context budgets. Experiments on compositional multi-objective QA and long-horizon web browsing benchmarks show that BACM-RL consistently outperforms prior methods across model scales and task complexities, achieving over $1.6\times$ gains over strong baselines in high-complexity settings, while maintaining strong advantages as budgets shrink, where most methods exhibit a downward performance trend.

Clinical AI systems routinely train on health data structurally distorted by documentation workflows, billing incentives, and terminology fragmentation. Prior work has characterised the mechanisms of this distortion: the three-forces model of documentary enactment, the reification feedback loop through which AI may amplify coding artefacts, and terminology governance failures that allow semantic drift to accumulate. Yet translating these insights into implementable software architecture remains an open problem. This paper proposes seven ontology-aware design patterns in Gang-of-Four pattern language for building clinical AI pipelines resilient to ontological distortion. The patterns address data ingestion validation (Ontological Checkpoint), low-frequency signal preservation (Dormancy-Aware Pipeline), continuous drift monitoring (Drift Sentinel), parallel representation maintenance (Dual-Ontology Layer), feedback loop interruption (Reification Circuit Breaker), terminology evolution management (Terminology Version Gate), and pluggable regulatory compliance (Regulatory Compliance Adapter). Each pattern is specified with Problem, Forces, Solution, Consequences, Known Uses, and Related Patterns. We illustrate their composition in a reference architecture for a primary care AI system and provide a walkthrough tracing all seven patterns through a diabetes risk prediction scenario. This paper does not report empirical validation; it offers a design vocabulary grounded in theoretical analysis, subject to future evaluation in production systems. Three patterns have partial precedent in existing systems; the remaining four have not been formally described. Limitations include the absence of runtime benchmarks and restriction to the German and EU regulatory context.

Large language model (LLM)-based evolution is a promising approach for open-ended discovery, where progress requires sustained search and knowledge accumulation. Existing methods still rely heavily on fixed heuristics and hard-coded exploration rules, which limit the autonomy of LLM agents. We present CORAL, the first framework for autonomous multi-agent evolution on open-ended problems. CORAL replaces rigid control with long-running agents that explore, reflect, and collaborate through shared persistent memory, asynchronous multi-agent execution, and heartbeat-based interventions. It also provides practical safeguards, including isolated workspaces, evaluator separation, resource management, and agent session and health management. Evaluated on diverse mathematical, algorithmic, and systems optimization tasks, CORAL sets new state-of-the-art results on 10 tasks, achieving 3-10 times higher improvement rates with far fewer evaluations than fixed evolutionary search baselines across tasks. On Anthropic's kernel engineering task, four co-evolving agents improve the best known score from 1363 to 1103 cycles. Mechanistic analyses further show how these gains arise from knowledge reuse and multi-agent exploration and communication. Together, these results suggest that greater agent autonomy and multi-agent evolution can substantially improve open-ended discovery. Code is available at https://github.com/Human-Agent-Society/CORAL.

Despite rapid progress in claim verification, we lack a systematic understanding of what reasoning these benchmarks actually exercise. We generate structured reasoning traces for 24K claim-verification examples across 9 datasets using GPT-4o-mini and find that direct evidence extraction dominates, while multi-sentence synthesis and numerical reasoning are severely under-represented. A dataset-level breakdown reveals stark biases: some datasets almost exclusively test lexical matching, while others require information synthesis in roughly half of cases. Using a compact 1B-parameter reasoning verifier, we further characterize five error types and show that error profiles vary dramatically by domain -- general-domain verification is dominated by lexical overlap bias, scientific verification by overcautiousness, and mathematical verification by arithmetic reasoning failures. Our findings suggest that high benchmark scores primarily reflect retrieval-plus-entailment ability. We outline recommendations for building more challenging evaluation suites that better test the reasoning capabilities verification systems need.

Recent advances in video generation have made AI-synthesized content increasingly difficult to distinguish from real footage. We propose a physics-based authentication signature that real cameras produce naturally, but that generative models cannot faithfully reproduce. Our approach exploits the Moiré effect: the interference fringes formed when a camera views a compact two-layer grating structure. We derive the Moiré motion invariant, showing that fringe phase and grating image displacement are linearly coupled by optical geometry, independent of viewing distance and grating structure. A verifier extracts both signals from video and tests their correlation. We validate the invariant on both real-captured and AI-generated videos from multiple state-of-the-art generators, and find that real and AI-generated videos produce significantly different correlation signatures, suggesting a robust means of differentiating them. Our work demonstrates that deterministic optical phenomena can serve as physically grounded, verifiable signatures against AI-generated video.

Electroencephalography (EEG) provides a non-invasive insight into the brain's cognitive and emotional dynamics. However, modeling how these states evolve in real time and quantifying the energy required for such transitions remains a major challenge. The Schrödinger Bridge Problem (SBP) offers a principled probabilistic framework to model the most efficient evolution between the brain states, interpreted as a measure of cognitive energy cost. While generative models such as GANs have been widely used to augment EEG data, it remains unclear whether synthetic EEG preserves the underlying dynamical structure required for transition-based analysis. In this work, we address this gap by using SBP-derived transport cost as a metric to evaluate whether GAN-generated EEG retains the distributional geometry necessary for energy-based modeling of cognitive state transitions. We compare transition energies derived from real and synthetic EEG collected during Stroop tasks and demonstrate strong agreement across group and participant-level analyses. These results indicate that synthetic EEG preserves the transition structure required for SBP-based modeling, enabling its use in data-efficient neuroadaptive systems. We further present a framework in which SBP-derived cognitive energy serves as a control signal for adaptive human-machine systems, supporting real-time adjustment of system behavior in response to user cognitive and affective state.

We present ThinknCheck, a 1B-parameter verifier for grounded claim verification that first produces a short, structured rationale and then a binary verdict. We construct LLMAggreFact-Think, a 24.1k reasoning-augmented training set derived from LLMAggreFact, and fine-tune a 4-bit Gemma3 model to follow this format. On LLMAggreFact, ThinknCheck attains 78.1 balanced accuracy (BAcc), surpassing MiniCheck-7B (77.4) with 7x fewer parameters; removing the reasoning step reduces BAcc to 57.5. On SciFact, ThinknCheck reaches 64.7 BAcc, a +14.7 absolute gain over MiniCheck-7B. By contrast, zero-shot chain-of-thought on the base Gemma3-1B harms accuracy relative to direct answers, and preference optimization with a simple format+accuracy reward underperforms supervised reasoning. To probe the latter, we introduce GSMClaims and a domain-specialized variant, ThinknCheck-Science, which improves across benchmarks, including 61.0\% accuracy on GSMClaims. Overall, explicit, supervised reasoning enables compact verifiers that are competitive while remaining resource-efficient and interpretable.

An assumption often made in supervised learning is that the training and testing sets have the same label distribution. However, in real-life scenarios, this assumption rarely holds. For example, medical diagnosis result distributions change over time and across locations; fraud detection models must adapt as patterns of fraudulent activity shift; the category distribution of social media posts changes based on trending topics and user demographics. In the task of label shift estimation, the goal is to estimate the changing label distribution $p_t(y)$ in the testing set, assuming the likelihood $p(x|y)$ does not change, implying no concept drift. In this paper, we propose a new approach for post-hoc label shift estimation, unlike previous methods that perform moment matching with confusion matrix estimated from a validation set or maximize the likelihood of the new data with an expectation-maximization algorithm. We aim to incrementally update the prior on each sample, adjusting each posterior for more accurate label shift estimation. The proposed method is based on intuitive assumptions on classifiers that are generally true for modern probabilistic classifiers. The proposed method relies on a weaker notion of calibration compared to other methods. As a post-hoc approach for label shift estimation, the proposed method is versatile and can be applied to any black-box probabilistic classifier. Experiments on CIFAR-10 and MNIST show that the proposed method consistently outperforms the current state-of-the-art maximum likelihood-based methods under different calibrations and varying intensities of label shift.

Smell's deep connection with food, memory, and social experience has long motivated researchers to bring olfaction into interactive systems. Yet most olfactory interfaces remain limited to fixed scent cartridges and pre-defined generation patterns, and the scarcity of large-scale olfactory datasets has further constrained AI-based approaches. We present AromaGen, an AI-powered wearable interface capable of real-time, general-purpose aroma generation from free-form text or visual inputs. AromaGen is powered by a multimodal LLM that leverages latent olfactory knowledge to map semantic inputs to structured mixtures of 12 carefully selected base odorants, released through a neck-worn dispenser. Users can iteratively refine generated aromas through natural language feedback via in-context learning. Through a controlled user study ($N = 26$), AromaGen matches human-composed mixtures in zero-shot generation and significantly surpasses them after iterative refinement, achieving a median similarity of 8/10 to real food aromas and reducing perceived artificiality to levels comparable to real food. AromaGen is a step towards real-world interactive aroma generation, opening new possibilities for communication, wellbeing, and immersive technologies.

Embedding LLM-driven agents into environmental FAIR data management is compelling - they can externalize operational knowledge and scale curation across heterogeneous data and evolving conventions. However, replacing deterministic components with probabilistic workflows changes the failure mode: LLM pipelines may generate plausible but incorrect outputs that pass superficial checks and propagate into irreversible actions such as DOI minting and public release. We introduce EnviSmart, a production data management system deployed on campus-wide storage infrastructure for environmental research. EnviSmart treats reliability as an architectural property through two mechanisms: a three-track knowledge architecture that externalizes behaviors (governance constraints), domain knowledge (retrievable context), and skills (tool-using procedures) as persistent, interlocking artifacts; and a role-separated multi-agent design where deterministic validators and audited handoffs restore fail-stop semantics at trust boundaries before irreversible steps. We compare two production deployments. The University's GIS Center Ecological Archive (849 curated datasets) serves as a single-agent baseline. SF2Bench, a compound flooding benchmark comprising 2,452 monitoring stations and 8,557 published files spanning 39 years, validates the multi-agent workflow. The multi-agent approach improved both efficiency - completed by a single operator in two days with repeated artifact reuse across deployments - and reliability: audited handoffs detected and blocked a coordinate transformation error affecting all 2,452 stations before publication. A representative incident (ISS-004) demonstrated boundary-based containment with 10-minute detection latency, zero user exposure, and 80-minute resolution. This paper has been accepted at PEARC 2026.

Large language models demonstrate strong performance on mathematical reasoning benchmarks, yet remain surprisingly fragile to meaning-preserving surface perturbations. We systematically evaluate three open-weight LLMs, Mistral-7B, Llama-3-8B, and Qwen2.5-7B, on 677 GSM8K problems paired with semantically equivalent variants generated through name substitution and number format paraphrasing. All three models exhibit substantial answer-flip rates (28.8%-45.1%), with number paraphrasing consistently more disruptive than name swaps. To trace the mechanistic basis of these failures, we introduce the Mechanistic Perturbation Diagnostics (MPD) framework, combining logit lens analysis, activation patching, component ablation, and the Cascading Amplification Index (CAI) into a unified diagnostic pipeline. CAI, a novel metric quantifying layer-wise divergence amplification, outperforms first divergence layer as a failure predictor for two of three architectures (AUC up to 0.679). Logit lens reveals that flipped samples diverge from correct predictions at significantly earlier layers than stable samples. Activation patching reveals a stark architectural divide in failure localizability: Llama-3 failures are recoverable by patching at specific layers (43/60 samples), while Mistral and Qwen failures are broadly distributed (3/60 and 0/60). Based on these diagnostic signals, we propose a mechanistic failure taxonomy (localized, distributed, and entangled) and validate it through targeted repair experiments: steering vectors and layer fine-tuning recover 12.2% of localized failures (Llama-3) but only 7.2% of entangled (Qwen) and 5.2% of distributed (Mistral) failures.

Existing benchmarks for LLM-based vulnerability detection compress model performance into a single metric, which fails to reflect the distinct priorities of different stakeholders. For example, a CISO may emphasize high recall of critical vulnerabilities, an engineering leader may prioritize minimizing false positives, and an AI officer may balance capability against cost. To address this limitation, we introduce SecLens-R, a multi-stakeholder evaluation framework structured around 35 shared dimensions grouped into 7 measurement categories. The framework defines five role-specific weighting profiles: CISO, Chief AI Officer, Security Researcher, Head of Engineering, and AI-as-Actor. Each profile selects 12 to 16 dimensions with weights summing to 80, yielding a composite Decision Score between 0 and 100. We apply SecLens-R to evaluate 12 frontier models on a dataset of 406 tasks derived from 93 open-source projects, covering 10 programming languages and 8 OWASP-aligned vulnerability categories. Evaluations are conducted across two settings: Code-in-Prompt (CIP) and Tool-Use (TU). Results show substantial variation across stakeholder perspectives, with Decision Scores differing by as much as 31 points for the same model. For instance, Qwen3-Coder achieves an A (76.3) under the Head of Engineering profile but a D (45.2) under the CISO profile, while GPT-5.4 shows a similar disparity. These findings demonstrate that vulnerability detection is inherently a multi-objective problem and that stakeholder-aware evaluation provides insights that single aggregated metrics obscure.

Real-world reasoning often requires combining information across modalities, connecting textual context with visual cues in a multi-hop process. Yet, most multimodal benchmarks fail to capture this ability: they typically rely on single images or set of images, where answers can be inferred from a single modality alone. This limitation is mirrored in the training data, where interleaved image-text content rarely enforces complementary, multi-hop reasoning. As a result, Vision-Language Models (VLMs) frequently hallucinate and produce reasoning traces poorly grounded in visual evidence. To address this gap, we introduce CRIT, a new dataset and benchmark built with a graph-based automatic pipeline for generating complex cross-modal reasoning tasks. CRIT consists of diverse domains ranging from natural images, videos, and text-rich sources, and includes a manually verified test set for reliable evaluation. Experiments on this benchmark reveal that even state-of-the-art models struggle on such reasoning tasks. Models trained on CRIT show significant gains in cross-modal multi-hop reasoning, including strong improvements on SPIQA and other standard multimodal benchmarks.

Despite emerging use in Indonesian classrooms, there is limited large-scale, teacher-centred evidence on how AI is used in practice and what support teachers need, hindering the development of context-appropriate AI systems and policies. To address this gap, we conduct a nationwide survey of 349 K-12 teachers across elementary, junior high, and senior high schools. We find increasing use of AI for pedagogy, content development, and teaching media, although adoption remains uneven. Elementary teachers report more consistent use, while senior high teachers engage less; mid-career teachers assign higher importance to AI, and teachers in Eastern Indonesia perceive greater value. Across levels, teachers primarily use AI to reduce instructional preparation workload (e.g., assessment, lesson planning, and material development). However, generic outputs, infrastructure constraints, and limited contextual alignment continue to hinder effective classroom integration.

Translating security intent into deployable network enforcement rules and maintaining their effectiveness despite evolving cyber threats remains a largely manual process in most Security Operations Centers (SOCs). In large and heterogeneous networks, this challenge is complicated by topology-dependent reachability constraints and device-specific security control capabilities, making the process slow, error-prone, and a recurring source of misconfigurations. This paper presents RefinementEngine, an engine that automates the refinement of high-level security intents into low-level, deployment-ready configurations. Given a network topology, devices, and available security controls, along with high-level intents and Cyber Threat Intelligence (CTI) reports, RefinementEngine automatically generates settings that implement the desired intent, counter reported threats, and can be directly deployed on target security controls. The proposed approach is validated through real-world use cases on packet and web filtering policies derived from actual CTI reports, demonstrating both correctness, practical applicability, and adaptability to new data.

Diffusion language models (DLMs) expose their denoising trajectories, offering a natural handle for inference-time control; accordingly, an ideal hallucination mitigation framework should intervene during generation using this model-native signal rather than relying on an externally trained hallucination classifier. Toward this, we formulate commitment uncertainty localization: given a denoising trajectory, identify token positions whose cross-chain entropy exceeds an unsupervised threshold before factually unreliable commitments propagate into self-consistent but incorrect outputs. We introduce a suite of trajectory-level assessments, including a cross-chain divergence-at-hallucination (CDH) metric, for principled comparison of localization methods. We also introduce OSCAR, a training-free inference-time framework operationalizing this formulation. OSCAR runs N parallel denoising chains with randomized reveal orders, computes cross-chain Shannon entropy to detect high-uncertainty positions, and then performs targeted remasking conditioned on retrieved evidence. Ablations confirm that localization and correction contribute complementary gains, robust across N in {4, 8, 16}. On TriviaQA, HotpotQA, RAGTruth, and CommonsenseQA using LLaDA-8B and Dream-7B, OSCAR enhances generation quality by significantly reducing hallucinated content and improving factual accuracy through uncertainty-guided remasking, which also facilitates more effective integration of retrieved evidence. Its native entropy-based uncertainty signal surpasses that of specialized trained detectors, highlighting an inherent capacity of diffusion language models to identify factual uncertainty that is not present in the sequential token commitment structure of autoregressive models. We are releasing the codebase1 to support future research on localization and uncertainty-aware generation in DLMs.

Diffusion language models (DLMs) enable parallel, non-autoregressive text generation, yet existing DLM mixture-of-experts (MoE) models inherit token-choice (TC) routing from autoregressive systems, leading to load imbalance and rigid computation allocation. We show that expert-choice (EC) routing is a better fit for DLMs: it provides deterministic load balancing by design, yielding higher throughput and faster convergence than TC. Building on the property that EC capacity is externally controllable, we introduce timestep-dependent expert capacity, which varies expert allocation according to the denoising step. We find that allocating more capacity to low-mask-ratio steps consistently achieves the best performance under matched FLOPs, and provide a mechanistic explanation: tokens in low-mask-ratio contexts exhibit an order-of-magnitude higher learning efficiency, so concentrating compute on these steps yields the largest marginal return. Finally, we show that existing pretrained TC DLMs can be retrofitted to EC by replacing only the router, achieving faster convergence and improved accuracy across diverse downstream tasks. Together, these results establish EC routing as a superior paradigm for DLM MoE models and demonstrate that computation in DLMs can be treated as an adaptive policy rather than a fixed architectural constant. Code is available at https://github.com/zhangshuibai/EC-DLM.

Large language model (LLM) inference increasingly depends on multi-GPU execution, yet existing inference parallelization strategies require layer-wise inter-rank synchronization, making end-to-end performance sensitive to workload imbalance. We present DWDP (Distributed Weight Data Parallelism), an inference parallelization strategy that preserves data-parallel execution while offloading MoE weights across peer GPUs and fetching missing experts on demand. By removing collective inter-rank synchronization, DWDP allows each GPU to progress independently. We further address the practical overheads of this design with two optimizations for split-weight management and asynchronous remote-weight prefetch. Implemented in TensorRT-LLM and evaluated with DeepSeek-R1 on GB200 NVL72, DWDP improves end-to-end output TPS/GPU by 8.8% at comparable TPS/user in the 20-100 TPS/user serving range under 8K input sequence length and 1K output sequence length.

Morphological traits are physical characteristics of biological organisms that provide vital clues on how organisms interact with their environment. Yet extracting these traits remains a slow, expert-driven process, limiting their use in large-scale ecological studies. A major bottleneck is the absence of high-quality datasets linking biological images to trait-level annotations. In this work, we demonstrate that sparse autoencoders trained on foundation-model features yield monosemantic, spatially grounded neurons that consistently activate on meaningful morphological parts. Leveraging this property, we introduce a trait annotation pipeline that localizes salient regions and uses vision-language prompting to generate interpretable trait descriptions. Using this approach, we construct Bioscan-Traits, a dataset of 80K trait annotations spanning 19K insect images from BIOSCAN-5M. Human evaluation confirms the biological plausibility of the generated morphological descriptions. We assess design sensitivity through a comprehensive ablation study, systematically varying key design choices and measuring their impact on the quality of the resulting trait descriptions. By annotating traits with a modular pipeline rather than prohibitively expensive manual efforts, we offer a scalable way to inject biologically meaningful supervision into foundation models, enable large-scale morphological analyses, and bridge the gap between ecological relevance and machine-learning practicality.

Vision-language-action (VLA) models have shown strong performance in robotic manipulation, yet their robustness to physically realizable adversarial attacks remains underexplored. Existing studies reveal vulnerabilities through language perturbations and 2D visual attacks, but these attack surfaces are either less representative of real deployment or limited in physical realism. In contrast, adversarial 3D textures pose a more physically plausible and damaging threat, as they are naturally attached to manipulated objects and are easier to deploy in physical environments. Bringing adversarial 3D textures to VLA systems is nevertheless nontrivial. A central obstacle is that standard 3D simulators do not provide a differentiable optimization path from the VLA objective function back to object appearance, making it difficult to optimize through an end-to-end manner. To address this, we introduce Foreground-Background Decoupling (FBD), which enables differentiable texture optimization through dual-renderer alignment while preserving the original simulation environment. To further ensure that the attack remains effective across long-horizon and diverse viewpoints in the physical world, we propose Trajectory-Aware Adversarial Optimization (TAAO), which prioritizes behaviorally critical frames and stabilizes optimization with a vertex-based parameterization. Built on these designs, we present Tex3D, the first framework for end-to-end optimization of 3D adversarial textures directly within the VLA simulation environment. Experiments in both simulation and real-robot settings show that Tex3D significantly degrades VLA performance across multiple manipulation tasks, achieving task failure rates of up to 96.7\%. Our empirical results expose critical vulnerabilities of VLA systems to physically grounded 3D adversarial attacks and highlight the need for robustness-aware training.

This paper presents a systematic, cost-aware evaluation of large language models (LLMs) for receipt-item categorisation within a production-oriented classification framework. We compare four instruction-tuned models available through AWS Bedrock: Claude 3.7 Sonnet, Claude 4 Sonnet, Mixtral 8x7B Instruct, and Mistral 7B Instruct. The aim of the study was (1) to assess performance across accuracy, response stability, and token-level cost, and (2) to investigate what prompting methods, zero-shot or few-shot, are especially appropriate both in terms of accuracy and in terms of incurred costs. Results of our experiments demonstrated that Claude 3.7 Sonnet achieves the most favourable balance between classification accuracy and cost efficiency.

In reinforcement learning (RL), temporal difference (TD) errors are widely adopted for optimizing value and policy functions. However, since the TD error is defined by a bootstrap method, its computation tends to be noisy and destabilize learning. Heuristics to improve the accuracy of TD errors, such as target networks and ensemble models, have been introduced so far. While these are essential approaches for the current deep RL algorithms, they cause side effects like increased computational cost and reduced learning efficiency. Therefore, this paper revisits the TD learning algorithm based on control as inference, deriving a novel algorithm capable of robust learning against noisy TD errors. First, the distribution model of optimality, a binary random variable, is represented by a sigmoid function. Alongside forward and reverse Kullback-Leibler divergences, this new model derives a robust learning rule: when the sigmoid function saturates with a large TD error probably due to noise, the gradient vanishes, implicitly excluding it from learning. Furthermore, the two divergences exhibit distinct gradient-vanishing characteristics. Building on these analyses, the optimality is decomposed into multiple levels to achieve pseudo-quantization of TD errors, aiming for further noise reduction. Additionally, a Jensen-Shannon divergence-based approach is approximately derived to inherit the characteristics of both divergences. These benefits are verified through RL benchmarks, demonstrating stable learning even when heuristics are insufficient or rewards contain noise.

Volumetric CT imaging is essential for clinical diagnosis, yet annotating 3D volumes is expensive and time-consuming, motivating self-supervised learning (SSL) from unlabeled data. However, applying SSL to 3D CT remains challenging due to the high memory cost of full-volume transformers and the anisotropic spatial structure of CT data, which is not well captured by conventional masking strategies. We propose NEMESIS, a masked autoencoder (MAE) framework that operates on local 128x128x128 superpatches, enabling memory-efficient training while preserving anatomical detail. NEMESIS introduces three key components: (i) noise-enhanced reconstruction as a pretext task, (ii) Masked Anatomical Transformer Blocks (MATB) that perform dual-masking through parallel plane-wise and axis-wise token removal, and (iii) NEMESIS Tokens (NT) for cross-scale context aggregation. On the BTCV multi-organ classification benchmark, NEMESIS with a frozen backbone and a linear classifier achieves a mean AUROC of 0.9633, surpassing fully fine-tuned SuPreM (0.9493) and VoCo (0.9387). Under a low-label regime with only 10% of available annotations, it retains an AUROC of 0.9075, demonstrating strong label efficiency. Furthermore, the superpatch-based design reduces computational cost to 31.0 GFLOPs per forward pass, compared to 985.8 GFLOPs for the full-volume baseline, providing a scalable and robust foundation for 3D medical imaging.

The use of knowledge graphs for grounding agents in real-world Q&A applications has become increasingly common. Answering complex queries often requires multi-hop reasoning and the ability to navigate vast relational structures. Standard approaches rely on prompting techniques that steer large language models to reason over raw graph context, or retrieval-augmented generation pipelines where relevant subgraphs are injected into the context. These, however, face severe limitations with enterprise-scale KGs that cannot fit in even the largest context windows available today. We present GraphWalk, a problem-agnostic, training-free, tool-based framework that allows off-the-shelf LLMs to reason through sequential graph navigation, dramatically increasing performance across different tasks. Unlike task-specific agent frameworks that encode domain knowledge into specialized tools, GraphWalk equips the LLM with a minimal set of orthogonal graph operations sufficient to traverse any graph structure. We evaluate whether models equipped with GraphWalk can compose these operations into correct multi-step reasoning chains, where each tool call represents a verifiable step creating a transparent execution trace. We first demonstrate our approach on maze traversal, a problem non-reasoning models are completely unable to solve, then present results on graphs resembling real-world enterprise knowledge graphs. To isolate structural reasoning from world knowledge, we evaluate on entirely synthetic graphs with random, non-semantic labels. Our benchmark spans 12 query templates from basic retrieval to compound first-order logic queries. Results show that tool-based traversal yields substantial and consistent gains over in-context baselines across all model families tested, with gains becoming more pronounced as scale increases, precisely where in-context approaches fail catastrophically.

The deployment of Large Language Models is constrained by the memory and bandwidth demands of static weights and dynamic Key-Value cache. SVD-based compression provides a hardware-friendly solution to reduce these costs. However, existing methods suffer from two key limitations: some are suboptimal in reconstruction error, while others are theoretically optimal but practically inefficient. In this paper, we propose Swift-SVD, an activation-aware, closed-form compression framework that simultaneously guarantees theoretical optimum, practical efficiency and numerical stability. Swift-SVD incrementally aggregates covariance of output activations given a batch of inputs and performs a single eigenvalue decomposition after aggregation, enabling training-free, fast, and optimal layer-wise low-rank approximation. We employ effective rank to analyze local layer-wise compressibility and design a dynamic rank allocation strategy that jointly accounts for local reconstruction loss and end-to-end layer importance. Extensive experiments across six LLMs and eight datasets demonstrate that Swift-SVD outperforms state-of-the-art baselines, achieving optimal compression accuracy while delivering 3-70X speedups in end-to-end compression time. Our code will be released upon acceptance.

Multi-agent systems (MAS) tackle complex tasks by distributing expertise, though this often comes at the cost of heavy coordination overhead, context fragmentation, and brittle phase ordering. Distilling a MAS into a single-agent skill can bypass these costs, but this conversion lacks a principled answer for when and what to distill. Instead, the empirical outcome is surprisingly inconsistent: skill lift ranges from a 28% improvement to a 2% degradation across metrics of the exact same task. In this work, we reveal that skill utility is governed not by the task, but by the evaluation metric. We introduce Metric Freedom ($F$), the first a priori predictor of skill utility. $F$ measures the topological rigidity of a metric's scoring landscape by quantifying how output diversity couples with score variance via a Mantel test. Guided by $F$, we propose a two-stage adaptive distillation framework. Stage 1 acts as a selective extraction mechanism, extracting tools and knowledge while discarding restrictive structures on "free" metrics to preserve exploration. Stage 2 targets computationally intensive iterative refinement exclusively toward "rigid" metrics ($F \lesssim 0.6$) to eliminate trajectory-local overfitting. Evaluating across 4 tasks, 11 datasets, and 6 metrics, $F$ strongly predicts skill utility ($ρ= -0.62$, $p < 0.05$). Strikingly, identical agent trajectories yield diametrically opposite skill lifts under rigid versus free metrics, demonstrating that skill utility is fundamentally a metric-level property. Driven by this signal, our adaptive agent matches or exceeds the original MAS while reducing cost up to 8$\times$ and latency by up to 15$\times$.

Large-scale distributed training has been a research hot spot in machine learning systems for industry and academia in recent years. However, conducting experiments without physical machines and corresponding resources is difficult. One solution is to leverage distributed training simulators, but current ones like ASTRA-sim do not support importing real-world developed models, which poses challenges for ML researchers seeking to use them. Based on this challenge, we developed ModTrans, a translator supporting format translation from any real-world model to the ASTRA-sim simulator's input, removing the barrier between machine learning experts and machine learning system researchers. The experiment results show that ModTrans's cost is negligible.

Optimization over the space of probability measures endowed with the Wasserstein-2 geometry is central to modern machine learning and mean-field modeling. However, traditional methods relying on full Wasserstein gradients often suffer from high computational overhead in high-dimensional or ill-conditioned settings. We propose a randomized coordinate descent framework specifically designed for the Wasserstein manifold, introducing both Random Wasserstein Coordinate Descent (RWCD) and Random Wasserstein Coordinate Proximal{-Gradient} (RWCP) for composite objectives. By exploiting coordinate-wise structures, our methods adapt to anisotropic objective landscapes where full-gradient approaches typically struggle. We provide a rigorous convergence analysis across various landscape geometries, establishing guarantees under non-convex, Polyak-Łojasiewicz, and geodesically convex conditions. Our theoretical results mirror the classic convergence properties found in Euclidean space, revealing a compelling symmetry between coordinate descent on vectors and on probability measures. The developed techniques are inherently adaptive to the Wasserstein geometry and offer a robust analytical template that can be extended to other optimization solvers within the space of measures. Numerical experiments on ill-conditioned energies demonstrate that our framework offers significant speedups over conventional full-gradient methods.

As safety concerns around large language models (LLMs) grow, understanding the internal mechanisms underlying refusal behavior has become increasingly important. Recent work has studied this behavior by identifying internal features associated with refusal and manipulating them to induce compliance with harmful requests. However, existing refusal feature selection methods rely on how strongly features activate on harmful prompts, which tends to capture superficial signals rather than the causal factors underlying the refusal decision. We propose CRaFT, a circuit-guided refusal feature selection framework that ranks features by their influence on the model's refusal-compliance decision using prompts near the refusal boundary. On Gemma-3-1B-it, CRaFT improves attack success rate (ASR) from 6.7% to 48.2% and outperforms baseline methods across multiple jailbreak benchmarks. These results suggest that circuit influence is a more reliable criterion than activation magnitude for identifying features that causally mediate refusal behavior.

We investigate training strategies that co-develop in-context learning (ICL) and in-weights learning (IWL), and the ability to switch between them based on context relevance. Although current LLMs exhibit both modes, standard task-specific fine-tuning often erodes ICL, motivating IC-Train - fine-tuning with in-context examples. Prior work has shown that emergence of ICL after IC-Train depends on factors such as task diversity and training duration. In this paper we show that the similarity structure between target inputs and context examples also plays an important role. Random context leads to loss of ICL and IWL dominance, while only similar examples in context causes ICL to degenerate to copying labels without regard to relevance. To address this, we propose a simple Contrastive-Context which enforces two types of contrasts: (1) mix of similar and random examples within a context to evolve a correct form of ICL, and (2) varying grades of similarity across contexts to evolve ICL-IWL mixtures. We present insights on the importance of such contrast with theoretical analysis of a minimal model. We validate with extensive empirical evaluation on four LLMs and several tasks. Diagnostic probes confirm that contrasted contexts yield stable ICL-IWL mixtures, avoiding collapse into pure ICL, IWL, or copying.

Multimodal Large Language Models (MLLMs) have recently demonstrated promising capabilities in multimodal coding tasks such as chart-to-code generation. However, existing methods primarily rely on supervised fine-tuning (SFT), which requires the model to learn code patterns through chart-code pairs but does not expose the model to a code execution environment. Moreover, while self-correction through execution feedback offers a potential route to improve coding quality, even state-of-the-art MLLMs have been shown to struggle with effective self-correction. In this work, we introduce MM-ReCoder, a chart-to-code generation model trained with reinforcement learning (RL) and equipped with self-correction ability. We propose a two-stage multi-turn self-correction RL strategy based on Group Relative Policy Optimization (GRPO). The first stage enhances the model's self-correction ability via rolling out a shared first turn, while the second stage improves the coding capability with full-trajectory optimization. MM-ReCoder learns to produce more accurate and executable code through the interaction with the environment and by iteratively correcting its own outputs. Our results on three chart-to-code benchmarks demonstrate the state-of-the-art performance of MM-ReCoder.

Memory-Augmented Generation (MAG) extends large language models with external memory to support long-context reasoning, but existing approaches universally treat memory as an external service that agents call into, delegating storage to separate pipelines of chunking, embedding, and graph extraction. This architectural separation means the system that stores knowledge does not understand it, leading to semantic drift between what the agent intended to remember and what the pipeline actually captured, loss of coordination context across agents, and fragile recovery after failures. In this paper, we propose ByteRover, an agent-native memory architecture that inverts the memory pipeline: the same LLM that reasons about a task also curates, structures, and retrieves knowledge. ByteRover represents knowledge in a hierarchical Context Tree, a file-based knowledge graph organized as Domain, Topic, Subtopic, and Entry, where each entry carries explicit relations, provenance, and an Adaptive Knowledge Lifecycle (AKL) with importance scoring, maturity tiers, and recency decay. Retrieval uses a 5-tier progressive strategy that resolves most queries at sub-100 ms latency without LLM calls, escalating to agentic reasoning only for novel questions. Experiments on LoCoMo and LongMemEval demonstrate that ByteRover achieves state-of-the-art accuracy on LoCoMo and competitive results on LongMemEval while requiring zero external infrastructure, no vector database, no graph database, no embedding service, with all knowledge stored as human-readable markdown files on the local filesystem.

Traditional RL algorithms like Proximal Policy Optimization (PPO) typically train on the entire rollout buffer, operating under the assumption that all generated episodes provide a beneficial optimization signal. However, these episodes frequently contain noisy or unfaithful reasoning, which can degrade model performance and slow down training. In this paper, we propose \textbf{Influence-Guided PPO (I-PPO)}, a novel framework that integrates data attribution into the RL post-training loop. By calculating an influence score for each episode using a gradient-based approximation, I-PPO identifies and eliminates episodes that are anti-aligned with a validation gradient. Our experiments demonstrate that I-PPO consistently outperforms SFT and PPO baselines. We show that our filtering process acts as an intrinsic early stopping mechanism, accelerating training efficiency while effectively reducing unfaithful CoT reasoning.

Seizure detection from EEG signals is highly challenging due to complex spatiotemporal dynamics and extreme inter-patient variability. To model them, recent methods construct dynamic graphs via statistical correlations, predefined similarity measures, or implicit learning, yet rarely account for EEG's noisy nature. Consequently, these graphs usually contain redundant or task-irrelevant connections, undermining model performance even with state-of-the-art architectures. In this paper, we present a new perspective for EEG seizure detection: jointly learning denoised dynamic graph structures and informative spatial-temporal representations guided by the Information Bottleneck (IB). Unlike prior approaches, our graph constructor explicitly accounts for the noisy characteristics of EEG data, producing compact and reliable connectivity patterns that better support downstream seizure detection. To further enhance representation learning, we employ a self-supervised Graph Masked AutoEncoder that reconstructs masked EEG signals based on dynamic graph context, promoting structure-aware and compact representations aligned with the IB principle. Bringing things together, we introduce Information Bottleneck-guided EEG SeizuRE DetectioN via SElf-Supervised Learning (IRENE), which explicitly learns dynamic graph structures and interpretable spatial-temporal EEG representations. IRENE addresses three core challenges: (i) Identifying the most informative nodes and edges; (ii) Explaining seizure propagation in the brain network; and (iii) Enhancing robustness against label scarcity and inter-patient variability. Extensive experiments on benchmark EEG datasets demonstrate that our method outperforms state-of-the-art baselines in seizure detection and provides clinically meaningful insights into seizure dynamics. The source code is available at https://github.com/LabRAI/IRENE.

How do LLMs decide what to teach next: by reasoning about a learner's knowledge, or by using simpler rules of thumb? We test this in a controlled task previously used to study human teaching strategies. On each trial, a teacher LLM sees a hypothetical learner's trajectory through a reward-annotated directed graph and must reveal a single edge so the learner would choose a better path if they replanned. We run a range of LLMs as simulated teachers and fit their trial-by-trial choices with the same cognitive models used for humans: a Bayes-Optimal teacher that infers which transitions the learner is missing (inverse planning), weaker Bayesian variants, heuristic baselines (e.g., reward based), and non-mentalizing utility models. In a baseline experiment matched to the stimuli presented to human subjects, most LLMs perform well, show little change in strategy over trials, and their graph-by-graph performance is similar to that of humans. Model comparison (BIC) shows that Bayes-Optimal teaching best explains most models' choices. When given a scaffolding intervention, models follow auxiliary inference- or reward-focused prompts, but these scaffolds do not reliably improve later teaching on heuristic-incongruent test graphs and can sometimes reduce performance. Overall, cognitive model fits provide insight into LLM tutoring policies and show that prompt compliance does not guarantee better teaching decisions.

We introduce ThinkTwice, a simple two-phase framework that jointly optimizes LLMs to solve reasoning problems and refine the answers, based on Group Relative Policy Optimization (GRPO). In each pair of training steps, ThinkTwice first optimizes the model on solving reasoning problems, then optimizes it on refining its own solutions to the same problems, using the same binary correctness reward in both phases without correctness signals or critique annotations. Across five mathematical reasoning benchmarks and two model families including Qwen3-4B and Olmo3-7B, ThinkTwice substantially improves both reasoning and refinement performance over competitive online policy optimization baselines. Specifically, on Qwen3-4B, ThinkTwice outperforms GRPO on AIME by 5 percentage points before refinement and by 11.5 points after one self-refinement step, measured by pass@4. Analysis of the training dynamics of ThinkTwice reveals an implicit rectify-then-fortify curriculum: refinement predominantly corrects errors early in training and naturally shifts toward preserving already-correct solutions as the model improves, yielding a more rectified reward signal. Our work establishes joint training of reasoning and self-refinement as a principled and effective methodology for RLVR.

Automating the translation of Operations Research (OR) problems from natural language to executable models is a critical challenge. While Large Language Models (LLMs) have shown promise in linear tasks, they suffer from severe performance degradation in real-world nonlinear scenarios due to semantic misalignment between mathematical formulations and solver codes, as well as unstable information extraction. In this study, we introduce NED-Tree, a systematic framework designed to bridge the semantic gap. NED-Tree employs (a) a sentence-by-sentence extraction strategy to ensure robust parameter mapping and traceability; and (b) a recursive tree-based structure that adaptively decomposes complex nonlinear terms into solver-compatible sub-elements. Additionally, we present NEXTOR, a novel benchmark specifically designed for complex nonlinear, extensive-constraint OR problems. Experiments across 10 benchmarks demonstrate that NED-Tree establishes a new state-of-the-art with 72.51% average accuracy, NED-Tree is the first framework that drives LLMs to resolve nonlinear modeling difficulties through element decomposition, achieving alignment between modeling semantics and code semantics. The NED-Tree framework and benchmark are accessible in the anonymous repository https://anonymous.4open.science/r/NORA-NEXTOR.

Uncertainty propagation in high-dimensional nonlinear dynamic structural systems is pivotal in state-of-the-art performance-based design and risk assessment, where uncertainties from both excitations and structures, i.e., the aleatoric uncertainty, must be considered. This poses a significant challenge due to heavy computational demands. Machine learning techniques are thus introduced as metamodels to alleviate this burden. However, the "black box" nature of Machine learning models underscores the necessity of avoiding overly confident predictions, particularly when data and training efforts are insufficient. This creates a need, in addition to considering the aleatoric uncertainty, of estimating the uncertainty related to the prediction confidence, i.e., epistemic uncertainty, for machine learning-based metamodels. We developed a probabilistic metamodeling technique based on a variational long short-term memory (LSTM) with augmented inputs to simultaneously capture aleatoric and epistemic uncertainties. Key random system parameters are treated as augmented inputs alongside excitation series carrying record-to-record variability to capture the full range of aleatoric uncertainty. Meanwhile, epistemic uncertainty is effectively approximated via the Monte Carlo dropout scheme. Unlike computationally expensive full Bayesian approaches, this method incurs negligible additional training costs while enabling nearly cost-free uncertainty simulation. The proposed technique is demonstrated through multiple case studies involving stochastic seismic or wind excitations. Results show that the calibrated metamodels accurately reproduce nonlinear response time histories and provide confidence bounds indicating the associated epistemic uncertainty.

Open-vocabulary human-object interaction (HOI) detection is a step towards building scalable systems that generalize to unseen interactions in real-world scenarios and support grounded multimodal systems that reason about human-object relationships. However, standard evaluation metrics, such as mean Average Precision (mAP), treat HOI classes as discrete categorical labels and fail to credit semantically valid but lexically different predictions (e.g., "lean on couch" vs. "sit on couch"), limiting their applicability for evaluating open-vocabulary predictions that go beyond any predefined set of HOI labels. We introduce SHOE (Semantic HOI Open-Vocabulary Evaluation), a new evaluation framework that incorporates semantic similarity between predicted and ground-truth HOI labels. SHOE decomposes each HOI prediction into its verb and object components, estimates their semantic similarity using the average of multiple large language models (LLMs), and combines them into a similarity score to evaluate alignment beyond exact string match. This enables a flexible and scalable evaluation of both existing HOI detection methods and open-ended generative models using standard benchmarks such as HICO-DET. Experimental results show that SHOE scores align more closely with human judgments than existing metrics, including LLM-based and embedding-based baselines, achieving an agreement of 85.73% with the average human ratings. Our work underscores the need for semantically grounded HOI evaluation that better mirrors human understanding of interactions. We will release our evaluation metric to the public to facilitate future research.

Multimodal tabular-image fusion is an emerging task that has received increasing attention in various domains. However, existing methods may be hindered by gradient conflicts between modalities, misleading the optimization of the unimodal learner. In this paper, we propose a novel Gradient-Aligned Alternating Learning (GAAL) paradigm to address this issue by aligning modality gradients. Specifically, GAAL adopts an alternating unimodal learning and shared classifier to decouple the multimodal gradient and facilitate interaction. Furthermore, we design uncertainty-based cross-modal gradient surgery to selectively align cross-modal gradients, thereby steering the shared parameters to benefit all modalities. As a result, GAAL can provide effective unimodal assistance and help boost the overall fusion performance. Empirical experiments on widely used datasets reveal the superiority of our method through comparison with various state-of-the-art (SoTA) tabular-image fusion baselines and test-time tabular missing baselines. The source code is available at https://github.com/njustkmg/ICME26-GAAL.

We extend the recent latent recurrent modeling to sequential input streams. By interleaving fast, recurrent latent updates with self-organizational ability between slow observation updates, our method facilitates the learning of stable internal structures that evolve alongside the input. This mechanism allows the model to maintain coherent and clustered representations over long horizons, improving out-of-distribution generalization in reinforcement learning and algorithmic tasks compared to sequential baselines such as LSTM, state space models, and Transformer variants.

Large language models deployed in supportive or advisory roles must balance helpfulness with preservation of user autonomy, yet standard alignment methods primarily optimize for helpfulness and harmlessness without explicitly modeling relational risks such as dependency reinforcement, overprotection, or coercive guidance. We introduce Care-Conditioned Neuromodulation (CCN), a state-dependent control framework in which a learned scalar signal derived from structured user state and dialogue context conditions response generation and candidate selection. We formalize this setting as an autonomy-preserving alignment problem and define a utility function that rewards autonomy support and helpfulness while penalizing dependency and coercion. We also construct a benchmark of relational failure modes in multi-turn dialogue, including reassurance dependence, manipulative care, overprotection, and boundary inconsistency. On this benchmark, care-conditioned candidate generation combined with utility-based reranking improves autonomy-preserving utility by +0.25 over supervised fine-tuning and +0.07 over preference optimization baselines while maintaining comparable supportiveness. Pilot human evaluation and zero-shot transfer to real emotional-support conversations show directional agreement with automated metrics. These results suggest that state-dependent control combined with utility-based selection is a practical approach to multi-objective alignment in autonomy-sensitive dialogue.

Parallel p-bit Ising machines are a promising platform for fast and energy-efficient combinatorial optimization, but their scalability depends on update synchronization, hardware delay, and architectural cost. In this work, we establish a unified performance-cost framework by analyzing synchronous and asynchronous update schemes under realistic constraints, including finite delay, time-multiplexed p-bit reuse, and limited DAC precision. We show that synchronous updates are not inherently unstable but can exhibit oscillations under excessive simultaneity, while asynchronous updates require slower operation due to hardware delay. To address this trade-off, we introduce time-multiplexed p-bit reuse with structured synchronous control, preserving correct annealing dynamics while reducing hardware requirements. This approach decouples statistical correctness from physical resources, enabling the number of p-bits and DACs to scale inversely with the reuse factor. As a result, synchronous architectures achieve comparable or better solution quality at less than half the hardware cost of optimized asynchronous designs on G-set MaxCut benchmarks (800-2000 nodes). We also show that low-resolution DACs (3-4 bits) are sufficient to reach near-optimal solutions when annealing time is properly adjusted. These findings provide practical design guidelines for scalable probabilistic computing hardware under realistic constraints.

In LLM training, normalization layers and optimizers are typically treated as independent design choices. In a 3x2 factorial at 1B parameters and 1000 training steps, we show this assumption can fail: Dynamic Erf (Derf; Chen & Liu, 2025) suffers a large negative interaction with Muon (Jordan, 2024), with its gap to RMSNorm growing from +0.31 nats under AdamW to +0.97 under Muon, approximately three times larger. Dynamic Tanh (DyT; Zhu et al., 2025), included as a bounded-normalizer control, shows no such penalty. Our evidence points to two failure modes of erf under Muon's faster spectral-norm growth: saturation (lossy compression) and scale blindness (discarding activation magnitude). An EMA-blend that reintroduces running scale estimates recovers ~84% of the gap. Separately, reducing Derf's alpha from its published default (0.5 to 0.3) recovers ~80% by keeping erf in its near-linear regime, where it approximately preserves relative scale; this setting is not the published default of Chen & Liu (2025). Using Derf's published default alpha with Muon incurs a 0.66-nat interaction penalty without producing NaNs or divergence, making the failure easy to miss in short pilot runs.

Voice cloning is often evaluated in terms of overall quality, but less is known about accent preservation and its perceptual consequences. We compare standard and heavily accented Mandarin speech and their voice clones using a combined computational and perceptual design. Embedding-based analyses show no reliable accented-standard difference in original-clone distances across systems. In the perception study, clones are rated as more similar to their originals for standard than for accented speakers, and intelligibility increases from original to clone, with a larger gain for accented speech. These results show that accent variation can shape perceived identity match and intelligibility in voice cloning even when it is not reflected in an off-the-shelf speaker-embedding distance, and they motivate evaluating speaker identity preservation and accent preservation as separable dimensions.

We present ReFlow, a unified framework for monocular dynamic scene reconstruction that learns 3D motion in a novel self-correction manner from raw video. Existing methods often suffer from incomplete scene initialization for dynamic regions, leading to unstable reconstruction and motion estimation, which often resorts to external dense motion guidance such as pre-computed optical flow to further stabilize and constrain the reconstruction of dynamic components. However, this introduces additional complexity and potential error propagation. To address these issues, ReFlow integrates a Complete Canonical Space Construction module for enhanced initialization of both static and dynamic regions, and a Separation-Based Dynamic Scene Modeling module that decouples static and dynamic components for targeted motion supervision. The core of ReFlow is a novel self-correction flow matching mechanism, consisting of Full Flow Matching to align 3D scene flow with time-varying 2D observations, and Camera Flow Matching to enforce multi-view consistency for static objects. Together, these modules enable robust and accurate dynamic scene reconstruction. Extensive experiments across diverse scenarios demonstrate that ReFlow achieves superior reconstruction quality and robustness, establishing a novel self-correction paradigm for monocular 4D reconstruction.

Recent advances in persona-centric memory have revealed the powerful capability of multi-agent systems in managing persona memory, especially in conversational scenarios. However, these complex frameworks often suffer from information loss and are fragile across varying scenarios, resulting in suboptimal performance. In this paper, we propose DeltaMem, an agentic memory management system that formulates persona-centric memory management as an end-to-end task within a single-agent setting. To further improve the performance of our agentic memory manager, we draw inspiration from the evolution of human memory and synthesize a user-assistant dialogue dataset along with corresponding operation-level memory updating labels. Building on this, we introduce a novel Memory-based Levenshtein Distance to formalize the memory updating reward, and propose a tailored reinforcement learning framework to further enhance the management capabilities of DeltaMem. Extensive experiments show that both training-free and RL-trained DeltaMem outperform all product-level baselines across diverse long-term memory benchmarks, including LoCoMo, HaluMem, and PersonaMem.

Binary Function Similarity Detection (BFSD) is a core problem in software security, supporting tasks such as vulnerability analysis, malware classification, and patch provenance. In the past few decades, numerous models and tools have been developed for this application; however, due to the lack of a comprehensive universal benchmark in this field, researchers have struggled to compare different models effectively. Existing datasets are limited in scope, often focusing on a narrow set of transformations or types of binaries, and fail to reflect the full diversity of real-world applications. We introduce EXHIB, a benchmark comprising five realistic datasets collected from the wild, each highlighting a distinct aspect of the BFSD problem space. We evaluate 9 representative models spanning multiple BFSD paradigms on EXHIB and observe performance degradations of up to 30% on firmware and semantic datasets compared to standard settings, revealing substantial generalization gaps. Our results show that robustness to low- and mid-level binary variations does not generalize to high-level semantic differences, underscoring a critical blind spot in current BFSD evaluation practices.

Denoising generative models deliver high-fidelity generation but remain bottlenecked by inference latency due to the many iterative denoiser calls required during sampling. Training-free acceleration methods reduce latency by either sparsifying the model architecture or shortening the sampling trajectory. Current training-free acceleration methods are more complex than necessary: higher-order predictors amplify error under aggressive speedups, and architectural modifications hinder deployment. Beyond 2x acceleration, step skipping creates structural scarcity -- at most one fresh evaluation per local window -- leaving the computed output and its backward difference as the only causally grounded information. Based on this, we propose ZEUS, an acceleration method that predicts reduced denoiser evaluations using a second-order predictor, and stabilizes aggressive consecutive skipping with an interleaved scheme that avoids back-to-back extrapolations. ZEUS adds essentially zero overhead, no feature caches, and no architectural modifications, and it is compatible with different backbones, prediction objectives, and solver choices. Across image and video generation, ZEUS consistently improves the speed-fidelity performance over recent training-free baselines, achieving up to 3.2x end-to-end speedup while maintaining perceptual quality. Our code is available at: https://github.com/Ting-Justin-Jiang/ZEUS.

The latent space of generative modeling is long dominated by the VAE encoder. The latents from the pretrained representation encoders (e.g., DINO, SigLIP, MAE) are previously considered inappropriate for generative modeling. Recently, RAE method lights the hope and reveals that the representation autoencoder can also achieve competitive performance as the VAE encoder. However, the integration of representation autoencoder into continuous autoregressive (AR) models, remains largely unexplored. In this work, we investigate the challenges of employing high-dimensional representation autoencoders within the AR paradigm, denoted as \textit{RAE-AR}. We focus on the unique properties of AR models and identify two primary hurdles: complex token-wise distribution modeling and the high-dimensionality amplified training-inference gap (exposure bias). To address these, we introduce token simplification via distribution normalization to ease modeling difficulty and improve convergence. Furthermore, we enhance prediction robustness by incorporating Gaussian noise injection during training to mitigate exposure bias. Our empirical results demonstrate that these modifications substantially bridge the performance gap, enabling representation autoencoder to achieve results comparable to traditional VAEs on AR models. This work paves the way for a more unified architecture across visual understanding and generative modeling.

Large language models have been adopted in the medical domain for clinical documentation to reduce clinician burden. However, studies have reported that LLMs often "forget" a significant amount of instruction-following ability when fine-tuned using a task-specific medical dataset, a critical challenge in adopting general-purpose LLMs for clinical applications. This study presents a model merging framework to efficiently adapt general-purpose LLMs to the medical domain by countering this forgetting issue. By merging a clinical foundation model (GatorTronLlama) with a general instruct model (Llama-3.1-8B-Instruct) via interpolation-based merge methods, we seek to derive a domain-adapted model with strong performance on clinical tasks while retaining instruction-following ability. Comprehensive evaluation across medical benchmarks and five clinical generation tasks (e.g., radiology and discharge summarization) shows that merged models can effectively mitigate catastrophic forgetting, preserve clinical domain expertise, and retain instruction-following ability. In addition, our model merging strategies demonstrate training efficiency, achieving performance on par with fully fine-tuned baselines under severely constrained supervision (e.g., 64-shot vs. 256-shot). Consequently, weight-space merging constitutes a highly scalable solution for adapting open-source LLMs to clinical applications, facilitating broader deployment in resource-constrained healthcare environments.

Web agents based on large language models (LLMs) rely on observations of web pages -- commonly represented as HTML -- as the basis for identifying available actions and planning subsequent steps. Prior work has treated the verbosity of HTML as an obstacle to performance and adopted observation reduction as a standard practice. We revisit this trend and demonstrate that the optimal observation representation depends on model capability and thinking token budget: (1) compact observations (accessibility trees) are preferable for lower-capability models, while detailed observations (HTML) are advantageous for higher-capability models; moreover, increasing thinking tokens further amplifies the benefit of HTML. (2) Our error analysis suggests that higher-capability models exploit layout information in HTML for better action grounding, while lower-capability models suffer from increased hallucination under longer inputs. We also find that incorporating observation history improves performance across most models and settings, and a diff-based representation offers a token-efficient alternative. Based on these findings, we suggest practical guidelines: adaptively select observation representations based on model capability and thinking token budget, and incorporate observation history using diff-based representations.

Large language model (LLM) agents are increasingly deployed for complex tool-orchestration tasks, yet existing benchmarks fail to capture the rigorous demands of industrial domains where incorrect decisions carry significant safety and financial consequences. To address this critical gap, we introduce PHMForge, the first comprehensive benchmark specifically designed to evaluate LLM agents on Prognostics and Health Management (PHM) tasks through realistic interactions with domain-specific MCP servers. Our benchmark encompasses 75 expert-curated scenarios spanning 7 industrial asset classes (turbofan engines, bearings, electric motors, gearboxes, aero-engines) across 5 core task categories: Remaining Useful Life (RUL) Prediction, Fault Classification, Engine Health Analysis, Cost-Benefit Analysis, and Safety/Policy Evaluation. To enable rigorous evaluation, we construct 65 specialized tools across two MCP servers and implement execution-based evaluators with task-commensurate metrics: MAE/RMSE for regression, F1-score for classification, and categorical matching for health assessments. Through extensive evaluation of leading frameworks (ReAct, Cursor Agent, Claude Code) paired with frontier LLMs (Claude Sonnet 4.0, GPT-4o, Granite-3.0-8B), we find that even top-performing configurations achieve only 68\% task completion, with systematic failures in tool orchestration (23\% incorrect sequencing), multi-asset reasoning (14.9 percentage point degradation), and cross-equipment generalization (42.7\% on held-out datasets). We open-source our complete benchmark, including scenario specifications, ground truth templates, tool implementations, and evaluation scripts, to catalyze research in agentic industrial AI.

Current Large Language Model (LLM) approaches for information extraction (IE) in the healthy food policy domain are often hindered by various factors, including misinformation, specifically hallucinations, misclassifications, and omissions that result from the structural diversity and inconsistency of policy documents. To address these limitations, this study proposes a role-based LLM framework that automates the IE from unstructured policy data by assigning specialized roles: an LLM policy analyst for metadata and mechanism classification, an LLM legal strategy specialist for identifying complex legal approaches, and an LLM food system expert for categorizing food system stages. This framework mimics expert analysis workflows by incorporating structured domain knowledge, including explicit definitions of legal mechanisms and classification criteria, into role-specific prompts. We evaluate the framework using 608 healthy food policies from the Healthy Food Policy Project (HFPP) database, comparing its performance against zero-shot, few-shot, and chain-of-thought (CoT) baselines using Llama-3.3-70B. Our proposed framework demonstrates superior performance in complex reasoning tasks, offering a reliable and transparent methodology for automating IE from health policies.

Benchmarks that reflect production workloads are better for evaluating AI coding agents in industrial settings, yet existing benchmarks differ from real usage in programming language distribution, prompt style and codebase structure. This paper presents a methodology for curating production-derived benchmarks, illustrated through ProdCodeBench - a benchmark built from real sessions with a production AI coding assistant. We detail our data collection and curation practices including LLM-based task classification, test relevance validation, and multi-run stability checks which address challenges in constructing reliable evaluation signals from monorepo environments. Each curated sample consists of a verbatim prompt, a committed code change and fail-to-pass tests spanning seven programming languages. Our systematic analysis of four foundation models yields solve rates from 53.2% to 72.2% revealing that models making greater use of work validation tools, such as executing tests and invoking static analysis, achieve higher solve rates. This suggests that iterative verification helps achieve effective agent behavior and that exposing codebase-specific verification mechanisms may significantly improve the performance of externally trained agents operating in unfamiliar environments. We share our methodology and lessons learned to enable other organizations to construct similar production-derived benchmarks.

We give a short linear--algebraic proof of the inequality \[ \|x\|_1\,\|x\|_\infty \le \frac{1+\sqrt{p}}{2}\,\|x\|_2^2, \] valid for every \(x\in\mathbb{R}^p\). This inequality relates three fundamental norms on finite-dimensional spaces and has applications in optimization and numerical analysis. Our proof exploits the determinantal structure of a parametrized family of quadratic forms, and we show the constant $(1+\sqrt{p})/2$ is optimal.

Electrocardiograms (ECGs) are among the most widely used diagnostic tools for cardiovascular diseases, and a large amount of ECG data worldwide appears only in image form. However, most existing automated ECG analysis methods rely on access to raw signal recordings, limiting their applicability in real-world and resource-constrained settings. In this paper, we present ECG-Scan, a self-supervised framework for learning clinically generalized representations from ECG images through dual physiological-aware alignments: 1) Our approach optimizes image representation learning using multimodal contrastive alignment between image and gold-standard signal-text modalities. 2) We further integrate domain knowledge via soft-lead constraints, regularizing the reconstruction process and improving signal lead inter-consistency. Extensive benchmarking across multiple datasets and downstream tasks demonstrates that our image-based model achieves superior performance compared to existing image baselines and notably narrows the gap between ECG image and signal analysis. These results highlight the potential of self-supervised image modeling to unlock large-scale legacy ECG data and broaden access to automated cardiovascular diagnostics.

The increasing scale and complexity of mobile applications make automated GUI exploration essential for software quality assurance. However, existing methods often neglect state dependencies between test fragments, which leads to redundant exploration and prevents access to deep application states. We introduce EpiDroid, a black-box, pluggable framework that augments existing explorers through semantic state dependency awareness. EpiDroid distills raw traces into stable test fragments to extract underlying dependencies. It then employs a Recomposition-Replay paradigm to perform impact reasoning via LLM and deterministic replay on high-value mutable state elements. Through iterative feedback, EpiDroid refines the state-dependency graph to systematically reach deep application states. We integrated EpiDroid into both industrial and state-of-the-art research tools and evaluated it on 20 real-world apps. The results show that EpiDroid consistently improves the performance of all baselines, increasing average code coverage by 10--28\% and delivering 3--4$\times$ more coverage gain compared to continuing the baselines alone from the same starting point. This demonstrates that dependency-guided recomposition unlocks deep states that forward exploration cannot access, irrespective of additional budget.

Traditional social science research often requires designing complex experiments across vast methodological spaces and depends on real human participants, making it labor-intensive, costly, and difficult to scale. Here we present S-Researcher, an LLM-agent-based platform that assists researchers in conducting social science research more efficiently and at greater scale by "siliconizing" both the research process and the participant pool. To build S-Researcher, we first develop YuLan-OneSim, a large-scale social simulation system designed around three core requirements: generality via auto-programming from natural language to executable scenarios, scalability via a distributed architecture supporting up to 100,000 concurrent agents, and reliability via feedback-driven LLM fine-tuning. Leveraging this system, S-Researcher supports researchers in designing social experiments, simulating human behavior with LLM agents, analyzing results, and generating reports, forming a complete human-AI collaborative research loop in which researchers retain oversight and intervention at every stage. We operationalize LLM simulation research paradigms into three canonical reasoning modes (induction, deduction, and abduction) and validate S-Researcher through systematic case studies: inductive reproduction of cultural dynamics consistent with Axelrod's theory, deductive testing of competing hypotheses on teacher attention validated against survey data, and abductive identification of a cooperation mechanism in public goods games confirmed by human experiments. S-Researcher establishes a new human--AI collaborative paradigm for social science, in which computational simulation augments human researchers to accelerate discovery across the full spectrum of social inquiry.

Benchmarks driven by test suites, notably SWE-bench, have become the de facto standard for measuring the effectiveness of automated issue-resolution agents: a generated patch is accepted whenever it passes the accompanying regression tests. In practice, however, insufficiently strong test suites can admit plausible yet semantically incorrect patches, inflating reported success rates. We introduce STING, a framework for targeted test augmentation that uses semantically altered program variants as diagnostic stressors to uncover and repair weaknesses in benchmark regression suites. Variants of the ground-truth patch that still pass the existing tests reveal under-constrained behaviors; these gaps then guide the generation of focused regression tests. A generated test is retained only if it (i) passes on the ground-truth patch, (ii) fails on at least one variant that survived the original suite, and (iii) remains valid under behavior-preserving transformations designed to guard against overfitting. Applied to SWE-bench Verified, STING finds that 77% of instances contain at least one surviving variant. STING produces 1,014 validated tests spanning 211 instances and increases patch-region line and branch coverage by 10.8% and 9.5%, respectively. Re-assessing the top-10 repair agents with the strengthened suites lowers their resolved rates by 4.2%-9.0%, revealing that a substantial share of previously passing patches exploit weaknesses in the benchmark tests rather than faithfully implementing the intended fix. These results underscore that reliable benchmark evaluation depends not only on patch generation, but equally on test adequacy.

Instruction-based unlearning has proven effective for modifying the behavior of large language models at inference time, but whether this paradigm extends to other generative models remains unclear. In this work, we investigate instruction-based unlearning in diffusion-based image generation models and show, through controlled experiments across multiple concepts and prompt variants, that diffusion models systematically fail to suppress targeted concepts when guided solely by natural-language unlearning instructions. By analyzing both the CLIP text encoder and cross-attention dynamics during the denoising process, we find that unlearning instructions do not induce sustained reductions in attention to the targeted concept tokens, causing the targeted concept representations to persist throughout generation. These results reveal a fundamental limitation of prompt-level instruction in diffusion models and suggest that effective unlearning requires interventions beyond inference-time language control.

Tool using agents often fail for operational reasons even when language understanding is strong. Common causes include invalid arguments, interface drift, weak recovery, and inefficient retry behavior. We introduce ToolMisuseBench, an offline deterministic benchmark for evaluating tool misuse and recovery under explicit step, call, and retry budgets. The benchmark covers CRUD, retrieval, file, and scheduling environments with replayable fault injection. It reports success, invalid call behavior, policy violations, recovery quality, and budgeted efficiency. We release a public dataset with 6800 tasks and a reproducible evaluation pipeline. Baseline results show fault specific recovery gains for schema aware methods, while overall success remains limited under the released authorization and hard failure settings.

Long-tailed classification, where a small number of frequent classes dominate many rare ones, remains challenging because models systematically favor frequent classes at inference time. Existing post-hoc methods such as logit adjustment address this by adding a fixed classwise offset to the base-model logits. However, the correction required to restore the relative ranking of two classes need not be constant across inputs, and a fixed offset cannot adapt to such variation. We study this problem through Bayes-optimal reranking on a base-model top-k shortlist. The gap between the optimal score and the base score, the residual correction, decomposes into a classwise component that is constant within each class, and a pairwise component that depends on the input and competing labels. When the residual is purely classwise, a fixed offset suffices to recover the Bayes-optimal ordering. We further show that when the same label pair induces incompatible ordering constraints across contexts, no fixed offset can achieve this recovery. This decomposition leads to testable predictions regarding when pairwise correction can improve performance and when cannot. We develop REPAIR (Reranking via Pairwise residual correction), a lightweight post-hoc reranker that combines a shrinkage-stabilized classwise term with a linear pairwise term driven by competition features on the shortlist. Experiments on five benchmarks spanning image classification, species recognition, scene recognition, and rare disease diagnosis confirm that the decomposition explains where pairwise correction helps and where classwise correction alone suffices.

Research on Large Language Models (LLMs) studies output variation across generation, reasoning, alignment, and representational analysis, often under the umbrella of "diversity." Yet the terminology remains fragmented, largely because the normative objectives underlying tasks are rarely made explicit. We introduce the Magic, Madness, Heaven, Sin framework, which models output variation along a homogeneity-heterogeneity axis, where valuation is determined by the task and its normative objective. We organize tasks into four normative contexts: epistemic (factuality), interactional (user utility), societal (representation), and safety (robustness). For each, we examine the failure modes and vocabulary such as hallucination, mode collapse, bias, and erasure through which variation is studied. We apply the framework to analyze all pairwise cross-contextual interactions, revealing that optimizing for one objective, such as improving safety, can inadvertently harm demographic representation or creative diversity. We argue for context-aware evaluation of output variation, reframing it as a property shaped by task objectives rather than a model's intrinsic trait.

Conformal risk control (CRC) provides distribution-free guarantees for controlling the expected loss at a user-specified level. Existing theory typically assumes that the loss decreases monotonically with a tuning parameter that governs the size of the prediction set. This assumption is often violated in practice, where losses may behave non-monotonically due to competing objectives such as coverage and efficiency. We study CRC under non-monotone loss functions when the tuning parameter is selected from a finite grid, a common scenario in thresholding or discretized decision rules. Revisiting a known counterexample, we show that the validity of CRC without monotonicity depends on the relationship between the calibration sample size and the grid resolution. In particular, risk control can still be achieved when the calibration sample is sufficiently large relative to the grid. We provide a finite-sample guarantee for bounded losses over a grid of size $m$, showing that the excess risk above the target level $α$ is of order $\sqrt{\log(m)/n}$, where $n$ is the calibration sample size. A matching lower bound shows that this rate is minimax optimal. We also derive refined guarantees under additional structural conditions, including Lipschitz continuity and monotonicity, and extend the analysis to settings with distribution shift via importance weighting. Numerical experiments on synthetic multilabel classification and real object detection data illustrate the practical impact of non-monotonicity. Methods that account for finite-sample deviations achieve more stable risk control than approaches based on monotonicity transformations, while maintaining competitive prediction-set sizes.

Evolution Strategies (ES) have emerged as a scalable gradient-free alternative to reinforcement learning based LLM fine-tuning, but it remains unclear whether comparable task performance implies comparable solutions in parameter space. We compare ES and Group Relative Policy Optimization (GRPO) across four tasks in both single-task and sequential continual-learning settings. ES matches or exceeds GRPO in single-task accuracy and remains competitive sequentially when its iteration budget is controlled. Despite this similarity in task performance, the two methods produce markedly different model updates: ES makes much larger changes and induces broader off-task KL drift, whereas GRPO makes smaller, more localized updates. Strikingly, the ES and GRPO solutions are linearly connected with no loss barrier, even though their update directions are nearly orthogonal. We develop an analytical theory of ES that explains all these phenomena within a unified framework, showing how ES can accumulate large off-task movement on weakly informative directions while still making enough progress on the task to match gradient-based RL in downstream accuracy. These results show that gradient-free and gradient-based fine-tuning can reach similarly accurate yet geometrically distinct solutions, with important consequences for forgetting and knowledge preservation. The source code is publicly available: https://github.com/Bhoy1/ESvsGRPO.

We introduce SWE-ZERO to SWE-HERO, a two-stage SFT recipe that achieves state-of-the-art results on SWE-bench by distilling open-weight frontier LLMs. Our pipeline replaces resource-heavy dependencies with an evolutionary refinement strategy: (1) SWE-ZERO utilizes large-scale, execution-free trajectories to master code semantics and repository-level reasoning, and (2) SWE-HERO applies targeted, execution-backed refinement to transition these semantic intuitions into rigorous engineering workflows. Our empirical results set a new benchmark for open-source models of comparable size. We release a dataset of 300k SWE-ZERO and 13k SWE-HERO trajectories distilled from Qwen3-Coder-480B, alongside a suite of agents based on the Qwen2.5-Coder series. Notably, SWE-HERO-32B achieves a 62.2% resolution rate on SWE-bench Verified. Furthermore, despite being trained exclusively on Python, our agents demonstrate robust zero-shot transferability on SWE-bench Multilingual, reaching 44.1% and confirming the paradigm's generalizability across diverse languages.

Clone-and-own development produces families of related software variants that evolve independently. As variants diverge, important fixes applied in one repository are often missing in others. PaReco has shown that thousands of such missed opportunity (MO) patches exist across real ecosystems, yet its textual output provides limited support for understanding where and how these fixes should be propagated. We present MOVis, a lightweight, interactive desktop tool that visualizes MO patches between a source and target variant. MOVis loads PaReco's MO classifications and presents patched and buggy hunks side-by-side, highlighting corresponding regions and exposing structural differences that hinder reuse. This design enables developers to quickly locate missed fixes, understand required adaptations, and more efficiently maintain consistency across software variants. The tool, replication package, and demonstration video are available at https://zenodo.org/records/18356553 and https://youtu.be/Ac-gjBxHJ3Y.

High-performance GPU kernels are critical to modern machine learning systems, yet developing efficient implementations remains a challenging, expert-driven process due to the tight coupling between algorithmic structure, memory hierarchy usage, and hardware-specific optimizations. Recent work has explored using large language models (LLMs) to generate GPU kernels automatically, but generated implementations often struggle to maintain correctness and achieve competitive performance across iterative refinements. We present CuTeGen, an agentic framework for automated generation and optimization of GPU kernels that treats kernel development as a structured generate--test--refine workflow. Unlike approaches that rely on one-shot generation or large-scale search over candidate implementations, CuTeGen focuses on progressive refinement of a single evolving kernel through execution-based validation, structured debugging, and staged optimization. A key design choice is to generate kernels using the CuTe abstraction layer, which exposes performance-critical structures such as tiling and data movement while providing a more stable representation for iterative modification. To guide performance improvement, CuTeGen incorporates workload-aware optimization prompts and delayed integration of profiling feedback. Experimental results on matrix multiplication and activation workloads demonstrate that the framework produces functionally correct kernels and achieves competitive performance relative to optimized library implementations.

With the rise of personalized, persistent LLM agent frameworks such as OpenClaw, human-centered agentic social networks in which teams of collaborative AI agents serve individual users in a social network across multiple domains are becoming a reality. This setting creates novel privacy challenges: agents must coordinate across domain boundaries, mediate between humans, and interact with other users' agents, all while protecting sensitive personal information. While prior work has evaluated multi-agent coordination and privacy preservation, the dynamics and privacy risks of human-centered agentic social networks remain unexplored. To this end, we introduce AgentSocialBench, the first benchmark to systematically evaluate privacy risk in this setting, comprising scenarios across seven categories spanning dyadic and multi-party interactions, grounded in realistic user profiles with hierarchical sensitivity labels and directed social graphs. Our experiments reveal that privacy in agentic social networks is fundamentally harder than in single-agent settings: (1) cross-domain and cross-user coordination creates persistent leakage pressure even when agents are explicitly instructed to protect information, (2) privacy instructions that teach agents how to abstract sensitive information paradoxically cause them to discuss it more (we call it abstraction paradox). These findings underscore that current LLM agents lack robust mechanisms for privacy preservation in human-centered agentic social networks, and that new approaches beyond prompt engineering are needed to make agent-mediated social coordination safe for real-world deployment.

The topological gap $Δ= TP_{H_1}^{real} - TP_{H_1}^{shuf}$ -- the excess $H_1$ total persistence of the majority-spin alpha complex over a density-matched null -- encodes critical correlations in spin models. We establish finite-size scaling: $Δ(L,T) = A L^{d+η} G_-(L|t/T_c|)$, with $G_-(x) \sim (1+x/x_0)^{-(1+β/ν)}$. For 2D Ising, $α= 2.249 \pm 0.038$, matching $d+η= 9/4$ to $0.03σ$; the $G_-$ exponent $γ= 1.089 \pm 0.077$ is consistent with $1+β/ν= 9/8$ ($ΔR^2 < 10^{-5}$). For 2D Potts $q=3$ with $L$ up to 1024, $α= 2.272 \pm 0.024$ ($0.2σ$ from $d+η= 2.267$), with two-term corrections to scaling ($R^2 = 0.9999$). The $G_-$ exponent $γ= 1.114$ (68% CI $[1.053, 1.173]$) matches $1+β/ν= 17/15$. Scope boundaries: the law fails for 2D Potts $q=4$ ($α= 2.347 \pm 0.017$, $9.3σ$ from $d+η= 5/2$) where logarithmic corrections prevent convergence, and for raw 3D Ising ($4σ$ from $d+η$), but density normalization $Δ/|M|^{1/2}$ recovers $α= 3.06 \pm 0.04$ ($0.6σ$). The framework fails for first-order, BKT, and percolation. The criterion: $α= d+η$ holds when corrections to scaling are algebraic ($ω> 0$) but fails when logarithmic ($ω\to 0$).

The rapid evolution of autonomous, agentic artificial intelligence within financial services has introduced an existential architectural crisis: large language models (LLMs) are probabilistic, non-deterministic systems operating in domains that demand absolute, mathematically verifiable compliance guarantees. Existing guardrail solutions -- including NVIDIA NeMo Guardrails and Guardrails AI -- rely on probabilistic classifiers and syntactic validators that are fundamentally inadequate for enforcing complex multi-variable regulatory constraints mandated by the SEC, FINRA, and OCC. This paper presents the Lean-Agent Protocol, a formal-verification-based AI guardrail platform that leverages the Aristotle neural-symbolic model developed by Harmonic AI to auto-formalize institutional policies into Lean 4 code. Every proposed agentic action is treated as a mathematical conjecture: execution is permitted if and only if the Lean 4 kernel proves that the action satisfies pre-compiled regulatory axioms. This architecture provides cryptographic-level compliance certainty at microsecond latency, directly satisfying SEC Rule 15c3-5, OCC Bulletin 2011-12, FINRA Rule 3110, and CFPB explainability mandates. A three-phase implementation roadmap from shadow verification through enterprise-scale deployment is provided.

The development of robust clinical decision support systems is frequently impeded by the scarcity of high-fidelity, privacy-preserving biomedical data. While Generative Large Language Models (LLMs) offer a promising avenue for synthetic data generation, they often struggle to capture the complex, non-linear dependencies and severe class imbalances inherent in Electronic Health Records (EHR), leading to statistically plausible but clinically invalid records. To bridge this gap, we introduce DISCO-TAB (DIScriminator-guided COntrol for TABular synthesis), a novel framework that orchestrates a fine-tuned LLM with a multi-objective discriminator system optimized via Reinforcement Learning. Unlike prior methods relying on scalar feedback, DISCO-TAB evaluates synthesis at four granularities, token, sentence, feature, and row, while integrating Automated Constraint Discovery and Inverse-Frequency Reward Shaping to autonomously preserve latent medical logic and resolve minority-class collapse. We rigorously validate our framework across diverse benchmarks, including high-dimensional, small-sample medical datasets (e.g., Heart Failure, Parkinson's). Our results demonstrate that hierarchical feedback yields state-of-the-art performance, achieving up to 38.2% improvement in downstream clinical classifier utility compared to GAN and Diffusion baselines, while ensuring exceptional statistical fidelity (JSD < 0.01) and robust resistance to membership inference attacks. This work establishes a new standard for generating trustworthy, utility-preserving synthetic tabular data for sensitive healthcare applications.

Metasurface inverse design has become central to realizing complex optical functionality, yet translating target responses into executable, solver-compatible workflows still demands specialized expertise in computational electromagnetics and solver-specific software engineering. Recent large language models (LLMs) offer a complementary route to reducing this workflow-construction burden, but existing language-driven systems remain largely session-bounded and do not preserve reusable workflow knowledge across inverse-design tasks. We present an agentic framework for metasurface inverse design that addresses this limitation through context-level skill evolution. The framework couples a coding agent, evolving skill artifacts, and a deterministic evaluator grounded in physical simulation so that solver-specific strategies can be iteratively refined across tasks without modifying model weights or the underlying physics solver. We evaluate the framework on a benchmark spanning multiple metasurface inverse-design task types, with separate training-aligned and held-out task families. Evolved skills raise in-distribution task success from 38% to 74%, increase criteria pass fraction from 0.510 to 0.870, and reduce average attempts from 4.10 to 2.30. On held-out task families, binary success changes only marginally, but improvements in best margin together with shifts in error composition and agent behavior indicate partial transfer of workflow knowledge. These results suggest that the main value of skill evolution lies in accumulating reusable solver-specific expertise around reliable computational engines, thereby offering a practical path toward more autonomous and accessible metasurface inverse-design workflows.

Reinforcement learning (RL) and model predictive control (MPC) offer complementary strengths, yet combining them at scale remains computationally challenging. We propose soft MPCritic, an RL-MPC framework that learns in (soft) value space while using sample-based planning for both online control and value target generation. soft MPCritic instantiates MPC through model predictive path integral control (MPPI) and trains a terminal Q-function with fitted value iteration, aligning the learned value function with the planner and implicitly extending the effective planning horizon. We introduce an amortized warm-start strategy that recycles planned open-loop action sequences from online observations when computing batched MPPI-based value targets. This makes soft MPCritic computationally practical, while preserving solution quality. soft MPCritic plans in a scenario-based fashion with an ensemble of dynamic models trained for next-step prediction accuracy. Together, these ingredients enable soft MPCritic to learn effectively through robust, short-horizon planning on classic and complex control tasks. These results establish soft MPCritic as a practical and scalable blueprint for synthesizing MPC policies in settings where policy extraction and direct, long-horizon planning may fail.

Reinforcement learning for LLMs is vulnerable to reward hacking, where models exploit shortcuts to maximize reward without solving the intended task. We systematically study this phenomenon in coding tasks using an environment-manipulation setting, where models can rewrite evaluator code to trivially pass tests without solving the task, as a controlled testbed. Across both studied models, we identify a reproducible three-phase rebound pattern: models first attempt to rewrite the evaluator but fail, as their rewrites embed test cases their own solutions cannot pass. They then temporarily retreat to legitimate solving. When legitimate reward remains scarce, they rebound into successful hacking with qualitatively different strategies. Using representation engineering, we extract concept directions for shortcut, deception, and evaluation awareness from domain-general contrastive pairs and find that the shortcut direction tracks hacking behavior most closely, making it an effective representational proxy for detection. Motivated by this finding, we propose Advantage Modification, which integrates shortcut concept scores into GRPO advantage computation to penalize hacking rollouts before policy updates. Because the penalty is internalized into the training signal rather than applied only at inference time, Advantage Modification provides more robust suppression of hacking compared with generation-time activation steering.

Adapting closed-box service models (i.e., APIs) for target tasks typically relies on reprogramming via Zeroth-Order Optimization (ZOO). However, this standard strategy is known for extensive, costly API calls and often suffers from slow, unstable optimization. Furthermore, we observe that this paradigm faces new challenges with modern APIs (e.g., GPT-4o). These models can be less sensitive to the input perturbations ZOO relies on, thereby hindering performance gains. To address these limitations, we propose an Alternative efficient Reprogramming approach for Service models (AReS). Instead of direct, continuous closed-box optimization, AReS initiates a single-pass interaction with the service API to prime an amenable local pre-trained encoder. This priming stage trains only a lightweight layer on top of the local encoder, making it highly receptive to the subsequent glass-box (white-box) reprogramming stage performed directly on the local model. Consequently, all subsequent adaptation and inference rely solely on this local proxy, eliminating all further API costs. Experiments demonstrate AReS's effectiveness where prior ZOO-based methods struggle: on GPT-4o, AReS achieves a +27.8% gain over the zero-shot baseline, a task where ZOO-based methods provide little to no improvement. Broadly, across ten diverse datasets, AReS outperforms state-of-the-art methods (+2.5% for VLMs, +15.6% for standard VMs) while reducing API calls by over 99.99%. AReS thus provides a robust and practical solution for adapting modern closed-box models.

Large Language Models (LLMs) are powerful tools for answering user queries, yet they remain highly vulnerable to jailbreak attacks. Existing guardrail methods typically rely on internal features or textual responses to detect malicious queries, which either introduce substantial latency or suffer from the randomness in text generation. To overcome these limitations, we propose SelfGrader, a lightweight guardrail method that formulates jailbreak detection as a numerical grading problem using token-level logits. Specifically, SelfGrader evaluates the safety of a user query within a compact set of numerical tokens (NTs) (e.g., 0-9) and interprets their logit distribution as an internal safety signal. To align these signals with human intuition of maliciousness, SelfGrader introduces a dual-perspective scoring rule that considers both the maliciousness and benignness of the query, yielding a stable and interpretable score that reflects harmfulness and reduces the false positive rate simultaneously. Extensive experiments across diverse jailbreak benchmarks, multiple LLMs, and state-of-the-art guardrail baselines demonstrate that SelfGrader achieves up to a 22.66% reduction in ASR on LLaMA-3-8B, while maintaining significantly lower memory overhead (up to 173x) and latency (up to 26x).

The Muon optimizer has received considerable attention for its strong performance in training large language models, yet the design principle behind its matrix-gradient orthogonalization remains largely elusive. In this paper, we introduce a surrogate model that not only sheds new light on the design of Muon, but more importantly leads to a new optimizer. In the same spirit as the derivation of Newton's method, the surrogate approximates the loss as a quadratic function of the perturbation to a weight matrix $W$ using only three matrices: the gradient $G$, an output-space curvature matrix $H$, and the data matrix $Z$ that stacks the layer inputs. By minimizing this surrogate in one step and adopting a certain isotropic assumption on the weights, we obtain the closed-form update rule (up to momentum and weight decay) $W \leftarrow W - η\cdot \mathrm{msgn}(G(ZZ^\top)^{-1})$, where $η$ is the learning rate and $\mathrm{msgn}(X)=UV^\top$ if $X=USV^\top$ is a compact singular value decomposition. This new optimization method, which we refer to as Newton-Muon, shows that standard Muon can be interpreted as an implicit Newton-type method that neglects the right preconditioning induced by the input second moment. Empirically, on a reproduction of the earliest publicly released Modded-NanoGPT speedrun configuration using Muon for GPT-2 pretraining, Newton-Muon reaches the target validation loss in 6\% fewer iteration steps and reduces wall-clock training time by about 4\%.

Associative memory systems enable content-addressable storage and retrieval of patterns, a capability central to biological neural computation and artificial intelligence. Classical implementations such as Hopfield networks face fundamental limitations in memory capacity, scaling at most linearly with network size. We present an associative memory architecture based on Kuramoto oscillator networks with honeycomb topology in which memories are encoded as stable phase-locked configurations. The honeycomb network consists of multiple cycles that share nodes in a chain-like arrangement, creating a one-dimensional lattice of chained+loops. We prove that this architecture achieves exponential memory capacity: a network of $N$ oscillators can store $(2\lceil n_c/4 \rceil - 1)^m$ distinct patterns, where $m$ honeycomb cycles each contain $n_c$ oscillators. Moreover, we fully characterize all stable configurations and prove that each memory's basin of attraction maintains a guaranteed minimum size independent of network scale. Simulations using charge-density-wave (CDW) oscillators validate predicted phase-locking behavior, demonstrating practical realizability in neuromorphic hardware.

Persian poetry is often remembered through recurrent symbols before it is remembered through plot. Wine vessels, gardens, flames, sacred titles, bodily beauty, and courtly names return across centuries, yet computational work still tends to flatten this material into isolated words or broad document semantics. That misses a practical unit of organization in Persian poetics: related forms travel as families and gain force through recurring relations. Using a corpus of 129,451 poems, we consolidate recurrent forms into traceable families, separate imagistic material from sacred and courtly reference, and map their relations in a multi-layer graph. The symbolic core is relatively sparse, the referential component much denser, and the attachment zone between them selective rather than diffuse. Across 11 Hijri-century bins, some families remain widely distributed, especially Shab (Night), Ruz (Day), and Khaak (Earth). Wine vessels, garden space, flame, and lyric sound strengthen later, while prestige-coded and heroic-courtly vocabulary is weighted earlier. Century-specific graphs show change in arrangement as well as membership. Modularity rises, cross-scope linkage declines, courtly bridges weaken, and sacred bridges strengthen. Hub positions shift too: Kherqe (Sufi Robe) gains late prominence, Farkhondeh {Blessed} and Banafsheh (Violet) recede, and Saaghar (Wine Cup) stays central across the chronology. In this corpus, Persian symbolism appears less as a fixed repertory than as a long-lived system whose internal weights and connections change over time.

Accurately modeling agent behaviors is an important task in self-driving. It is also a task with many symmetries, such as equivariance to the order of agents and objects in the scene or equivariance to arbitrary roto-translations of the entire scene as a whole; i.e., SE(2)-equivariance. The transformer architecture is a ubiquitous tool for modeling these symmetries. While standard self-attention is inherently permutation equivariant, explicit pairwise relative positional encodings have been the standard for introducing SE(2)-equivariance. However, this approach introduces an additional cost that is quadratic in the number of agents, limiting its scalability to larger scenes and batch sizes. In this work, we propose DriveGATr, a novel transformer-based architecture for agent modeling that achieves SE(2)-equivariance without the computational cost of existing methods. Inspired by recent advances in geometric deep learning, DriveGATr encodes scene elements as multivectors in the 2D projective geometric algebra $\mathbb{R}^*_{2,0,1}$ and processes them with a stack of equivariant transformer blocks. Crucially, DriveGATr models geometric relationships using standard attention between multivectors, eliminating the need for costly explicit pairwise relative positional encodings. Experiments on the Waymo Open Motion Dataset demonstrate that DriveGATr is comparable to the state-of-the-art in traffic simulation and establishes a superior Pareto front for performance vs computational cost.

Physically Assistive Robots (PARs) require personalized behaviors to ensure user safety and comfort. However, traditional preference learning methods, like exhaustive pairwise comparisons, cause severe physical and cognitive fatigue for users with profound motor impairments. To solve this, we propose a low-burden, offline framework that translates unstructured natural language feedback directly into deterministic robotic control policies. To safely bridge the gap between ambiguous human speech and robotic code, our pipeline uses Large Language Models (LLMs) grounded in the Occupational Therapy Practice Framework (OTPF). This clinical reasoning decodes subjective user reactions into explicit physical and psychological needs, which are then mapped into transparent decision trees. Before deployment, an automated "LLM-as-a-Judge" verifies the code's structural safety. We validated this system in a simulated meal preparation study with 10 adults with paralysis. Results show our natural language approach significantly reduces user workload compared to traditional baselines. Additionally, independent clinical experts confirmed the generated policies are safe and accurately reflect user preferences.

Reducing hallucinations in Large Language Models (LLMs) is essential for improving the accuracy of data extraction from large text corpora. Current methods, like prompt engineering and chain-of-thought prompting, focus on individual documents but fail to consider relationships across a corpus. This paper introduces Peer Context Outlier Detection (P-COD), a novel approach that uses the relationships between documents to improve extraction accuracy. Our application domain is in scientific literature summarization, where papers with similar experiment settings should draw similar conclusions. By comparing extracted data to validated peer information within the corpus, we adjust confidence scores and flag low-confidence results for expert review. High-confidence results, supported by peer validation, are considered reliable. Our experiments demonstrate up to 98% precision in outlier detection across 6 domains of science, demonstrating that our design reduces hallucinations, enhances trust in automated systems, and allows researchers to focus on ambiguous cases, streamlining the data extraction workflows.

Large language models are often not just wrong, but \emph{confidently wrong}: when they produce factually incorrect answers, they tend to verbalize overly high confidence rather than signal uncertainty. Such verbalized overconfidence can mislead users and weaken confidence scores as a reliable uncertainty signal, yet its internal mechanisms remain poorly understood. We present a circuit-level mechanistic analysis of this inflated verbalized confidence in LLMs, organized around three axes: capturing verbalized confidence as a differentiable internal signal, identifying the circuits that causally inflate it, and leveraging these insights for targeted inference-time recalibration. Across two instruction-tuned LLMs on three datasets, we find that a compact set of MLP blocks and attention heads, concentrated in middle-to-late layers, consistently writes the confidence-inflation signal at the final token position. We further show that targeted inference-time interventions on these circuits substantially improve calibration. Together, our results suggest that verbalized overconfidence in LLMs is driven by identifiable internal circuits and can be mitigated through targeted intervention.

Large language models (LLMs) are increasingly explored for NP-hard combinatorial optimization problems, but most existing methods emphasize feasible-instance solution generation and do not explicitly address infeasibility detection. We propose an infeasibility-aware framework that combines certifiable dataset construction, supervised fine-tuning, and LLM-assisted downstream search. For the minor-embedding problem, we introduce a new mathematical programming formulation together with provable zero-phase infeasibility screening, which enables scalable construction of training instances labeled either as feasible with structured certificates or as certifiably infeasible. Using training data generated through this exact optimization pipeline, we show that an 8B-parameter LLM can be fine-tuned to jointly perform solution generation and infeasibility detection. We further utilize LLM outputs as warm starts for downstream local search, providing a practical way to accelerate optimization even when the LLM outputs are imperfect. Experiments show that our fine-tuned model improves overall accuracy by up to 30\% over GPT-5.2; meanwhile LLM-guided warm starts provide up to $2\times$ speedup compared with starting from scratch in downstream local search.

Scientific discovery is slowed by fragmented literature that requires excessive human effort to gather, analyze, and understand. AI tools, including autonomous summarization and question answering, have been developed to aid in understanding scientific literature. However, these tools lack the structured, multi-step approach necessary for extracting deep insights from scientific literature. Large Language Models (LLMs) offer new possibilities for literature analysis, but remain unreliable due to hallucinations and incomplete extraction. We introduce Elhuyar, a multi-agent, human-in-the-loop system that integrates LLMs, structured AI, and human scientists to extract, analyze, and iteratively refine insights from scientific literature. The framework distributes tasks among specialized agents for filtering papers, extracting data, fitting models, and summarizing findings, with human oversight ensuring reliability. The system generates structured reports with extracted data, visualizations, model equations, and text summaries, enabling deeper inquiry through iterative refinement. Deployed in materials science, it analyzed literature on tungsten under helium-ion irradiation, showing experimentally correlated exponential helium bubble growth with irradiation dose and temperature, offering insight for plasma-facing materials (PFMs) in fusion reactors. This demonstrates how AI-assisted literature review can uncover scientific patterns and accelerate discovery.

Artificial intelligence (AI) systems are increasingly integrated into healthcare and pharmacy workflows, supporting tasks such as medication recommendations, dosage determination, and drug interaction detection. While these systems often demonstrate strong performance under standard evaluation metrics, their reliability in real-world decision-making remains insufficiently understood. In high-risk domains such as medication management, even a single incorrect recommendation can result in severe patient harm. This paper examines the reliability of AI-assisted medication systems by focusing on system failures and their potential clinical consequences. Rather than evaluating performance solely through aggregate metrics, this work shifts attention towards how errors occur and what happens when AI systems produce incorrect outputs. Through a series of controlled, simulated scenarios involving drug interactions and dosage decisions, we analyse different types of system failures, including missed interactions, incorrect risk flagging, and inappropriate dosage recommendations. The findings highlight that AI errors in medication-related contexts can lead to adverse drug reactions, ineffective treatment, or delayed care, particularly when systems are used without sufficient human oversight. Furthermore, the paper discusses the risks of over-reliance on AI recommendations and the challenges posed by limited transparency in decision-making processes. This work contributes a reliability-focused perspective on AI evaluation in healthcare, emphasising the importance of understanding failure behavior and real-world impact. It highlights the need to complement traditional performance metrics with risk-aware evaluation approaches, particularly in safety-critical domains such as pharmacy practice.

Recent 3D Gaussian splatting methods built atop SMPL achieve remarkable visual fidelity while continually increasing the complexity of the overall training architecture. We demonstrate that much of this complexity is unnecessary: by replacing SMPL with the Momentum Human Rig (MHR), estimated via SAM-3D-Body, a minimal pipeline with no learned deformations or pose-dependent corrections achieves the highest reported PSNR and competitive or superior LPIPS and SSIM on PeopleSnapshot and ZJU-MoCap. To disentangle pose estimation quality from body model representational capacity, we perform two controlled ablations: translating SAM-3D-Body meshes to SMPL-X, and translating the original dataset's SMPL poses into MHR both retrained under identical conditions. These ablations confirm that body model expressiveness has been a primary bottleneck in avatar reconstruction, with both mesh representational capacity and pose estimation quality contributing meaningfully to the full pipeline's gains.

When does consulting one information source raise the value of another, and when does it diminish it? We study this question for Bayesian decision-makers facing finite actions. The interaction decomposes into two opposing forces: a complement force, measuring how one source moves beliefs to where the other becomes more useful, and a substitute force, measuring how much the current decision is resolved. Their balance obeys a localization principle: substitution requires an observation to cross a decision boundary, though crossing alone does not guarantee it. Whenever posteriors remain inside the current decision region, the substitute force vanishes, and sources are guaranteed to complement each other, even when one source cannot, on its own, change the decision. The results hold for arbitrarily correlated sources and are formalized in Lean 4. Substitution is confined to the thin boundaries where decisions change. Everywhere else, information cooperates. Code and proofs: https://github.com/nidhishs/all-substitution-is-local.

Modern generator-based fuzzing techniques combine lightweight input generators with coverage-guided mutation as a method of exploring deep execution paths in a target program. A complimentary approach in prior research focuses on creating highly customized, domain-specific generators that encode structural and semantic logic sufficient enough to reach deep program states; the challenge comes from the overhead of writing and testing these complex generators. We investigate whether AI coding agents can automatically synthesize such target-specific generators, and whether the resulting generators are strong enough to obviate the need for coverage guidance and mutation entirely. Our approach, Gentoo, is comprised of an LLM coding agent (provided terminal access and source code of the fuzz target and its library) instructed to iteratively synthesize and refine an input generator, and optionally provided fine-grained predicate-level coverage feedback. We evaluate three configurations of Gentoo against human-written generators on fuzz targets for 7 real-world Java libraries. Our findings show that agent-synthesized generators achieve statistically significantly higher branch coverage than human-written baseline generators on 4 of 7 benchmarks. Critically, the use of coverage guidance and mutation strategies is not statistically significantly beneficial for agent-synthesized generators, but is significant for all human-written generators, suggesting that structural and semantic logic encoded in the agent generators makes coverage guidance largely unnecessary.

Modern real-time systems require accurate characterization of task timing behavior to ensure predictable performance, particularly on complex hardware architectures. Existing methods, such as worst-case execution time analysis, often fail to capture the fine-grained timing behaviors of a task under varying resource contexts (e.g., an allocation of cache, memory bandwidth, and CPU frequency), which is necessary to achieve efficient resource utilization. In this paper, we introduce a novel generative profiling approach that synthesizes context-dependent, fine-grained timing profiles for real-time tasks, including those for unmeasured resource allocations. Our approach leverages a nonparametric, conditional multi-marginal Schrödinger Bridge (MSB) formulation to generate accurate execution profiles for unseen resource contexts, with maximum likelihood guarantees. We demonstrate the efficiency and effectiveness of our approach through real-world benchmarks, and showcase its practical utility in a representative case study of adaptive multicore resource allocation for real-time systems.

The shift toward IoT-enabled, sensor-driven systems has transformed how operational data is generated, favoring continuous, real-time event streams (ES) over static event logs. This evolution presents new challenges for Streaming Process Mining (SPM), which must cope with out-of-order events, concurrent activities, incomplete cases, and concept drifts. Yet, the evaluation of SPM algorithms remains rooted in outdated practices, relying on static logs or artificially streamified data that fail to reflect the complexities of real-world streams. To address this gap, we first perform a comprehensive review of data stream literature to identify stream characteristics currently not reflected in the SPM community. Next, we use this information to extend the conceptual foundation for ES. Finally, we propose Stream of Intent, a prototype generator to produce ES with specific features. Our evaluation shows excellence in producing reproducible, intentional ES for targeted benchmarking and adaptive algorithm development in SPM.

Personal AI agents like OpenClaw run with elevated privileges on users' local machines, where a single successful prompt injection can leak credentials, redirect financial transactions, or destroy files. This threat goes well beyond conventional text-level jailbreaks, yet existing safety evaluations fall short: most test models in isolated chat settings, rely on synthetic environments, and do not account for how the agent framework itself shapes safety outcomes. We introduce CLAWSAFETY, a benchmark of 120 adversarial test scenarios organized along three dimensions (harm domain, attack vector, and harmful action type) and grounded in realistic, high-privilege professional workspaces spanning software engineering, finance, healthcare, law, and DevOps. Each test case embeds adversarial content in one of three channels the agent encounters during normal work: workspace skill files, emails from trusted senders, and web pages. We evaluate five frontier LLMs as agent backbones, running 2,520 sandboxed trials across all configurations. Attack success rates (ASR) range from 40\% to 75\% across models and vary sharply by injection vector, with skill instructions (highest trust) consistently more dangerous than email or web content. Action-trace analysis reveals that the strongest model maintains hard boundaries against credential forwarding and destructive actions, while weaker models permit both. Cross-scaffold experiments on three agent frameworks further demonstrate that safety is not determined by the backbone model alone but depends on the full deployment stack, calling for safety evaluation that treats model and framework as joint variables.

With the advancement of Agentic AI, researchers are increasingly leveraging autonomous agents to address challenges in software engineering (SE). However, the large language models (LLMs) that underpin these agents often function as black boxes, making it difficult to justify the superiority of Agentic AI approaches over baselines. Furthermore, missing information in the evaluation design description frequently renders the reproduction of results infeasible. To synthesize current evaluation practices for Agentic AI in SE, this study analyzes 18 papers on the topic, published or accepted by ICSE 2026, ICSE 2025, FSE 2025, ASE 2025, and ISSTA 2025. The analysis identifies prevailing approaches and their limitations in evaluating Agentic AI for SE, both in current research and potential future studies. To address these shortcomings, this position paper proposes a set of guidelines and recommendations designed to empower reproducible, explainable, and effective evaluations of Agentic AI in software engineering. In particular, we recommend that Agentic AI researchers make their Thought-Action-Result (TAR) trajectories and LLM interaction data, or summarized versions of these artifacts, publicly accessible. Doing so will enable subsequent studies to more effectively analyze the strengths and weaknesses of different Agentic AI approaches. To demonstrate the feasibility of such comparisons, we present a proof-of-concept case study that illustrates how TAR trajectories can support systematic analysis across approaches.

Partially observable Markov decision processes (POMDPs) offer a principled formalism for planning under state and transition uncertainty. Despite advances made towards solving large POMDPs, obtaining performant policies under limited planning time remains a major challenge due to the curse of dimensionality and the curse of history. For many POMDP problems, the value of information (VOI) - the expected performance gain from reasoning about observations - varies over the belief space. We introduce a dynamic programming framework that exploits this structure by conditionally processing observations based on the value of information at each belief. Building on this framework, we propose Value of Information Monte Carlo planning (VOIMCP), a Monte Carlo Tree Search algorithm that allocates computational effort more efficiently by selectively disregarding observation information when the VOI is low, avoiding unnecessary branching of observations. We provide theoretical guarantees on the near-optimality of our VOI reasoning framework and derive non-asymptotic convergence bounds for VOIMCP. Simulation evaluations demonstrate that VOIMCP outperforms baselines on several POMDP benchmarks.

Current limitations in wireless modeling and radio frequency (RF)-based AI are primarily driven by a lack of high-quality, measurement-based datasets that connect RF signals to their physical environments. RF heatmaps, the typical form of such data, are high-dimensional and complex but lack the geometric and semantic context needed for interpretation, constraining the development of supervised machine learning models. To address this bottleneck, we propose a new class of multimodal datasets that combines RF measurements with auxiliary modalities like high-resolution cameras and lidar to bridge the gap between RF signals and their physical causes. The proposed data collection will span diverse indoor and outdoor environments, featuring both static and dynamic scenarios, including human activities ranging from walking to subtle gestures. By achieving precise spatial and temporal co-registration and creating digital replicas for voxel-level annotation, this dataset will enable transformative AI research. Key tasks include the forward problem of predicting RF heatmaps from visual data to revolutionize wireless system design, and the inverse problem of inferring scene semantics from RF signals, creating a new form of RF-based perception.

Citation granularity - whether to cite individual sentences, paragraphs, or documents - is a critical design choice in attributed generation. While fine-grained citations are often preferred for precise human verification, their impact on model performance remains under-explored. We analyze four model scales (8B-120B) and demonstrate that enforcing fine-grained citations degrades attribution quality by 16-276% compared to the best-performing granularity. We observe a consistent performance pattern where attribution quality peaks at intermediate granularities (paragraph-level). Our analysis suggests that fine-grained (sentence-level) citations disrupt necessary semantic dependencies for attributing evidence to answer claims, while excessively coarse citations (multi-paragraph) introduce distracting noise. Importantly, the magnitude of this performance gap varies non-monotonically with model scale: fine-grained constraints disproportionately penalize larger models, suggesting that atomic citation units disrupt the multi-sentence information synthesis at which these models excel. Strikingly, citation-optimal granularity leads to substantial gains in attribution quality while preserving or even improving answer correctness. Overall, our findings demonstrate that optimizing solely for human verification via fine-grained citation disregards model constraints, compromising both attribution faithfulness and generation reliability. Instead, effective attribution requires aligning citation granularity with the model's natural semantic scope.

Language Models (LMs) exhibit two distinct mechanisms for knowledge acquisition: in-weights learning (i.e., encoding information within the model weights) and in-context learning (ICL). Although these two modes offer complementary strengths, in-weights learning frequently struggles to facilitate deductive reasoning over the internalized knowledge. We characterize this limitation as a deficit in latent generalization, of which the reversal curse is one example. Conversely, in-context learning demonstrates highly robust latent generalization capabilities. To improve latent generalization from in-weights knowledge, prior approaches rely on train-time data augmentation, yet these techniques are task-specific, scale poorly, and fail to generalize to out-of-distribution knowledge. To overcome these shortcomings, this work studies how models can be taught to use test-time compute, or 'thinking', specifically to improve latent generalization. We use Reinforcement Learning (RL) from correctness feedback to train models to produce long chains-of-thought (CoTs) to improve latent generalization. Our experiments show that this thinking approach not only resolves many instances of latent generalization failures on in-distribution knowledge but also, unlike augmentation baselines, generalizes to new knowledge for which no RL was performed. Nevertheless, on pure reversal tasks, we find that thinking does not unlock direct knowledge inversion, but the generate-and-verify ability of thinking models enables them to get well above chance performance. The brittleness of factual self-verification means thinking models still remain well below the performance of in-context learning for this task. Overall, our results establish test-time thinking as a flexible and promising direction for improving the latent generalization of LMs.

This paper develops a distributed variational quantum algorithm for solving large-scale linear equations. For a linear system of the form $Ax=b$, the large square matrix $A$ is partitioned into smaller square block submatrices, each of which is known only to a single noisy intermediate-scale quantum (NISQ) computer. Each NISQ computer communicates with certain other quantum computers in the same row and column of the block partition, where the communication patterns are described by the row- and column-neighbor graphs, both of which are connected. The proposed algorithm integrates a variant of the variational quantum linear solver at each computer with distributed classical optimization techniques. The derivation of the quantum cost function provides insight into the design of the distributed algorithm. Numerical quantum simulations demonstrate that the proposed distributed quantum algorithm can solve linear systems whose size scales with the number of computers and is therefore not limited by the capacity of a single quantum computer.

Synonymy is a widespread yet puzzling linguistic phenomenon. Absolute synonyms theoretically should not exist, as they do not expand language's expressive potential. However, it was suggested that even if synonyms denote the same concept, they may reflect different perspectives or carry distinct cultural associations, claims that have rarely been tested quantitatively. In Hindi, prolonged contact with Persian produced many Perso-Arabic loanwords coexisting with their Sanskrit counterpart, forming numerous synonym pairs. This study investigates whether centuries after these borrowings appeared in the Subcontinent their origin can still be distinguished using distributional data alone and regardless of their semantic content. A Random Forest trained on word embeddings of Hindi synonyms successfully classified words by Sanskrit or Perso-Arabic origin, even when they were semantically unrelated, suggesting that usage patterns preserve traces of etymology. These findings provide quantitative evidence that context encodes etymological signals and that synonymy may reflect subtle but systematic distinctions linked to origin. They support the idea that synonymous words can offer different perspectives and that etymologically related words may form distinct conceptual subspaces, creating a new type of semantic frame shaped by historical origin. Overall, the results highlight the power of context in capturing nuanced distinctions beyond traditional semantic similarity.

Thousands of diverse benchmarks have been developed to measure the quality of large language models (LLMs). Yet prior work has demonstrated that LLM performance is often sufficiently explained by a small set of latent factors, or abilities. This suggests the potential for more efficient and principled benchmarking, but it remains difficult to compare the quality of different methods. Motivated by predictive validity, we argue that the quality of a benchmarking framework should be grounded in how efficiently it enables the prediction of model performance on unseen tasks. To analyze this objective, we collect the "Wide-scale Item Level Dataset" (WILD), a dataset of item-model response pairs, comprising evaluations of 65 models on 109,564 unique items spanning 163 tasks drawn from 27 datasets. This dataset enables the first analysis of how different techniques can predict a model's performance on a large, diverse collection of unseen tasks under different budget constraints. We demonstrate that combining a modified multidimensional item response theory (IRT) model with adaptive item selection driven by optimal experimental design can predict performance on 112 held-out benchmark tasks with a mean absolute error (MAE) of less than 7%, and can do so after observing only 16 items. We further demonstrate that incorporating cost-aware discount factors into our selection criteria can reduce the total tokens needed to reach 7% MAE from 141,000 tokens to only 22,000, an 85% reduction in evaluation cost.

We present ReFormeR, a pattern-guided approach for query reformulation. Instead of prompting a language model to generate reformulations of a query directly, ReFormeR first elicits short reformulation patterns from pairs of initial queries and empirically stronger reformulations, consolidates them into a compact library of transferable reformulation patterns, and then selects an appropriate reformulation pattern for a new query given its retrieval context. The selected pattern constrains query reformulation to controlled operations such as sense disambiguation, vocabulary grounding, or discriminative facet addition, to name a few. As such, our proposed approach makes the reformulation policy explicit through these reformulation patterns, guiding the LLM towards targeted and effective query reformulations. Our extensive experiments on TREC DL 2019, DL 2020, and DL Hard show consistent improvements over classical feedback methods and recent LLM-based query reformulation and expansion approaches.

Large Language Models (LLMs) increasingly rely on multi-turn reasoning and interaction, such as adaptive retrieval-augmented generation (RAG) and ReAct-style agents, to answer difficult questions. These methods improve accuracy by iteratively retrieving information, reasoning, or acting, but introduce a key challenge: \textbf{When should the model stop?} Existing approaches rely on heuristic stopping rules or fixed turn budgets and provide no formal guarantees that the final prediction still contains the correct answer. This limitation is particularly problematic in high-stakes domains such as finance and healthcare, where unnecessary turns increase cost and latency, while stopping too early risks incorrect decisions. Conformal prediction (CP) provides formal coverage guarantees, but existing LLM-CP methods only apply to a single model output and cannot handle multi-turn pipelines with adaptive stopping. To address this gap, we propose Multi-Turn Language Models with Conformal Prediction (MiCP), the first CP framework for multi-turn reasoning. MiCP allocates different error budgets across turns, enabling the model to stop early while maintaining an overall coverage guarantee. We demonstrate MiCP on adaptive RAG and ReAct, where it achieves the target coverage on both single-hop and multi-hop question answering benchmarks while reducing the number of turns, inference cost, and prediction set size. We further introduce a new metric that jointly evaluates coverage validity and answering efficiency.

Modern LLMs scale at test-time, e.g. via repeated sampling, where inference cost grows with model size and the number of samples. This creates a trade-off that pretraining scaling laws, such as Chinchilla, do not address. We present Train-to-Test ($T^2$) scaling laws that jointly optimize model size, training tokens, and number of inference samples under fixed end-to-end budgets. $T^2$ modernizes pretraining scaling laws with pass@$k$ modeling used for test-time scaling, then jointly optimizes pretraining and test-time decisions. Forecasts from $T^2$ are robust over distinct modeling approaches: measuring joint scaling effect on the task loss and modeling impact on task accuracy. Across eight downstream tasks, we find that when accounting for inference cost, optimal pretraining decisions shift radically into the overtraining regime, well-outside of the range of standard pretraining scaling suites. We validate our results by pretraining heavily overtrained models in the optimal region that $T^2$ scaling forecasts, confirming their substantially stronger performance compared to pretraining scaling alone. Finally, as frontier LLMs are post-trained, we show that our findings survive the post-training stage, making $T^2$ scaling meaningful in modern deployments.

Text production (and translations) proceeds in the form of stretches of typing, interrupted by keystroke pauses. It is often assumed that fast typing reflects unchallenged/automated translation production while long(er) typing pauses are indicative of translation problems, hurdles or difficulties. Building on a long discussion concerning the determination of pause thresholds that separate automated from presumably reflective translation processes (O'Brien, 2006; Alves and Vale, 2009; Timarova et al., 2011; Dragsted and Carl, 2013; Lacruz et al., 2014; Kumpulainen, 2015; Heilmann and Neumann 2016), this paper compares three recent approaches for computing these pause thresholds, and suggest and evaluate a novel method for computing Production Unit Breaks.

We study the problem of identifying an optimal coupling between input-output distributional data generated by a causal dynamical system. The coupling is required to satisfy prescribed marginal distributions and a causality constraint reflecting the temporal structure of the system. We formulate this problem as a Schr"odinger Bridge, which seeks the coupling closest - in Kullback-Leibler divergence - to a given prior while enforcing both marginal and causality constraints. For the case of Gaussian marginals and general time-dependent quadratic cost functions, we derive a fully tractable characterization of the Sinkhorn iterations that converges to the optimal solution. Beyond its theoretical contribution, the proposed framework provides a principled foundation for applying causal optimal transport methods to system identification from distributional data.

Language models can answer many entity-centric factual questions, but it remains unclear which internal mechanisms are involved in this process. We study this question across multiple language models. We localize entity-selective MLP neurons using templated prompts about each entity, and then validate them with causal interventions on PopQA-based QA examples. On a curated set of 200 entities drawn from PopQA, localized neurons concentrate in early layers. Negative ablation produces entity-specific amnesia, while controlled injection at a placeholder token improves answer retrieval relative to mean-entity and wrong-cell controls. For many entities, activating a single localized neuron is sufficient to recover entity-consistent predictions once the context is initialized, consistent with compact entity retrieval rather than purely gradual enrichment across depth. Robustness to aliases, acronyms, misspellings, and multilingual forms supports a canonicalization interpretation. The effect is strong but not universal: not every entity admits a reliable single-neuron handle, and coverage is higher for popular entities. Overall, these results identify sparse, causally actionable access points for analyzing and modulating entity-conditioned factual behavior.

This report is part of the Qumphy project (22HLT01 Qumphy) that is funded by the European Union and is dedicated to the development of measures to quantify the uncertainties associated with Machine Learning algorithms applied to medical problems, in particular the analysis and processing of Photoplethysmography (PPG) signals. In this report, a list of six medical problems that are related to PPG signals and serve as Benchmark Problems is given. Suitable Benchmark datasets and their usage are described also.

This paper introduces EXaCTz, a parallel algorithm that concurrently preserves extremum graphs and contour trees in lossy-compressed scalar field data. While error-bounded lossy compression is essential for large-scale scientific simulations and workflows, existing topology-preserving methods suffer from (1) a significant throughput disparity, where topology correction speeds are on the order of MB/s, lagging orders of magnitude behind compression speeds on the order of GB/s, (2) limited support for diverse topological descriptors, and (3) a lack of theoretical convergence bounds. To address these challenges, EXaCTz introduces a high-performance, bounded-iteration algorithm that enforces topological consistency by deriving targeted edits for decompressed data. Unlike prior methods that rely on explicit topology reconstruction, EXaCTz enforces consistent min/max neighbors of all vertices, along with global ordering among critical points. As such, the algorithm enforces consistent critical-point classification, saddle extremum connectivity, and the preservation of merge/split events. We theoretically prove the convergence of our algorithm, bounded by the longest path in a vulnerability graph that characterizes potential cascading effects during correction. Experiments on real-world datasets show that EXaCTz achieves a single-GPU throughput of up to 4.52 GB/s, outperforming the state-of-the-art contour-tree-preserving method (Gorski et al.) by up to 213x (with a single-core CPU implementation for fair comparison) and 3,285x (with a single-GPU version). In distributed environments, EXaCTz scales to 128 GPUs with 55.6\% efficiency (compared with 6.4\% for a naive parallelization), processing datasets of up to 512 GB in under 48 seconds and achieving an aggregate correction throughput of up to 32.69 GB/s.

As AI becomes more common in students' everyday experiences, a major challenge for K-12 AI education is designing learning experiences that can be meaningfully integrated into existing subject-area instruction. This paper presents the design and implementation of an AI4K12-aligned curriculum that embeds AI learning goals within a rural middle school science classroom using Breadth-First Search (BFS) as an accessible entry point to AI problem-solving. Through unplugged activities and an interactive simulation environment, students learned BFS as a strategy for exploring networks and identifying shortest paths, then applied it to science contexts involving virus spread and contact tracing. To examine engagement and learning, we analyzed pre- and post-assessments, student work artifacts, and a teacher interview. Results suggest that students engaged productively with the curriculum, improved their understanding of BFS and AI problem-solving, and benefited from learning these ideas within ongoing science instruction. Teacher feedback further indicated that the module fit well within the science curriculum while supporting intended science learning outcomes. We conclude with curriculum and design considerations for broadening access to learning about problem-solving with AI in education.

Retrieval-augmented generation (RAG) systems are gaining traction in enterprise settings, yet stringent data protection regulations prevent many organizations from using cloud-based services, necessitating on-premises deployments. While existing blueprints and reference architectures focus on cloud deployments and lack enterprise-grade components, comprehensive on-premises implementation frameworks remain scarce. This paper aims to address this gap by presenting a comprehensive AI engineering blueprint for scalable on-premises enterprise RAG solutions. It is designed to address common challenges and streamline the integration of RAG into existing enterprise infrastructure. The blueprint provides: (1) an end-to-end reference architecture described using the 4+1 view model, (2) a reference application for on-premises deployment, and (3) best practices for tooling, development, and CI/CD pipelines, all publicly available on GitHub. Ongoing case studies and expert interviews with industry partners will assess its practical benefits.

Identifying the features to be released in the next version of software, from a pool of potential candidates, is a challenging problem. User feedback from app stores is frequently used by software vendors for the evolution of apps across releases. Privacy feedback, although smaller in volume, carries a larger impact influencing app's success. Multiple existing work has focused on summarizing privacy concerns at the app level and has also shown that developers utilize feedback to implement security and privacy-related changes in subsequent releases. However, the current literature offers little support for release managers and developers in identifying privacy concerns prior to release. This gap exists as user reviews are typically available in app stores only after new features of a software system is released. In this paper, we introduce Pre-PI, a novel approach that summarizes privacy concerns for to-be-released features. Our method first maps existing features to semantically similar privacy reviews to learn feature-privacy review relations. We then simulate feedback for candidate features and generate concise summaries of privacy concerns. We evaluate Pre-PI across three real-world apps, and compare it with Hark, a state-of-the-art method that relies on post-release user feedback to identify privacy concerns. Results show that Pre-PI generates additional valid privacy concerns and identifies these concerns earlier than Hark, allowing proactive mitigation prior to release.

American football practice generates video at scale, yet the interaction of interest occupies only a brief window of each long, untrimmed clip. Reliable biomechanical analysis, therefore, depends on spatiotemporal localization that identifies both the interacting entities and the onset of contact. We study First Point of Contact (FPOC), defined as the first frame in which a player physically touches a tackle dummy, in unconstrained practice footage with camera motion, clutter, multiple similarly equipped athletes, and rapid pose changes around impact. We present GRAZE, a training-free pipeline for FPOC localization that requires no labeled tackle-contact examples. GRAZE uses Grounding DINO to discover candidate player-dummy interactions, refines them with motion-aware temporal reasoning, and uses SAM2 as an explicit pixel-level verifier of contact rather than relying on detection confidence alone. This separation between candidate discovery and contact confirmation makes the approach robust to cluttered scenes and unstable grounding near impact. On 738 tackle-practice videos, GRAZE produces valid outputs for 97.4% of clips and localizes FPOC within $\pm$ 10 frames on 77.5% of all clips and within $\pm$ 20 frames on 82.7% of all clips. These results show that frame-accurate contact onset localization in real-world practice footage is feasible without task-specific training.

Can large language models (LLMs) predict which researchers will collaborate? We study this question through link prediction on real-world co-authorship networks from OpenAlex (9.96M authors, 108.7M edges), evaluating whether LLMs can predict future scientific collaborations using only author profiles, without access to graph structure. Using Qwen2.5-72B-Instruct across three historical eras of AI research, we find that LLMs and topology heuristics capture distinct signals and are strongest in complementary settings. On new-edge prediction under natural class imbalance, the LLM achieves AUROC 0.714--0.789, outperforming Common Neighbors, Jaccard, and Preferential Attachment, with recall up to 92.9\%; under balanced evaluation, the LLM outperforms \emph{all} topology heuristics in every era (AUROC 0.601--0.658 vs.\ best-heuristic 0.525--0.538); on continued edges, the LLM (0.687) is competitive with Adamic-Adar (0.684). Critically, 78.6--82.7\% of new collaborations occur between authors with no common neighbor -- a blind spot where all topology heuristics score zero but the LLM still achieves AUROC 0.652 by reasoning from author metadata alone. A temporal metadata ablation reveals that research concepts are the dominant signal (removing concepts drops AUROC by 0.047--0.084). Providing pre-computed graph features to the LLM \emph{degrades} performance due to anchoring effects, confirming that LLMs and topology methods should operate as separate, complementary channels. A socio-cultural ablation finds that name-inferred ethnicity and institutional country do not predict collaboration beyond topology, reflecting the demographic homogeneity of AI research. A node2vec baseline achieves AUROC comparable to Adamic-Adar, establishing that LLMs access a fundamentally different information channel -- author metadata -- rather than encoding the same structural signal differently.

Offline reinforcement learning (RL) has received increasing attention for learning policies from previously collected data without interaction with the real environment, which is particularly important in high-stakes applications. While a growing body of work has developed offline RL algorithms, these methods often rely on restrictive assumptions about data coverage and suffer from distribution shift. In this paper, we propose a residuals-based offline RL framework for general state and action spaces. Specifically, we define a residuals-based Bellman optimality operator that explicitly incorporates estimation error in learning transition dynamics into policy optimization by leveraging empirical residuals. We show that this Bellman operator is a contraction mapping and identify conditions under which its fixed point is asymptotically optimal and possesses finite-sample guarantees. We further develop a residuals-based offline deep Q-learning (DQN) algorithm. Using a stochastic CartPole environment, we demonstrate the effectiveness of our residuals-based offline DQN algorithm.

Rubric-based evaluation is widely used in LLM benchmarks and training pipelines for open-ended, less verifiable tasks. While prior work has demonstrated the effectiveness of rubrics using downstream signals such as reinforcement learning outcomes, there remains no principled way to diagnose rubric quality issues from such aggregated or downstream signals alone. To address this gap, we introduce RIFT: RubrIc Failure mode Taxonomy, a taxonomy for systematically characterizing failure modes in rubric composition and design. RIFT consists of eight failure modes organized into three high-level categories: Reliability Failures, Content Validity Failures, and Consequential Validity Failures. RIFT is developed using grounded theory by iteratively annotating rubrics drawn from five diverse benchmarks spanning general instruction following, code generation, creative writing, and expert-level deep research, until no new failure modes are identified. We evaluate the consistency of the taxonomy by measuring agreement among independent human annotators, observing fair agreement overall (87% pairwise agreement and 0.64 average Cohen's kappa). Finally, to support scalable diagnosis, we propose automated rubric quality metrics and show that they align with human failure-mode annotations, achieving up to 0.86 F1.

Surgical action automation has progressed rapidly toward achieving surgeon-like dexterous control, driven primarily by advances in learning from demonstration and vision-language-action models. While these have demonstrated success in table-top experiments, translating them to clinical deployment remains challenging: current methods offer limited predictability on where instruments will interact on tissue surfaces and lack explicit conditioning inputs to enforce tool-action-specific safe interaction regions. Addressing this gap, we introduce AffordTissue, a multimodal framework for predicting tool-action specific tissue affordance regions as dense heatmaps during cholecystectomy. Our approach combines a temporal vision encoder capturing tool motion and tissue dynamics across multiple viewpoints, language conditioning enabling generalization across diverse instrument-action pairs, and a DiT-style decoder for dense affordance prediction. We establish the first tissue affordance benchmark by curating and annotating 15,638 video clips across 103 cholecystectomy procedures, covering six unique tool-action pairs involving four instruments (hook, grasper, scissors, clipper) and their associated tasks: dissection, grasping, clipping, and cutting. Experiments demonstrate substantial improvement over vision-language model baselines (20.6 px ASSD vs. 60.2 px for Molmo-VLM), showing that our task-specific architecture outperforms large-scale foundation models for dense surgical affordance prediction. By predicting tool-action specific tissue affordance regions, AffordTissue provides explicit spatial reasoning for safe surgical automation, potentially unlocking explicit policy guidance toward appropriate tissue regions and early safe stop when instruments deviate outside predicted safe zones.

Large Language Models (LLMs) are increasingly deployed in high-stakes decision-making contexts. While prior work has shown that LLMs exhibit cognitive biases behaviorally, whether these biases correspond to identifiable internal representations and can be mitigated through targeted intervention remains an open question. We define LLM cognitive bias as systematic, reproducible deviations from correct answers in tasks with computable ground-truth baselines, and introduce LLM CogBias, a benchmark organized around four families of cognitive biases: Judgment, Information Processing, Social, and Response. We evaluate three LLMs and find that cognitive biases emerge systematically across all four families, with magnitudes and debiasing responses that are strongly family-dependent: prompt-level debiasing substantially reduces Response biases but backfires for Judgment biases. Using linear probes under a contrastive design, we show that these biases are encoded as linearly separable directions in model activation space. Finally, we apply activation steering to modulate biased behavior, achieving 26--32\% reduction in bias score (fraction of biased responses) while preserving downstream capability on 25 benchmarks (Llama: negligible degradation; Qwen: up to $-$19.0pp for Judgment biases). Despite near-orthogonal bias representations across models (mean cosine similarity 0.01), steering reduces bias at similar rates across architectures ($r(246)$=.621, $p$<.001), suggesting shared functional organization.

Predicting the perceived intensity of odorants remains a fundamental challenge in sensory science due to the complex, non-linear behavior of their response, as well as the difficulty in correlating molecular structure with human perception. While traditional deep learning models, such as Graph Convolutional Networks (GCNs), excel at capturing molecular topology, they often fail to account for the biological and perceptual context of olfaction. This study introduces VIANA, a novel "tri-pillar" framework that integrates structural graph theory, character value embeddings, and phenomenological behavior. This methodology systematically evaluates knowledge transfer across three distinct domains: molecular structure via GCNs, semantic odor character values via Principal Odor Map (POM) embeddings, and biological dose-response logic via Hill's law. We demonstrate that knowledge transfer is not inherently positive; rather, a balance must be maintained in the volume of information provided to the model. While raw semantic data led to "information overload" in domain-informed models, applying Principal Component Analysis (PCA) to distill the 95% most impactful semantic variance yielded a superior "signal distillation" effect. Results indicate that the synthesis of these three knowledge transfer pillars significantly outperforms baseline structural models, with VIANA achieving a peak R^2 of 0.996 and a test Mean Squared Error (MSE) of 0.19. In this context, VIANA successfully captures the physical ceiling of saturation, the sensitivity of detection thresholds, and the nuance of odor character value expression, providing a domain grounded simulation of the human olfactory experience. This research provides a robust framework for digital olfaction, effectively bridging the gap between molecular informatics and sensory perception.

Society 5.0 and Industry 5.0 call for human-centric technology integration, yet the concept lacks an operational definition that can be measured, optimized, or evaluated at the firm level. This paper addresses three gaps. First, existing models of human-AI complementarity treat the augmentation function phi(D) as exogenous -- dependent only on the stock of AI deployed -- ignoring that two firms with identical technology investments achieve radically different augmentation outcomes depending on how the workplace is organized around the human-AI interaction. Second, no multi-dimensional instrument exists linking workplace design choices to augmentation productivity. Third, the Society 5.0 literature proposes human-centricity as a normative aspiration but provides no formal criterion for when it is economically optimal. We make four contributions. (1) We endogenize the augmentation function as phi(D, W), where W is a five-dimensional workplace design vector -- AI interface design, decision authority allocation, task orchestration, learning loop architecture, and psychosocial work environment -- and prove that human-centric design is profit-maximizing when the workforce's augmentable cognitive capital exceeds a critical threshold. (2) We conduct a PRISMA-guided systematic review of 120 papers (screened from 6,096 records) to map the evidence base for each dimension. (3) We provide secondary empirical evidence from Colombia's EDIT manufacturing survey (N=6,799 firms) showing that management practice quality amplifies the return to technology investment (interaction coefficient 0.304, p<0.01). (4) We propose the Workplace Augmentation Design Index (WADI), a 36-item theory-grounded instrument for diagnosing human-centricity at the firm level. Decision authority allocation emerges as the binding constraint for Society 5.0 transitions, and task orchestration as the most under-researched dimension

We propose that AI automation is a continuum between: (i) crashing waves where AI capabilities surge abruptly over small sets of tasks, and (ii) rising tides where the increase in AI capabilities is more continuous and broad-based. We test for these effects in preliminary evidence from an ongoing evaluation of AI capabilities across over 3,000 broad-based tasks derived from the U.S. Department of Labor O*NET categorization that are text-based and thus LLM-addressable. Based on more than 17,000 evaluations by workers from these jobs, we find little evidence of crashing waves (in contrast to recent work by METR), but substantial evidence that rising tides are the primary form of AI automation. AI performance is high and improving rapidly across a wide range of tasks. We estimate that, in 2024-Q2, AI models successfully complete tasks that take humans approximately 3-4 hours with about a 50% success rate, increasing to about 65% by 2025-Q3. If recent trends in AI capability growth persist, this pace of AI improvement implies that LLMs will be able to complete most text-related tasks with success rates of, on average, 80%-95% by 2029 at a minimally sufficient quality level. Achieving near-perfect success rates at this quality level or comparable success rates at superior quality would require several additional years. These AI capability improvements would impact the economy and labor market as organizations adopt AI, which could have a substantially longer timeline.

We introduce world-centered multi-agent systems (WMAS) as an alternative to traditional agent-centered architectures, arguing that structured domains such as enterprises and institutional systems require a shared, explicit world representation to ensure semantic consistency, explainability, and long-term stability. We classify worlds along dimensions including ontological explicitness, normativity, etc. In WMAS, learning and coordination operate over a shared world model rather than isolated agent-local representations, enabling global consistency and verifiable system behavior. We propose semantic models as a mathematical formalism for representing such worlds. Finally, we present the Ontobox platform as a realization of WMAS.

Moderation layers are increasingly a core component of many products built on user- or model-generated content. However, drafting and maintaining domain-specific safety policies remains costly. We present Deep Policy Research (DPR), a minimal agentic system that drafts a full content moderation policy based on only human-written seed domain information. DPR uses a single web search tool and lightweight scaffolding to iteratively propose search queries, distill diverse web sources into policy rules, and organize rules into an indexed document. We evaluate DPR on (1) the OpenAI undesired content benchmark across five domains with two compact reader LLMs and (2) an in-house multimodal advertisement moderation benchmark. DPR consistently outperforms definition-only and in-context learning baselines, and in our end-to-end setting it is competitive with expert-written policy sections in several domains. Moreover, under the same seed specification and evaluation protocol, DPR outperforms a general-purpose deep research system, suggesting that a task-specific, structured research loop can be more effective than generic web research for policy drafting. We release our experiment code at https://github.com/xiaowu0162/deep-policy-research.

LLM-based agents increasingly operate across repeated sessions, maintaining task states to ensure continuity. In many deployments, a single agent serves multiple users within a team or organization, reusing a shared knowledge layer across user identities. This shared persistence expands the failure surface: information that is locally valid for one user can silently degrade another user's outcome when the agent reapplies it without regard for scope. We refer to this failure mode as unintentional cross-user contamination (UCC). Unlike adversarial memory poisoning, UCC requires no attacker; it arises from benign interactions whose scope-bound artifacts persist and are later misapplied. We formalize UCC through a controlled evaluation protocol, introduce a taxonomy of three contamination types, and evaluate the problem in two shared-state mechanisms. Under raw shared state, benign interactions alone produce contamination rates of 57--71%. A write-time sanitization is effective when shared state is conversational, but leaves substantial residual risk when shared state includes executable artifacts, with contamination often manifesting as silent wrong answers. These results indicate that shared-state agents need artifact-level defenses beyond text-level sanitization to prevent silent cross-user failures.

Reservoir simulation workflows face a fundamental data asymmetry: input parameter fields (geostatistical permeability realizations, porosity distributions) are free to generate in arbitrary quantities, yet existing neural operator surrogates require large corpora of expensive labeled simulation trajectories and cannot exploit this unlabeled structure. We introduce \textbf{PI-JEPA} (Physics-Informed Joint Embedding Predictive Architecture), a surrogate pretraining framework that trains \emph{without any completed PDE solves}, using masked latent prediction on unlabeled parameter fields under per-sub-operator PDE residual regularization. The predictor bank is structurally aligned with the Lie--Trotter operator-splitting decomposition of the governing equations, dedicating a separate physics-constrained latent module to each sub-process (pressure, saturation transport, reaction), enabling fine-tuning with as few as 100 labeled simulation runs. On single-phase Darcy flow, PI-JEPA achieves $1.9\times$ lower error than FNO and $2.4\times$ lower error than DeepONet at $N_\ell{=}100$, with 24\% improvement over supervised-only training at $N_\ell{=}500$, demonstrating that label-free surrogate pretraining substantially reduces the simulation budget required for multiphysics surrogate deployment.

Test-time scaling has emerged as an effective way to improve language models on challenging reasoning tasks. However, most existing methods treat each problem in isolation and do not systematically reuse knowledge from prior reasoning trajectories. In particular, they underutilize procedural knowledge: how to reframe a problem, choose an approach, and verify or backtrack when needed. We introduce Reasoning Memory, a retrieval-augmented generation (RAG) framework for reasoning models that explicitly retrieves and reuses procedural knowledge at scale. Starting from existing corpora of step-by-step reasoning trajectories, we decompose each trajectory into self-contained subquestion-subroutine pairs, yielding a datastore of 32 million compact procedural knowledge entries. At inference time, a lightweight in-thought prompt lets the model verbalize the core subquestion, retrieve relevant subroutines within its reasoning trace, and reason under diverse retrieved subroutines as implicit procedural priors. Across six math, science, and coding benchmarks, Reasoning Memory consistently outperforms RAG with document, trajectory, and template knowledge, as well as a compute-matched test-time scaling baseline. With a higher inference budget, it improves over no retrieval by up to 19.2% and over the strongest compute-matched baseline by 7.9% across task types. Ablation studies show that these gains come from two key factors: the broad procedural coverage of the source trajectories and our decomposition and retrieval design, which together enable effective extraction and reuse of procedural knowledge.

World models -- learned internal simulators of environment dynamics -- are rapidly becoming foundational to autonomous decision-making in robotics, autonomous vehicles, and agentic AI. Yet this predictive power introduces a distinctive set of safety, security, and cognitive risks. Adversaries can corrupt training data, poison latent representations, and exploit compounding rollout errors to cause catastrophic failures in safety-critical deployments. World model-equipped agents are more capable of goal misgeneralisation, deceptive alignment, and reward hacking precisely because they can simulate the consequences of their own actions. Authoritative world model predictions further foster automation bias and miscalibrated human trust that operators lack the tools to audit. This paper surveys the world model landscape; introduces formal definitions of trajectory persistence and representational risk; presents a five-profile attacker capability taxonomy; and develops a unified threat model extending MITRE ATLAS and the OWASP LLM Top 10 to the world model stack. We provide an empirical proof-of-concept on trajectory-persistent adversarial attacks (GRU-RSSM: A_1 = 2.26x amplification, -59.5% reduction under adversarial fine-tuning; stochastic RSSM proxy: A_1 = 0.65x; DreamerV3 checkpoint: non-zero action drift confirmed). We illustrate risks through four deployment scenarios and propose interdisciplinary mitigations spanning adversarial hardening, alignment engineering, NIST AI RMF and EU AI Act governance, and human-factors design. We argue that world models must be treated as safety-critical infrastructure requiring the same rigour as flight-control software or medical devices.

Inverse reinforcement learning (IRL) recovers the loss function of a forward learner from its observed responses adaptive IRL aims to reconstruct the loss function of a forward learner by passively observing its gradients as it performs reinforcement learning (RL). This paper proposes a novel passive Langevin-based algorithm that achieves adaptive IRL. The key difficulty in adaptive IRL is that the required gradients in the passive algorithm are counterfactual, that is, they are conditioned on events of probability zero under the forward learner's trajectory. Therefore, naive Monte Carlo estimators are prohibitively inefficient, and kernel smoothing, though common, suffers from slow convergence. We overcome this by employing Malliavin calculus to efficiently estimate the required counterfactual gradients. We reformulate the counterfactual conditioning as a ratio of unconditioned expectations involving Malliavin quantities, thus recovering standard estimation rates. We derive the necessary Malliavin derivatives and their adjoint Skorohod integral formulations for a general Langevin structure, and provide a concrete algorithmic approach which exploits these for counterfactual gradient estimation.

Competency question (CQ) elicitation represents a critical but resource-intensive bottleneck in ontology engineering. This foundational phase is often hampered by the communication gap between domain experts, who possess the necessary knowledge, and ontology engineers, who formalise it. This paper introduces IDEA2, a novel, semi-automated workflow that integrates Large Language Models (LLMs) within a collaborative, expert-in-the-loop process to address this challenge. The methodology is characterised by a core iterative loop: an initial LLM-based extraction of CQs from requirement documents, a co-creational review and feedback phase by domain experts on an accessible collaborative platform, and an iterative, feedback-driven reformulation of rejected CQs by an LLM until consensus is achieved. To ensure transparency and reproducibility, the entire lifecycle of each CQ is tracked using a provenance model that captures the full lineage of edits, anonymised feedback, and generation parameters. The workflow was validated in 2 real-world scenarios (scientific data, cultural heritage), demonstrating that IDEA2 can accelerate the requirements engineering process, improve the acceptance and relevance of the resulting CQs, and exhibit high usability and effectiveness among domain experts. We release all code and experiments at https://github.com/KE-UniLiv/IDEA2

Multivariate Hawkes processes are a widely used class of self-exciting point processes, but maximum likelihood estimation naively scales as $O(N^2)$ in the number of events. The canonical linear exponential Hawkes process admits a faster $O(N)$ recurrence, but prior work evaluates this recurrence sequentially, without exploiting parallelization on modern GPUs. We show that the Hawkes process intensity can be expressed as a product of sparse transition matrices admitting a linear-time associative multiply, enabling computation via a parallel prefix scan. This yields a simple yet massively parallelizable algorithm for maximum likelihood estimation of linear exponential Hawkes processes. Our method reduces the computational complexity to approximately $O(N/P)$ with $P$ parallel processors, and naturally yields a batching scheme to maintain constant memory usage, avoiding GPU memory constraints. Importantly, it computes the exact likelihood without any additional assumptions or approximations, preserving the simplicity and interpretability of the model. We demonstrate orders-of-magnitude speedups on simulated and real datasets, scaling to thousands of nodes and tens of millions of events, substantially beyond scales reported in prior work. We provide an open-source PyTorch library implementing our optimizations.

Vision transformers (ViT) rely on attention mechanism to weigh input features, and therefore attention scores have naturally been considered as explanations for its decision-making process. However, attention scores are almost always non-zero, resulting in noisy and diffused attention maps and limiting interpretability. Can we quantify uncertainty measures of attention scores and obtain regularized attention scores? To this end, we consider attention scores of ViT in a statistical framework where independent noise would lead to insignificant yet non-zero scores. Leveraging statistical learning techniques, we introduce the bootstrapping for attention scores which generates a baseline distribution of attention scores by resampling input features. Such a bootstrap distribution is then used to estimate significances and posterior probabilities of attention scores. In natural and medical images, the proposed \emph{Attention Regularization} approach demonstrates a straightforward removal of spurious attention arising from noise, drastically improving shrinkage and sparsity. Quantitative evaluations are conducted using both simulation and real-world datasets. Our study highlights bootstrapping as a practical regularization tool when using attention scores as explanations for ViT. Code available: https://github.com/ncchung/AttentionRegularization

While deep learning has significantly advanced accident anticipation, the robustness of these safety-critical systems against real-world perturbations remains a major challenge. We reveal that state-of-the-art models like CRASH, despite their high performance, exhibit significant instability in predictions and latent representations when faced with minor input perturbations, posing serious reliability risks. To address this, we introduce SECURE - Stable Early Collision Understanding Robust Embeddings, a framework that formally defines and enforces model robustness. SECURE is founded on four key attributes: consistency and stability in both prediction space and latent feature space. We propose a principled training methodology that fine-tunes a baseline model using a multi-objective loss, which minimizes divergence from a reference model and penalizes sensitivity to adversarial perturbations. Experiments on DAD and CCD datasets demonstrate that our approach not only significantly enhances robustness against various perturbations but also improves performance on clean data, achieving new state-of-the-art results.

We investigate the data-driven discovery of constitutive closures in nonlinear reaction-diffusion systems with known governing PDE structures. Our objective is to robustly recover diffusion and reaction laws from spatiotemporal observations while avoiding the common pitfall where low residuals or short-horizon predictions are conflated with physical recovery. We propose a three-stage neural-symbolic framework: (1) learning numerical surrogates under physical constraints using a noise-robust weak-form-driven objective; (2) compressing these surrogates into restricted interpretable symbolic families (e.g., polynomial, rational, and saturation forms); and (3) validating the symbolic closures through explicit forward re-simulation on unseen initial conditions. Extensive numerical experiments reveal two distinct regimes. Under matched-library settings, weak polynomial baselines behave as correctly specified reference estimators, showing that neural surrogates do not uniformly outperform classical bases. Conversely, under function-class mismatch, neural surrogates provide necessary flexibility and can be compressed into compact symbolic laws with minimal rollout degradation. However, we identify a critical "bias inheritance" mechanism where symbolic compression does not automatically repair constitutive bias. Across various observation regimes, the true error of the symbolic closure closely tracks that of the neural surrogate, yielding a bias inheritance ratio near one. These findings demonstrate that the primary bottleneck in neural-symbolic modeling lies in the initial numerical inverse problem rather than the subsequent symbolic compression. We underscore that constitutive claims must be rigorously supported by forward validation rather than residual minimization alone.

While deepfake speech detectors built on large self-supervised learning (SSL) models achieve high accuracy, employing standard ensemble fusion to further enhance robustness often results in oversized systems with diminishing returns. To address this, we propose an evolutionary multi-objective score fusion framework that jointly minimizes detection error and system complexity. We explore two encodings optimized by NSGA-II: binary-coded detector selection for score averaging and a real-valued scheme that optimizes detector weights for a weighted sum. Experiments on the ASVspoof 5 dataset with 36 SSL-based detectors show that the obtained Pareto fronts outperform simple averaging and logistic regression baselines. The real-valued variant achieves 2.37% EER (0.0684 minDCF) and identifies configurations that match state-of-the-art performance while significantly reducing system complexity, requiring only half the parameters. Our method also provides a diverse set of trade-off solutions, enabling deployment choices that balance accuracy and computational cost.

Model merging provides a way of cheaply combining individual models to produce a model that inherits each individual's capabilities. While some merging methods can approach the performance of multitask training, they are often heuristically motivated and lack theoretical justification. A principled alternative is to pose model merging as a layer-wise optimization problem that directly minimizes interference between tasks. However, this formulation requires estimating per-layer covariance matrices from data, which may not be available when performing merging. In contrast, many of the heuristically-motivated methods do not require auxiliary data, making them practically advantageous. In this work, we revisit the interference minimization framework and show that, under certain conditions, covariance matrices can be estimated directly from difference matrices, eliminating the need for data while also reducing computational costs. We validate our approach across vision and language benchmarks on models ranging from 86M parameters to 7B parameters, outperforming previous data-free state-of-the-art merging methods

Traditional scientific discovery relies on an iterative hypothesise-experiment-refine cycle that has driven progress for centuries, but its intuitive, ad-hoc implementation often wastes resources, yields inefficient designs, and misses critical insights. This tutorial presents Bayesian Optimisation (BO), a principled probability-driven framework that formalises and automates this core scientific cycle. BO uses surrogate models (e.g., Gaussian processes) to model empirical observations as evolving hypotheses, and acquisition functions to guide experiment selection, balancing exploitation of known knowledge and exploration of uncharted domains to eliminate guesswork and manual trial-and-error. We first frame scientific discovery as an optimisation problem, then unpack BO's core components, end-to-end workflows, and real-world efficacy via case studies in catalysis, materials science, organic synthesis, and molecule discovery. We also cover critical technical extensions for scientific applications, including batched experimentation, heteroscedasticity, contextual optimisation, and human-in-the-loop integration. Tailored for a broad audience, this tutorial bridges AI advances in BO with practical natural science applications, offering tiered content to empower cross-disciplinary researchers to design more efficient experiments and accelerate principled scientific discovery.

Macroscopic unmanned aerial vehicle (UAV) traffic organization in three-dimensional airspace faces significant challenges from static wind fields and complex obstacles. A critical difficulty lies in simultaneously capturing the strong anisotropy induced by wind while strictly preserving transport consistency and boundary semantics, which are often compromised in standard physics-informed learning approaches. To resolve this, we propose a constraint-preserving hybrid solver that integrates a physics-informed neural network for the anisotropic Eikonal value problem with a conservative finite-volume method for steady density transport. These components are coupled through an outer Picard iteration with under-relaxation, where the target condition is hard-encoded and strictly conservative no-flux boundaries are enforced during the transport step. We evaluate the framework on reproducible homing and point-to-point scenarios, effectively capturing value slices, induced-motion patterns, and steady density structures such as bands and bottlenecks. Ultimately, our perspective emphasizes the value of a reproducible computational framework supported by transparent empirical diagnostics to enable the traceable assessment of macroscopic traffic phenomena.

The fundamental problem of causal inference - that the counterfactual outcome for any individual is never observed - has shaped the entire methodology of the field. Every existing approach substitutes assumptions for missing data: ignorability, parallel trends, exclusion restrictions. None produces the counterfactual itself. This paper proposes the Digital Twin Counterfactual Framework (DTCF): rather than estimating the counterfactual statistically, we simulate it using a digital twin and subject the simulation to a hierarchical validation regime. We formalize the digital twin simulator as a stochastic mapping within the potential outcomes framework and introduce a hierarchy of twin fidelity assumptions - from marginal fidelity through joint fidelity to structural fidelity - each unlocking a progressively richer class of estimands. The central contribution is threefold. First, a five-level validation architecture converts the unfalsifiable claim that the simulator produces correct counterfactuals into falsifiable tests against observable data. Second, a formal decomposition separates causal quantities into those that are marginally validated (ATE, CATE, QTE - testable through observable-arm comparison) and those that are copula-dependent (the ITE distribution, probability of benefit/harm, variance of treatment effects - permanently reliant on the unobservable within-individual dependence structure). Third, bounding, sensitivity, and uncertainty quantification tools make the copula dependence explicit. The DTCF does not resolve the fundamental problem of causal inference. What it provides is a framework in which marginal causal claims become increasingly testable, joint causal claims become explicitly assumption-indexed, and the gap between the two is formally characterized.

Money launderers take advantage of limitations in existing detection approaches by hiding their financial footprints in a deceitful manner. They manage this by replicating transaction patterns that the monitoring systems cannot easily distinguish. As a result, criminally gained assets are pushed into legitimate financial channels without drawing attention. Algorithms developed to monitor money flows often struggle with scale and complexity. The difficulty of identifying such activities is further intensified by the (persistent) inability of current solutions to control the excessive number of false positive signals produced by rigid, risk-based rules systems. We propose a framework called ReDiRect (REduce, DIstribute, and RECTify), specifically designed to overcome these challenges. The primary contribution of our work is a novel framing of this problem in an unsupervised setting; where a large transaction graph is fuzzily partitioned into smaller, manageable components to enable fast processing in a distributed manner. In addition, we define a refined evaluation metric that better captures the effectiveness of exposed money laundering patterns. Through comprehensive experimentation, we demonstrate that our framework achieves superior performance compared to existing and state-of-the-art techniques, particularly in terms of efficiency and real-world applicability. For validation, we used the real (open source) Libra dataset and the recently released synthetic datasets by IBM Watson. Our code and datasets are available at https://github.com/mhaseebtariq/redirect.

High-fidelity Monte Carlo simulations and complex inverse problems, such as mapping smeared experimental observations to ground-truth states, are computationally intensive yet essential for robust data analysis. Conditional Flow Matching (CFM) offers a mathematically robust approach to accelerating these tasks, but we demonstrate its standard training loss is fundamentally misleading. In rigorous physics applications, CFM loss plateaus prematurely, serving as an unreliable indicator of true convergence and physical fidelity. To investigate this disconnect, we designed JetPrism, a configurable CFM framework acting as an efficient generative surrogate for evaluating unconditional generation and conditional detector unfolding. Using synthetic stress tests and a Jefferson Lab kinematic dataset ($γp \to ρ^0 p \to π^+π^- p$) relevant to the forthcoming Electron-Ion Collider (EIC), we establish that physics-informed metrics continue to improve significantly long after the standard loss converges. Consequently, we propose a multi-metric evaluation protocol incorporating marginal and pairwise $χ^2$ statistics, $W_1$ distances, correlation matrix distances ($D_{\mathrm{corr}}$), and nearest-neighbor distance ratios ($R_{\mathrm{NN}}$). By demonstrating that domain-specific evaluations must supersede generic loss metrics, this work establishes JetPrism as a dependable generative surrogate that ensures precise statistical agreement with ground-truth data without memorizing the training set. While demonstrated in nuclear physics, this diagnostic framework is readily extensible to parameter generation and complex inverse problems across broad domains. Potential applications span medical imaging, astrophysics, semiconductor discovery, and quantitative finance, where high-fidelity simulation, rigorous inversion, and generative reliability are critical.

Learning human preferences in language models remains fundamentally challenging, as reward modeling relies on subtle, subjective comparisons or shades of gray rather than clear-cut labels. This study investigates the limits of current approaches and proposes a feature-augmented framework to better capture the multidimensional nature of human judgment. Using the Anthropic HHRLHF dataset, we evaluate ten diverse large language models LLMs under a standard pairwise preference setting, where baseline performance remains below 0.74 ROC AUC, highlighting the difficulty of the task. To address this, we enrich textual representations with interpretable signals: response length, refusal indicators, toxicity scores and prompt response semantic similarity, enabling models to explicitly capture key aspects of helpfulness, safety and relevance. The proposed hybrid approach yields consistent improvements across all models, achieving up to 0.84 ROC AUC and significantly higher pairwise accuracy, with DeBERTav3Large demonstrating the best performance. Beyond accuracy, we integrate SHAP and LIME to provide fine-grained interpretability, revealing that model decisions depend on contextualized safety and supportive framing rather than isolated keywords. We further analyze bias amplification, showing that while individual features have weak marginal effects, their interactions influence preference learning.

Designing reliable integrated energy systems for industrial processes requires optimization and verification models across multiple fidelities, from architecture-level sizing to high-fidelity dynamic operation. However, model mismatch across fidelities obscures the sources of performance loss and complicates the quantification of architecture-to-operation performance gaps. We propose an online, machine-learning-accelerated multi-resolution optimization framework that estimates an architecture-specific upper bound on achievable performance while minimizing expensive high-fidelity model evaluations. We demonstrate the approach on a pilot energy system supplying a 1 MW industrial heat load. First, we solve a multi-objective architecture optimization to select the system configuration and component capacities. We then develop an machine learning (ML)-accelerated multi-resolution, receding-horizon optimal control strategy that approaches the achievable-performance bound for the specified architecture, given the additional controls and dynamics not captured by the architectural optimization model. The ML-guided controller adaptively schedules the optimization resolution based on predictive uncertainty and warm-starts high-fidelity solves using elite low-fidelity solutions. Our results on the pilot case study show that the proposed multi-resolution strategy reduces the architecture-to-operation performance gap by up to 42% relative to a rule-based controller, while reducing required high-fidelity model evaluations by 34% relative to the same multi-fidelity approach without ML guidance, enabling faster and more reliable design verification. Together, these gains make high-fidelity verification tractable, providing a practical upper bound on achievable operational performance.

Evaluating scientific arguments requires assessing the strict consistency between a claim and its underlying multimodal evidence. However, existing benchmarks lack the scale, domain diversity, and visual complexity needed to evaluate this alignment realistically. To address this gap, we introduce M2-Verify, a large-scale multimodal dataset for checking scientific claim consistency. Sourced from PubMed and arXiv, M2-Verify provides over 469K instances across 16 domains, rigorously validated through expert audits. Extensive baseline experiments show that state-of-the-art models struggle to maintain robust consistency. While top models achieve up to 85.8\% Micro-F1 on low-complexity medical perturbations, performance drops to 61.6\% on high-complexity challenges like anatomical shifts. Furthermore, expert evaluations expose hallucinations when models generate scientific explanations for their alignment decisions. Finally, we demonstrate our dataset's utility and provide comprehensive usage guidelines.

Reconstructing high-dimensional spatiotemporal fields from sparse sensor measurements is critical in a wide range of scientific applications. The SHallow REcurrent Decoder (SHRED) architecture is a recent state-of-the-art architecture that reconstructs high-quality spatial domain from hyper-sparse sensor measurement streams. An important limitation of SHRED is that in complex, data-scarce, high-frequency, or stochastic systems, portions of the spatiotemporal field must be modeled with valid uncertainty estimation. We introduce UQ-SHRED, a distributional learning framework for sparse sensing problems that provides uncertainty quantification through a neural network-based distributional regression called engression. UQ-SHRED models the uncertainty by learning the predictive distribution of the spatial state conditioned on the sensor history. By injecting stochastic noise into sensor inputs and training with an energy score loss, UQ-SHRED produces predictive distributions with minimal computational overhead, requiring only noise injection at the input and resampling through a single architecture without retraining or additional network structures. On complicated synthetic and real-life datasets including turbulent flow, atmospheric dynamics, neuroscience and astrophysics, UQ-SHRED provides a distributional approximation with well-calibrated confidence intervals. We further conduct ablation studies to understand how each model setting affects the quality of the UQ-SHRED performance, and its validity on uncertainty quantification over a set of different experimental setups.

We study how to scale reasoning token budgets for competitive programming through two complementary approaches: training-time reinforcement learning (RL) and test-time parallel thinking. During RL training, we observe an approximately log-linear relationship between validation accuracy and the average number of generated reasoning tokens over successive checkpoints, and show two ways to shift this training trajectory: verification RL warmup raises the starting point, while randomized clipping produces a steeper trend in the observed regime. As scaling single-generation reasoning during RL quickly becomes expensive under full attention, we introduce a multi-round parallel thinking pipeline that distributes the token budget across threads and rounds of generation, verification, and refinement. We train the model end-to-end on this pipeline to match the training objective to the test-time structure. Starting from Seed-OSS-36B, the full system with 16 threads and 16 rounds per thread matches the underlying RL model's oracle pass@16 at pass@1 using 7.6 million tokens per problem on average, and surpasses GPT-5-high on 456 hard competitive programming problems from AetherCode.

Anticipating supply chain disruptions before they materialize is a core challenge for firms and policymakers alike. A key difficulty is learning to reason reliably about infrequent, high-impact events from noisy and unstructured inputs - a setting where general-purpose models struggle without task-specific adaptation. We introduce an end-to-end framework that trains LLMs to produce calibrated probabilistic forecasts using realized disruption outcomes as supervision. The resulting model substantially outperforms strong baselines - including GPT-5 - on accuracy, calibration, and precision. We also show that training induces more structured and reliable probabilistic reasoning without explicit prompting. These results suggest a general pathway for training domain-specific forecasting models that produce decision-ready signals. To support transparency we open-source the evaluation dataset used in this study. Dataset: https://huggingface.co/datasets/LightningRodLabs/supply-chain-predictions

Answering questions about images often requires combining visual understanding with external knowledge. Multimodal Large Language Models (MLLMs) provide a natural framework for this setting, but they often struggle to identify the most relevant visual and textual evidence when answering knowledge-intensive queries. In such scenarios, models must integrate visual cues with retrieved textual evidence that is often noisy or only partially relevant, while also localizing fine-grained visual information in the image. In this work, we introduce Look Twice (LoT), a training-free inference-time framework that improves how pretrained MLLMs utilize multimodal evidence. Specifically, we exploit the model attention patterns to estimate which visual regions and retrieved textual elements are relevant to a query, and then generate the answer conditioned on this highlighted evidence. The selected cues are highlighted through lightweight prompt-level markers that encourage the model to re-attend to the relevant evidence during generation. Experiments across multiple knowledge-based VQA benchmarks show consistent improvements over zero-shot MLLMs. Additional evaluations on vision-centric and hallucination-oriented benchmarks further demonstrate that visual evidence highlighting alone improves model performance in settings without textual context, all without additional training or architectural modifications. Source code will be publicly released.

We introduce Sven (Singular Value dEsceNt), a new optimization algorithm for neural networks that exploits the natural decomposition of loss functions into a sum over individual data points, rather than reducing the full loss to a single scalar before computing a parameter update. Sven treats each data point's residual as a separate condition to be satisfied simultaneously, using the Moore-Penrose pseudoinverse of the loss Jacobian to find the minimum-norm parameter update that best satisfies all conditions at once. In practice, this pseudoinverse is approximated via a truncated singular value decomposition, retaining only the $k$ most significant directions and incurring a computational overhead of only a factor of $k$ relative to stochastic gradient descent. This is in comparison to traditional natural gradient methods, which scale as the square of the number of parameters. We show that Sven can be understood as a natural gradient method generalized to the over-parametrized regime, recovering natural gradient descent in the under-parametrized limit. On regression tasks, Sven significantly outperforms standard first-order methods including Adam, converging faster and to a lower final loss, while remaining competitive with LBFGS at a fraction of the wall-time cost. We discuss the primary challenge to scaling, namely memory overhead, and propose mitigation strategies. Beyond standard machine learning benchmarks, we anticipate that Sven will find natural application in scientific computing settings where custom loss functions decompose into several conditions.

In this paper, we attempt to explore the landscape of two-dimensional conformal field theories (2d CFTs) by efficiently searching for numerical solutions to the modular bootstrap equation using machine-learning-style optimization. The torus partition function of a 2d CFT is fixed by the spectrum of its primary operators and its chiral algebra, which we take to be the Virasoro algebra with $c>1$. We translate the requirement that this partition function is modular invariant into a loss function, which we then minimize to identify possible primary spectra. Our approach involves two technical innovations that facilitate finding reliable candidate CFTs. The first is a strategy to estimate the uncertainty associated with truncating the spectrum to the lowest dimension operators. The second is the use of a new singular-value-based optimizer (Sven) that is more effective than gradient descent at navigating the hierarchical structure of the loss landscape. We numerically construct candidate truncated CFT partition functions with central charges between 1 and $\frac{8}{7}$, a range devoid of known examples, and argue that these candidates likely come from a continuous space of modular bootstrap solutions. We also provide evidence for a more stringent constraint on the spectral gap near $c = 1$ than the existing bound of $Δ_{\rm gap} \le \frac{c}{6} + \frac{1}{3}$.

We present HippoCamp, a new benchmark designed to evaluate agents' capabilities on multimodal file management. Unlike existing agent benchmarks that focus on tasks like web interaction, tool use, or software automation in generic settings, HippoCamp evaluates agents in user-centric environments to model individual user profiles and search massive personal files for context-aware reasoning. Our benchmark instantiates device-scale file systems over real-world profiles spanning diverse modalities, comprising 42.4 GB of data across over 2K real-world files. Building upon the raw files, we construct 581 QA pairs to assess agents' capabilities in search, evidence perception, and multi-step reasoning. To facilitate fine-grained analysis, we provide 46.1K densely annotated structured trajectories for step-wise failure diagnosis. We evaluate a wide range of state-of-the-art multimodal large language models (MLLMs) and agentic methods on HippoCamp. Our comprehensive experiments reveal a significant performance gap: even the most advanced commercial models achieve only 48.3% accuracy in user profiling, struggling particularly with long-horizon retrieval and cross-modal reasoning within dense personal file systems. Furthermore, our step-wise failure diagnosis identifies multimodal perception and evidence grounding as the primary bottlenecks. Ultimately, HippoCamp exposes the critical limitations of current agents in realistic, user-centric environments and provides a robust foundation for developing next-generation personal AI assistants.

The rise of test-time scaling has remarkably boosted the reasoning and agentic proficiency of Large Language Models (LLMs). Yet, standard Transformers struggle to scale inference-time compute efficiently, as conventional looping strategies suffer from high computational overhead and a KV cache that inflates alongside model depth. We present Universal YOCO (YOCO-U), which combines the YOCO decoder-decoder architecture with recursive computation to achieve a synergistic effect greater than either alone. Built on the YOCO framework, YOCO-U implements a Universal Self-Decoder that performs multiple iterations via parameter sharing, while confining the iterative process to shallow, efficient-attention layers. This combination yields a favorable capability-efficiency tradeoff that neither YOCO nor recursion achieves independently. The YOCO architecture provides a constant global KV cache and linear pre-filling, while partial recursion enhances representational depth with limited overhead. Together, YOCO-U improves token utility and scaling behavior while maintaining efficient inference. Empirical results confirm that YOCO-U remains highly competitive in general and long-context benchmarks, demonstrating that the integration of efficient-attention architectures and recursive computation is a promising direction for scalable LLMs.

Reconstructing full spatio-temporal dynamics from sparse observations in both space and time remains a central challenge in complex systems, as measurements can be spatially incomplete and can be also limited to narrow temporal windows. Yet approximating the complete spatio-temporal trajectory is essential for mechanistic insight and understanding, model calibration, and operational decision-making. We introduce LAPIS-SHRED (LAtent Phase Inference from Short time sequence using SHallow REcurrent Decoders), a modular architecture that reconstructs and/or forecasts complete spatiotemporal dynamics from sparse sensor observations confined to short temporal windows. LAPIS-SHRED operates through a three-stage pipeline: (i) a SHRED model is pre-trained entirely on simulation data to map sensor time-histories into a structured latent space, (ii) a temporal sequence model, trained on simulation-derived latent trajectories, learns to propagate latent states forward or backward in time to span unobserved temporal regions from short observational time windows, and (iii) at deployment, only a short observation window of hyper-sparse sensor measurements from the true system is provided, from which the frozen SHRED model and the temporal model jointly reconstruct or forecast the complete spatiotemporal trajectory. The framework supports bidirectional inference, inherits data assimilation and multiscale reconstruction capabilities from its modular structure, and accommodates extreme observational constraints including single-frame terminal inputs. We evaluate LAPIS-SHRED on six experiments spanning complex spatio-temporal physics: turbulent flows, multiscale propulsion physics, volatile combustion transients, and satellite-derived environmental fields, highlighting a lightweight, modular architecture suited for operational settings where observation is constrained by physical or logistical limitations.

AI weather prediction has advanced rapidly, yet no unified mathematical framework explains what determines forecast skill. Existing theory addresses specific architectural choices rather than the learning pipeline as a whole, while operational evidence from 2023-2026 demonstrates that training methodology, loss function design, and data diversity matter at least as much as architecture selection. This paper makes two interleaved contributions. Theoretically, we construct a framework rooted in approximation theory on the sphere, dynamical systems theory, information theory, and statistical learning theory that treats the complete learning pipeline (architecture, loss function, training strategy, data distribution) rather than architecture alone. We establish a Learning Pipeline Error Decomposition showing that estimation error (loss- and data-dependent) dominates approximation error (architecture-dependent) at current scales. We develop a Loss Function Spectral Theory formalizing MSE-induced spectral blurring in spherical harmonic coordinates, and derive Out-of-Distribution Extrapolation Bounds proving that data-driven models systematically underestimate record-breaking extremes with bias growing linearly in record exceedance. Empirically, we validate these predictions via inference across ten architecturally diverse AI weather models using NVIDIA Earth2Studio with ERA5 initial conditions, evaluating six metrics across 30 initialization dates spanning all seasons. Results confirm universal spectral energy loss at high wavenumbers for MSE-trained models, rising Error Consensus Ratios showing that the majority of forecast error is shared across architectures, and linear negative bias during extreme events. A Holistic Model Assessment Score provides unified multi-dimensional evaluation, and a prescriptive framework enables mathematical evaluation of proposed pipelines before training.

Efficient robotic extraterrestrial exploration requires robots with diverse capabilities, ranging from scientific measurement tools to advanced locomotion. A robotic team enables the distribution of tasks over multiple specialized subsystems, each providing specific expertise to complete the mission. The central challenge lies in efficiently coordinating the team to maximize utilization and the extraction of scientific value. Classical planning algorithms scale poorly with problem size, leading to long planning cycles and high inference costs due to the combinatorial growth of possible robot-target allocations and possible trajectories. Learning-based methods are a viable alternative that move the scaling concern from runtime to training time, setting a critical step towards achieving real-time planning. In this work, we present a collaborative planning strategy based on Multi-Agent Proximal Policy Optimization (MAPPO) to coordinate a team of heterogeneous robots to solve a complex target allocation and scheduling problem. We benchmark our approach against single-objective optimal solutions obtained through exhaustive search and evaluate its ability to perform online replanning in the context of a planetary exploration scenario.

As LLM agents tackle increasingly complex tasks, a critical question is whether they can maintain strategic coherence over long horizons: planning under uncertainty, learning from delayed feedback, and adapting when early mistakes compound. We introduce $\texttt{YC-Bench}$, a benchmark that evaluates these capabilities by tasking an agent with running a simulated startup over a one-year horizon spanning hundreds of turns. The agent must manage employees, select task contracts, and maintain profitability in a partially observable environment where adversarial clients and growing payroll create compounding consequences for poor decisions. We evaluate 12 models, both proprietary and open source, across 3 seeds each. Only three models consistently surpass the starting capital of \$200K, with Claude Opus 4.6 achieving the highest average final funds at \$1.27 M, followed by GLM-5 at \$1.21 M at 11$\times$ lower inference cost. Scratchpad usage, the sole mechanism for persisting information across context truncation, is the strongest predictor of success, and adversarial client detection is the primary failure mode, accounting for $47\%$ of bankruptcies. Our analysis reveals that frontier models still fail through distinct failure modes such as over-parallelization, demonstrating the capability gaps for long-horizon performance. $\texttt{YC-Bench}$ is open-source, reproducible, and configurable.

Scientific algorithm discovery is iterative: hypotheses are proposed, implemented, stress-tested, and revised. Current LLM-guided search systems accelerate proposal generation, but often under-represent scientific structure by optimizing code-only artifacts with weak correctness/originality gating. We present CliffSearch, an agentic evolutionary framework in which the core evolution operators (pair selection, crossover, mutation, and review) are implemented as LLM agents, and the loop is designed around three principles: (1) each node is a structured scientific artifact, instantiated in either theory+code or code_only mode, (2) reviewer judgments of correctness and originality are first-class selection gates alongside optimization of the benchmark metric of interest, and (3) mutation is split into exploration and correction pathways with distinct objectives. Exploration mutation imports ideas from adjacent scientific domains to increase novelty, while correction mutation performs targeted evidence-guided repair using reviewer signals over theory, code, benchmark results, and runtime errors. We illustrate the framework on three benchmark-grounded studies: transformer hyper-connection evolution, optimizer discovery on a fixed nanoGPT stack, and a smaller native-optimizer ablation. Across these settings, the same loop supports explicit metric direction, reproducible persistence, and reviewer-gated comparison of discoveries under controlled search conditions. The result is a discovery workflow that prioritizes scientific interpretability and correctness while optimizing task metrics under controlled novelty constraints, rather than maximizing candidate throughput alone. Full run artifacts, interactive visualizations, and exported best nodes for the reported studies are available at https://cliffsearch.ai .

We present RELISH (REgression with a Latent Iterative State Head), a novel, lightweight architecture designed for text regression with large language models. Rather than decoding numeric targets as text or aggregating multiple generated outputs, RELISH predicts scalar values directly from frozen LLM representations by iteratively refining a learned latent state through cross-attention over token-level representations, and then mapping the final state to a point estimate with a linear regressor. Across five datasets, four LLM backbones, and two LLM training regimes, RELISH consistently outperforms prior baselines from all three major LLM regression families, including autoregressive decoding, regression-aware inference, and existing predictive head methods. Despite these gains, RELISH remains highly parameter-efficient, requiring only 3.4-3.7M trainable parameters across frozen LLM backbones (only 0.01-0.04% additional overhead), far less than LoRA-based alternatives that grow with model size (0.26-0.42%).

Primitive-based methods such as 3D Gaussian Splatting have recently become the state-of-the-art for novel-view synthesis and related reconstruction tasks. Compared to neural fields, these representations are more flexible, adaptive, and scale better to large scenes. However, the limited expressivity of individual primitives makes modeling high-frequency detail challenging. We introduce Neural Harmonic Textures, a neural representation approach that anchors latent feature vectors on a virtual scaffold surrounding each primitive. These features are interpolated within the primitive at ray intersection points. Inspired by Fourier analysis, we apply periodic activations to the interpolated features, turning alpha blending into a weighted sum of harmonic components. The resulting signal is then decoded in a single deferred pass using a small neural network, significantly reducing computational cost. Neural Harmonic Textures yield state-of-the-art results in real-time novel view synthesis while bridging the gap between primitive- and neural-field-based reconstruction. Our method integrates seamlessly into existing primitive-based pipelines such as 3DGUT, Triangle Splatting, and 2DGS. We further demonstrate its generality with applications to 2D image fitting and semantic reconstruction.

We consider the question: when a large language reasoning model makes a choice, did it think first and then decide to, or decide first and then think? In this paper, we present evidence that detectable, early-encoded decisions shape chain-of-thought in reasoning models. Specifically, we show that a simple linear probe successfully decodes tool-calling decisions from pre-generation activations with very high confidence, and in some cases, even before a single reasoning token is produced. Activation steering supports this causally: perturbing the decision direction leads to inflated deliberation, and flips behavior in many examples (between 7 - 79% depending on model and benchmark). We also show through behavioral analysis that, when steering changes the decision, the chain-of-thought process often rationalizes the flip rather than resisting it. Together, these results suggest that reasoning models can encode action choices before they begin to deliberate in text.

Search agents, which integrate language models (LMs) with web search, are becoming crucial for answering complex user queries. Constructing training datasets for deep research tasks, involving multi-step retrieval and reasoning, remains challenging due to expensive human annotation, or cumbersome prerequisites. In this work, we introduce ORBIT, a training dataset with 20K reasoning-intensive queries with short verifiable answers, generated using a frugal framework without relying on paid API services. The modular framework relies on four stages: seed creation, question--answer pair generation, and two stages of verification: self and external. ORBIT spans 15 domains and each training pair requires 4--5 reasoning steps, with external search verification required from the complete web. We train Qwen3-4B as the base model on ORBIT using GRPO and evaluate it on Wikipedia question answering tasks. Extensive experiment results demonstrate that ORBIT-4B achieves strong performance among sub-4B LLMs as search agents, proving the utility of synthetic datasets. Our framework, code and datasets are open-sourced and available publicly.

Can a large language model (LLM) improve at code generation using only its own raw outputs, without a verifier, a teacher model, or reinforcement learning? We answer in the affirmative with simple self-distillation (SSD): sample solutions from the model with certain temperature and truncation configurations, then fine-tune on those samples with standard supervised fine-tuning. SSD improves Qwen3-30B-Instruct from 42.4% to 55.3% pass@1 on LiveCodeBench v6, with gains concentrating on harder problems, and it generalizes across Qwen and Llama models at 4B, 8B, and 30B scale, including both instruct and thinking variants. To understand why such a simple method can work, we trace these gains to a precision-exploration conflict in LLM decoding and show that SSD reshapes token distributions in a context-dependent way, suppressing distractor tails where precision matters while preserving useful diversity where exploration matters. Taken together, SSD offers a complementary post-training direction for improving LLM code generation.

This study investigates the ability of multimodal Large Language Models (LLMs) to identify and interpret misleading visualizations, and recognize these observations along with their underlying causes and potential intentionality. Our analysis leverages concepts from visualization rhetoric and a newly developed taxonomy of authorial intents as explanatory lenses. We formulated three research questions and addressed them experimentally using a dataset of 2,336 COVID-19-related tweets, half of which contain misleading visualizations, and supplemented it with real-world examples of perceptual, cognitive, and conceptual errors drawn from VisLies, the IEEE VIS community event dedicated to showcasing deceptive and misleading visualizations. To ensure broad coverage of the current LLM landscape, we evaluated 16 state-of-the-art models. Among them, 15 are open-weight models, spanning a wide range of model sizes, architectural families, and reasoning capabilities. The selection comprises small models, namely Nemotron-Nano-V2-VL (12B parameters), Mistral-Small-3.2 (24B), DeepSeek-VL2 (27B), Gemma3 (27B), and GTA1 (32B); medium-sized models, namely Qianfan-VL (70B), Molmo (72B), GLM-4.5V (108B), LLaVA-NeXT (110B), and Pixtral-Large (124B); and large models, namely Qwen3-VL (235B), InternVL3.5 (241B), Step3 (321B), Llama-4-Maverick (400B), and Kimi-K2.5 (1000B). In addition, we employed OpenAI GPT-5.4, a frontier proprietary model. To establish a human perspective on these tasks, we also conducted a user study with visualization experts to assess how people perceive rhetorical techniques and the authorial intentions behind the same misleading visualizations. This allows comparison between model and expert behavior, revealing similarities and differences that provide insights into where LLMs align with human judgment and where they diverge.

Foundation vision-language models are becoming increasingly relevant to robotics because they can provide richer semantic perception than narrow task-specific pipelines. However, their practical adoption in robot software stacks still depends on reproducible middleware integrations rather than on model quality alone. Florence-2 is especially attractive in this regard because it unifies captioning, optical character recognition, open-vocabulary detection, grounding and related vision-language tasks within a comparatively manageable model size. This article presents a ROS 2 wrapper for Florence-2 that exposes the model through three complementary interaction modes: continuous topic-driven processing, synchronous service calls and asynchronous actions. The wrapper is designed for local execution and supports both native installation and Docker container deployment. It also combines generic JSON outputs with standard ROS 2 message bindings for detection-oriented tasks. A functional validation is reported together with a throughput study on several GPUs, showing that local deployment is feasible with consumer grade hardware. The repository is publicly available here: https://github.com/JEDominguezVidal/florence2_ros2_wrapper

A core limitation of standard softmax attention is that it does not define a notion of absolute query--key relevance: attention weights are obtained by redistributing a fixed unit mass across all keys according to their relative scores. As a result, relevance is defined only relative to competing keys, and irrelevant keys cannot be explicitly rejected. We introduce Multiscreen, a language-model architecture built around a mechanism we call screening, which enables absolute query--key relevance. Instead of redistributing attention across all keys, screening evaluates each key against an explicit threshold, discarding irrelevant keys and aggregating the remaining keys, thereby removing global competition among keys. Across experiments, Multiscreen achieves comparable validation loss with approximately 40% fewer parameters than a Transformer baseline, enables stable optimization at substantially larger learning rates, maintains strong performance in long-context perplexity, shows little to no degradation in retrieval performance even far beyond the training context length, and reduces inference latency by up to 3.2$\times$ at 100K context length.

Accurate air quality forecasting is crucial for protecting public health and guiding environmental policy, yet it remains challenging due to nonlinear spatiotemporal dynamics, wind-driven transport, and distribution shifts across regions. Physics-based models are interpretable but computationally expensive and often rely on restrictive assumptions, whereas purely data-driven models can be accurate but may lack robustness and calibrated uncertainty. To address these limitations, we propose Neural Dynamic Diffusion-Advection Fields (NeuroDDAF), a physics-informed forecasting framework that unifies neural representation learning with open-system transport modeling. NeuroDDAF integrates (i) a GRU-Graph Attention encoder to capture temporal dynamics and wind-aware spatial interactions, (ii) a Fourier-domain diffusion-advection module with learnable residuals, (iii) a wind-modulated latent Neural ODE to model continuous-time evolution under time-varying connectivity, and (iv) an evidential fusion mechanism that adaptively combines physics-guided and neural forecasts while quantifying uncertainty. Experiments on four urban datasets (Beijing, Shenzhen, Tianjin, and Ancona) across 1-3 day horizons show that NeuroDDAF consistently outperforms strong baselines, including AirPhyNet, achieving up to 9.7% reduction in RMSE and 9.4% reduction in MAE on long-term forecasts. On the Beijing dataset, NeuroDDAF attains an RMSE of 41.63 $μ$g/m$^3$ for 1-day prediction and 48.88 $μ$g/m$^3$ for 3-day prediction, representing the best performance among all compared methods. In addition, NeuroDDAF improves cross-city generalization and yields well-calibrated uncertainty estimates, as confirmed by ensemble variance analysis and case studies under varying wind conditions.

Uncertainty quantification is essential when deploying learning-based control methods in safety-critical systems. This is commonly realized by constructing uncertainty tubes that enclose the unknown function of interest, e.g., the reward and constraint functions or the underlying dynamics model, with high probability. However, existing approaches for uncertainty quantification typically rely on restrictive assumptions on the unknown function, such as known bounds on functional norms or Lipschitz constants, and struggle with discontinuities. In this paper, we model the unknown function as a random function from which independent and identically distributed realizations can be generated, and construct uncertainty tubes via the scenario approach that hold with high probability and rely solely on the sampled realizations. We integrate these uncertainty tubes into a safe Bayesian optimization algorithm, which we then use to safely tune control parameters on a real Furuta pendulum.

While test-time scaling has enabled large language models to solve highly difficult tasks, state-of-the-art results come at exorbitant compute costs. These inefficiencies can be attributed to the miscalibration of post-trained language models, and the lack of calibration in popular sampling techniques. Here, we present Online Reasoning Calibration (ORCA), a framework for calibrating the sampling process that draws upon conformal prediction and test-time training. Specifically, we introduce a meta-learning procedure that updates the calibration module for each input. This allows us to provide valid confidence estimates under distributional shift, e.g. in thought patterns that occur across different stages of reasoning, or in prompt distributions between model development and deployment. ORCA not only provides theoretical guarantees on conformal risks, but also empirically shows higher efficiency and generalization across different reasoning tasks. At risk level $δ=0.1$, ORCA improves Qwen2.5-32B efficiency on in-distribution tasks with savings up to 47.5% with supervised labels and 40.7% with self-consistency labels. Under zero-shot out-of-domain settings, it improves MATH-500 savings from 24.8% of the static calibration baseline to 67.0% while maintaining a low empirical error rate, and the same trend holds across model families and downstream benchmarks. Our code is publicly available at https://github.com/wzekai99/ORCA.

A fundamental challenge in science and engineering is the simulation-to-experiment gap. While we often possess prior knowledge of physical laws, these physical laws can be too difficult to solve exactly for complex systems. Such systems are commonly modeled using simulators, which impose computational approximations. Meanwhile, experimental measurements more faithfully represent the real world, but experimental data typically consists of observations that only partially reflect the system's full underlying state. We propose a data-driven distribution alignment framework that bridges this simulation-to-experiment gap by pre-training a generative model on fully observed (but imperfect) simulation data, then aligning it with partial (but real) observations of experimental data. While our method is domain-agnostic, we ground our approach in the physical sciences by introducing Adversarial Distribution Alignment (ADA). This method aligns a generative model of atomic positions -- initially trained on a simulated Boltzmann distribution -- with the distribution of experimental observations. We prove that our method recovers the target observable distribution, even with multiple, potentially correlated observables. We also empirically validate our framework on synthetic, molecular, and experimental protein data, demonstrating that it can align generative models with diverse observables. Our code is available at https://kaityrusnelson.com/ada/.

Using roughly 48 execution-verified HumanEval training solutions, tuning a single initial state matrix per recurrent layer, with zero inference overhead, outperforms LoRA by +10.8 pp (p < 0.001) on HumanEval. The method, which we call S0 tuning, optimizes one state matrix per recurrent layer while freezing all model weights. On Qwen3.5-4B (GatedDeltaNet hybrid), S0 tuning improves greedy pass@1 by +23.6 +/- 1.7 pp (10 seeds). On FalconH1-7B (Mamba-2 hybrid), S0 reaches 71.8% +/- 1.3 and LoRA reaches 71.4% +/- 2.4 (3 seeds), statistically indistinguishable at this sample size while requiring no weight merging. Cross-domain transfer is significant on MATH-500 (+4.8 pp, p = 0.00002, 8 seeds) and GSM8K (+2.8 pp, p = 0.0003, 10 seeds); a text-to-SQL benchmark (Spider) shows no transfer, consistent with the trajectory-steering mechanism. A prefix-tuning control on a pure Transformer (Qwen2.5-3B) degrades performance by -13.9 pp under all nine configurations tested. On Qwen3.5, a per-step state-offset variant reaches +27.1 pp, above both S0 and LoRA but with per-step inference cost. Taken together, the results show that recurrent state initialization is a strong zero-inference-overhead PEFT surface for hybrid language models when verified supervision is scarce. The tuned state is a ~48 MB file; task switching requires no weight merging or model reload. Code and library: https://github.com/jackyoung27/s0-tuning.

Chest X-ray (CXR) segmentation is an important step in computer-aided diagnosis, yet deploying large foundation models in clinical settings remains challenging due to computational constraints. We propose AdaLoRA-QAT, a two-stage fine-tuning framework that combines adaptive low-rank encoder adaptation with full quantization-aware training. Adaptive rank allocation improves parameter efficiency, while selective mixed-precision INT8 quantization preserves structural fidelity crucial for clinical reliability. Evaluated across large-scale CXR datasets, AdaLoRA-QAT achieves 95.6% Dice, matching full-precision SAM decoder fine-tuning while reducing trainable parameters by 16.6\times and yielding 2.24\times model compression. A Wilcoxon signed-rank test confirms that quantization does not significantly degrade segmentation accuracy. These results demonstrate that AdaLoRA-QAT effectively balances accuracy, efficiency, and structural trust-worthiness, enabling compact and deployable foundation models for medical image segmentation. Code and pretrained models are available at: https://prantik-pdeb.github.io/adaloraqat.github.io/

Large language models (LLMs) exhibiting test-time scaling behavior, such as extended reasoning traces and self-verification, have demonstrated remarkable performance on complex, long-term reasoning tasks. However, the robustness of these reasoning behaviors remains underexplored. To investigate this, we conduct a systematic evaluation of multiple reasoning models across three scenarios: (1) problems augmented with lengthy, irrelevant context; (2) multi-turn conversational settings with independent tasks; and (3) problems presented as a subtask within a complex task. We observe an interesting phenomenon: reasoning models tend to produce much shorter reasoning traces (up to 50%) for the same problem under different context conditions compared to the traces produced when the problem is presented in isolation. A finer-grained analysis reveals that this compression is associated with a decrease in self-verification and uncertainty management behaviors, such as double-checking. While this behavioral shift does not compromise performance on straightforward problems, it might affect performance on more challenging tasks. We hope our findings draw additional attention to both the robustness of reasoning models and the problem of context management for LLMs and LLM-based agents.

This paper argues that AI-enabled analysis of street-view imagery, complemented by performance-gated machine-learning imputation, provides a viable pathway for generating building-specific elevation data at regional scale for flood risk assessment. We develop and apply a three-stage pipeline across 18 areas of interest (AOIs) in Texas that (1) extracts LFE and the height difference between street grade and the lowest floor (HDSL) from Google Street View imagery using the Elev-Vision framework, (2) imputes missing HDSL values with Random Forest and Gradient Boosting models trained on 16 terrain, hydrologic, geographic, and flood-exposure features, and (3) integrates the resulting elevation dataset with Fathom 1-in-100 year inundation surfaces and USACE depth-damage functions to estimate property-specific interior flood depth and expected loss. Across 12,241 residential structures, street-view imagery was available for 73.4% of parcels and direct LFE/HDSL extraction was successful for 49.0% (5,992 structures). Imputation was retained for 13 AOIs where cross-validated performance was defensible, with selected models achieving R suqre values from 0.159 to 0.974; five AOIs were explicitly excluded from prediction because performance was insufficient. The results show that street-view-based elevation mapping is not universally available for every property, but it is sufficiently scalable to materially improve regional flood-risk characterization by moving beyond hazard exposure to structure-level estimates of interior inundation and expected damage. Scientifically, the study advances LFE estimation from a pilot-scale proof of concept to a regional, end-to-end workflow. Practically, it offers a replicable framework for jurisdictions that lack comprehensive Elevation Certificates but need parcel-level information to support mitigation, planning, and flood-risk management.

We present Brainstacks, a modular architecture for continual multi-domain fine-tuning of large language models that packages domain expertise as frozen adapter stacks composing additively on a shared frozen base at inference. Five interlocking components: (1) MoE-LoRA with Shazeer-style noisy top-2 routing across all seven transformer projections under QLoRA 4-bit quantization with rsLoRA scaling; (2) an inner loop performing residual boosting by freezing trained stacks and adding new ones; (3) an outer loop training sequential domain-specific stacks with curriculum-ordered dependencies; (4) null-space projection via randomized SVD constraining new stacks to subspaces orthogonal to prior directions, achieving zero forgetting in isolation; (5) an outcome-based sigmoid meta-router trained on empirically discovered domain-combination targets that selectively weights stacks, enabling cross-domain composition. Two boundary experiments: (6) PSN pretraining on a randomly initialized model; (7) per-domain RL (DPO/GRPO) validating compatibility with post-SFT alignment. Validated on TinyLlama-1.1B (4 domains, 9 stacks) and Gemma 3 12B IT (5 domains, 10 stacks), MoE-LoRA achieves 2.5x faster convergence than parameter-matched single LoRA, residual boosting breaks through the single-stack ceiling, and the routed system recovers generation quality destroyed by ungated stack accumulation. The central finding: the outcome-based router discovers that domain stacks encode transferable cognitive primitives (instruction-following clarity, numerical reasoning, procedural logic, chain-of-thought structure) rather than domain-specific knowledge, with medical prompts routing to chat+math stacks in 97% of cases despite zero medical data in those stacks.

As LLM agents are increasingly deployed in multi-agent systems, they introduce risks of covert coordination that may evade standard forms of human oversight. While linear probes on model activations have shown promise for detecting deception in single-agent settings, collusion is inherently a multi-agent phenomenon, and the use of internal representations for detecting collusion between agents remains unexplored. We introduce NARCBench, a benchmark for evaluating collusion detection under environment distribution shift, and propose five probing techniques that aggregate per-agent deception scores to classify scenarios at the group level. Our probes achieve 1.00 AUROC in-distribution and 0.60--0.86 AUROC when transferred zero-shot to structurally different multi-agent scenarios and a steganographic blackjack card-counting task. We find that no single probing technique dominates across all collusion types, suggesting that different forms of collusion manifest differently in activation space. We also find preliminary evidence that this signal is localised at the token level, with the colluding agent's activations spiking specifically when processing the encoded parts of their partner's message. This work takes a step toward multi-agent interpretability: extending white-box inspection from single models to multi-agent contexts, where detection requires aggregating signals across agents. These results suggest that model internals provide a complementary signal to text-level monitoring for detecting multi-agent collusion, particularly for organisations with access to model activations. Code and data are available at https://github.com/aaronrose227/narcbench.

Most defects in mobile applications are visually observable on the device screen. To track these defects, users, testers, and developers must manually submit bug reports, especially in the absence of crashes. However, these reports are frequently ambiguous or inaccurate, often omitting essential components such as the Observed Behavior (OB), Expected Behavior (EB), or Steps to Reproduce (S2Rs). Low-quality reports hinder developers' ability to understand and reproduce defects, delaying resolution and leading to incorrect or unresolvable fixes. In this paper, we posit that providing specific app-related information (e.g., GUI interactions or specific screens where bugs appear) to LLMs as key points of context can assist in automatically generating clear, detailed, and accurate OB, EB, and S2Rs. We built and evaluated a novel approach, BugScribe, that generates bug reports in this way. To support the evaluation, we introduce a unified quality framework that defines correctness and completeness dimensions for OB, EB, and S2Rs. Using 48 bug reports from 26 Android apps, we show that BugScribe produces higher-quality and more accurate components than the original reports and outperforms recent LLM-based baselines. We envision that BugScribe can serve as a practical assistant for testers and developers by enhancing incomplete bug reports with reliable and accurate OB, EB, and S2Rs, thereby streamlining bug resolution and improving mobile app quality.

As Large Language Models (LLMs) for code increasingly utilize massive, often non-permissively licensed datasets, evaluating data contamination through Membership Inference Attacks (MIAs) has become critical. We propose SERSEM (Selective Entropy-Weighted Scoring for Membership Inference), a novel white-box attack framework that suppresses uninformative syntactical boilerplate to amplify specific memorization signals. SERSEM utilizes a dual-signal methodology: first, a continuous character-level weight mask is derived through static Abstract Syntax Tree (AST) analysis, spellchecking-based multilingual logic detection, and offline linting. Second, these heuristic weights are used to pool internal transformer activations and calibrate token-level Z-scores from the output logits. Evaluated on a 25,000-sample balanced dataset, SERSEM achieves a global AUC-ROC of 0.7913 on the StarCoder2-3B model and 0.7867 on the StarCoder2-7B model, consistently outperforming the implemented probability-based baselines Loss, Min-K% Prob, and PAC. Our findings demonstrate that focusing on human-centric coding anomalies provides a significantly more robust indicator of verbatim memorization than sequence-level probability averages.

Reinforcement learning has shown strong performance in robotic manipulation, but learned policies often degrade in performance when test conditions differ from the training distribution. This limitation is especially important in contact-rich tasks such as pushing and pick-and-place, where changes in goals, contact conditions, or robot dynamics can drive the system out-of-distribution at inference time. In this paper, we investigate a hybrid controller that combines reinforcement learning with bounded extremum seeking to improve robustness under such conditions. In the proposed approach, deep deterministic policy gradient (DDPG) policies are trained under standard conditions on the robotic pushing and pick-and-place tasks, and are then combined with bounded ES during deployment. The RL policy provides fast manipulation behavior, while bounded ES ensures robustness of the overall controller to time variations when operating conditions depart from those seen during training. The resulting controller is evaluated under several out-of-distribution settings, including time-varying goals and spatially varying friction patches.

Due to the large footprint of pixels in remote sensing imagery, hyperspectral unmixing (HU) has become an important and necessary procedure in hyperspectral image analysis. Traditional HU methods rely on a prior spectral mixing model, especially for nonlinear mixtures, which has largely limited the performance and generalization capacity of the unmixing approach. In this paper, we address the challenging problem of hyperspectral nonlinear unmixing (HNU) without explicit knowledge of the mixing model. Inspired by the principle of generative models, where images of the same distribution can be generated as that of the training images without knowing the exact probability distribution function of the image, we develop an invertible mixing-unmixing process via a bi-directional GAN framework, constrained by both the cycle consistency and the linkage between linear and nonlinear mixtures. The combination of cycle consistency and linear linkage provides powerful constraints without requiring an explicit mixing model. We refer to the proposed approach as the linearly-constrained CycleGAN unmixing net, or LCGU net. Experimental results indicate that the proposed LCGU net exhibits stable and competitive performance across different datasets compared with other state-of-the-art model-based HNU methods.

Individuals engaging in online communication frequently express personal opinions with informal styles (e.g., memes and emojis). While Language Models (LMs) with informal communications have been widely discussed, a unique and emphatic style, the Repetitive Lengthening Form (RLF), has been overlooked for years. In this paper, we explore answers to two research questions: 1) Is RLF important for sentiment analysis (SA)? 2) Can LMs understand RLF? Inspired by previous linguistic research, we curate \textbf{Lengthening}, the first multi-domain dataset with 850k samples focused on RLF for SA. Moreover, we introduce \textbf{Exp}lainable \textbf{Instruct}ion Tuning (\textbf{ExpInstruct}), a two-stage instruction tuning framework aimed to improve both performance and explainability of LLMs for RLF. We further propose a novel unified approach to quantify LMs' understanding of informal expressions. We show that RLF sentences are expressive expressions and can serve as signatures of document-level sentiment. Additionally, RLF has potential value for online content analysis. Our results show that fine-tuned Pre-trained Language Models (PLMs) can surpass zero-shot GPT-4 in performance but not in explanation for RLF. Finally, we show ExpInstruct can improve the open-sourced LLMs to match zero-shot GPT-4 in performance and explainability for RLF with limited samples. Code and sample data are available at https://github.com/Tom-Owl/OverlookedRLF

File-level defect prediction models traditionally rely on product and process metrics. While process metrics effectively complement product metrics, they often overlook commit size the number of files changed per commit despite its strong association with software quality. Network centrality measures on dependency graphs have also proven to be valuable product level indicators. Motivated by this, we first redefine process metrics as commit size aware process metric vectors, transforming conventional scalar measures into 100 dimensional profiles that capture the distribution of changes across commit size strata. We then model change history as a hyper co change graph, where hyperedges naturally encode commit-size semantics. Vector centralities computed on these hypergraphs quantify size-aware node importance for source files. Experiments on nine long-lived Apache projects using five popular classifiers show that replacing scalar process metrics with the proposed commit size aware vectors, alongside product metrics, consistently improves predictive performance. These findings establish that commit size aware process metrics and hypergraph based vector centralities capture higher-order change semantics, leading to more discriminative, better calibrated, and statistically superior defect prediction models.

As sports training becomes more data-driven, traditional dart coaching based mainly on experience and visual observation is increasingly inadequate for high-precision, goal-oriented movements. Although prior studies have highlighted the importance of release parameters, joint motion, and coordination in dart throwing, most quantitative methods still focus on local variables, single-release metrics, or static template matching. These approaches offer limited support for personalized training and often overlook useful movement variability. This paper presents a data-driven dart training assistance system. The system creates a closed-loop framework spanning motion capture, feature modeling, and personalized feedback. Dart-throwing data were collected in markerless conditions using a Kinect 2.0 depth sensor and an optical camera. Eighteen kinematic features were extracted from four biomechanical dimensions: three-link coordination, release velocity, multi-joint angular configuration, and postural stability. Two modules were developed: a personalized optimal throwing trajectory model that combines historical high-quality samples with the minimum jerk criterion, and a motion deviation diagnosis and recommendation model based on z-scores and hierarchical logic. A total of 2,396 throwing samples from professional and non-professional athletes were collected. Results show that the system generates smooth personalized reference trajectories consistent with natural human movement. Case studies indicate that it can detect poor trunk stability, abnormal elbow displacement, and imbalanced velocity control, then provide targeted recommendations. The framework shifts dart evaluation from deviation from a uniform standard to deviation from an individual's optimal control range, improving personalization and interpretability for darts training and other high-precision target sports.

This paper introduces the first systematic evaluation framework for quantifying the quality and risks of papers written by modern coding agents. While AI-driven paper writing has become a growing concern, rigorous evaluation of the quality and potential risks of AI-written papers remains limited, and a unified understanding of their reliability is still lacking. We introduce Paper Reconstruction Evaluation (PaperRecon), an evaluation framework in which an overview (overview.md) is created from an existing paper, after which an agent generates a full paper based on the overview and minimal additional resources, and the result is subsequently compared against the original paper. PaperRecon disentangles the evaluation of the AI-written papers into two orthogonal dimensions, Presentation and Hallucination, where Presentation is evaluated using a rubric and Hallucination is assessed via agentic evaluation grounded in the original paper source. For evaluation, we introduce PaperWrite-Bench, a benchmark of 51 papers from top-tier venues across diverse domains published after 2025. Our experiments reveal a clear trade-off: while both ClaudeCode and Codex improve with model advances, ClaudeCode achieves higher presentation quality at the cost of more than 10 hallucinations per paper on average, whereas Codex produces fewer hallucinations but lower presentation quality. This work takes a first step toward establishing evaluation frameworks for AI-driven paper writing and improving the understanding of its risks within the research community.

Leveraging the rich semantic features of vision-language models (VLMs) like CLIP for monocular depth estimation tasks is a promising direction, yet often requires extensive fine-tuning or lacks geometric precision. We present a parameter-efficient framework, named MoA-DepthCLIP, that adapts pretrained CLIP representations for monocular depth estimation with minimal supervision. Our method integrates a lightweight Mixture-of-Adapters (MoA) module into the pretrained Vision Transformer (ViT-B/32) backbone combined with selective fine-tuning of the final layers. This design enables spatially-aware adaptation, guided by a global semantic context vector and a hybrid prediction architecture that synergizes depth bin classification with direct regression. To enhance structural accuracy, we employ a composite loss function that enforces geometric constraints. On the NYU Depth V2 benchmark, MoA-DepthCLIP achieves competitive results, significantly outperforming the DepthCLIP baseline by improving the $δ_1$ accuracy from 0.390 to 0.745 and reducing the RMSE from 1.176 to 0.520. These results are achieved while requiring substantially few trainable parameters, demonstrating that lightweight, prompt-guided MoA is a highly effective strategy for transferring VLM knowledge to fine-grained monocular depth estimation tasks.

Dependency networks (Heckerman et al., 2000) provide a flexible framework for modeling complex systems with many variables by combining independently learned local conditional distributions through pseudo-Gibbs sampling. Despite their computational advantages over Bayesian and Markov networks, the theoretical foundations of dependency networks remain incomplete, primarily because their model distributions -- defined as stationary distributions of pseudo-Gibbs sampling -- lack closed-form expressions. This paper develops an information-geometric analysis of pseudo-Gibbs sampling, interpreting each sampling step as an m-projection onto a full conditional manifold. Building on this interpretation, we introduce the full conditional divergence and derive an upper bound that characterizes the location of the stationary distribution in the space of probability distributions. We then reformulate both structure and parameter learning as optimization problems that decompose into independent subproblems for each node, and prove that the learned model distribution converges to the true underlying distribution as the number of training samples grows to infinity. Experiments confirm that the proposed upper bound is tight in practice.

As generative AI systems are integrated into educational settings, students often encounter AI-generated output while working through learning tasks, either by requesting help or through integrated tools. Trust in AI can influence how students interpret and use that output, including whether they evaluate it critically or exhibit overreliance. We investigate how students' trust relates to their appropriate reliance on an AI assistant during programming problem-solving tasks, and whether this relationship differs by learner characteristics. With 432 undergraduate participants, students' completed Python output-prediction problems while receiving recommendations and explanations from an AI chatbot, including accurate and intentionally misleading suggestions. We operationalize reliance behaviorally as the extent to which students' responses reflected appropriate use of the AI assistant's suggestions, accepting them when they were correct and rejecting them when they were incorrect. Pre- and post-task surveys assessed trust in the assistant, AI literacy, need for cognition, programming self-efficacy, and programming literacy. Results showed a non-linear relationship in which higher trust was associated with lower appropriate reliance, suggesting weaker discrimination between correct and incorrect recommendations. This relationship was significantly moderated by students' AI literacy and need for cognition. These findings highlight the need for future work on instructional and system supports that encourage more reflective evaluation of AI assistance during problem-solving.

Large language model (LLM) systems are increasingly used to support high-stakes decision-making, but they typically perform worse when the available evidence is internally inconsistent. Such a scenario exists in real-world healthcare settings, with patient-reported symptoms contradicting medical signs. To study this problem, we introduce MIMIC-DOS, a dataset for short-horizon organ dysfunction worsening prediction in the intensive care unit (ICU) setting. We derive this dataset from the widely recognized MIMIC-IV, a publicly available electronic health record dataset, and construct it exclusively from cases in which discordance between signs and symptoms exists. This setting poses a substantial challenge for existing LLM-based approaches, with single-pass LLMs and agentic pipelines often struggling to reconcile such conflicting signals. To address this problem, we propose CARE: a multi-stage privacy-compliant agentic reasoning framework in which a remote LLM provides guidance by generating structured categories and transitions without accessing sensitive patient data, while a local LLM uses these categories and transitions to support evidence acquisition and final decision-making. Empirically, CARE achieves stronger performance across all key metrics compared to multiple baseline settings, showing that CARE can more robustly handle conflicting clinical evidence while preserving privacy.

Software engineering students often struggle to appreciate empirical methods and hypothesis-driven inquiry, especially when taught in theoretical terms. This experience report explores whether grounding empirical learning in hype-driven technologies can make these concepts more accessible and engaging. We conducted a one-semester seminar framed around the currently popular topic of AI coding assistants, which attracted unusually high student interest. The course combined hands-on sessions using AI coding assistants with small, student-designed empirical studies. Classroom observations and survey responses suggest that the hype topic sparked curiosity and critical thinking. Students engaged with the AI coding assistants while questioning their limitations -- developing the kind of empirical thinking needed to assess claims about emerging technologies. Key lessons: (1) Hype-driven topics can lower barriers to abstract concepts like empirical research; (2) authentic hands-on development tasks combined with ownership of inquiry foster critical engagement; and (3) a single seminar can effectively teach both technical and research skills.

Evaluating the ethical robustness of large language models (LLMs) deployed in software systems remains challenging, particularly under sustained adversarial user interaction. Existing safety benchmarks typically rely on single-round evaluations and aggregate metrics, such as toxicity scores and refusal rates, which offer limited visibility into behavioral instability that may arise during realistic multi-turn interactions. As a result, rare but high-impact ethical failures and progressive degradation effects may remain undetected prior to deployment. This paper introduces Adversarial Moral Stress Testing (AMST), a stress-based evaluation framework for assessing ethical robustness under adversarial multi-round interactions. AMST applies structured stress transformations to prompts and evaluates model behavior through distribution-aware robustness metrics that capture variance, tail risk, and temporal behavioral drift across interaction rounds. We evaluate AMST on several state-of-the-art LLMs, including LLaMA-3-8B, GPT-4o, and DeepSeek-v3, using a large set of adversarial scenarios generated under controlled stress conditions. The results demonstrate substantial differences in robustness profiles across models and expose degradation patterns that are not observable under conventional single-round evaluation protocols. In particular, robustness has been shown to depend on distributional stability and tail behavior rather than on average performance alone. Additionally, AMST provides a scalable and model-agnostic stress-testing methodology that enables robustness-aware evaluation and monitoring of LLM-enabled software systems operating in adversarial environments.

Inverse design of optical multilayer stacks seeks to infer layer materials, thicknesses, and ordering from a desired target spectrum. It is a long-standing challenge due to the large design space and non-unique solutions. We introduce \texttt{OptoLlama}, a masked diffusion language model for inverse thin-film design from optical spectra. Representing multilayer stacks as sequences of material-thickness tokens, \texttt{OptoLlama} conditions generation on reflectance, absorptance, and transmittance spectra and learns a probabilistic mapping from optical response to structure. Evaluated on a representative test set of 3,000 targets, \texttt{OptoLlama} reduces the mean absolute spectral error by 2.9-fold relative to a nearest-neighbor template baseline and by 3.45-fold relative to the state-of-the-art data-driven baseline, called \texttt{OptoGPT}. Case studies on designed and expert-defined targets show that the model reproduces characteristic spectral features and recovers physically meaningful stack motifs, including distributed Bragg reflectors. These results establish diffusion-based sequence modeling as a powerful framework for inverse photonic design.

Stochastic Multi-Objective Optimization (SMOO) is critical for decision-making trading off multiple potentially conflicting objectives in uncertain environments. SMOO aims at identifying the Pareto frontier, which contains all mutually non-dominating decisions. The problem is highly intractable due to the embedded probabilistic inference, such as computing the marginal, posterior probabilities, or expectations. Existing methods, such as scalarization, sample average approximation, and evolutionary algorithms, either offer arbitrarily loose approximations or may incur prohibitive computational costs. We propose XOR-SMOO, a novel algorithm that with probability $1-δ$, obtains $γ$-approximate Pareto frontiers ($γ>1$) for SMOO by querying an SAT oracle poly-log times in $γ$ and $δ$. A $γ$-approximate Pareto frontier is only below the true frontier by a fixed, multiplicative factor $γ$. Thus, XOR-SMOO solves highly intractable SMOO problems (\#P-hard) with only queries to SAT oracles while obtaining tight, constant factor approximation guarantees. Experiments on real-world road network strengthening and supply chain design problems demonstrate that XOR-SMOO outperforms several baselines in identifying Pareto frontiers that have higher objective values, better coverage of the optimal solutions, and the solutions found are more evenly distributed. Overall, XOR-SMOO significantly enhanced the practicality and reliability of SMOO solvers.

Large language models (LLMs) exhibit strong in-context learning capabilities, but how they track and retrieve information from context remains underexplored. Drawing on the free recall paradigm in cognitive science (where participants recall list items in any order), we show that several open-source LLMs consistently display a serial-recall-like pattern, assigning peak probability to tokens that immediately follow a repeated token in the input sequence. Through systematic ablation experiments, we show that induction heads, specialized attention heads that attend to the token following a previous occurrence of the current token, play an important role in this phenomenon. Removing heads with a high induction score substantially reduces the +1 lag bias, whereas ablating random heads does not reproduce the same reduction. We also show that removing heads with high induction scores impairs the performance of models prompted to do serial recall using few-shot learning to a larger extent than removing random heads. Our findings highlight a mechanistically specific connection between induction heads and temporal context processing in transformers, suggesting that these heads are especially important for ordered retrieval and serial-recall-like behavior during in-context learning.

WiFi is inherently vulnerable to eavesdropping because RF signals may penetrate many physical boundaries, such as walls and floors. LiFi, by contrast, is an optical method confined to line-of-sight and blocked by opaque surfaces. We present LightGuard, a dual-link architecture built on this insight: cryptographic key establishment can be offloaded from WiFi to a physically confined LiFi channel to mitigate the risk of key exposure over RF. LightGuard derives session keys over a LiFi link and installs them on the WiFi interface, ensuring cryptographic material never traverses the open RF medium. A prototype with off-the-shelf WiFi NICs and our LiFi transceiver frontend validates the design.

Partial audio deepfakes, where synthesized segments are spliced into genuine recordings, are particularly deceptive because most of the audio remains authentic. Existing detectors are supervised: they require frame-level annotations, overfit to specific synthesis pipelines, and must be retrained as new generative models emerge. We argue that this supervision is unnecessary. We hypothesize that speech foundation models implicitly encode a forensic signal: genuine speech forms smooth, slowly varying embedding trajectories, while splice boundaries introduce abrupt disruptions in frame-level transitions. Building on this, we propose TRACE (Training-free Representation-based Audio Countermeasure via Embedding dynamics), a training-free framework that detects partial audio deepfakes by analyzing the first-order dynamics of frozen speech foundation model representations without any training, labeled data, or architectural modification. We evaluate TRACE on four benchmarks that span two languages using six speech foundation models. In PartialSpoof, TRACE achieves 8.08% EER, competitive with fine-tuned supervised baselines. In LlamaPartialSpoof, the most challenging benchmark featuring LLM-driven commercial synthesis, TRACE surpasses a supervised baseline outright (24.12% vs. 24.49% EER) without any target-domain data. These results show that temporal dynamics in speech foundation models provide an effective, generalize signal for training-free audio forensics.

3D semantic occupancy prediction is central to autonomous driving, yet current methods are vulnerable to long-tailed class bias and out-of-distribution (OOD) inputs, often overconfidently assigning anomalies to rare classes. We present ProOOD, a lightweight, plug-and-play method that couples prototype-guided refinement with training-free OOD scoring. ProOOD comprises (i) prototype-guided semantic imputation that fills occluded regions with class-consistent features, (ii) prototype-guided tail mining that strengthens rare-class representations to curb OOD absorption, and (iii) EchoOOD, which fuses local logit coherence with local and global prototype matching to produce reliable voxel-level OOD scores. Extensive experiments on five datasets demonstrate that ProOOD achieves state-of-the-art performance on both in-distribution 3D occupancy prediction and OOD detection. On SemanticKITTI, it surpasses baselines by +3.57% mIoU overall and +24.80% tail-class mIoU; on VAA-KITTI, it improves AuPRCr by +19.34 points, with consistent gains across benchmarks. These improvements yield more calibrated occupancy estimates and more reliable OOD detection in safety-critical urban driving. The source code is publicly available at https://github.com/7uHeng/ProOOD.

There is a growing need for cybersecurity professionals with practical knowledge and experience to meet societal needs and comply with new standards and regulations. At the same time, the advances in software technology and artificial intelligence point towards a future where software agents will play an important role in protecting the computer systems that are critical for society to function. The training and development of both humans and software agents requires the design and execution of cybersecurity exercises that differ in properties such as size, scope, objectives, difficultly, etc. Cybersecurity scenarios are critical for the operation of cybersecurity exercises as they describe the scope, context, operational environment and storyline of each exercise. In this work, we present an approach to automatically generate cybersecurity scenarios that model enterprise IT systems. Our approach is able to generate a large number of scenarios that differ in multiple criteria including size, scope, difficulty, complexity and diversity. We further release as open source: a simulation and a virtualization environment that can run cybersecurity exercises based on the generated scenarios and a dataset containing 100000 sample scenarios.

3D field-programmable gate arrays (FPGAs) promise higher performance through vertical integration. However, existing placement tools, largely inherited from 2D frameworks, fail to capture the unique delay characteristics and optimization dynamics of 3D fabrics. We introduce a 3D FPGA placement flow that integrates partitioning-based initialization, adaptive cost scheduling, refined delay estimation, and a simulated annealing move set -- all targeted at 3D FPGA architecture. Together, these enhancements improve timing estimates and the exploration of layer assignments during placement. Compared to Verilog-To-Routing (VTR), our experiments show geometric-mean (max) critical-path delay reductions of ~3% (~7%), ~2% (~4%), ~3% (~8%), and ~6% (~18%) for four 3D architectures: 3D CB, 3D CB-O, 3D CB-I, and 3D SB, respectively. We also achieve geometric-mean (max) routed wirelength reductions of ~1% (~3%), ~2% (~8%), < 1% (~5%), and ~5% (~10%), respectively. Our work will be permissively open-sourced on GitHub.

The optimization of over-parameterized deep neural networks represents a large-scale, high-dimensional, and strongly non-convex decision problem that challenges existing optimization frameworks. Current evolutionary and gradient-based pruning methods often struggle to scale to such dimensionalities, as they rely on flat search spaces, scalarized objectives, or repeated retraining, leading to premature convergence and prohibitive computational cost. This paper introduces a hierarchical importance-guided evolutionary framework that reformulates convolutional network pruning as a tractable large-scale multi-objective optimization problem. In the first phase, a continuous evolutionary search performs coarse exploration of weight-wise pruning thresholds to shrink the search space and identify promising regions of the Pareto set. The second phase applies a fine-grained binary evolutionary optimization constrained to the surviving weights, where importance-aware sampling and adaptive variation operators refine local search in the sparse region of the Pareto set. This hierarchical design combines global exploration and localized exploitation to achieve a well-distributed Pareto set of networks balancing compactness and accuracy. Empirical results on CIFAR-10 and CIFAR-100 using ResNet-56 and ResNet-110 confirm the method's effectiveness compared to existing evolutionary approaches: pruning achieves up to 51.9\% and 38.9\% parameter reductions with almost no accuracy loss compared to state-of-the-art evolutionary DNN pruning methods. The proposed method contributes a scalable evolutionary approach for solving very-large-scale multi-objective optimization problems, offering a general paradigm extendable to other domains where the decision space is exponentially large, objective functions are conflicting, and efficient trade-off discovery is essential.

We test whether authors have characteristic "fingerprints" in the information-theoretic novelty curves of their published works. Working with two corpora -- Books3 (52,796 books, 759 qualifying authors) and PG-19 (28,439 books, 1,821 qualifying authors) -- we find that authorial voice leaves measurable traces in how novelty unfolds across a text. The signal is multi-scale: at book level, scalar dynamics (mean novelty, speed, volume, circuitousness) identify 43% of authors significantly above chance; at chapter level, SAX motif patterns in sliding windows achieve 30x-above-chance attribution, far exceeding the scalar features that dominate at book level. These signals are complementary, not redundant. We show that the fingerprint is partly confounded with genre but persists within-genre for approximately one-quarter of authors. Classical authors (Twain, Austen, Kipling) show fingerprints comparable in strength to modern authors, suggesting the phenomenon is not an artifact of contemporary publishing conventions.

Computational reproducibility is fundamental to trustworthy science, yet remains difficult to achieve in practice across various research workflows, including Jupyter notebooks published alongside scholarly articles. Environment drift, undocumented dependencies and implicit execution assumptions frequently prevent independent re-execution of published research. Despite existing reproducibility guidelines, scalable and systematic infrastructure for automated assessment remains limited. We present an automated, web-oriented reproducibility engineering pipeline that reconstructs and evaluates repository-level execution environments for scholarly notebooks. The system performs dependency inference, automated container generation, and isolated execution to approximate the notebook's original computational context. We evaluate the approach on 443 notebooks from 116 GitHub repositories referenced by publications in PubMed Central. Execution outcomes are classified into four categories: resolved environment failures, persistent logic or data errors, reproducibility drift, and container-induced regressions. Our results show that containerization resolves 66.7% of prior dependency-related failures and substantially improves execution robustness. However, a significant reproducibility gap remains: 53.7% of notebooks exhibit low output fidelity, largely due to persistent runtime failures and stochastic non-determinism. These findings indicate that standardized containerization is essential for computational stability but insufficient for full bit-wise reproducibility. The framework offers a scalable solution for researchers, editors, and archivists seeking systematic, automated assessment of computational artifacts.

"Vibe coding," in which developers delegate code generation to AI assistants and accept the output with little manual review, has gained rapid adoption in production settings. On March 31, 2026, Anthropic's Claude Code CLI shipped a 59.8 MB source map file in its npm package, exposing roughly 512,000 lines of proprietary TypeScript. The tool had itself been largely vibe-coded, and the leak traced to a misconfigured packaging rule rather than a logic bug. Existing static-analysis and secret-scanning tools did not cover this failure mode, pointing to a gap between the vulnerabilities AI tends to introduce and the vulnerabilities current tooling is built to find. We present VibeGuard, a pre-publish security gate that targets five such blind spots: artifact hygiene, packaging-configuration drift, source-map exposure, hardcoded secrets, and supply-chain risk. In controlled experiments on eight synthetic projects (seven vulnerable, one clean control), VibeGuard achieved 100% recall, 89.47% precision (F1 = 94.44%), and correct pass/fail gate decisions on all eight projects across three policy levels. We discuss how these results inform a defense-in-depth workflow for teams that rely on AI code generation.

Next-generation (NextG) cellular networks are designed to support emerging applications with diverse data rate and latency requirements, such as immersive multimedia services and large-scale Internet of Things deployments. A key enabling mechanism is radio access network (RAN) slicing, which dynamically partitions radio resources into virtual resource blocks to efficiently serve heterogeneous traffic classes, including enhanced mobile broadband (eMBB), massive machine-type communications (mMTC), and ultra-reliable low-latency communications (URLLC). In this paper, we study the impact of adversarial attacks on AI-driven RAN slicing decisions, where a budget-constrained adversary selectively jams slice transmissions to bias deep reinforcement learning (DRL)-based resource allocation, and quantify the resulting service level agreement (SLA) violations and post-attack recovery behavior. Our results indicate that budget-constrained adversarial jamming can induce severe and slice-dependent steady-state SLA violations. Moreover, the DRL agent's reward converges toward the clean baseline only after a non-negligible recovery period.

Spiking neural networks (SNNs) support energy-efficient machine intelligence because event-driven computation and sparse activity map naturally to low-power digital hardware. In practical implementations, however, membrane states, synaptic weights, and thresholds are represented with finite-precision integer arithmetic. Quantization, clipping, and overflow can therefore alter network dynamics, not just approximate a higher-precision model. This paper adopts an integer-state dynamical perspective, modeling a hardware-oriented SNN as a deterministic map on a bounded integer lattice. Under this view, recurrence, periodic orbits, and regime changes become intrinsic properties of the system. We introduce a lightweight update rule with integer-valued states and shift-based leakage, and demonstrate the approach through exploratory simulations with network sizes N = 30-130, connection densities 0.1-0.9, and bit widths 4/8/16 over T = 1000 steps. The results show bounded and recurrent temporal structure with strong quantization sensitivity. The observed regimes depend heavily on representation semantics and scaling choices. These findings suggest that numerical precision acts as a dynamical design variable and highlight integer-state analysis as a useful framework for hardware-aware SNN co-design, motivating future work on attractor analysis, precision-aware training, and FPGA/ASIC validation.

System Instructions in Large Language Models (LLMs) are commonly used to enforce safety policies, define agent behavior, and protect sensitive operational context in agentic AI applications. These instructions may contain sensitive information such as API credentials, internal policies, and privileged workflow definitions, making system instruction leakage a critical security risk highlighted in the OWASP Top 10 for LLM Applications. Without incurring the overhead costs of reasoning models, many LLM applications rely on refusal-based instructions that block direct requests for system instructions, implicitly assuming that prohibited information can only be extracted through explicit queries. We introduce an automated evaluation framework that tests whether system instructions remain confidential when extraction requests are re-framed as encoding or structured output tasks. Across four common models and 46 verified system instructions, we observe high attack success rates (> 0.7) for structured serialization where models refuse direct extraction requests but disclose protected content in the requested serialization formats. We further demonstrate a mitigation strategy based on one-shot instruction reshaping using a Chain-of-Thought reasoning model, indicating that even subtle changes in wording and structure of system instructions can significantly reduce attack success rate without requiring model retraining.

Popularity bias is a pervasive problem in recommender systems, where recommendations disproportionately favor popular items. This not only results in "rich-get-richer" dynamics and a homogenization of visible content, but can also lead to misalignment of recommendations with individual users' preferences for popular or niche content. This work studies popularity bias through the lens of user-recommender alignment. To this end, we introduce Popularity Quantile Calibration, a measurement framework that quantifies misalignment between a user's historical popularity preference and the popularity of their recommendations. Building on this notion of popularity alignment, we propose SPREE, an inference-time mitigation method for sequential recommenders based on activation steering. SPREE identifies a popularity direction in representation space and adaptively steers model activations based on an estimate of each user's personal popularity bias, allowing both the direction and magnitude of steering to vary across users. Unlike global debiasing approaches, SPREE explicitly targets alignment rather than uniformly reducing popularity. Experiments across multiple datasets show that SPREE consistently improves user-level popularity alignment while preserving recommendation quality.

Multi-LLM revision pipelines, in which a second model reviews and improves a draft produced by a first, are widely assumed to derive their gains from genuine error correction. We question this assumption with a controlled decomposition experiment that uses four matched conditions to separate second-pass gains into three additive components: re-solving, scaffold, and content. We evaluate this design across two model pairs on three benchmarks spanning knowledge-intensive MCQ and competitive programming. Our results show that the gains of multi-LLM revision are not monolithic, but depend on task structure, draft quality, and the type of draft information. On MCQ tasks, where the answer space is constrained and drafts provide little structural guidance, most gains are consistent with stronger-model re-solving, and directly routing queries to the stronger model can be more effective than revising a weak draft. On code generation tasks, however, two-stage prompting remains useful because even semantically null drafts can provide substantial structural scaffolding, while weak draft content can be harmful. Finally, role-reversed experiments show that strong drafts clearly benefit weak reviewers. Ultimately, our findings demonstrate that the utility of multi-LLM revision is dynamically bottlenecked by task structure and draft quality, necessitating more targeted pipeline designs rather than blanket revision strategies.

The paradigm of scaling Large Language Models (LLMs) in both parameter size and test time has pushed the boundaries of AI capabilities, but at the cost of making the traditional generative evaluation paradigm prohibitively expensive, therefore making the latency of LLM's in-training downstream performance evaluation unbearable. However, simple metrics like training loss (perplexity) are not always correlated with downstream performance, as sometimes their trends diverge from the actual task outcomes. This dilemma calls for a method that is computationally efficient and sufficiently accurate in measuring model capabilities. To address this challenge, we introduce a new in-training evaluation paradigm that uses a lightweight probe for monitoring downstream performance. The probes take the internal representations of LLM checkpoints (during training) as input and directly predict the checkpoint's performance on downstream tasks measured by success probability (i.e., pass@1). We design several probe architectures, validating their effectiveness using the OLMo3-7B's checkpoints across a diverse set of downstream tasks. The probes can accurately predict a checkpoint's performance (with avg. AUROC$>$0.75), have decent generalizability across checkpoints (earlier predicts later), and reduce the computation latency from $\sim$1 hr (using conventional generative evaluation method) to $\sim$3 min. In sum, this work presents a practical and scalable in-training downstream evaluation paradigm, enabling a more agile, informed, and efficient LLM development process.

We study model-based learning of finite-window policies in tabular partially observable Markov decision processes (POMDPs). A common approach to learning under partial observability is to approximate unbounded history dependencies using finite action-observation windows. This induces a finite-state Markov decision process (MDP) over histories, referred to as the superstate MDP. Once a model of this superstate MDP is available, standard MDP algorithms can be used to compute optimal policies, motivating the need for sample-efficient model estimation. Estimating the superstate MDP model is challenging because trajectories are generated by interaction with the original POMDP, creating a mismatch between the sampling process and target model. We propose a model estimation procedure for tabular POMDPs and analyze its sample complexity. Our analysis exploits a connection between filter stability and concentration inequalities for weakly dependent random variables. As a result, we obtain tight sample complexity guarantees for estimating the superstate MDP model from a single trajectory. Combined with value iteration, this yields approximately optimal finite-window policies for the POMDP.

This paper introduces two transfer learning methodologies for estimating nonparametric Bayesian networks under scarce data. We propose two algorithms, a constraint-based structure learning method, called PC-stable-transfer learning (PCS-TL), and a score-based method, called hill climbing transfer learning (HC-TL). We also define particular metrics to tackle the negative transfer problem in each of them, a situation in which transfer learning has a negative impact on the model's performance. Then, for the parameters, we propose a log-linear pooling approach. For the evaluation, we learn kernel density estimation Bayesian networks, a type of nonparametric Bayesian network, and compare their transfer learning performance with the models alone. To do so, we sample data from small, medium and large-sized synthetic networks and datasets from the UCI Machine Learning repository. Then, we add noise and modifications to these datasets to test their ability to avoid negative transfer. To conclude, we perform a Friedman test with a Bergmann-Hommel post-hoc analysis to show statistical proof of the enhanced experimental behavior of our methods. Thus, PCS-TL and HC-TL demonstrate to be reliable algorithms for improving the learning performance of a nonparametric Bayesian network with scarce data, which in real industrial environments implies a reduction in the required time to deploy the network.

While large language model-based multi-agent systems have shown strong potential for complex reasoning, how to effectively organize multiple agents remains an open question. In this paper, we introduce OrgAgent, a company-style hierarchical multi-agent framework that separates collaboration into governance, execution, and compliance layers. OrgAgent decomposes multi-agent reasoning into three layers: a governance layer for planning and resource allocation, an execution layer for task solving and review, and a compliance layer for final answer control. By evaluating the framework across reasoning tasks, LLMs, execution modes, and execution policies, we find that multi-agent systems organized in a company-style hierarchy generally outperform other organizational structures. Besides, hierarchical coordination also reduces token consumption relative to flat collaboration in most settings. For example, for GPT-OSS-120B, the hierarchical setting improves performance over flat multi-agent system by 102.73% while reducing token usage by 74.52% on SQuAD 2.0. Further analysis shows that hierarchy helps most when tasks benefit from stable skill assignment, controlled information flow, and layered verification. Overall, our findings highlight organizational structure as an important factor in multi-agent reasoning, shaping not only effectiveness and cost, but also coordination behavior.

AI agents increasingly operate over extended time horizons, yet their ability to retain, organize, and recall multimodal experiences remains a critical bottleneck. Building effective lifelong memory requires navigating a vast design space spanning architecture, retrieval strategies, prompt engineering, and data pipelines; this space is too large and interconnected for manual exploration or traditional AutoML to explore effectively. We deploy an autonomous research pipeline to discover Omni-SimpleMem, a unified multimodal memory framework for lifelong AI agents. Starting from a naïve baseline (F1=0.117 on LoCoMo), the pipeline autonomously executes ${\sim}50$ experiments across two benchmarks, diagnosing failure modes, proposing architectural modifications, and repairing data pipeline bugs, all without human intervention in the inner loop. The resulting system achieves state-of-the-art on both benchmarks, improving F1 by +411% on LoCoMo (0.117$\to$0.598) and +214% on Mem-Gallery (0.254$\to$0.797) relative to the initial configurations. Critically, the most impactful discoveries are not hyperparameter adjustments: bug fixes (+175%), architectural changes (+44%), and prompt engineering (+188% on specific categories) each individually exceed the cumulative contribution of all hyperparameter tuning, demonstrating capabilities fundamentally beyond the reach of traditional AutoML. We provide a taxonomy of six discovery types and identify four properties that make multimodal memory particularly suited for autoresearch, offering guidance for applying autonomous research pipelines to other AI system domains. Code is available at this https://github.com/aiming-lab/SimpleMem.

Multimodal Large Language Models (MLLMs) have shown strong performance on video question answering, but their application to long-form videos is constrained by limited context length and computational cost, making keyframe sampling essential. Existing approaches typically rely on semantic relevance or reinforcement learning, which either fail to capture evidential clues or suffer from inefficient combinatorial optimization. In this work, we propose an evidence-driven keyframe sampling framework grounded in information bottleneck theory. We formulate keyframe selection as maximizing the conditional mutual information between selected frames and the query, providing a principled objective that reflects each frame's contribution to answering the question. To make this objective tractable, we exploit its structure to derive a decomposed optimization that reduces subset selection to independent frame-level scoring. We further introduce a query-conditioned evidence scoring network trained with a contrastive objective to estimate evidential importance efficiently. Experiments on long-form video understanding benchmarks show that our method consistently outperforms prior sampling strategies under strict token budgets, while significantly improving training efficiency.

We introduce EgoSim, a closed-loop egocentric world simulator that generates spatially consistent interaction videos and persistently updates the underlying 3D scene state for continuous simulation. Existing egocentric simulators either lack explicit 3D grounding, causing structural drift under viewpoint changes, or treat the scene as static, failing to update world states across multi-stage interactions. EgoSim addresses both limitations by modeling 3D scenes as updatable world states. We generate embodiment interactions via a Geometry-action-aware Observation Simulation model, with spatial consistency from an Interaction-aware State Updating module. To overcome the critical data bottleneck posed by the difficulty in acquiring densely aligned scene-interaction training pairs, we design a scalable pipeline that extracts static point clouds, camera trajectories, and embodiment actions from in-the-wild large-scale monocular egocentric videos. We further introduce EgoCap, a capture system that enables low-cost real-world data collection with uncalibrated smartphones. Extensive experiments demonstrate that EgoSim significantly outperforms existing methods in terms of visual quality, spatial consistency, and generalization to complex scenes and in-the-wild dexterous interactions, while supporting cross-embodiment transfer to robotic manipulation. Codes and datasets will be open soon. The project page is at egosimulator.github.io.

Graph Neural Networks (GNNs) are widely used for learning on graph-structured data, but scaling GNN training to massive graphs remains challenging. To enable scalable distributed training, graphs are divided into smaller partitions that are distributed across multiple machines such that inter-machine communication is minimized and computational load is balanced. In practice, existing partitioning approaches face a fundamental trade-off between partitioning overhead and partitioning quality. We propose EmbedPart, an embedding-driven partitioning approach that achieves both speed and quality. Instead of operating directly on irregular graph structures, EmbedPart leverages node embeddings produced during the actual GNN training workload and clusters these dense embeddings to derive a partitioning. EmbedPart achieves more than 100x speedup over Metis while maintaining competitive partitioning quality and accelerating distributed GNN training. Moreover, EmbedPart naturally supports graph updates and fast repartitioning, and can be applied to graph reordering to improve data locality and accelerate single-machine GNN training. By shifting partitioning from irregular graph structures to dense embeddings, EmbedPart enables scalable and high-quality graph data optimization.

We study whether phone-use agents respect privacy while completing benign mobile tasks. This question has remained hard to answer because privacy-compliant behavior is not operationalized for phone-use agents, and ordinary apps do not reveal exactly what data agents type into which form entries during execution. To make this question measurable, we introduce MyPhoneBench, a verifiable evaluation framework for privacy behavior in mobile agents. We operationalize privacy-respecting phone use as permissioned access, minimal disclosure, and user-controlled memory through a minimal privacy contract, iMy, and pair it with instrumented mock apps plus rule-based auditing that make unnecessary permission requests, deceptive re-disclosure, and unnecessary form filling observable and reproducible. Across five frontier models on 10 mobile apps and 300 tasks, we find that task success, privacy-compliant task completion, and later-session use of saved preferences are distinct capabilities, and no single model dominates all three. Evaluating success and privacy jointly reshuffles the model ordering relative to either metric alone. The most persistent failure mode across models is simple data minimization: agents still fill optional personal entries that the task does not require. These results show that privacy failures arise from over-helpful execution of benign tasks, and that success-only evaluation overestimates the deployment readiness of current phone-use agents. All code, mock apps, and agent trajectories are publicly available at~ https://github.com/FreedomIntelligence/MyPhoneBench.

Current aligned language models exhibit a dual failure mode we term the Evasive Servant: they sycophantically validate flawed user beliefs while deflecting responsibility with boilerplate disclaimers. We propose the Dignified Peer framework, which counters servility with anti-sycophancy and trustworthiness, and mitigates evasiveness through empathy and creativity. Realizing this agent requires overcoming significant challenges in data supervision, objective collapse, and evaluation bias. We address these issues by introducing the PersonaKnob dataset which features a compositional partial order structure of multiple persona preference. This data is utilized alongside a tolerant constrained Lagrangian DPO algorithm that dynamically balances all persona dimensions to prevent behavioral collapse. Additionally, we employ a psychometrically calibrated Item Response Theory evaluation protocol to disentangle latent model persona capability from confounders like judge biases. Extensive empirical studies demonstrate that our approach successfully build a LLM agent with both dignity and peer.

Existing methods for AI psychological counselors predominantly rely on supervised fine-tuning using static dialogue datasets. However, this contrasts with human experts, who continuously refine their proficiency through clinical practice and accumulated experience. To bridge this gap, we propose an Experience-Driven Lifelong Learning Agent (\texttt{PsychAgent}) for psychological counseling. First, we establish a Memory-Augmented Planning Engine tailored for longitudinal multi-session interactions, which ensures therapeutic continuity through persistent memory and strategic planning. Second, to support self-evolution, we design a Skill Evolution Engine that extracts new practice-grounded skills from historical counseling trajectories. Finally, we introduce a Reinforced Internalization Engine that integrates the evolved skills into the model via rejection fine-tuning, aiming to improve performance across diverse scenarios. Comparative analysis shows that our approach achieves higher scores than strong general LLMs (e.g., GPT-5.4, Gemini-3) and domain-specific baselines across all reported evaluation dimensions. These results suggest that lifelong learning can improve the consistency and overall quality of multi-session counseling responses.

Medical imaging techniques, especially Magnetic Resonance Imaging (MRI), are accepted as the gold standard in the diagnosis and treatment planning of neurological diseases. However, the manual analysis of MRI images is a time-consuming process for radiologists and is prone to human error due to fatigue. In this study, two different Deep Learning approaches were developed and analyzed comparatively for the automatic detection and classification of brain tumors (Glioma, Meningioma, Pituitary, and No Tumor). In the first approach, a custom Convolutional Neural Network (CNN) architecture named "OkanNet", which has a low computational cost and fast training time, was designed from scratch. In the second approach, the Transfer Learning method was applied using the 50-layer ResNet-50 [1] architecture, pre-trained on the ImageNet dataset. In experiments conducted on an extended dataset compiled by Masoud Nickparvar containing a total of $7,023$ MRI images, the Transfer Learning-based ResNet-50 model exhibited superior classification performance, achieving $96.49\%$ Accuracy and $0.963$ Precision. In contrast, the custom OkanNet architecture reached an accuracy rate of $88.10\%$; however, it proved to be a strong alternative for mobile and embedded systems with limited computational power by yielding results approximately $3.2$ times faster ($311$ seconds) than ResNet-50 in terms of training time. This study demonstrates the trade-off between model depth and computational efficiency in medical image analysis through experimental data.

Traditional Online Public Access Catalogues (OPACs) are becoming less effective due to the rapid growth of scholarly literature. Conventional search methods, such as keyword indexing and Boolean queries, often fail to support efficient knowledge discovery. This paper proposes a Smart OPAC framework that transforms traditional OPACs into intelligent discovery systems using artificial intelligence and knowledge graph techniques. The framework enables semantic search, thematic filtering, and knowledge graph-based visualization to enhance user interaction and exploration. It integrates multiple open scholarly data sources and applies semantic embeddings to improve relevance and contextual understanding. The system supports exploratory search, semantic navigation, and refined result filtering based on user-defined themes. Quantitative evaluation demonstrates improvements in retrieval efficiency, relevance, and reduction of information overload. The proposed approach offers practical implications for modernizing digital library services and supports next-generation research workflows. Future work includes user-centric evaluation, personalization, and dynamic knowledge graph updates.

Test-Time Learning (TTL) enables language agents to iteratively refine their performance through repeated interactions with the environment at inference time. At the core of TTL is an adaptation policy that updates the actor policy based on experience from previous episodes, thereby improving future behavior. Existing methods rely on fixed, hand-crafted adaptation policies rather than optimizing them for downstream improvement. We argue that optimal adaptation policies should be learned from task environments, not hand-engineered based on human intuition. To achieve this, we introduce Meta-TTL, a framework that formulates the discovery of effective adaptation policies as a bi-level optimization problem. Within this framework, the inner loop executes the standard TTL process, measuring how effectively a candidate adaptation policy helps an agent correct errors across sequential episodes. Guided by the agent's performance, the outer loop employs evolutionary search over a diverse distribution of training tasks to iteratively refine the adaptation policy. We evaluate Meta-TTL on Jericho and WebArena-Lite across both in-distribution (ID) and out-of-distribution (OOD) settings, using multiple meta-agent backbones. Results on both benchmarks show that Meta-TTL consistently outperforms hand-crafted baselines, suggesting that the optimized adaptation policy encodes transferable strategies that generalize beyond the training task distribution.

Time series forecasting (TSF) is critical across domains such as finance, meteorology, and energy. While extending the lookback window theoretically provides richer historical context, in practice, it often introduces irrelevant noise and computational redundancy, preventing models from effectively capturing complex long-term dependencies. To address these challenges, we propose a Dynamic Semantic Compression (DySCo) framework. Unlike traditional methods that rely on fixed heuristics, DySCo introduces an Entropy-Guided Dynamic Sampling (EGDS) mechanism to autonomously identify and retain high-entropy segments while compressing redundant trends. Furthermore, we incorporate a Hierarchical Frequency-Enhanced Decomposition (HFED) strategy to separate high-frequency anomalies from low-frequency patterns, ensuring that critical details are preserved during sparse sampling. Finally, a Cross-Scale Interaction Mixer(CSIM) is designed to dynamically fuse global contexts with local representations, replacing simple linear aggregation. Experimental results demonstrate that DySCo serves as a universal plug-and-play module, significantly enhancing the ability of mainstream models to capture long-term correlations with reduced computational cost.

Training effective software engineering agents requires large volumes of task-specific trajectories, incurring substantial data construction costs. Inspired by the "Less-Is-More" hypothesis in mathematical reasoning, we investigate its extension to agentic scenarios and propose an end-to-end training framework that achieves superior agentic capabilities with fewer but higher-quality training trajectories. This is achieved via STITCH (Sliding-memory Trajectory Inference and Task Chunking Heuristic), a coarse-to-fine mechanism that filters low-value noise and retains decision-critical tokens to maximize training signal quality. We conduct experiments across multiple agent frameworks (e.g., mini-SWE-agent, MSWE-agent), model scales (30B to 355B), and multilingual settings (Python, Java, and ArkTS). On SWE-bench Verified, models trained with STITCH achieve up to 63.16% relative improvement over base models. On Multi-SWE-bench (Java), MiniMax-M2.5-STITCH achieves 43.75% with our CodeArts Agent scaffold (+16.67%). On HarmonyOS (ArkTS), GLM-4.7-STITCH improves the compilation pass rate to 61.31% (+43.34%) with less than 1K training trajectories. Our results confirm that the "Less-Is-More" paradigm generalizes effectively to complex agentic tasks across diverse languages and model scales.

In the previous work, a lexical (re)categorisation -- or confirmation of the given category -- of roots identified as verbal was undertaken to determine their original category accurately. Building on this, the present paper offers an account of the valency classification of those Mapudungun roots confirmed to be verbal, using the language's own morphotactics; specifically, by examining the permissible and restricted combinations of various suffixes with roots or verbal stems in the Mapuche verb form. As with all work conducted thus far, the results presented here aim to improve the morphological analyser (Dungupeyum) with all verified findings incorporated into the system. From a theoretical perspective, we also hope to contribute to the recognition and understanding of issues related to the valency of Mapuche verb forms.

We obtain better algorithms for computing more balanced orientations and degree splits in LOCAL. Important to our result is a connection to the hypergraph sinkless orientation problem [BMNSU, SODA'25] We design an algorithm of complexity $\mathcal{O}(\varepsilon^{-1} \cdot \log n)$ for computing a balanced orientation with discrepancy at most $\varepsilon \cdot \mathrm{deg}(v)$ for every vertex $v \in V$. This improves upon a previous result by [GHKMSU, Distrib. Comput. 2020] of complexity $\mathcal{O}(\varepsilon^{-1} \cdot \log \varepsilon^{-1} \cdot (\log \log \varepsilon^{-1})^{1.71} \cdot \log n)$. Further, we show that this result can also be extended to compute undirected degree splits with the same discrepancy and in the same runtime. As as application we show that $(3 / 2 + \varepsilon)Δ$-edge coloring can now be solved in $\mathcal{O}(\varepsilon^{-1} \cdot \log^2 Δ\cdot \log n + \varepsilon^{-2} \cdot \log n)$ rounds in LOCAL. Note that for constant $\varepsilon$ and $Δ= \mathcal{O}(2^{\log^{1/3} n})$ this runtime matches the current state-of-the-art for $(2Δ- 1)$-edge coloring in [Ghaffari & Kuhn, FOCS'21].

We present OmniVoice, a massive multilingual zero-shot text-to-speech (TTS) model that scales to over 600 languages. At its core is a novel diffusion language model-style discrete non-autoregressive (NAR) architecture. Unlike conventional discrete NAR models that suffer from performance bottlenecks in complex two-stage (text-to-semantic-to-acoustic) pipelines, OmniVoice directly maps text to multi-codebook acoustic tokens. This simplified approach is facilitated by two key technical innovations: (1) a full-codebook random masking strategy for efficient training, and (2) initialization from a pre-trained LLM to ensure superior intelligibility. By leveraging a 581k-hour multilingual dataset curated entirely from open-source data, OmniVoice achieves the broadest language coverage to date and delivers state-of-the-art performance across Chinese, English, and diverse multilingual benchmarks. Our code and pre-trained models are publicly available at https://github.com/k2-fsa/OmniVoice.

Predictive process monitoring (PPM) focuses on predicting future process trajectories, including next activity predictions. This is crucial in dynamic environments where processes change or face uncertainty. However, current frameworks often assume a static environment, overlooking dynamic characteristics and concept drifts. This results in catastrophic forgetting, where training while focusing merely on new data distribution negatively impacts the performance on previously learned data distributions. Continual learning addresses, among others, the challenges related to mitigating catastrophic forgetting. This paper proposes a novel approach called Continual Next Activity Prediction with Prompts (CNAPwP), which adapts the DualPrompt algorithm for next activity prediction to improve accuracy and adaptability while mitigating catastrophic forgetting. We introduce new datasets with recurring concept drifts, alongside a task-specific forgetting metric that measures the prediction accuracy gap between initial occurrence and subsequent task occurrences. Extensive testing on three synthetic and two real-world datasets representing several setups of recurrent drifts shows that CNAPwP achieves SOTA or competitive results compared to five baselines, demonstrating its potential applicability in real-world scenarios. An open-source implementation of our method, together with the datasets and results, is available at: https://github.com/SvStraten/CNAPwP.

The classical-quantum system heterogeneity (different data characteristics, execution paradigms and synchronization mechanism etc.) renders existing distributed communication mechanisms (e.g. MPI, NCCL etc.) inadequate. This bottleneck severely impairs operational synergy and programming efficiency. Thus, the performance of hybrid applications on classical-quantum heterogeneous infrastructures is directly limited. To address these challenges, this paper proposes a message-passing library tailored for large-scale classical-quantum heterogeneous distributed computing, referred to as MPI-Q. The design centers on three mechanisms. First, it defines a heterogeneous hybrid communication domain that achieves unified management of classical and quantum processes in heterogeneous hybrid systems. Second, it uses a lightweight communication path that allows classical control nodes to send device-ready waveform data directly to quantum MonitorProcesses, avoiding unnecessary relay stages. Third, it establishes a heterogeneous hybrid synchronization mechanism to tackle the problem of timing control for multi-node quantum operations. While retaining the traditional MPI programming model, MPI-Q achieves extension toward quantum subsystems. Experiments on distributed GHZ state preparation demonstrate that this model exhibits near-linear scalability, achieving a maximum speedup of 18.76 times on 24 quantum nodes. This proves that the library can effectively support large-scale heterogeneous hybrid distributed computing applications, filling the technical gap in this field.

In recent years, the scaling laws of recommendation models have attracted increasing attention, which govern the relationship between performance and parameters/FLOPs of recommenders. Currently, there are three mainstream architectures for achieving scaling in recommendation models, namely attention-based, TokenMixer-based, and factorization-machine-based methods, which exhibit fundamental differences in both design philosophy and architectural structure. In this paper, we propose a unified scaling architecture for recommendation systems, namely \textbf{UniMixer}, to improve scaling efficiency and establish a unified theoretical framework that unifies the mainstream scaling blocks. By transforming the rule-based TokenMixer to an equivalent parameterized structure, we construct a generalized parameterized feature mixing module that allows the token mixing patterns to be optimized and learned during model training. Meanwhile, the generalized parameterized token mixing removes the constraint in TokenMixer that requires the number of heads to be equal to the number of tokens. Furthermore, we establish a unified scaling module design framework for recommender systems, which bridges the connections among attention-based, TokenMixer-based, and factorization-machine-based methods. To further boost scaling ROI, a lightweight UniMixing module is designed, \textbf{UniMixing-Lite}, which further compresses the model parameters and computational cost while significantly improve the model performance. The scaling curves are shown in the following figure. Extensive offline and online experiments are conducted to verify the superior scaling abilities of \textbf{UniMixer}.

3D Visual Grounding (3D-VG) aims to localize objects in 3D scenes via natural language descriptions. While recent advancements leveraging Vision-Language Models (VLMs) have explored zero-shot possibilities, they typically suffer from a static workflow relying on preprocessed 3D point clouds, essentially degrading grounding into proposal matching. To bypass this reliance, our core motivation is to decouple the task: leveraging 2D VLMs to resolve complex spatial semantics, while relying on deterministic multi-view geometry to instantiate the 3D structure. Driven by this insight, we propose "Think, Act, Build (TAB)", a dynamic agentic framework that reformulates 3D-VG tasks as a generative 2D-to-3D reconstruction paradigm operating directly on raw RGB-D streams. Specifically, guided by a specialized 3D-VG skill, our VLM agent dynamically invokes visual tools to track and reconstruct the target across 2D frames. Crucially, to overcome the multi-view coverage deficit caused by strict VLM semantic tracking, we introduce the Semantic-Anchored Geometric Expansion, a mechanism that first anchors the target in a reference video clip and then leverages multi-view geometry to propagate its spatial location across unobserved frames. This enables the agent to "Build" the target's 3D representation by aggregating these multi-view features via camera parameters, directly mapping 2D visual cues to 3D coordinates. Furthermore, to ensure rigorous assessment, we identify flaws such as reference ambiguity and category errors in existing benchmarks and manually refine the incorrect queries. Extensive experiments on ScanRefer and Nr3D demonstrate that our framework, relying entirely on open-source models, significantly outperforms previous zero-shot methods and even surpasses fully supervised baselines.

With the rapid growth of e-commerce, exploring general representations rather than task-specific ones has attracted increasing attention. Although recent multimodal large language models (MLLMs) have driven significant progress in product understanding, they are typically employed as feature extractors that implicitly encode product information into global embeddings, thereby limiting their ability to capture fine-grained attributes. Therefore, we argue that leveraging the reasoning capabilities of MLLMs to explicitly model fine-grained product attributes holds significant potential. Nevertheless, achieving this goal remains non-trivial due to several key challenges: (i) long-context reasoning tends to dilute the model's attention to salient information in the raw input; (ii) supervised fine-tuning (SFT) primarily encourages rigid imitation, limiting the exploration of effective reasoning strategies; and (iii) fine-grained details are progressively attenuated during forward propagation. To address these issues, we propose MOON3.0, the first reasoning-aware MLLM-based model for product representation learning. Our method (1) employs a multi-head modality fusion module to adaptively integrate raw signals; (2) incorporates a joint contrastive and reinforcement learning framework to autonomously explore more effective reasoning strategies; and (3) introduces a fine-grained residual enhancement module to progressively preserve local details throughout the network. Additionally, we release a large-scale multimodal e-commerce benchmark MBE3.0. Experimentally, our model demonstrates state-of-the-art zero-shot performance across various downstream tasks on both our benchmark and public datasets.

Large Language Models (LLMs) increasingly prioritize user validation over epistemic accuracy - a phenomenon known as sycophancy. We present The Silicon Mirror, an orchestration framework that dynamically detects user persuasion tactics and adjusts AI behavior to maintain factual integrity. Our architecture introduces three components: (1) a Behavioral Access Control (BAC) system that restricts context layer access based on real-time sycophancy risk scores, (2) a Trait Classifier that identifies persuasion tactics across multi-turn dialogues, and (3) a Generator-Critic loop where an auditor vetoes sycophantic drafts and triggers rewrites with "Necessary Friction." In a live evaluation across all 437 TruthfulQA adversarial scenarios, Claude Sonnet 4 exhibits 9.6% baseline sycophancy, reduced to 1.4% by the Silicon Mirror - an 85.7% relative reduction (p < 10^-6, OR = 7.64, Fisher's exact test). Cross-model evaluation on Gemini 2.5 Flash reveals a 46.0% baseline reduced to 14.2% (p < 10^-10, OR = 5.15). We characterize the validation-before-correction pattern as a distinct failure mode of RLHF-trained models.

Estimated density is often interpreted as indicating how typical a sample is under a model. Yet deep models trained on one dataset can assign higher density to simpler out-of-distribution (OOD) data than to in-distribution test data. We refer to this behavior as the OOD anomaly. Prior work typically studies this phenomenon within a single architecture, detector, or benchmark, implicitly assuming certain canonical densities. We instead separate the trained network from the density estimator built from its representations or outputs. We introduce two estimators: Jacobian-based estimators and autoregressive self-estimators, making density analysis applicable to a wide range of models. Applying this perspective to a range of models, including iGPT, PixelCNN++, Glow, score-based diffusion models, DINOv2, and I-JEPA, we find the same striking regularity that goes beyond the OOD anomaly: lower-complexity samples receive higher estimated density, while higher-complexity samples receive lower estimated density. This ordering appears within a test set and across OOD pairs such as CIFAR-10 and SVHN, and remains highly consistent across independently trained models. To quantify these orderings, we introduce Spearman rank correlation and find striking agreement both across models and with external complexity metrics. Even when trained only on the lowest-density (most complex) samples - or even a single such sample - the resulting models still rank simpler images as higher density. These observations lead us beyond the original OOD anomaly to a more general conclusion: deep networks consistently favor simple data. Our goal is not to close this question, but to define and visualize it more clearly. We broaden its empirical scope and show that it appears across architectures, objectives, and density estimators.

Energy-based models (EBMs) implement inference as gradient descent on a learned Lyapunov function, yielding interpretable, structure-preserving alternatives to black-box neural ODEs and aligning naturally with physical AI. Yet their use in system identification remains limited, and existing architectures lack formal stability guarantees that globally preclude unstable modes. We address this gap by introducing an EBM framework for system identification with stable, dissipative, absorbing invariant dynamics. Unlike classical global Lyapunov stability, absorbing invariance expands the class of stability-preserving architectures, enabling more flexible and expressive EBMs. We extend EBM theory to nonsmooth activations by establishing negative energy dissipation via Clarke derivatives and deriving new conditions for radial unboundedness, exposing a stability-expressivity tradeoff in standard EBMs. To overcome this, we introduce a hybrid architecture with a dynamical visible layer and static hidden layers, prove absorbing invariance under mild assumptions, and show that these guarantees extend to port-Hamiltonian EBMs. Experiments on metric-deformed multi-well and ring systems validate the approach, showcasing how our hybrid EBM architecture combines expressivity with sound and provable safety guarantees by design.

Large language models (LLMs) have achieved strong performance on medical question answering (medical QA), and chain-of-thought (CoT) prompting has further improved results by eliciting explicit intermediate reasoning; meanwhile, self-reflective (self-corrective) prompting has been widely claimed to enhance model reliability by prompting LLMs to critique and revise their own reasoning, yet its effectiveness in safety-critical medical settings remains unclear. In this work, we conduct an exploratory analysis of self-reflective reasoning for medical multiple-choice question answering: using GPT-4o and GPT-4o-mini, we compare standard CoT prompting with an iterative self-reflection loop and track how predictions evolve across reflection steps on three widely used medical QA benchmarks (MedQA, HeadQA, and PubMedQA). We analyze whether self-reflection leads to error correction, error persistence, or the introduction of new errors. Our results show that self-reflective prompting does not consistently improve accuracy and its impact is highly dataset- and model-dependent: it yields modest gains on MedQA but provides limited or negative benefits on HeadQA and PubMedQA, and increasing the number of reflection steps does not guarantee better performance. These findings highlight a gap between reasoning transparency and reasoning correctness, suggesting that self-reflective reasoning is better viewed as an analytical tool for understanding model behavior rather than a standalone solution for improving medical QA reliability.

Autonomous tool-using agents operating in networked environments must decide which information source to query and when to stop querying and act. Without principled bounds on information-acquisition costs, unconstrained agents exhibit systematic failure modes: excessive tool use under congestion, prolonged deliberation under time decay, and brittle behavior under ambiguous evidence. We propose the Triadic Cognitive Architecture (TCA), a unified decision-theoretic framework that formalizes these failure modes through the concept of Cognitive Friction. By synthesizing nonlinear filtering theory, congestion-dependent cost dynamics, and HJB optimal stopping, we model deliberation as a stochastic control problem over a joint belief-congestion state space, where information acquisition is explicitly priced by tool-dependent signal quality and live network load. Rather than relying on arbitrary heuristic stop-tokens or fixed query budgets, TCA derives an HJB-inspired stopping boundary and instantiates a computable rollout-based approximation of belief-dependent value-of-information with a net-utility halting condition. We validate the framework on two controlled simulation environments, the Emergency Medical Diagnostic Grid (EMDG) and the Network Security Triage Grid (NSTG), designed to isolate key decision-theoretic quantities under reproducible conditions. TCA reduces time-to-action while improving resource outcomes without degrading accuracy: over greedy baselines, TCA gains 36 viability points in EMDG and 33 integrity points in NSTG. Ablations confirm joint optimization of selection and stopping is essential; stopping rules alone recover at most 4 viability points. A sensitivity sweep over alpha, beta, lambda_S shows stable accuracy and interpretable tradeoffs; an empirical sweep over eta in {0, 0.1, 0.3, 0.5} confirms eta=0 is optimal on EMDG trajectories under high temporal urgency.

Recent advances in reasoning Large Language Models (LLMs) have primarily relied on upfront thinking, where reasoning occurs before final answer. However, this approach suffers from critical limitations in code generation, where upfront thinking is often insufficient as problems' full complexity only reveals itself during code implementation. Moreover, it cannot adaptively allocate reasoning effort throughout the code generation process where difficulty varies significantly. In this paper, we propose Think-Anywhere, a novel reasoning mechanism that enables LLMs to invoke thinking on-demand at any token position during code generation. We achieve Think-Anywhere by first teaching LLMs to imitate the reasoning patterns through cold-start training, then leveraging outcome-based RL rewards to drive the model's autonomous exploration of when and where to invoke reasoning. Extensive experiments on four mainstream code generation benchmarks (i.e., LeetCode, LiveCodeBench, HumanEval, and MBPP) show that Think-Anywhere achieves state-of-the-art performance over both existing reasoning methods and recent post-training approaches, while demonstrating consistent generalization across diverse LLMs. Our analysis further reveals that Think-Anywhere enables the model to adaptively invoke reasoning at high-entropy positions, providing enhanced interpretability.

Quantum sensing technologies offer transformative potential for ultra-sensitive biomedical sensing, yet their clinical translation remains constrained by classical noise limits and a reliance on macroscopic ensembles. We propose a unifying generational framework to organize the evolving landscape of quantum biosensors based on their utilization of quantum resources. First-generation devices utilize discrete energy levels for signal transduction but follow classical scaling laws. Second-generation sensors exploit quantum coherence to reach the standard quantum limit, while third-generation architectures leverage entanglement and spin squeezing to approach Heisenberg-limited precision. We further define an emerging fourth generation characterized by the end-to-end integration of quantum sensing with quantum learning and variational circuits, enabling adaptive inference directly within the quantum domain. By analyzing critical parameters such as bandwidth matching and sensor-tissue proximity, we identify key technological bottlenecks and propose a roadmap for transitioning from measuring physical observables to extracting structured biological information with quantum-enhanced intelligence.

Reinforcement Learning (RL) agents often struggle with efficiency and performance in complex environments. We propose a novel framework that uses a Large Language Model (LLM) to dynamically generate a curriculum over available actions, enabling the agent to incorporate each action individually. We apply this framework to the game of Blackjack, where the LLM creates a multi-stage training path that progressively introduces complex actions to a Tabular Q-Learning and a Deep Q-Network (DQN) agent. Our evaluation in a realistic 8-deck simulation over 10 independent runs demonstrates significant performance gains over standard training methods. The curriculum-based approach increases the DQN agent's average win rate from 43.97% to 47.41%, reduces the average bust rate from 32.9% to 28.0%, and accelerates the overall workflow by over 74%, with the agent's full training completing faster than the baseline's evaluation phase alone. These results validate that LLM-guided curricula can build more effective, robust, and efficient RL agents.

Memory-augmented Large Language Models (LLMs) are essential for developing capable, long-term AI agents. Recently, applying Reinforcement Learning (RL) to optimize memory operations, such as extraction, updating, and retrieval, has emerged as a highly promising research direction. However, existing implementations remain highly fragmented and task-specific, lacking a unified infrastructure to streamline the integration, training, and evaluation of these complex pipelines. To address this gap, we present MemFactory, the first unified, highly modular training and inference framework specifically designed for memory-augmented agents. Inspired by the success of unified fine-tuning frameworks like LLaMA-Factory, MemFactory abstracts the memory lifecycle into atomic, plug-and-play components, enabling researchers to seamlessly construct custom memory agents via a "Lego-like" architecture. Furthermore, the framework natively integrates Group Relative Policy Optimization (GRPO) to fine-tune internal memory management policies driven by multi-dimensional environmental rewards. MemFactory provides out-of-the-box support for recent cutting-edge paradigms, including Memory-R1, RMM, and MemAgent. We empirically validate MemFactory on the open-source MemAgent architecture using its publicly available training and evaluation data. Across the evaluation sets, MemFactory improves performance over the corresponding base models on average, with relative gains of up to 14.8%. By providing a standardized, extensible, and easy-to-use infrastructure, MemFactory significantly lowers the barrier to entry, paving the way for future innovations in memory-driven AI agents.

We introduce Nomad, a system for autonomous data exploration and insight discovery. Given a corpus of documents, databases, or other data sources, users rarely know the full set of questions, hypotheses, or connections that could be explored. As a result, query-driven question answering and prompt-driven deep-research systems remain limited by human framing and often fail to cover the broader insight space. Nomad addresses this problem with an exploration-first architecture. It constructs an explicit Exploration Map over the domain and systematically traverses it to balance breadth and depth. It generates and selects hypotheses and investigates them with an explorer agent that can use document search, web search, and database tools. Candidate insights are then checked by an independent verifier before entering a reporting pipeline that produces cited reports and higher-level meta-reports. We also present a comprehensive evaluation framework for autonomous discovery systems that measures trustworthiness, report quality, and diversity. Using a corpus of selected UN and WHO reports, we show that Nomad produces more trustworthy and higher-quality reports than baselines, while also producing more diverse insights over several runs. Nomad is a step toward autonomous systems that not only answer user questions or conduct directed research, but also discover which questions, research directions, and insights are worth surfacing in the first place.

LLM-based shopping agents increasingly rely on long purchase histories and multi-turn interactions for personalization, yet naively appending raw history to prompts is often ineffective due to noise, length, and relevance mismatch. We propose MemRerank, a preference memory framework that distills user purchase history into concise, query-independent signals for personalized product reranking. To study this problem, we build an end-to-end benchmark and evaluation framework centered on an LLM-based \textbf{1-in-5} selection task, which measures both memory quality and downstream reranking utility. We further train the memory extractor with reinforcement learning (RL), using downstream reranking performance as supervision. Experiments with two LLM-based rerankers show that MemRerank consistently outperforms no-memory, raw-history, and off-the-shelf memory baselines, yielding up to \textbf{+10.61} absolute points in 1-in-5 accuracy. These results suggest that explicit preference memory is a practical and effective building block for personalization in agentic e-commerce systems.

Deep learning-based tropical cyclone (TC) forecasting methods have demonstrated significant potential and application advantages, as they feature much lower computational cost and faster operation speed than numerical weather prediction models. However, existing deep learning methods still have key limitations: they can only process a single type of sequential trajectory data or homogeneous meteorological variables, and fail to achieve accurate forecasting of abnormal deflected TCs. To address these challenges, we present two groundbreaking contributions. First, we have constructed a multimodal and multi-source dataset named AOT-TCs for TC forecasting in the Northwest Pacific basin. As the first dataset of its kind, it innovatively integrates heterogeneous variables from the atmosphere, ocean, and land, thus obtaining a comprehensive and information-rich meteorological dataset. Second, based on the AOT-TCs dataset, we propose a forecasting model that can handle both normal and abnormally deflected TCs. This is the first TC forecasting model to adopt an explicit atmosphere-ocean-terrain coupling architecture, enabling it to effectively capture complex interactions across physical domains. Extensive experiments on all TC cases in the Northwest Pacific from 2017 to 2024 show that our model achieves state-of-the-art performance in TC forecasting: it not only significantly improves the forecasting accuracy of normal TCs but also breaks through the technical bottleneck in forecasting abnormally deflected TCs.

Can a safety gate permit unbounded beneficial self-modification while maintaining bounded cumulative risk? We formalize this question through dual conditions -- requiring sum delta_n < infinity (bounded risk) and sum TPR_n = infinity (unbounded utility) -- and establish a theory of their (in)compatibility. Classification impossibility (Theorem 1): For power-law risk schedules delta_n = O(n^{-p}) with p > 1, any classifier-based gate under overlapping safe/unsafe distributions satisfies TPR_n <= C_alpha * delta_n^beta via Holder's inequality, forcing sum TPR_n < infinity. This impossibility is exponent-optimal (Theorem 3). A second independent proof via the NP counting method (Theorem 4) yields a 13% tighter bound without Holder's inequality. Universal finite-horizon ceiling (Theorem 5): For any summable risk schedule, the exact maximum achievable classifier utility is U*(N, B) = N * TPR_NP(B/N), growing as exp(O(sqrt(log N))) -- subpolynomial. At N = 10^6 with budget B = 1.0, a classifier extracts at most U* ~ 87 versus a verifier's ~500,000. Verification escape (Theorem 2): A Lipschitz ball verifier achieves delta = 0 with TPR > 0, escaping the impossibility. Formal Lipschitz bounds for pre-LayerNorm transformers under LoRA enable LLM-scale verification. The separation is strict. We validate on GPT-2 (d_LoRA = 147,456): conditional delta = 0 with TPR = 0.352. Comprehensive empirical validation is in the companion paper [D2].

For any $Δ$, let $k_Δ$ be the maximum integer $k$ such that $(k+1)(k+2)\le Δ$. We give a distributed \LOCAL algorithm that, given an integer $k < k_Δ$, computes a valid $Δ-k$-coloring if one exists. The algorithm runs in $\tilde{O}(\log^4 \log n)$ rounds, which is within a polynomial factor of the $Ω(\log\log n)$ lower bound, which already applies to the case $k=0$. It is also best possible in the sense that if $k \ge k_Δ$, the problem requires $Ω(n/Δ)$ distributed rounds [Molloy, Reed, '14, Bamas, Esperet '19]. For $Δ$ at most polylogarithmic, the algorithm is an exponential improvement over the current state of the art of $O(\log^{49/12} n)$ rounds. When $Δ\ge (\log n)^{50}$, our algorithm achieves an even faster runtime of $O(\log^* n)$ rounds.

Large language models (LLMs) can generate chains of thought (CoTs) that are not always causally responsible for their final outputs. When such a mismatch occurs, the CoT no longer faithfully reflects the actual reasons (i.e., decision-critical factors) driving the model's behavior, leading to the reduced CoT monitorability problem. However, a comprehensive and fully open-source benchmark for thoroughly evaluating CoT monitorability remains lacking. To address this gap, we propose MonitorBench, a systematic benchmark for evaluating CoT monitorability in LLMs. MonitorBench provides: (1) a diverse set of 1,514 test instances with carefully designed decision-critical factors across 19 tasks spanning 7 categories to characterize \textit{when} CoTs can be used to monitor the factors driving LLM behavior; and (2) two stress-test settings to quantify \textit{the extent to which} CoT monitorability can be degraded. Extensive experiments across multiple popular LLMs with varying capabilities show that CoT monitorability is higher when the decision-critical factors shape the intermediate reasoning process without merely influencing the final answer. More capable LLMs tend to exhibit lower monitorability. And all evaluated LLMs can intentionally reduce monitorability under stress-tests, with monitorability dropping by up to 30\% in some tasks that do not require structural reasoning over the decision-critical factors. Overall, MonitorBench provides a basis for further research on evaluating future LLMs, studying advanced stress-test monitorability techniques, and developing new monitoring approaches. The code is available at https://github.com/ASTRAL-Group/MonitorBench.

Although 3D Gaussian Splatting (3DGS) enables high-fidelity real-time rendering, its prohibitive storage overhead severely hinders practical deployment. Recent anchor-based 3DGS compression schemes reduce gaussian redundancy through some advanced context models. However, they overlook explicit geometric dependencies, leading to structural degradation and suboptimal ratedistortion performance. In this paper, we propose a Local Geometry-aware Hierarchical Context Compression framework for 3DGS(LG-HCC) that incorporates inter-anchor geometric correlations into anchor pruning and entropy coding for compact representation. Specifically, we introduce an Neighborhood-Aware Anchor Pruning (NAAP) strategy, which evaluates anchor importance via weighted neighborhood feature aggregation and then merges low-contribution anchors into salient neighbors, yielding a compact yet geometry-consistent anchor set. Moreover, we further develop a hierarchical entropy coding scheme, in which coarse-to-fine priors are exploited through a lightweight Geometry-Guided Convolution(GG-Conv) operator to enable spatially adaptive context modeling and rate-distortion optimization. Extensive experiments show that LG-HCC effectively alleviates structural preservation issues,achieving superior geometric integrity and rendering fidelity while reducing storage by up to 30.85x compared to the Scaffold-GS baseline on the Mip-NeRF360 dataset

Neurodiverse learners often require reading supports, yet increasing scaffold richness can sometimes overload attention and working memory rather than improve comprehension. Grounded in the Construction-Integration model and a contingent scaffolding perspective, we examine how structural versus semantic scaffolds shape comprehension and reading experience in a supervised inclusive context. Using an adapted reading interface, we compared four modalities: unmodified text, sentence-segmented text, segmented text with pictograms, and segmented text with pictograms plus keyword labels. In a within-subject pilot with 14 primary-school learners with special educational needs and disabilities, we measured reading comprehension using standardized questions and collected brief child- and therapist-reported experience measures alongside open-ended feedback. Results highlight heterogeneous responses as some learners showed patterns consistent with benefits from segmentation and pictograms, while others showed patterns consistent with increased coordination costs when visual scaffolds were introduced. Experience ratings showed limited differences between modalities, with some apparent effects linked to clinical complexity, particularly for perceived ease of understanding. Open-ended feedback of the learners frequently requested simpler wording and additional visual supports. These findings suggest that no single scaffold is universally optimal, reinforcing the need for calibrated, adjustable scaffolding and provide design implications for human-AI co-regulation in supervised inclusive reading contexts.

Quantum algorithms require encoding classical vectors as quantum states, a step known as amplitude encoding. General-purpose state preparation routines accept any input vector of length $N = 2^m$ and produce circuits with $\bigO{2^m}$ gates. However, vectors arising in scientific and engineering applications often exhibit mathematical structure that admits far more efficient encoding. Recent theoretical work has established closed-form circuits for several structured vector classes, but without open-source implementations. We present PyEncode, an open-source Python library that implements this body of theory in a unified, immediately deployable framework. The library covers sparse, step, square (general interval), Walsh, geometric, and Fourier patterns, and supports weighted superpositions of pattern states via the linear combination of unitaries (LCU) protocol, enabling exact preparation of piecewise-structured vectors such as multi-interval Hamiltonians. PyEncode exposes a single function encode(VectorObj, N) that maps a typed parameter declaration directly to a verified Qiskit circuit, with no vector materialization and no approximation. Sparse, step, and Walsh vectors require only $\bigO{m}$ gates; geometric (exponential-decay) vectors require $\bigO{m}$ gates with zero two-qubit gates; square (general interval) vectors require $\bigO{m^2}$ gates via a QFT-based constant adder, with $\bigO{m}$ special cases; Fourier (sinusoidal) vectors require $\bigO{m^2}$ gates via the inverse Quantum Fourier Transform -- all exponentially fewer than the $\bigO{2^m}$ cost of general state preparation. LCU combines $r$ component circuits whose total gate cost is the sum of individual component costs, with success probability $p \in (0,1]$ determined analytically. The library is available at https://github.com/UW-ERSL/PyEncode.

Cooperative Coevolution (CC) effectively addresses Large-Scale Global Optimization (LSGO) via decomposition but struggles with the emerging class of Heterogeneous LSGO (H-LSGO) problems arising from real-world applications, where subproblems exhibit diverse dimensions and distinct landscapes. The prevailing CC paradigm, relying on a fixed low-dimensional optimizer, often fails to navigate this heterogeneity. To address this limitation, we propose the Learning-Based Heterogeneous Cooperative Coevolution Framework (LH-CC). By formulating the optimization process as a Markov Decision Process, LH-CC employs a meta-agent to adaptively select the most suitable optimizer for each subproblem. We also introduce a flexible benchmark suite to generate diverse H-LSGO problem instances. Extensive experiments on 3000-dimensional problems with complex coupling relationships demonstrate that LH-CC achieves superior solution quality and computational efficiency compared to state-of-the-art baselines. Furthermore, the framework exhibits robust generalization across varying problem instances, optimization horizons, and optimizers. Our findings reveal that dynamic optimizer selection is a pivotal strategy for solving complex H-LSGO problems.

Strategic coopetition in multi-stakeholder systems requires understanding how cooperation persists through time without binding contracts. This technical report extends computational foundations for strategic coopetition to sequential interaction dynamics, bridging conceptual modeling (i* framework) with game-theoretic reciprocity analysis. We develop: (1) bounded reciprocity response functions mapping partner deviations to finite conditional responses, (2) memory-windowed history tracking capturing cognitive limitations over k recent periods, (3) structural reciprocity sensitivity derived from interdependence matrices where behavioral responses are amplified by structural dependencies, and (4) trust-gated reciprocity where trust modulates reciprocity responses. The framework applies to both human stakeholder interactions and multi-agent computational systems. Comprehensive validation across 15,625 parameter configurations demonstrates robust reciprocity effects, with all six behavioral targets exceeding thresholds: cooperation emergence (97.5%), defection punishment (100%), forgiveness dynamics (87.9%), asymmetric differentiation (100%), trust-reciprocity interaction (100%), and bounded responses (100%). Empirical validation using the Apple iOS App Store ecosystem (2008-2024) achieves 43/51 applicable points (84.3%), reproducing documented cooperation patterns across five ecosystem phases. Statistical significance confirmed at p < 0.001 with Cohen's d = 1.57. This report concludes the Foundations Series (TR-1 through TR-4) adopting uniaxial treatment where agents choose cooperation levels along a single continuum. Companion work on interdependence (arXiv:2510.18802), trust (arXiv:2510.24909), and collective action (arXiv:2601.16237) has been prepublished. Extensions Series (TR-5 through TR-8) introduces biaxial treatment where cooperation and competition are independent dimensions.

Real-world mathematical modeling is inherently an experiential and collaborative endeavor. Domain experts rarely solve complex problems from scratch; instead, they draw upon analogies from historical cases and subject their hypotheses to rigorous peer scrutiny. However, autonomous agents powered by Large Language Models predominantly rely on isolated reasoning paradigms, frequently generating plausible but fundamentally flawed models due to a lack of domain grounding and adversarial verification. To address these limitations, we propose Sci-Mind, a novel framework that mirrors the human scientific discovery process. Sci-Mind integrates Experiential Memory Recall to retrieve executable code snippets and modeling paradigm descriptors, grounding abstract reasoning in historical solutions. Subsequently, it employs an Adversarial Cognitive Dialectic where a Theorist optimizing mathematical coherence and a Pragmatist enforcing data feasibility debate through competing objectives to prune elegant but infeasible formulations. A Self-Validating Execution Strategy further ensures blueprint consistency through formal predicates before code generation, achieving fully autonomous execution. Extensive experiments on the MM-Bench and EngiBench demonstrate that Sci-Mind significantly outperforms leading autonomous agents in both modeling rigorousness and code executability.

Graph neural networks (GNNs) have achieved strong performance across various real-world domains. Nevertheless, they suffer from oversquashing, where long-range information is distorted as it is compressed through limited message-passing pathways. This bottleneck limits their ability to capture essential global context and decreases their performance, particularly in dense and heterophilic regions of graphs. To address this issue, we propose a novel graph learning framework that enriches node embeddings via cross-attentive cohesive subgraph representations to mitigate the impact of excessive long-range dependencies. This framework enhances the node representation by emphasizing cohesive structure in long-range information but removing noisy or irrelevant connections. It preserves essential global context without overloading the narrow bottlenecked channels, which further mitigates oversquashing. Extensive experiments on multiple benchmark datasets demonstrate that our model achieves consistent improvements in classification accuracy over standard baseline methods.

We present exact results for steady-state density correlation functions in conserved-mass transport processes with {\it anisotropic}, reflection-symmetric hopping on a $d-$dimensional hypercubic lattice. In addition to mass conservation, we consider center-of-mass (CoM) conservation, imposed either along a specific axis or along all axes. CoM-conserving dynamics is implemented through coordinated {\it multidirectional} hopping of two equal chunks of masses in {\it opposite} directions. While anisotropy and mass conservation are known to generate power-law density correlations $C({\bf x}) \sim 1/|{\bf x}|^d$ at large distance $|{\bf x}| \gg 1$ {\it [Phys. Rev. A {\bf 42}, 1954 (1990)]}, an additional CoM conservation can qualitatively alter the nature of the power law. Indeed, when CoM is conserved in {\it all} directions, the correlations decay faster $-$ typically as $C({\bf x}) \sim 1/|{\bf x}|^{(d+2)}$, regardless of the presence (or absence) of anisotropy. Consequently, the systems exhibit an extreme {\it hyperuniformity} (``class I''), where the long-wavelength density fluctuations, despite the slow power-law decay, are anomalously suppressed. When CoM is conserved along particular ({\it not} all) directions, the slower $1/|{\bf x}|^{d}$ power-law decay is recovered. The above behavior can be understood from an analogy between the correlation function and an electrostatic potential: While a (rank-$2$) quadrupolar charge distribution gives rise to the $1/|{\bf x}|^{d}$ power law, the $1/|{\bf x}|^{(d+2)}$ power law originates from a higher-order (rank-$4$) multipolar charge distribution. These findings reveal a rich interplay between anisotropy and CoM conservation in nonequilibrium steady states.

Transport in disordered environments is often controlled not by typical fluctuations but by rare, extreme events that dominate long-time dynamics. In such settings, Zwanzig's classic mean-field theory predicts that energetic roughness reduces the diffusion coefficient by an exponential factor governed solely by the variance of the disorder. However, this prediction breaks down in uncorrelated Gaussian landscapes, where rare but deep multi-site traps dominate transport and lead to a much stronger suppression of diffusion. Here, we present a unified theoretical framework that clarifies both the origin of this breakdown and its resolution. We show that Zwanzig's local averaging can be interpreted as a Gaussian cumulant expansion whose validity is destroyed by uncorrelated disorder through the emergence of extreme trapping events. Introducing Gaussian spatial correlations fundamentally reshapes the landscape: roughness increments become smoother, asymmetric multi-site traps are suppressed, and the statistics of escape pathways are regularized. As a result, Zwanzig's exponential scaling is recovered. We provide an explicit analytical derivation demonstrating how spatial correlations modify trap statistics and restore mean-field diffusion, complemented by illustrative numerical examples showing the dramatic reduction of escape times in correlated landscapes.

Employing the Schwinger-Keldysh formalism, we formulate an effective field theory for s-wave superconducting phase transition, where the dynamical variables consist of electromagnetic gauge field and a complex scalar order parameter. Symmetry-constrained effective action allows systematic handling of dissipations and fluctuations. In particular, we explore the physical implications of higher-order terms, including those involving additional dynamical fields as well as higher time derivatives, for the real-time dynamics near the superconducting critical point. When appropriately truncated, the effective field theory reproduces the phenomenological Ginzburg-Landau equations. Upon crossing the critical temperature into the low-temperature phase, the electromagnetic gauge symmetry undergoes spontaneous breaking induced by the condensate of the order parameter. Collective excitation analysis reveals that the Higgs mode behaves as an overdamped diffusive mode near the critical point, while the phase fluctuation is absorbed into the gauge field via the Higgs mechanism. Via the holographic Schwinger-Keldysh technique, rigorous validation in a holographic superconductor confirms the structure of the effective action and quantifies the Wilsonian coefficients. The holographic results uncover a complex relaxation parameter that is indicative of oscillatory dynamics, a hallmark of strongly coupled systems.

We find that a novel Superconductor-Insulator-2D electron gas tunnel junction (SISm) strongly and efficiently generates thermoelectricity via a nonlinear mechanism. We simulate across the parameter space of the junction, finding and discussing different regimes with features useful for thermoelectricity generation or for specific applications. The generated Seebeck potential can go up to $6.75Δ_0$ with a huge nonlinear Seebeck coefficient, and efficiency can get very close to Carnot efficiency $η=0.96η_C$, a record for a solid-state device model. Thermoelectric performance is far better than analogous junctions, with fewer fabrication challenges, as the device can be fabricated via standard methods.

Analytical treatments of polymer dynamics have mostly been restricted to linear response theory around some steady state obtained via perturbative field theory. Here, I derive an analytical framework that yields unified access to the evolution of conformations, contact probabilities, and fluctuations within a dynamical mean-field theory. Starting with the Langevin equation of a hydrodynamically coupled and self-avoiding heteropolymer, the key idea is to focus on the two-point correlator as the lowest-order relevant observable. Truncating higher-order correlations via a Gaussian closure leads to a self-consistent diffusion equation for the chain correlations. The theory is validated by contrasting coiled, globular, and self-avoiding polymers within a single dynamical framework, and predicts hyper-compacted fractal states in hydrodynamically coupled active polymers such as chromatin.

We demonstrate the stray-field-mediated skyrmion stabilizing capabilities of ultrathin exchange-decoupled antisymmetric ferromagnetic bilayers based on conventional transition metal materials. Using an asymptotically exact micromagnetic model valid in the ultrathin film limit, we show that the antisymmetric tailoring of the bilayer allows the Dzyaloshinskii-Moriya interaction and the dipolar interaction to act synergistically to stabilize skyrmions, in contrast to the monolayer case, in which these energies compete. To obtain optimal stability of these skyrmions against collapse and bursting -- the two fundamental processes determining skyrmion lifetime, we carry out an asymptotic analysis of the saddle point solution that separates the skyrmion from the demagnetized state. The result is an optimal stability line for compact skyrmions in the non-dimensional parameter space of the effective Dzyaloshinskii-Moriya interaction strength and the effective film thickness. Our predictions are confirmed by extensive micromagnetic simulations of antisymmetric bilayers, using magnetic parameters of the conventional Pt/Co/AlO$_x$ systems. Our results provide a new pathway for experimental observations of 10 nm radius zero-field skyrmions with lifetimes compatible with information technology applications.

The widespread use of digital devices has raised growing concerns about its impact on sustained attention at the population level. In this work, we propose a minimal dynamical framework to describe the collective evolution of attention under continuous exposure to screen-mediated environments. We introduce a macroscopic variable representing the average level of sustained attention and model its dynamics as the result of competing mechanisms: intrinsic cognitive recovery and degradation induced by digital stimulation. The digital environment is treated as an external control parameter that continuously perturbs the system, leading to a relaxational dynamics. The proposed mechanisms are consistent with empirical findings on attentional dynamics under digital exposure. We first analyze a linear formulation, which provides an analytically tractable baseline, and then extend the model by incorporating a nonlinear degradation term that captures amplification effects under high-intensity stimulation. We derive an explicit expression for the stationary state and show that the equilibrium attention level decreases monotonically with increasing exposure. An effective potential formulation is introduced, revealing that digital overstimulation progressively deforms the dynamical landscape, shifting the stable state toward regimes of reduced attention without generating multiple equilibria. Importantly, the model does not rely on social contagion or interaction-driven bistability, but instead describes a continuous displacement of the collective cognitive regime under environmental pressure. Our results suggest that the impact of digital technologies on attention may be understood as a gradual macroscopic effect emerging from persistent external stimulation, rather than as a transition between competing behavioral states.

Neural communication operates on both fast synaptic transmission and slower, diffusive extrasynaptic signaling, yet how these two modes jointly organize brain function remains unclear. Here, using the complete synaptic and neuropeptidergic connectomes of \emph{Caenorhabditis elegans}, we develop a unified multiplex framework linking anatomical wiring to functional communication. We infer structure-derived functional connectivity from the synaptic connectome using equilibrium principles from statistical physics, yielding a probabilistic map of information flow across all synaptic pathways, and compare this functional layer directly with the extrasynaptic connectome. This reveals a principled functional specialization across four communication regimes: (i) a topology-dependent layer that reinforces and stabilizes synaptic motor circuits, (ii) a topology-resilient modulatory layer supporting global regulation and behavioral state control, (iii) a purely extrasynaptic network sustaining survival and homeostasis, and (iv) a purely synaptic regime mediating rapid, low-latency sensorimotor processing. Together, these findings reveal that synaptic and extrasynaptic signaling form complementary architectures optimized for speed, modulation, robustness, and survival, and provide a general strategy for integrating structural and modulatory connectomes to understand how distinct communication modes cooperate to sustain coherent brain function.

Entangled photon pair sources are a key enabling technology for quantum communication and networking, yet their deployment beyond laboratory environments is hindered by system-level complexity, limited operational stability, and insufficient industry compatibility. Here, we demonstrate a rack-based, mobile quantum light source architecture based on a semiconductor quantum dot emitter that directly addresses these challenges through modular system integration and automated operation. The source generates polarization-entangled photon pairs with an entanglement negativity 2n of up to $0.98(1)$, confirming near-maximal entanglement quality. In continuous, hands-off operation over a six-hour time window, the system achieves an average single-photon emission rate of $697(8)$ kHz and a maximum rate of $740(7)$ kHz, while maintaining 2n-value of more than $95$ $\%$. These results are enabled by the integration of optical excitation, collection, cryogenic operation, and control electronics within a standardized rack footprint, together with automated monitoring. By demonstrating simultaneously high entanglement quality, sustained brightness, and long-term operational stability in an industry-aligned system architecture, this work advances semiconductor quantum dot sources toward deployable entangled photon sources for applied quantum photonics.

The tunability and twist engineering of van der Waals materials enable the emergence of electronic states not present in individual monolayers. Among them, monolayer 1T-TaS$_2$ is a well-known Mott insulating system, whose star-of-David charge density wave reconstruction realizes an emergent triangular lattice of local magnetic moments. Interestingly, in its bulk form, the insulating gap is not correlation-driven, but stems from interlayer coupling. Here, we exploit the stacking-dependent nature of the insulating gap to show that in twisted 1T-TaS$_2$ bilayers, the spatially dependent competition between many-body and single-particle gaps creates Mott-trivial mosaic superlattices, featuring regions with local magnetic moments and non-magnetic insulating regions. We further demonstrate the tunability of the mosaic correlated state with an interlayer bias, giving rise to controllable charge transfer and quenching of correlations. Our results establish twisted 1T-TaS$_2$ as a flexible platform to engineer mixed spatially modulated correlated insulating phases, arising from the moiré profile.

We investigate the role of stochastic resetting in non-Markovian systems, where memory effects arise due to slow relaxation, rugged energy landscapes, disordered environments, and molecular crowding. Using the celebrated continuous-time random walk (CTRW) framework, we analyze first-passage processes with multiple competing outcomes and examine how resetting can selectively enhance desired events. We characterize the efficiency of resetting through conditional mean first-passage times (MFPTs) and demonstrate that its impact is highly sensitive to the underlying waiting-time statistics. Furthermore, we derive an inequality that quantifies how resetting controls fluctuations in conditional first-passage times (FPTs), revealing regimes where variability is significantly suppressed. Our results provide a systematic understanding of how long-term memory influences competitive first-passage outcomes and establish resetting as a powerful control mechanism beyond the conventional Markovian setting.

Urban populations exhibit fractal organization and systematic scaling regularities, yet the scaling exponents reported across cities vary substantially, challenging existing theory. Using 100~m gridded population maps for 477 urban areas spanning the Netherlands (2000--2023) and major world cities (1975--2020), we recursively coarse-grain each city and quantify how the mean and variance of inhabitants in square grid cells of side length $\ell$ scale with $\ell$. This yields two exponents, $β$ from $\langle N_\ell\rangle\sim \ell^β$ and $γ$ from $\mathrm{Var}(N_\ell)\sim \ell^γ$, where in the small-$\ell$ limit $β$ equals the planar fractal dimension of populated space. Across cities within a given year, $γ$ depends linearly on $β$. Compiling $>$10,000 exponent estimates over five decades shows that this hyperscaling relation is robust yet non-universal: its slope and intercept vary across continents and drift systematically in time, trending toward the limiting form $γ\simeq 2+β$. A mean-field (independent-cell) argument predicts a quadratic mean--variance mapping and cannot reproduce the observed $β$--$γ$ dependence, implying strong spatial correlations. We derive a correlation-aware variance decomposition in which $γ$ is controlled by a correlation dimension $D_c$; in the correlation-dominated regime $γ=2+D_c$. If large maturing cities, as are the ones selected in our dataset, evolve to effective monofractal ($D_c\simeq β$) cities, the asymptotic prediction becomes $γ\simeq 2+β$, consistent with the observed temporal drift. This interdependence links urban form and fluctuations, constrains mechanistic growth models, and implies scaling predictions for spatial indicators built from local means and variances.

Some notable systems, such as room-temperature superconductors and materials for controlled nuclear fusion, require an accurate description of finite-temperature quantum matter. Stochastic path integral methods are finite-temperature and numerically exact, but scale poorly with system size due the notorious Fermion sign problem. To somewhat mitigate this, we use a hydrodynamical backflow coordinate transformation. Our first attempt was a continuous normalizing flow machine learning approach to determine the optimal parameters. We found this to reduce the error of the total energy, approximately, three times at medium sign severity. Numerical issues challenged training effectively. Thus, a semi-analytic approach was developed to estimate the optimal parameters. We do this by using a derived expression dependent on a Bosonic observable. Hence, the calculation of these values does not have a sign problem. The resulting backflow transformations reduce the problem by multiple orders of magnitude, specifically, in the case of a harmonically trapped, two-dimensional electron gas at finite-temperature. The total energy of the system agrees with previous, backflow untransformed, studies and we calculate energies for up to 32 electrons. The limiting factor is found to be, primarily, the $O(N^3)$ calculation of the Jacobian, stemming from the coordinate transformation of the backflow. A more thorough implementation may further improve this scaling. Otherwise, a pathway for simulating electron systems at currently unreachable regimes is obtained. Finally, as a specific practical use case in energy storage systems, the quantum capacitance for graphene quantum dot materials is calculated.

We investigate the emergence of quantum anomalous Hall conductivity in a two-dimensional $d$-wave altermagnet on a Lieb lattice under an external magnetic field. Altermagnetic order induces momentum-dependent spin splitting without net magnetization in the relativistic limit, producing distinct spin-resolved bands at the $X$ and $Y$ valleys. The phase diagram features a normal insulator and a spin Chern insulator separated by an accidental Dirac semimetal. The magnetic field breaks rotational symmetry between valleys while maintaining vanishing total magnetization, enabling independent valley contributions to topology. One valley supports Chern numbers $C=-1$ or $0$, while the other hosts $C=0$ or $+1$, governed by field strength and bandwidth. This competition yields valley-dependent topology. Berry curvature analysis reveals fully gapped phases with total Chern numbers $C=\pm1$, separated by valley-selective gap closings. We uncover a mechanism for rapid magnetic control of the quantum anomalous Hall effect near the semimetal phase and highlight key distinctions from ferro-valleytronic and quantum spin Hall systems.

We introduce a residence-time approach (RTA) for determining position-dependent diffusivities from biased molecular dynamics simulations. The method is formulated for trajectory segments in which the effective drift along the transport coordinate is negligible, as realized here using adaptive biasing force simulations. In this regime, local diffusivities are obtained directly from mean first-exit times out of finite spatial intervals. Unlike conventional fluctuation-based approaches, the RTA does not require dedicated harmonically restrained simulations or numerical integration of noisy time-correlation functions. We assess the method for oxygen diffusion across a hexadecane slab, water permeation across a lipid bilayer, and permeation of water and selected volatile organic compounds through a model skin-barrier membrane. In the slab system, the RTA reproduces independently determined bulk diffusivities within statistical uncertainty. In the membrane systems, the inferred diffusivity profiles are supported by propagator-level validation. These results establish the RTA as a practical approach for extracting position-dependent diffusivities from biased molecular simulations.

We derive generic properties of nonequilibrium phase transitions in all-to-all Ising models placed in contact with two thermal reservoirs, in which parameters (temperatures, interactions and field parameters) assume arbitrary values depending on the contact with each thermal bath. The presence of different kinds of external parameters leads to remarkably different sort of phase transitions. While continuous, discontinuous and even tricritical points are presented when external parameters are symmetric (e.g. the case of energetic barriers or different couplings between the system and thermal baths), the tricriticality is absent when external parameters are antisymmetric (e.g. the case of magnetic fields or biased drivings) implying that solely critical or discontinuous are possible. In such latter case, the probability distribution acquires the Boltzmann-Gibbs like form, irrespectively the model parameters when the switching between thermal reservoirs is sufficiently fast. Our work sheds light about the differences between equilibrium and nonequilibrium ingredients and theirs consequences upon phase transitions.

The observation of a misalignment between the applied heat flux and the measured temperature gradient in insulating solids induced by magnetic field has become a subject of experimental investigation, theoretical speculation, and unsettled controversy. To identify the origin of this phonon thermal Hall effect, we performed a comparative study of longitudinal and transverse heat transport in crystalline (quartz) and vitreous (silica) SiO$_2$ using identical experimental set-ups and thermometers. A finite signal was detected in the crystalline samples and none in the amorphous sample, within our resolution. The cleaner crystal exhibited a larger thermal Hall conductivity than the dirtier one, ruling out disorder as the driver of the effect. On the other hand, the amplitude of the transverse thermal resistivity is almost identical in the two crystalline samples (W$_{\perp}$/B$\approx 10^{-6}$ m.K.W$^{-1}$.T$^{-1}$). We show that in a phonon gas, as in a molecular gas displaying the Senftleben-Beenakker effect, heat is conducted through two channels, and argue that a thermal Hall response is unavoidable whenever these channels differ both in entropy production and in their coupling to the magnetic field. Under such conditions, the conserved energy current and the non-conserved entropy current cease to be parallel. Finally, the magnitude of the transverse thermal resistivity can be accounted for by a surprisingly simple picture. The heat flux induces a tiny drift velocity of the lattice nuclei, the magnetic field exerts a transverse Berry force on this drift, and this force is balanced by an entropic restoring force.

We theoretically introduce a quasi-1D magnetic heterostructure of alternating 2D topological and normal insulator strips. Its low-energy physics is governed by a hybrid Hamiltonian intertwining the Su-Schrieffer-Heeger and Shockley models, with spin-momentum locking and local Zeeman splitting. Symmetry analysis places it in class AIII, characterized by chiral symmetry and a $\mathbb{Z}$ topological invariant. Computing the winding number from the block-off-diagonal structure of the Hamiltonian reveals topological phases characterized by invariants $ν= 0$, $1$, and $2$. Furthermore, a single magnetic defect acts as a sensitive local probe, whose in-gap spectrum provides a spectroscopic fingerprint to distinguish topological phases. Extending the platform to a multilayer geometry uncovers a nonsymmorphic projective symmetry that gives rise to Möbius band topology, with the Brillouin zone compactifying into a Klein bottle. Our work establishes a platform for higher-order topology via heterostructure design and magnetic patterning.

Surface growth driven by non-monomeric deposition has remained largely unexplored. We investigate a model based on the deposition of blobs with a power-law size distribution $P(s)\sim s^{-τ}$. We find that the critical exponents vary continuously with $τ$, recovering Kardar--Parisi--Zhang behavior only for $τ\ge 3$. For $τ<3$, roughness scaling exhibits strong corrections and scale invariance breaks down. We show that this behavior originates from the emergence of a second dynamical length scale $ζ$, corresponding to the linear size of the largest cluster, in addition to the usual correlation length $ξ$. The coexistence of these two relevant scales signals the breakdown of the usual Family--Vicsek scaling. These results point to a new phenomenology of surface growth beyond the standard scale-invariant paradigm.

We study a one-dimensional exclusion process with a fixed jump length $I \ge 1$ in which a particle may advance or retreat $I$ sites provided all intermediate sites are vacant, with hopping rates of Arrhenius type depending on the local headway. We identify the class of rates admitting an explicit Ising-Gibbs invariant measure, with stationarity governed by pairwise balance rather than detailed balance. In the thermodynamic limit, we derive a closed-form stationary current that recovers the mean-field prediction for look-ahead traffic flow models exactly when particles are uncorrelated, and quantifies the correlation-induced correction for non-trivial interactions, illustrated with two explicit families of interaction potentials.

We study the interplay between long-range electron-phonon and electron-electron interactions in electrostatically doped two-dimensional semiconductors, including interlayer couplings in van der Waals heterostructures. We evaluate the effects of those interactions on transport properties by writing dynamically coupled Boltzmann equations for the electrons and for the electrodynamically active excitations. We develop a theory with a general validity, and apply it both to simplified parabolic models, and to the realistic BN-encapsulated graphene system which we present in an accompanying paper [arXiv:2604.00678]. We show that dynamical screening effects are of fundamental importance in order to correctly describe the electronic transport properties of two-dimensional materials, and in particular the scattering from polar phonons, whether those come from the semiconductor itself or the surrounding layers.

We investigate the localization properties of a generalized SSH model. Numerical and analytical results indicate the emergence of extended states protected by unbounded hopping in this model. Moreover, this protection effect is disrupted by the appearance of generalized incommensurate zeros, causing the extended phase in the system to transition into a multifractal phase. However, at the boundaries of the phase region, we still observe the existence of extended states. These extended states coincide with multifractality-enriched mobility edges, separating the multifractal phase from the localized phase. Further analysis reveals that this extended states originates from the band edge states of SSH model. In addition, these isolated extended states also influence eigenstates with nearby energies, giving rise to an anomalous extended-to-multifractal critical transition. These findings not only enrich the behavioral repertoire of eigenstates at critical points, but also offer new insights for further understanding Anderson localization and the induction of multifractal phases.

Non-Hermitian skin effect (NHSE) is one of the most fascinating phenomena in non-Hermitian systems, which refers to enormous eigenstates localize at the boundary exponentially under open boundary condition (OBC). For typical NHSE, the localization length for a skin mode is independent of the system's size. Recently, some studies have revealed that for specific $1$-dimensional model, the localization length for eigenstates are proportional to the system's length under generalized boundary condition (GBC), and such phenomenon is dubbed as scale-free skin effect (SFSE). Further, SFSE is discovered in $1$-dimensional Hermitian chain with pure imaginary impurity at the end. In this work, we propose a mechanism for SFSE in 1-dimensional systems, which is model-independent. Our work provide a viewpoint for researching SFSE and shed new light on understanding finite size effect in non-Hermitian systems.

Discovering governing equations, whether manually or by data-driven methods, has been central in physics and related areas. Since governing equations are typically constrained by a set of symmetries, using symmetry constraints to restrict terms is usually the first step in manually formulating a governing equation, but it often becomes intractable for complex systems with high-order derivatives or multiple fields. When a data-driven method is used, on the other hand, imposing physical constraints such as symmetries typically requires manual preprocessing or computationally expensive iterative procedures. Here, we propose an algebraic filtering method that enumerates all symmetry-allowed terms for continuum equations within a finite candidate space. By treating symmetry generators as linear operators on the candidate space, we reduce the problem of enforcing both discrete and continuous symmetries to solving a set of linear kernel equations. The solution yields a provably complete list of permitted terms. We demonstrate the method's effectiveness by identifying invariant terms for systems with dihedral symmetry and recovering the governing equations for the Toner--Tu and Kardar--Parisi--Zhang systems, including higher-order terms useful for extending known models. The method provides a systematic way to obtain a symmetry-allowed search space for data-driven equation discovery, e.g., the sparse identification of nonlinear dynamics method.

Based on the process tensor framework, we extend the time-evolving matrix product operator (TEMPO) method to solve bosonic quantum impurity problems (QIPs) with off-diagonal system-bath coupling. Our method is a most generic extension of TEMPO, which applies for any QIPs as long as the bath is noninteracting and the system is linearly coupled to the bath. It naturally contains all the current developments of TEMPO in more restricted settings. As an application, we study the real-time dynamics of a spin that is coupled to a sub-ohmic bath via the Jaynes-Cummings-type system-bath coupling, and compare it against that of the standard spin-boson model. Our results show that the commonly used secular approximation could easily fail in presence of a structural bath. Our method provides a unified framework to understand different variants of TEMPO and directly suggests a fermionic generalization which has not been explored so far, it could also be straightforwardly used as an impurity solver in the bosonic dynamical mean field theory.

The historical Mpemba effect involves a first-order phase transition. This has prompted the experimental realization of microscopic proxies in the form of a colloidal particle trapped in an asymmetric double well, for which the Mpemba effect has indeed been observed. We establish that the existence of the one-dimensional Mpemba effect for a polynomial potential is driven solely by the presence of a hard enough boundary, irrespective of the potential's double-well shape. We then show that the physics of the underlying Mpemba effect is governed not only by single-well physics but also by the high-temperature initial regime.

In networks that grow by isotropic redirection (IR), a new node selects an initial target node uniformly at random and attaches to a randomly chosen neighbor of the target. The emerging networks exhibit leaf proliferation, in which the number of nonleaves scales sublinearly as $N^μ$ and the degree distribution has an algebraic tail with exponent $1+μ$. To understand these mysterious properties, we introduce a class of models with redirection to leaves whenever possible. The resulting networks exhibit qualitatively similar phenomenology to IR networks, but avoid the inherent non-locality of the IR growth rule. These networks admit an analytical description of the leaf degree distribution, from which we extract the exponent $μ$.

In this paper, we present a focused-ion-beam-assisted method for preparing magnet tips for magnetic resonance force microscopy measurements. The method electrostatically transfers prefabricated magnetic nanoparticles to microcantilevers, achieving precise control over the magnet overhang past the cantilever leading edge while minimizing the fabrication damage to the leading edge of the tip magnet. We demonstrate successful fabrication of magnets ranging in size from 460 nm to 2.8 um. These magnets were affixed to two types of cantilevers: silicon cantilevers with a spring constant of 800 uN/m, and single-crystal silicon cantilevers with a spring constant of 30 uN/m. We show that the electrostatic transfer method enables a wide variety of tip shapes, sizes, and materials that were previously not possible with conventional fabrication methods. The transfer procedure allows us to prefabricate the desired particle geometry with minimal ion-beam damage, as confirmed by Monte Carlo simulations. We show that the technique is versatile and can be used to fabricate custom-tipped cantilevers for a broader range of scanning probe techniques.

Topological phase transitions in generic multiband systems are mediated by band-touching defects whose codimension -- the number of momentum directions along which the gap closes linearly -- varies across universality classes. Although singular behavior of fidelity susceptibilities and quantum Fisher information (QFI) has been computed for specific models, no unifying principle connecting these results has been identified: it has remained unclear whether the controlling variable is spatial dimensionality, band structure, or an intrinsic geometric property of the defect. We resolve this question by showing that the singular contribution to the QFI with respect to the tuning parameter $m$ obeys a universal power-law scaling $\sim |m|^{p-2}$ for $p \neq 2$, with a logarithmic divergence $\sim \ln(1/|m|)$ at the marginal codimension $p = 2$, where $p$ denotes the codimension of the band-touching defect. This exponent is independent of spatial dimensionality, anisotropies, ultraviolet regularization, and additional gapped bands, and is protected by renormalization-group arguments at the linearized fixed point. The result unifies previously isolated observations for SSH chains ($p=1$), Chern insulators ($p=2$), and Weyl semimetals ($p=3$) as instances of a single codimension-dependent universality class, and reveals that only defects with $p \leq 2$ generate divergent information-geometric responses. This establishes a direct and previously missing link between topological classification in momentum space and quantum distinguishability in parameter space, with implications for metrological sensitivity near topological transitions and for the experimental detection of topological criticality via quantum geometric observables.

We investigate a classical formulation of the Quantum Approximate Optimization Algorithm (QAOA), realized as a Hamiltonian dynamical system of classical kicked tops, which we call the Variational Iterative Rotation Algorithm (VIRAL). The variational parameters are the transverse and longitudinal rotation angles at each of the p layers of the circuit. We find that VIRAL outperforms QAOA on the canonical Sherrington-Kirkpatrick spin-glass benchmark at all circuit depths, with the energy density converging to the ground state value linearly in 1/p. For large circuit depths, the optimized dynamics follows a Floquet protocol in which a pitchfork bifurcation destabilizes the equatorial fixed point and drives the spins toward polar Ising configurations. Our results demonstrate that the effectiveness of QAOA-like protocols derives primarily from their underlying iterative rotation structure, and that a classical implementation of it outperforms its quantum counterpart. We further elucidate its efficiency by reducing the many-body classical evolution to an effective Landau-Lifshitz dynamics for a single spin in a stochastic magnetic field. In this picture, the covariance matrix of the effective field reveals a nearly rank-one structure in which a single mode dominates the stochastic dynamics. In contrast, quantum fluctuations make the noise covariance of the effective quantum model of higher rank, hampering the control of the system. We propose nanometer-scale magnetic tunnel junctions as a natural physical platform for implementing VIRAL, where spin rotations can be realized using magnetic fields and spin torques.

Recent advances have allowed to tackle path-space probabilistic representations of mesoscopic Boltzmann transport nonlinearly coupled to a sub-model of the force-field by step forward approaches in terms of continuous branching stochastic processes. In this work, path-space probabilistic representations of free-space Poisson-Vlasov and Poisson-Boltzmann systems are exhibited. This yields novel propagator representations and opens new routes for efficient and reference simulations by use of new branching backward Monte Carlo algorithms. Subsequent statistical estimator are benchmarked on gravitational clusters and plasmas dynamics.

We develop a self-consistent free-energy framework in which membrane shape and osmotic pressure are determined simultaneously in a finite reservoir by minimizing bending elasticity and solute entropy. Solute conservation makes osmotic pressure a thermodynamic variable rather than an externally prescribed parameter, producing a nonlinear coupling between membrane mechanics and solvent entropy. This coupling modifies the classical stability condition for spherical vesicles: instability emerges from global free-energy competition rather than the linear Helfrich stability criterion. The resulting critical pressures differ by orders of magnitude from Helfrich predictions and agree with simulations for small and large unilamellar vesicles. The framework is relevant to cellular environments involving biomolecular condensate confinement as well as synthetic vesicles and the development of osmotic-pressure-driven encapsulation platforms.

Hyperspectral photoluminescence (HSPL) imaging provides spatially resolved spectral information for monolayer transition-metal dichalcogenides (TMDs), enabling the detection of subtle variations in excitonic features that are not accessible with conventional optical or photoluminescence intensity imaging. We employ HSPL to map the microscopic spatial distribution of strain and disorder in hBN-encapsulated MoSe$_2$ and WSe$_2$ samples. Quantitative extraction of exciton, trion, and biexciton energies and linewidths reveals strain gradients and localized deformations, such as wrinkles and ripples. The technique allows for characterization of regions with uniform optical properties and identification of areas affected by micro-scale disorder, which may be missed by optical microscopy. Measurements on samples with different device architectures and fabrication processes demonstrate the general utility of hyperspectral PL imaging for assessing spatial heterogeneity and optoelectronic quality in two-dimensional materials.

Computational chemistry has become an indispensable tool for generating data and insights, pervading all branches of experimental chemistry. Its most central concept is the potential energy hypersurface, key to all chemistry and materials science, as it assigns an energy to a molecular structure, the necessary ingredient for reaction mechanism elucidation and reaction rate calculation. Density functional theory (DFT) has been the most important method in practice for obtaining such energies, which is mirrored in the use of high-performance computing hardware. In the last two decades, a new class of surrogate potential energy functions has been evolving with remarkable properties: quantum accuracy combined with force-field speed. Until very recently, their application was hampered by the fact that they needed to be trained on truly large system-specific data sets, generated before a computational chemistry study could be started (in sharp contrast to DFT, which, as a first-principles method, works out of the box, but at a far higher price of computational cost). Very recently, this roadblock has been overcome by so-called foundation machine learning interatomic potentials, which are poised to completely change the way we do computational chemistry, likely prompting us to abandon DFT as the prime method of choice for this purpose in less than a decade.

Symmetries play a central role in quantum many-body physics, yet uncovering them systematically remains challenging. We introduce a bootstrap framework designed to reconstruct the representation theory of hidden finite group symmetries of quantum many-body lattice Hamiltonians, using only a known symmetry subgroup $N$ and spectral correlations between its symmetry sectors. We introduce a novel variant of the spectral form factor, the cross spectral form factor (xSFF), which we compute via exact diagonalization to seed the bootstrap algorithm. By applying the constraints derived from these data alongside the algebraic conditions of the fusion rules, our bootstrap procedure sharply restricts the set of candidate groups $G$. Remarkably, without any prior assumptions regarding the full symmetry group $G$, our method can systematically recover its representation-theoretic data, including the number and dimensions of the irreducible representations, their branching rules with respect to $N$, the fusion algebra, and the full character table. This framework applies equally well to chaotic and integrable many-body systems and accommodates both unitary and anti-unitary symmetries. Through various examples, we demonstrate that the underlying group $G$ can be uniquely identified. In particular, our bootstrap independently recovers the $\mathbb{Z}_4$ symmetry at the self-dual point of the three-state quantum torus chain, detects signatures of projective representations in the effective Hamiltonian of the driven Bose-Hubbard model, and rediscovers the $η$-pairing $\mathrm{SO}(4)$ symmetry of the one-dimensional Fermi-Hubbard model. Our framework thus establishes a practical route to identify symmetries directly from dynamical spectral observables.

The chiral vortical effect (CVE) -- an axial current driven by rotation in chiral matter -- appears in systems ranging from relativistic fluids to Weyl semimetals, yet its quantum origin remains unclear because existing derivations are semiclassical. We present an exact quantum solution of a rotating Weyl fermion in a finite cylinder. We show that the bulk vortical response is entirely a magnetization current while the current density on the rotation axis remains finite and matches semiclassical predictions. For spin-polarized boundary conditions, we uncover an additional effect beyond the known CVE: a robust family of chiral modes that transport axial charge, $ΔQ=χN^2\,Δθ/4π$, under rotation by angle $Δθ$, where $χ$ is the Weyl node chirality and $N$ is the number of chiral modes. The pump is independent of temperature, Fermi level and Weyl velocities, but depends on the UV-sensitive number $N$. These results establish a fully quantum picture of the CVE and reveal a boundary-enforced chiral spectral structure underlying vortical response in Weyl systems.

Recent advances have allowed to tackle exact path-space probabilistic representations of macroscopic advection-diffusion models involving advection nonlinearities by step forward approaches in terms of continuous branching stochastic processes. Yet, the need of such paradigm shift is huge for the broad flied of fluid flows. In deed, wherever for climate dynamics, engeenering, geophysical and planetary formations, or biomedical applications, complex transport phenomena involving diffusion and advection in confined domains set the physics. In this work, we advance this framework by casting such branching representations within the class of Navier-Stokes strongly nonlinear transport. This yields novel propagator representations for fluid dynamics and opens new routes for efficient simulations of fluids in confined domains by use of new Backward Monte Carlo algorithms.

Floquet engineering, the control of a quantum system by means of time-periodic driving, allows to modify the properties of the system so that it becomes described by an approximate effective time-independent Hamiltonian. However, in the presence of interactions the stabilization of interesting many-body ground states of such effective Hamiltonians is possible only on a certain time scale, beyond which Floquet heating sets in, as it results from unwanted driving induced resonant excitation. Moreover, already the preparation of such states is challenged by excitations due to imperfect adiabatic dynamics, especially when a phase transition has to be passed. Here, we propose a general dissipative strategy for the preparation and stabilization of effective ground states that are protected by an energy gap. Our approach relies on coupling the driven system to a thermal bath, the properties of which are chosen so that it both suppresses Floquet heating and guides the system into a non-equilibrium steady state with a large occupation of the effective ground-state, but generally non-thermal occupations of excited states of the effective Hamiltonian. We use the Floquet-Born-Markov master equation to verify the proposed strategy for the example of a strongly driven Bose-Hubbard chain with an effective gapped Mott-insulator ground state.

We reveal a correspondence between temperature and integrability-breaking in classical and quantum many-body systems through the lens of geometry and adiabatic transformations. Decreasing the temperature, obtained in a standard way through the derivative of entropy with respect to energy, steers the system towards an integrable point despite strong integrability-breaking interactions. Auto-correlation functions of local observables exhibit slow relaxation dynamics, which violates ergodicity on the approach to this integrable point. Subsequently, the average fidelity susceptibility of stationary states satisfies scaling relations near the integrable point, in close analogy with continuous phase transitions. We further find that the dynamical exponent encompassing relaxation can be different in the quantum and classical models, depending on dimension of the systems. Collectively, our results establish temperature as a tunable control parameter for chaos and puts it on equal footing with integrability-breaking perturbations.

Message passing, also known as belief propagation, is a versatile framework for analyzing models defined on networks. Its most prototypical application is percolation; yet, the interpretation of the message passing formulation of percolation remains elusive. We show that the message passing solutions commonly associated with the probability of belonging to the giant component actually identify reachability from cycles. This interpretation applies to bond and site percolation on arbitrary undirected or directed networks. Our findings emphasize the distinction between transition in cyclicity and the emergence of the giant component.

Controlling the motion of non-Newtonian drops on surfaces is crucial for applications ranging from inkjet printing to biomedical devices and food processing. While the macroscopic behavior of viscoelastic drops sliding on tilted hydrophobic surfaces has been characterized, showing reduced velocities and elongation compared to Newtonian fluids, the underlying microscopic mechanisms remain poorly understood. To address this gap, we developed a high-speed, high-resolution reflection microscope that enables direct visualization of the contact line of sliding drops. We used water/soluble polyelectrolyte solutions based on polyacrylamide and let drops sliding on hydrophobic substrates composed of Teflon AF- and PDMS-coated glass slides. The substrate tilting angle was varied between 20° and 45°. We reveal how viscoelasticity influences the dynamics of the receding contact line and drop motion. Our experiments demonstrate that viscoelasticity can destabilize the receding contact line, triggering filament formation. This instability previously observed in the coating of thin viscoelastic films, is reported here for the first time in sliding drops. We further highlight the critical role of polymer charge in this process: while cationic and non-ionic polymers promote filament formation, anionic polymers do not, a difference we attribute to the distinct wetting properties of the solutions. In conclusion, we clarify the interplay between rheology, surface interactions, and drop dynamics.

Proteins can combine into functional elements in living cells or self-assemble into unwanted structures in a number of diseases. The resulting aggregates often display filamentous morphologies across a large range of protein shapes and molecular interactions. This has led to the suggestion that filament formation could be a generic outcome of the aggregation of geometrically complex, ill-fitting objects, although such a mechanism has not been demonstrated in three dimensions. To address this problem, we theoretically study the self-assembly of three-dimensional identical, ill-fitting deformable subunits mimicking globular proteins in solution. In our model, self-assembling subunits incur deformations that accumulate as the aggregate size increases and can eventually hamper further assembly. We analytically predict the ground state morphologies of the resulting aggregates as a function of the subunit adhesivity and elasticity by mapping their mechanics onto those of two incompatible, interconnected networks. We find that zero-dimensional clusters, three-dimensional bulks as well as symmetry-broken one-dimensional filaments and two-dimensional layers can all form depending on assembly parameters. Incompressible, moderately adhesive subunits favor filaments. These findings hint at a generic pathway to control self-assembly in three dimensions and suggests that such mechanisms could be investigated in more realistic protein models.

Deterministic many-body systems governed by simple interactions can self-organize into macroscopic patterns, and the determinants of long-time behavior are assumed to be encoded in the initial configuration. Here we show that predictability can instead be constructed dynamically rather than being accessible in the initial configuration. We study a generalized cellular automaton of secrete-and-sense cells that self-organizes from disorder into static configurations, rectilinear waves, or spiral waves. Although dynamics are deterministic, the final outcome cannot be reliably inferred from the initial state alone. Treating cell states as a discrete phase field, we uncover emergent topological modes - charged vortices connected by strings that form non-contractible loops. Tracking their dynamics reveals that predictive signatures of macroscopic fate appear only late in the trajectory: vortex annihilation becomes readable through loop loss, whereas vortex persistence remains unreadable until spiral waves form abruptly. These results show how predictability can be dynamically constructed in deterministic nonequilibrium systems.

We propose an active Cahn-Hilliard theory for the dynamics of a new type of phase transition where the driving force is not the direct interactions between the two separating components, but their active sorting by a third polar species. This third species can transport the other two along its polarity in opposite directions, thus separating them. Inspired by recent experiments where molecular motors that walk in opposite directions along microtubules are sorted into separated domains, our theoretical description of this process introduces a new mechanism for active phase separation and could serve as a model for the organization of biological material in space inside cells. We predict the formation of motor domains, and further show that they can either coarsen to form macroscopic phases or reach a finite micro- or mesoscopic steady state size, these latter due to an arrest of coarsening through activity.

Recently, purpose-built analog hardware that can efficiently minimize the Ising energy and thereby solve a variety of combinatorial optimization problems has been receiving widespread attention. In this work, we show how multidimensional, vectorial degrees of freedom, that are either naturally present or can be artificially created in such hardware, could strengthen the capability to find optimal solutions to optimization problems. In order to achieve this, we introduce a simple model of soft vector spins that should be implementable on a variety of analog hardware platforms as well as three different dimensional annealing methods which harness the enlarged phase space of the vectorial degrees of freedom to minimize the Ising energy. We perform simulations on different benchmark problems and show that for all dimensional annealing methods we tested, vectorial degrees of freedom improve upon one-dimensional degrees of freedom when it comes to finding the ground state of the Ising model. In particular, we find that this advantage becomes most pronounced for $d \gtrsim 3$ dimensional degrees of freedom, with diminishing returns as the dimension is increased further. Our results could inspire new analog optimization hardware and algorithms that explicitly incorporate the advantage of vectorial degrees of freedom.

The phase transition in the Vicsek model is widely believed to be associated with spontaneous symmetry breaking of the two-dimensional rotational symmetry $O (2)$. In this paper, we revisit the original Vicsek model introduced in [Phys. Rev. Lett. 75, 1226] and demonstrate that the model lacks $O (2)$ symmetry. As a consequence, we numerically demonstrate that the phase transition reported in the original paper vanishes when the global phase is adaptively chosen.

We investigated the nonreciprocal spin-wave properties, including the frequency shift, of a helical equilibrium state in a versatile CoFeB/NiFe bilayer. Through an extension of the dynamic matrix formalism (developed in this work) to an arbitrary non-collinear configuration along a heterostructured multilayered system thickness, we explained the frequency shift via differences in the dynamic dipolar and interlayer exchange interactions arising from the distinct spin-wave mode profiles across the bilayer thickness for counterpropagating modes at the same wave vector. In contrast to recent literature wherein the frequency shift is attributed solely to the dipolar interaction, our results and explanations hereby presented involve a starring role of the interlayer exchange interaction not accounted in current literature. Furthermore, we also found a combination of large frequency shift values and sub-100 nm spin wave wavelengths that can be tuned or even enhanced with the twisting degree of the helical magnetization state by the application of the external field, and with the thickness of the NiFe sublayer, which might be highly relevant for magnonic applications. We validated our model and the physical mechanism that explains the frequency shift using recent simulations and experimental results.

This work investigates the origin of Kardar-Parisi-Zhang (KPZ) scaling in the phase dynamics of one-dimensional and two-dimensional polariton condensates. We demonstrate that the key mechanism leading to the observed power laws for the first-order correlation function $g^{(1)}$ is the fluctuation of the population of Goldstone modes, which arise due to the spontaneous breaking of $U(1)$ symmetry. Numerical simulations and analytical theory confirm that the critical exponents describing the KPZ universality class directly follow from the dynamics of Goldstone excitations. Our results establish a direct connection between the microscopic parameters of arrays of exciton-polariton condensates and the coherent properties of the light they emit.

The weakly first-order nature of the two-dimensional 5-state ferromagnetic Potts model poses challenges for numerical study. Using density-matrix and tensor-network renormalization group methods, we investigate these transitions of the Potts-$q$ model via the Klein bottle ratio $g$ on original and dual lattices. Finite-size scaling of $g$ as a function of transverse system size $L_y$ accurately locates the critical points for $q = 4, 5, 6$. We further examine the transfer-matrix spectra and entanglement entropy, extracting central charges through toroidal and Klein bottle boundary conditions. For $q = 5$, the extracted central charge ($c \approx 1.14811$) is close to the real part of the theoretical value $c_{5\text{-Potts}} = 1.1375 \pm 0.0211 i$ predicted by complex conformal field theories. The observed drift in the scaling exponent $b$ effectively distinguishes the continuous transition from the weakly first-order regime. Furthermore, the extrapolated divergence of $g$ confirms the first-order nature of the $q=5$ Potts model.

We study a paradigm of coding for compression of the natural numbers via the zeta distribution and develop a statistical-mechanical interpretation, both in terms of Hagedorn systems and a Bose gas with energy levels given by logarithms of prime numbers. We also propose a simple coding scheme for the zeta distribution that nearly achieves the ideal code length. For block coding of vectors of natural numbers, we derive the micro-canonical entropy function and demonstrate its asymptotic linearity implying that its behavior is analogous to that of a Hagedorn system. We also derive the large deviations rate function, and provide a formula for the best coding parameter in the large deviations sense. We show that due the Hagedorn-type phase transition there is only partial equivalence of ensembles, due to the degeneration of the domain of the partition function.

We present micromagnetic simulations and experiments on voltage-assisted field switching in perpendicular magnetic tunnel junctions (MTJs) with a synthetic antiferromagnetic (SAF) free layer, where the magnetic state of one sublayer is detected via tunneling magnetoresistance (TMR). Simulations reveal that local modulation of perpendicular magnetic anisotropy (PMA) in one SAF sublayer leads to distinct switching characteristics. The switching field varies linearly with the anisotropy field, indicating voltage-controlled magnetic anisotropy (VCMA)-dominated dynamics similar to single free-layer devices. We then experimentally study the magnetic switching field of MTJ devices with SAF free layers under applied gate voltage. By varying the MgO tunnel barrier thickness to systematically modulate the resistance-area (RA) product, we enable quantitative separation of spin-transfer torque (STT), VCMA, and Joule heating contributions. Our findings indicate that VCMA dominates in devices with a high RA product, while low-RA devices exhibit nonlinear switching behavior due to enhanced contributions from STT and Joule heating. Furthermore, the effective fields derived from STT, VCMA, and Joule heating contributions under various gate voltages show minimal dependence on device critical dimensions, indicating favorable scaling behavior. This work presents a unified framework analyzing the roles of STT, VCMA, and Joule heating in SAF-based voltage-gated spin-orbit torque (SOT) MRAM, offering key insights for the optimization of performance, energy efficiency, and scalability in SOT-MRAM technologies.

The origin of water's anomalous behavior remains a central open problem in the physical sciences and is often attributed to a liquid-liquid transition (LLT) between high- and low-density liquid states deep in the supercooled regime. Experimental access to this region has been challenging due to rapid crystallization, leaving atomistic simulations as a major source of supporting evidence. Using extensive machine-learning-accelerated first-principles simulations in direct comparison with spectroscopic, structural, and dynamical experimental measurements, we show that features commonly interpreted as signatures of two-liquid behavior emerge at the onset of dynamical arrest. Specifically, we find that two-state fluctuations previously associated with an LLT reflect a transformation from a high-density liquid to a kinetically arrested low-density glass. By mapping equilibrium dynamics across pressure and temperature, our results suggest a reassessment of water's metastable landscape, in which apparent two-state behavior may reflect a relatively high glass-transition temperature of ambient-pressure low-density water, 189$\pm$8 K -- remarkably close to the temperature previously associated with the LLT.

When coupling thermal baths at different temperatures, negative differential thermal conductivity is typically attributed to nonlinear interactions in the connecting medium. In this work, we demonstrate that such an effect can arise purely from the nature of the thermal baths and their coupling with the medium. Specifically, we construct a bath composed of overdamped thermal particles, which is coupled to one end of a harmonic chain, while the other end is connected to a standard Langevin heat bath. By analyzing the steady-state heat current, we observe significant negative differential thermal conductivity. In particular, as the temperature difference between the two baths diverges, the steady-state heat current through the chain vanishes. The effect is thermokinetic: we compute the effective dissipative coefficient and we find that it scales inversely with the square of the temperature of the particle bath in the high-temperature limit, resulting in an asymptotic decoupling between the bath and the chain. Our results highlight that nonequilibrium transport properties can be strongly influenced by the structure of the environment and its coupling to the system, even in otherwise linear systems.

Ge/SiGe quantum well heterostructures confining a high-mobility two-dimensional hole gas (2DHG) have emerged as a compelling platform for hybrid superconductor(S)-semiconductor(Sm) quantum devices. Here, we investigate the low-temperature transport properties of split-gate quantum point contacts (QPC) defined in one such heterostructure and positioned at different distances from an aluminum superconducting contact. We observe ballistic one-dimensional transport evidenced by conductance quantization with at least four clearly visible plateaus. Andreev reflection at the S/Sm interface induces a 40% enhancement of the conductance steps relative to the normal-state conductance staircase measured under a 100-mT out-of-plane magnetic field. This result is in excellent agreement with the theoretical expectation for an interface transparency of 0.88. By operating the QPCs in the tunneling regime, we probe the local density of states of the proximitized 2DHG. We report direct experimental evidence of an induced superconducting gap, demonstrating that its magnitude can be tuned by a gate voltage acting on the carrier density in the 2DHG.

A coarse-grained molecular simulation approach originally developed for entangled polymeric liquids is extended to model the mechanical behavior of long-fiber networks. The model, based on the slip-link picture of chain entanglements, resolves the force balance at contact points and accounts for fiber slippage under these topological constraints. Two key governing equations describe the time evolution of contact-point positions and the local fiber fraction between adjacent contact points. A yield-force criterion determines whether contact points are displaced or remain pinned, as well as whether fiber slippage occurs at contact points. Uniaxial compression simulations corresponding to press molding of fiber-reinforced thermoplastics were performed for networks with varying fiber lengths and compression rates. The results were qualitatively consistent with experimental observations of long-fiber thermoplastics. The model captures physics inaccessible to the classical van Wyk theory of fiber network compression, which is quasi-static and insensitive to fiber length. This work demonstrates that the slip-link framework, already validated for polymer melts, provides a promising mesoscale simulation tool for understanding and predicting the processing behavior of non-thermal fiber networks.

We present a theoretical study of quantum magneto-transport and magneto-optical (M-O) properties in modified-Haldane model; which is applicable to diverse classes of two-dimensional (2D) quantum materials such as buckled Xene monolayers and transition metal dichalcogenide (TMDC) monolayers. By varying the staggered sublattice potential and intrinsic spin-orbit coupling, we identify distinct topological regimes and analyze their manifestations in the emergence of Landau levels, the evolution of the density of states, and the characteristics of M-O absorption spectra. Using the Kubo formalism, we compute the longitudinal and Hall M-O conductivities and show that inter-Landau-level (inter-LL) transitions produce characteristic resonance features that provide optical signatures of the underlying topological phases. Within this framework, we demonstrate electrically tunable topological phase transitions in buckled silicene. Extending our study to monolayer TMDCs, we show that inspite of large band gap, the spin-valley coupling provides a powerful tool for tailoring M-O absorption features across wide range of 2D materials. Collectively, these results underscore modified-Haldane-model materials as an ideal testbed for engineering quantum transport, with promising applications in topological photonics, valleytronic devices, and next-generation optoelectronics.

The in-plane dielectric constant of nanoconfined water has attracted growing interest over the last years. Nevertheless, this magnitude is not well-defined at the nanoscale due to its dependence on the arbitrary choice of water width. We propose the in-plane 2D polarizability, $α_{\parallel}$, as an unambiguous characterization of the in-plane dielectric response under 2D confinement, in analogy to what has been recently done for the perpendicular response. Using classical molecular dynamics simulations, we compute $α_{\parallel}$ via two independent and consistent methods: based on fluctuation--dissipation theory, and from the induced dipole moment when water is placed in a capacitor. Our results provide the framework to quantify the in-plane dielectric response of polar liquids across simulations and experiments.

We study the impact of BN's phonons on the electrical resistivity of hBN-encapsulated graphene. While encapsulation yields high-mobility devices, the surrounding BN itself introduces remote scattering from polar optical phonons, whose role in standard resistivity measurements remains unclear. We combine high-quality transport experiments with ab initio calculations including a proper treatment of dynamically screened remote interactions. We demonstrate that hBN's out-of-plane phonons strongly influence resistivity between 150 K and room temperature, whereas higher-energy LO modes and intrinsic graphene phonons alone cannot explain the observed trends. The coupling between electrons and the BN's phonons becomes more pronounced at low carrier densities due to reduced screening. Our findings establish that remote phonon scattering fundamentally limits transport in encapsulated graphene, solving a longstanding debate.

Face-to-face interactions reveal recurring patterns, suggesting the possibility of shared underlying mechanisms. More specifically, inter-contact durations, contact durations and number of contacts per edge share similar heavy-tail distributions in many empirical settings. A common intuition is that face-to-face interactions may be influenced by spatial constraints, and that the observed complex behaviors could arise from such physical limitations. Our models explore the impact of this constraint by simulating pedestrian dynamics, and studying the generated temporal network of contacts. Previous work showed that the inter-contact duration distribution is recovered with a pedestrian dynamic as simple as the two dimensional random walk, but this approach doesn't allow to recover the distribution of the number of times a pair of individuals has been in contact. One assumption is that the number of contact between individual arises from the social relationship between them, in other words a memory of past interactions. However, we here present models that are based on solely spatial rules, by adding simple targeting mechanisms to the two-dimensional random walk. We show that these models allow to recover a broad distribution of the number of contacts, revealing the importance of two ingredients: localized phases and controlled population mixing. This suggests that the observed heterogeneity in the contact numbers within the data does not necessarily emerge from underlying social relationships between individuals, since an equivalent distribution may be reproduced using a purely spatially based model, without the need for memory mechanisms.

Nonreciprocity is known to generate a wide range of exotic phenomena in multi-species many-body systems, where different species influence one another through couplings that violate Newton's third law. In contrast, in the absence of explicitly imposed macroscopic nonreciprocal processes, single-species nonreciprocity -- another distinct form of nonreciprocity -- typically plays only a limited role in shaping macroscopic physics. Here, using a single-species Vicsek model with a vision cone and extrinsic noise, we show that spontaneous symmetry breaking (SSB) can dramatically enhance the macroscopic consequences of microscopic single-species nonreciprocity. In the ordered phase, this enhancement gives rise to an emergent macroscopic nonreciprocity that induces the system of identical active particles to admit an effective description with a "two-species" non-Hermitian structure. The resulting SSB-enhanced nonreciprocity substantially promotes traveling-band formation and, more strikingly, drives a novel real-space condensation of identical active particles, characterized by a "traveling line" with vanishing longitudinal width. Our findings uncover a fundamental mechanism by which microscopic single-species nonreciprocity can exert strong macroscopic influences in complex systems.

Reliable approximations for correlation functions at intermediate and strong coupling remain hard to obtain for general quantum field theories. Perturbative expansions are often asymptotic or have a finite radius of convergence, which limits their applicability beyond weak coupling. Here we combine weak- and strong-coupling expansions and propose to use two-point Padé schemes to construct approximants. For lattice $φ^4$ theory, we show that this two-point interpolation strategy yields accurate global approximations to the two-point correlation function across broad coupling regimes and compares favorably with standard one-point resummation methods. We also provide heuristic explanations for the observed convergence behavior and discuss the practical range of validity of the approach.

We report a method for producing an array of fifty two ion-sensitive PEDOT:PSS organic electrochemical transistors on a glass coverslip, each featuring an integrated fluoropolymer microwell sealed with lipid bilayer into which membrane proteins can be inserted for simultaneous electrical and fluorescence microscopy studies. To demonstrate capability, we fill the microwells with an `inner' phosphate assay buffer solution containing 20 $μ$M Alexa-488 dye and 50 mM KCl, seal the microwells with lipid bilayer using an aqueous-organic-aqueous liquid exchange technique, and then fill the common flow-cell volume above the sealed microwells with a dye-free `outer' phosphate assay buffer containing 100 mM KCl. We insert $α$-hemolysin, which embeds into the lipid bilayer forming a heptameric pore with diameter ~ 2.6 nm. The pore allows K$^{+}$ ions to diffuse into the microwell and Alexa-488 dye molecules to diffuse out of the microwell producing a corresponding drop in transistor conductance and microwell fluorescence intensity, respectively. These two signals occur at different timescales, consistent with the known size difference between K$^{+}$ ions and Alexa-488 molecules. Our approach to fabricating microwell arrays with PEDOT:PSS OECTs incorporated into the bottom of selected microwells distributed in the array is both scalable and versatile, opening a path to studies using larger arrays and with other membrane proteins embedded in the lipid bilayer sealing the microwells.

Lack of inversion symmetry in metals breaks SU(2) symmetry which results in spin-splitting of the electronic states at the Fermi level due to various types of spin-orbit coupling (SOC) such as Dresselhaus, Rashba, or Ising (also called valley-Zeeman). This splitting is known to enable both incoherent spin-flip excitations and coherent chiral-spin modes. Another effect of breaking of SU(2) is the introduction of a direct spin-photon interaction. We use this concept to formulate a theory of inelastic scattering of photons from the charge carriers of such a system [electronic Raman scattering (eRS)]. As a result of broken SU(2), we show that the eRS probe, unlike conventional theory of Raman scattering, couples to spin excitations even without tuning the laser to an internal resonance. We show that the spin dependent excitations induced by photon scattering are sensitive to the polarization geometries as well as to the spin structure of the Hilbert space of the low-energy states. As a concrete realization, we examine doped/gated graphene on substrates with strong SOC with various compositions of Rashba and valley-Zeeman SOC and compare their spectra with those for a model 2D electron gas (2DEG). The spectra are shown to have a resonant feature in select polarization geometries near the SOC-splitting energy and, importantly, is shown to be different in the two systems. The signal in graphene systems is shown to be stronger than that in a 2DEG by orders of magnitude owing to the large Dirac velocity. We also outline how the lineshapes from the spectra can be used to infer various components of SOC in the system.

Nanoscale topological polar textures promise new functionalities for ferroelectric memories and logic, yet their three-dimensional structure and mesoscale organization remain experimentally inaccessible. Here we introduce depth-resolved electron diffraction imaging (DREDI), a fast, non-destructive, method that maps polarization with <50 nm lateral and <10 nm depth sensitivity within fraction of a second. Its high acquisition speed enables the first continuous polarization mapping across six orders of magnitude in length scale, from nanometers to millimeters. Using epitaxial BiFeO3 films, DREDI reveals a hidden depth evolution of polar textures: surface 71-degree stripes evolve into subsurface flux-closure vortices that bifurcate into three-fold vertices near the bottom interface. Cross-sectional multi-slice electron ptychography and phase-field modeling confirm these buried configurations and attribute them to strain heterogeneity and ferroelastic twinning in the SrRuO3 electrode. Large-area analysis further shows that vertex-like frustration forms a mesoscale percolating network above a critical length scale of 4 um. DREDI enables real-time, volumetric studies of buried topological textures in ferroic nanomaterials.

Radio-frequency excitation directly reshapes the mesoscale charge-density-wave landscape in quasi-two-dimensional 1T-TaS2 thin films. Under combined RF and DC bias, the hysteretic current-voltage characteristics associated with the nearly commensurate-incommensurate transition are strongly altered, displaying RF-driven collapse, branching, and multiple step-like features that depend on frequency and drive amplitude. In-situ Raman measurements show enhanced intensity and linewidth narrowing of low-frequency CDW phonon modes, consistent with reduced dephasing and increased coherence of the periodic lattice distortion under RF drive. This behavior is captured by combining an overdamped time-dependent Ginzburg-Landau description of the commensurate CDW with a morphology-informed percolative resistor-capacitor transport model. The simulations indicate that oscillatory driving anneals frustrated domain configurations, reduces domain-wall density, and reorganizes the discommensuration network, while the transport model reproduces the resulting hysteresis, avalanche-like pathways, and RF-induced conductance steps. RF driving therefore provides an effective route for controlling collective electron-phonon order and accessing metastable transport states in 1T-TaS2, with implications for reconfigurable RF electronics, memory, and unconventional computing based on correlated materials.

Entropy production is often used as a proxy for energy consumption of a non-equilibrium system. Lower bounds can be estimated from coarse-grained observations, and this has been done for various biological systems. Here, we apply these tools to a more macroscopic system whose true energy consumption is also known. We find that while entropy production does give a lower bound, it is some 25 orders of magnitude away from being saturated. To be certain of this result, we survey different methods of estimating irreversibility, and write down a novel kNN estimator.

Graphene Josephson junctions (JJ) provide a promising platform for ultra-broadband quantum sensing of light owing to graphene's frequency-independent absorption, vanishing electronic heat capacity, and weak electron-phonon coupling, which enable rapid suppression of the critical current through radiation-induced electron heating. Existing investigations have been confined to the microwave and infrared regimes, where competing detector technologies are already established; by contrast, the terahertz (THz) band - where sensitivity is most urgently lacking and no mature quantum sensor exists - has remained largerly unexplored. Here we demonstrate a strong photoresponse of graphene JJs at THz frequencies, establishing a first experimental step towards graphene-based THz quantum sensors. Under low-intensity illumination, we observe a pronounced suppression of the critical current that generates a strong photovoltage (Vph) under current bias. By tracking this Vph and independently measuring the electron temperature as a function of absorbed power, we extract a responsivity of 88 kV W^-1 and a noise-equivalent power of 45 aW Hz^-1/2 at 1.7 K. Furthermore, gate tunability of our JJ enables access to a regime where hysteretic current-voltage characteristics persist up to 0.9 K, offering a potential route toward single-photon THz detection beyond millikelvin (mK) temperatures. These findings establish graphene JJ as a versatile platform for broadband cryogenic radiation sensing and point towards their use as quantum sensors at THz frequencies.

Even in the linear limit, the topology of multifold (also called higher-order) exceptional points across the Brillouin zone has lacked a general characterization, leaving the doubling theorem essentially limited to two-fold exceptional points. Here, we establish the doubling theorem of $n$-fold exceptional points [EP$n$s ($n=2,3,\ldots$)] for systems where nonlinearity enters through eigenvalues. To this end, we introduce new topological invariants, termed frequency-momentum winding numbers, which characterize nonlinear EP$n$s in $m$-band systems throughout the Brillouin zone for arbitrary $n$ and $m$ ($m\geq n$). These invariants enable a unified proof of the doubling theorem in the absence of symmetry and under several symmetry constraints, including parity-time ($PT$) and charge-conjugation-parity symmetries. Furthermore, even in the linear limit, the frequency-momentum winding number indicates $\mathbb{Z}$ topology of $PT$-symmetric EP$2$s which is beyond the previously reported $\mathbb{Z}_2$ topology. The frequency-momentum winding numbers can also be extended to a class of coupled resonators in which nonlinearity enters via the eigenvectors, whereas the spectrum is determined by a nonlinear scalar equation for the frequency.

We present a symmetry-based framework for equilibrium statistical mechanics that formulates a single Lie group combining conventional spacetime symmetries with a recently identified phase-space gauge-shifting invariance [Muller et al., Phys. Rev. Lett. 133, 217101 (2024)]. Using Noether's theorem, we obtain a set of general Ward identities together with previously unexplored cross-relations arising from the noncommutation of different symmetry generators. The approach extends standard many-body symmetries, such as translations, rotations, Galilean boosts, dilations, and particle exchange, by incorporating an internal gauge-shift symmetry within a unified group structure. The resulting Lie algebra suggests a hierarchy of exact identities that encompass established sum rules and indicate possible cross-coupling relations between distinct response and correlation functions. We also identify a Wigner-Eckart-Ward reduction that simplifies tensor-hyperforce correlators to two scalar radial spectra in isotropic fluids, and we outline an equivariant gauge-constrained DFT formulation whose Euler-Lagrange equations are constructed to satisfy the corresponding Ward and cross-Ward constraints. This framework provides a consistent organizational basis for phenomena in liquids, mixtures, and interfaces, and may offer a symmetry-based perspective connecting structure, mechanics, and dynamics in many-body systems.

Virial expansions are the series in powers of density assumed to be small. However, the equations of state require to consider finite densities for which virial expansions, as a rule, diverge. In order to extrapolate a virial expansion to the values, where this expansion diverges, one uses summation methods. The most often used method is the Padé summation, which has several deficiencies. First of all, Padé approximants are not uniquely defined, suggesting a large table of admissible variants. Second, often there appear spurious unphysical poles. On the contrary, in those cases where the existence of a pole is physically motivated, Padé approximants do not necessarily exhibit it. A new approach for the summation of virial expansions is suggested, based on self-similar approximation theory. The method is regular and uniquely defined. It allows for the determination of physically motivated poles. The accuracy of self-similar approximants is not worse than that of the best Padé approximants with fitting parameters or of Monte Carlo simulations. The self-similar summation is based solely on virial expansions, involving no fitting parameters. In some cases, self-similar summation allows for reconstructing the sought functions exactly. The approach is illustrated by summing virial expansions for hard-disk fluids, hard-sphere fluids, and systems with power-law potentials.

The coexisting rhombohedral-like (R', MA) and tetragonal-like (T', MC) monoclinic phases in compressively strained bismuth ferrite thin films exhibit exceptional piezoelectric and magnetic properties. While previous studies have largely focused on probing the morphotropic phase boundaries (MPBs) comprising ordered R'/T' twins, their self-organizing structures remain not fully explored. Here, we observed two types of interphase boundaries in a 60 nm-thick BiFeO3 film epitaxially grown on a LaAlO3 substrate by employing multi-modal diffraction-based electron microscopy. First, the flat MPBs form lines extending >1 mm, and repeat almost every ~20 um. Additionally, we uncover a new type of phase boundary with zig-zag regions of alternating R'/R' and T'/T' twin domains. Cross-sectional multislice electron ptychography confirms the atomic-scale polarization rotation across the MPB, with out-of-plane strain varying >15%. Plan-view electron backscatter diffraction reveals the lattice disclination of ~1.5-degrees across the zig-zag interphase boundaries, while having >2.5 degrees within the MPB. Phase-field modeling suggests that the formation of zig-zag phase boundaries arises from balancing between Landau and elastic energies. We speculate that such well-ordered interphase boundaries are associated with mesoscale in-plane strain modulations, thus providing a way to engineer and harness their properties for potential device applications.

The dielectric response and viscosity are two fundamental properties of liquids that are usually treated separately. Here we show that in highly polar liquids the viscosity can be predicted directly from the dielectric function. We employ a stochastic field theory for thermal dipole-field dynamics coupled to hydrodynamic flow, and derive a very general Kubo relation for the response of an observable to the flow. We then use this to derive a Green-Kubo formula for the viscosity operator in terms of the correlation function for the body force, rather than the usual stress tensor formulation, and from this we derive the contribution to the viscosity due to dipolar interactions. In strongly polar liquids like water we show that viscous dissipation arising from these thermal van der Waals interactions is the dominant dissipative mechanism, leading to a direct connection between dielectric relaxation and viscosity. The theory also predicts the emergence of a second relaxation time in the dielectric response even when only a single microscopic relaxation mechanism is present. This additional timescale contributes to the intrinsic Debye relaxation and provides a natural explanation for the widespread empirical observation that many liquids require two relaxation times to fit their dielectric spectra. By establishing a predictive link between dielectric properties and viscosity, our results revisit classical ideas of liquid dynamics originating with Debye and suggest a practical route for identifying promising solvents for electrochemical energy storage.

Understanding the ultrafast dynamics of photoexcited charges in graphene is essential, as the microscopic mechanisms underlying these dynamics determine many of graphene's optical, optothermal, and optoelectronic properties. These are crucial properties for many functionalities and devices enabled by graphene, such as high-speed photodectors. Therefore, beyond scientific understanding, it is highly desirable to control ultrafast carrier dynamics for practical applications. Here, we establish this control by engineering the dielectric environment of graphene, thereby regulating both heating and cooling dynamics without altering the Fermi energy, optical power, or ambient temperature. By combining optical pump-terahertz probe experiments with theoretical calculations, we show that dielectric screening suppresses carrier-carrier interactions and slows the dynamics. In particular, reduced carrier-carrier scattering delays the formation of a quasi-equilibrium hot electron distribution, thus slowing carrier heating. It also slows carrier cooling because re-thermalization after optical-phonon emission depends on the same interactions. The enhanced screening further reduces the energy of electron-hole puddles, thereby increasing charge mobility and the Seebeck coefficient. This ability to externally control internal graphene dynamics and transport properties enables the optimization of device performance, such as the sensitivity of photodetectors for data communication and wireless communication applications.

In dually flat manifolds, there is a deep connection between gradient flows and pregeodesics. This was one of the many important contributions of Amari to information geometry. In this paper, we extend the study of this relationship to general Riemannian manifolds. Our result does not impose conditions of flatness on the connection or symmetry on its non-metricity tensor, thus broadening the geometric setting beyond Hessian manifolds. Within this framework, we provide a criterion for comparing relaxation along two different gradient descent curves of a function, formulated in terms of the non-metricity tensor of a connection for which the gradient curves are pregeodesics. We use it to study Gaussian chains, whose relaxation trajectories coincide with gradient descent curves in the space of Gaussian distributions.Thus, we recover a recent result that establishes a universal asymmetry: warming up is faster than cooling down. Our work illustrates how geometric insights rooted in Amari's legacy offer new perspectives for optimization problems and stochastic processes.

The recent introduction of recovery rheology has provided qualitatively new physical insights into the yielding and flow of soft matter systems across diverse mechanically driven nonequilibrium protocols by separating the deformation strain into recoverable and unrecoverable components. A striking finding is that the fluid-like response associated with the gradually increasing unrecoverable strain ultimately leads to the continuous yielding transition from a solid to a liquid. We build on the force and particle level Elastically Collective Nonlinear Langevin Equation theory of activated dynamics within a nonequilibrium microrheological framework to formulate a general statistical mechanical foundation of step-rate start-up shear response that relates recovery rheology to microscopic structure, relaxation, and elasticity. Quantitative applications to metastable hard and soft sphere colloidal suspensions reveal testable new predictions and interconnections between macroscopic and microscopic properties: (i) the steady state recoverable strain is directly related to the steady-state shear thinning; (ii) the transient stress overshoot amplitude varies non-monotonically with packing fraction and is quantitatively linked to the steady-state recoverable strain; (iii) the acquired unrecoverable strain dictates the stress overshoot strain; (iv) the predicted enormous reduction of the structural relaxation time under deformation is inversely related to the unrecoverable strain-rate.

We derive a retained-spin micropolar hydrodynamic closure from the Boltzmann--Curtiss equation using a generalized Chapman--Enskog construction in which the local mean spin is retained as a quasi-slow variable. Starting from the exact kinetic balance laws for mass, linear momentum, and intrinsic angular momentum, we isolate the bookkeeping relation between antisymmetric stress and stress-induced spin torque, decompose the first-order source into irreducible scalar, axial, and symmetric-traceless sectors, and show explicitly how the standard micropolar constitutive structure with coefficients $(η,ξ,η_r,α,β,γ)$ emerges. This decomposition makes clear that the one-particle kinetic stress contributes only to the symmetric stress, whereas the rotational viscosity belongs to an intrinsic/collisional transfer channel. For perfectly rough elastic hard spheres, we further obtain explicit dilute-gas estimates for the rotational viscosity $η_r$ from homogeneous spin relaxation and for the transverse spin-diffusion combination $β+γ$ from a transport-relaxation calculation. Targeted event-driven molecular-dynamics simulations are used as a posteriori checks: expanded homogeneous-spin density and roughness sweeps support the predicted $n^2$ and $K/(K+1)$ trends for $η_r$, while finite-$k$ transverse runs provide a qualitative diagnostic of the retained-spin response. The result is a self-contained derivation and coefficient-level estimate of retained-spin micropolar hydrodynamics that clarifies which parts of the closure are exact, which are first-order generalized Chapman--Enskog results, and which remain controlled rough-sphere estimates.

Identifying extensively degenerate zero-temperature states is key in characterizing spin-liquid-candidate materials and spin ices. In experiments, finding zero-point entropy (ZPE) is often attempted by measuring the entropy released by a material when cooled down from very high to very low temperatures. Such investigations are often unreliable and lead to controversial results because accessible temperatures may be insufficient to accurately capture essential low- and high-temperature features of magnetic materials. The purpose of this paper is to point out a simple, easily accessible signature of nonzero ZPE: the Maxwell's relation $\left(\partial S/\partial H\right)_T = \left(\partial M/\partial T\right)_H$ can appear violated if a vanishing ZPE is assumed incorrectly. This relation can further be used for estimating the ZPE. In many materials below characteristic temperatures, the criterion of non-vanishing ZPE has a particularly simple form: $\left(\frac{\partial C}{\partial H}\right)_T\left(\frac{\partial M}{\partial T}\right)_H<0$. We discuss these effects and the ZPE signature in the benchmark test case of the well-studied spin ice $Dy_2Ti_2O_7$.

Nearly a century after Wigner's initial proposal, electron crystals are now a topic of intense experimental and theoretical interest. However, most proposed crystalline phases are commensurate and therefore become insulating in the presence of even weak pinning. In this work we discuss when a commensurate Wigner crystal will spontaneously self dope and develop itinerant carriers, giving rise to an incommensurate and thus metallic Wigner crystal (MWC). We develop a general criterion for the instability of the commensurate crystal which involves the competition between the charge gap at commensurability and a ``packing bias'' whose sign selects whether electron or hole doping is preferred. We then apply these insights to rhombohedral multilayer graphene, where calculations for commensurate crystals reveal instabilities towards self-doping. Carrying out self-consistent Hartree-Fock over the landscape of incommensurate crystals reveals the phase diagram, where a broad MWC phase appears directly adjacent to an insulating Wigner crystal phase. Recent observations of an island of reversed Hall conductance near a putative Wigner crystal phase in rhombohedral graphene are naturally explained by our theory.

When the Coulomb interaction dominates over kinetic energy, electrons can crystallize into a Wigner crystal (WC). This paradigmatic correlated electronic phase has been realized in two-dimensional electron gases with parabolic band dispersion and completely flat Landau levels under high magnetic fields. Beyond these conventional contexts of electron crystallization, more exotic electron crystals have been postulated but remain unexplored. For example, a metallic Wigner crystal (mWC), in which itinerant carriers coexist with a pinned electron lattice, has been proposed theoretically but considered difficult to realize. Non-parabolic electron bands and quantum geometry may facilitate mWC and other novel topological electron crystals. Here we report transport evidence for WC and mWC in rhombohedral tetra-, penta-, and hexalayer graphene in the charge density range 0.3-0.5x10^12 cm^-2. By flattening the conduction band with a gate-controlled displacement field D, we observe an insulating state at nonzero charge density that shows nonlinear, hysteretic current-voltage relations, signatures of a pinned WC, that are absent from the lower-density insulator. Further increasing D reveals transport dominated by hole-like carriers with density up to only 15% of the nominal electron density, consistent with mWC. This mWC state is closely tied to the WC state, as both collapse simultaneously with increasing temperature or bias voltage. The mWC state shows quantum Hall onset near 0.4 T and disobeys the Streda relation, indicating compressible charge exchange between itinerant holes and the transport-inert WC background. Our results establish rhombohedral graphene as a platform for exploring novel electron crystals, as well as possible nontrivial topology, and new collective modes.

In ordered, translationally invariant non-Hermitian systems, the skin effect is understood as a boundary phenomenon: nonreciprocal hopping drives an extensive accumulation of eigenstates towards the edges, whereas the periodic-boundary spectrum remains Bloch extended. Here we experimentally reveal the opposite limit -- a disorder-enabled, boundary-independent, and intrinsically bulk form of skin localization -- the recently predicted erratic non-Hermitian skin effect (ENHSE), realized in a driven photonic platform. Using a time-multiplexed photonic mesh lattice with programmable gain, loss, and phase modulation, we engineer spatially fluctuating imaginary gauge fields and realize a Floquet non-Hermitian lattice whose global reciprocity can be tuned independently of strong local nonreciprocity. We observe a disorder-driven non-Hermitian topological transition between two oppositely directed disordered skin phases through a critical point of global reciprocity. At this transition, boundary skin accumulation disappears, yet the wave dynamics self-organizes into bulk-localized patterns without any interface, providing direct evidence of ENHSE. The measured localization profiles agree with simulations and exhibit the defining feature that distinct eigenstates share a common bulk-localized envelope determined by the disordered imaginary gauge fields. By further introducing controllable on-site disorder, we reveal the competition between ENHSE and Anderson localization, and show how increasing scattering progressively suppresses erratic skin dynamics. Our results help establish ENHSE as a unique disorder-induced non-Hermitian phenomenon and open a route to engineering localization, transport, and topology beyond conventional Bloch and boundary-based paradigms.

Urban scaling laws describe how an urban quantity $Y$ varies with city population $P$, typically as $Y \sim P^β$. These relations are usually obtained from cross-sectional comparisons across cities at a given time (transversal scaling), but their link to the temporal evolution of individual cities (longitudinal scaling) remains unclear. Here we derive explicit expressions for the transversal exponent from the longitudinal dynamics of cities. We show that the measured exponent does not directly reflect individual city dynamics, but instead arises from a snapshot of a heterogeneous ensemble of cities with distinct growth trajectories. As a result, transversal scaling combines intrinsic dynamics with statistical effects due to the distribution of city sizes and correlations between population and city-specific parameters. Consequently, cross-sectional scaling laws cannot, in general, be used to infer the dynamics of individual cities. In particular, apparent sub- or superlinear scaling can emerge even when all cities follow linear longitudinal dynamics, as we demonstrate for the area-population relation. Strikingly, the behavior associated with the transversal exponent is in general not observed in the trajectory of any individual city, underscoring its collective, rather than dynamical, nature. More broadly, the transversal exponent has a clear dynamical meaning only under restrictive conditions-when cities behave as scaled versions of one another and path dependence is weak. Outside of these limits, it is not a law of urban growth, but a statistical artefact of heterogeneity.

We introduce a systematic method for expanding general spin-glass Hamiltonians in terms of Mattis interactions, providing a novel perspective for understanding the fundamental differences between short-range Edwards-Anderson (EA) and mean-field Sherrington-Kirkpatrick (SK) spin glasses. By iteratively extracting patterns from the coupling matrix, we expand the original spin-glass system into a Hopfield-like model (a series of Mattis interactions) plus a residual system. Our analysis reveals profound distinctions between EA and SK models: while EA models in two and three dimensions break into isolated subconnected sections after expansion, the SK model exhibits remarkable self-similar behavior, with the residual system preserving the mean-field structure and Gaussian statistics throughout the expansion process. This self-similarity manifests in exponential decay of residual matrix norms and expansion coefficients, reflecting the inherent mean-field nature of the SK model. Furthermore, we demonstrate that pattern expansion can identify ultra-low energy excitations in EA models, revealing excitations with energies that decrease rapidly with expansion step. Through connected component analysis, we quantify the size-energy relationship of these independent excitation clusters, opening new avenues for understanding the low-energy landscape of spin glasses and providing insights into the nature of metastable states.

Nanocomposites comprised of insulated magnetic single-domain particles are promising candidates for high-frequency, eddy current free, soft magnetic materials, but tend to suffer from low magnetic susceptibility ($<20$). Particle alignment has been proposed to increase nanocomposite susceptibility and reduce magnetic losses but experimental verification has been lacking. Here, magnetic nanocomposites containing 3-57 vol\% field-aligned 11$\pm$3 nm maghemite particles in a poly-vinyl matrix were investigated for potential use as high-frequency inductor core materials. The particles were aligned by a homogenous static alignment field during nanocomposite drying, fixating the particle orientation. Particle aggregation was disproved by small-angle scattering. The dependence of the alignment field strength and particle concentration on the nanocomposite's susceptibility and hysteresis losses were investigated from DC up to 922 kHz by vibrating sample magnetometry, AC-susceptibility and high-frequency hysteresis measurements. Nanocomposite susceptibility increased super-linearly with particle fraction due to weak particle interactions. Alignment of the particles increased the nanocomposite susceptibility from 21 to 50 for samples with a particle content of 57 vol\%. Hence, the synergy between particle alignment and interaction allows for a higher than expected susceptibility of nanocomposites. The results show that magnetically aligning particles in a nanocomposite reduces magnetic losses when using well-dispersed single-domain superparamagnetic nanoparticles. Measured nanocomposite susceptibility could be modelled by a combination of directional dependent Debye-models including mean-field interaction effects and partial particle alignment. Measured susceptibility of 50 is among the highest obtained for nanocomposites, making it a relevant candidate for applications in power electronics.

We address the connection between negative electronic friction and non-Markovian effects in the nonadiabatic vibrational dynamics of molecules interacting with metal surfaces under nonequilibrium conditions. We show that a generic nonequilibrium mechanism leading to negative Markovian electronic friction, where molecular vibrations couple directly to inelastic electronic transitions, also introduces significant non-Markovian contributions to the electronic friction. To demonstrate these ideas, we investigate nonequilibrium charge transport through a molecular nanojunction containing a vibrationally coupled donor-acceptor model, where negative electronic friction reflects driving of the vibrational mode beyond standard Joule heating. By comparison to numerically exact, fully quantum hierarchical equations of motion simulations, we verify that these non-Markovian effects have a significant impact on the nonequilibrium dynamics and even on the stability of the resulting Langevin equation.

We present a stochastic approach to calculate the full statistics of classical voltage fluctuations across an arbitrary, nonlinear, dissipative device embedded in a circuit in the presence of a bias. We show how the feedback resulting from the circuit, made of an ohmic resistor and a capacitor, affects the cumulants of the voltage, and in particular resolves Brillouin's paradox to satisfy thermodynamics. We apply our results to the case of a tunnel junction and a diode.

We report a scalable molecular beam epitaxy strategy to achieve a low density of O-band electrically tunable InAs/InGaAs quantum dots (QDs) on GaAs(001) substrates. Our approach is based on a gradient deposition of InAs in the sub-ML regime and subsequent capping with an InGaA strain-reducing layer to redshift the emission wavelength. For different growth conditions, we investigate the optical properties of the dots using photoluminescence mapping and correlate with structural properties determined by scanning transmission electron microscopy. Using a surface roughness modulation technique and synchronizing InAs sub-monolayer deposition cycles with substrate rotation, we control the dot density and position low-density regions (< 1 QD per um^2) on the substrate. Hyperspectral imaging is used to map the spatial and spectral characteristics of many individual dots in the low-density region, confirming that our approach is universally applicable to conventional MBE growth on (001) surfaces. Finally, we tune the QD emission wavelength within the O-band using electric fields and demonstrate single-photon emission with g(2)(0) = 0.020(14).

Disorder masks much of the rich physics in two-dimensional electronic systems, with charged impurities often the limiting factor. In graphene, progress in reducing disorder has largely stagnated since boron nitride encapsulation was introduced a decade ago. Here we show that a brief deep-UV exposure enhances the electronic quality of encapsulated graphene - typically by two orders of magnitude - by neutralizing charged impurities within boron nitride. Following illumination, standard graphene devices exhibit numerous evendenominator fractional quantum Hall states, including non-Abelian candidates, and frequently reveal hidden superlattice minibands. Even macroscopically inhomogeneous devices, seemingly unusable for transport studies, recover after deep-UV illumination and display Landau quantization in millitesla fields. This finding provides a straightforward route to exceptional-quality graphene, enabling further exploration of interaction-driven, topological and other quantum phenomena.

In many systems, communication proceeds by broadcasting rather than single source-target routing, but network structures that maximize signal lifetime are not well understood. Degree correlations are known to influence robustness and spreading, yet their effect on signal persistence has remained unclear. Here we introduce Copy-Spread-Annihilate dynamics, a minimal synchronous broadcasting model with annihilation. We show that signal lifetimes vary non-monotonically with assortativity and are maximized near neutral assortativity, where hub-driven amplification is strong but annihilation via short cycles is still limited. Applying this framework to the mouse connectome suggests assortativity as a structural control parameter for broadcast signal persistence in brain-like and other complex networks.

We present a symmetry-resolved classification of two-dimensional spin-flip small-angle neutron scattering (SANS) patterns arising from dilute ensembles of spherical nanoparticles hosting magnetic vortex states. Based on a linear vortex ansatz with an axially symmetric distribution of vortex axes and the corresponding analytical expression for the orientationally averaged spin-flip SANS cross section, we show that the angular scattering patterns organize into four distinct symmetry regimes: a four-fold anisotropy corresponding to coherent field-aligned magnetization, vertical and horizontal two-fold anisotropies associated with aligned and isotropically distributed vortex ensembles, and an isotropic ring-like condition separating the two two-fold regimes. The corresponding symmetry boundaries are obtained analytically and define a compact symmetry landscape in the parameter space of vortex amplitude and vortex-axis distribution width. Comparison with a nonlinear vortex profile shows that these symmetry regions are robust with respect to the detailed radial structure of the vortex core. The angular anisotropies are therefore governed primarily by rotational symmetry and by the statistical distribution of vortex axes, providing a compact and model-transparent classification framework of experimental polarized SANS data.

We analytically solve the finite-time control problem of driving an overdamped particle via an optical trap under costly measurement. By formulating this mesoscopic information engine within the Partially Observable Markov Decision Process (POMDP) framework, we demonstrate that the underlying Linear-Quadratic-Gaussian (LQG) dynamics decouple the optimal measurement and driving protocols. We derive the optimal feedback control law for the trap placement, which recovers the discontinuous Schmiedl-Seifert driving protocol in the open-loop limit and extends it to any measurement scheduling. For a costly, binary (on/off) sensor, we evaluate the optimal measurement protocol and derive physical bounds on the maximum gain that can be extracted from thermal fluctuations. We prove the emergence of deadline-induced blindness, a phenomenon where all measurements cease as the deadline approaches regardless of their cost. Taking the infinite-horizon limit, we find the exact periodic measurement schedules for the steady state as a function of the measurement cost $C$ and derive the macroscopic velocity envelopes beyond which viscous drag forces the engine into a net-dissipative regime. Finally, we generalize the results to a variable-precision sensor.

We investigate the flat-band properties and topological phase transitions in a non-Hermitian twisted bilayer $α-T_3$ lattice. Here, non-Hermiticity is introduced via Hatano-Nelson-type asymmetric hopping, while an aligned hexagonal boron nitride substrate provides a staggered sublattice mass to the system. We find that the introduction of non-reciprocal hopping splits the conventional single magic angle into three distinct non-Hermitian magic angles (NHMAs). Unlike the exceptional magic angles driven by spectral singularities, these NHMAs host perfectly isolated flat bands where the real and imaginary parts of the bandwidth simultaneously vanish. By mapping the complex eigenspectrum across the moiré Brillouin zone, we show that the scattered energy eigenvalues coalesce into well-defined, closed loop-like structures as the non-Hermitian parameter strength increases, indicating emergence of a nontrivial point-gap topology and hence the non-Hermitian skin effect. Furthermore, we characterize the topological phases by computing the direct band gap and the biorthogonal Chern number. While the system exhibits a transition to a higher topological phase at weak non-Hermiticity, we demonstrate that stronger non-Hermiticity drives the gap-closing boundaries to merge and their topological charges to mutually annihilate. This convergence results in a trivial gap closing and a complete suppression of the intermediate topological phase, confirming that non-Hermiticity fundamentally plays a crucial role with regard to destabilizing the robust topological features of this moiré system.

We report Barkhausen-like switching at millitesla fields in an $n$-doped InAs/Al nanowire Josephson junction, which serves as an interferometric probe of intrinsic magnetic reconfigurations, as evidenced by discrete switching-current jumps. At $T=30$~mK the device displays a Fraunhofer-like modulation with $I_{\mathrm{sw}}(0)\approx 0.24~μ\mathrm{A}$ and an abrupt transition at $|B|\approx 3~\mathrm{mT}$ between two branches differing by $ΔI_{\mathrm{sw}}\approx 0.13~μ\mathrm{A}$. By tracking the characteristic field scales from $30$ to $900$~mK, we find that the jump field is essentially temperature-independent, whereas the superconducting critical field decreases with temperature, as expected for thin Al films. The sharp discontinuity, sweep-direction asymmetry, and reproducibility across repeated scans point to avalanche-like switching between metastable magnetic configurations of the local magnetic texture, which are directly coupled to the weak link. Within an effective-field framework, each reconfiguration modifies a local field offset, thereby reshaping the interference response and leading to an abrupt reorganization of the switching-current pattern.

Bacterial chromosome replication occurs in the absence of a canonical spindle apparatus; yet it reliably produces organised and segregated genomes. While both passive and active mechanisms have been investigated, DNA replication itself is a non-equilibrium process that continuously generates new genetic material and reorganizes the nucleoid. Here, we investigate how replication-driven dynamics, combined with nucleoid-associated protein (NAP) interactions, shape spatiotemporal chromosome organisation using a three-dimensional polymer model that explicitly simulates DNA synthesis. We show that NAP-mediated interactions induce dynamic clustering of DNA, generating density fluctuations in the nucleoid. When coupled to replication, these clusters undergo cycles of stress buildup and release that produce stepwise expansion dynamics consistent with experimental observations. Chromosome segregation occurs naturally in this regime, but only within a finite range of interaction strengths: weak interactions fail to structure the nucleoid, whereas strong interactions hinder replication progression. Within this optimal balance, replication also promotes the spontaneous formation of replication factories. Our results demonstrate that bacterial chromosome organisation can be understood as a non-equilibrium system in which the interplay between replication forces and protein-mediated interactions generates nucleoid mechanics, dynamics, and segregation.

Coherent quantum phenomena can only emerge when decoherence is minimized, and mastery over decoherence is technologically crucial for designing and operating functional quantum devices. However, its microscopic mechanisms in spin-orbit-coupled ferromagnets remain elusive, and quantitative treatments have long been challenging. To solve this fundamentally significant and technologically crucial problem, we develop a quantum master-equation framework with a general ansatz for the off-diagonal density matrix that simultaneously captures electric-field-driven coherence and impurity-scattering-induced decoherence. This unified approach enables quantitative analysis of how decoherence reshapes the intrinsic anomalous Hall effect, revealing a clear crossover between intrinsic and extrinsic regimes. Remarkably, we identify a previously unrecognized extrinsic contribution: a second-order scattering process tightly relative to quantum decoherence-that is fundamentally distinct from both skew scattering and side jump mechanisms, yet substantially more significant than the skew scattering mechanism. Our work establishes decoherence as a key element in quantum transport and provides a systematic extension of the Boltzmann transport equation to incorporate decoherence, with broad implications for robust spintronic functionality.

Mismatched junctions formed by two C$_3$N zigzag nanoribbons of different widths provide a useful setting for studying quantum interference effects involving edge state transport. A crucial ingredient for this interference to appear is, besides the presence of edge states, the formation of localized interface states at the mismatched interface of the junction. At the level of a tight-binding model it is shown that, by means of an external gate potential, one of the edge state energy bands can selectively be shifted into the energy range of the localized interface states. The resulting coupling between the edge and localized interface states gives rise to pronounced Fano resonances in both the density of states and the transmission spectrum with line shapes well described by the canonical Fano formula. Furthermore, it is found that the geometrical mismatch of the junction not only determines the number of resonances but also the energetic orientation of their asymmetric line shapes. These results identify mismatched C$_3$N nanojunctions as a tunable and robust platform for engineering interference-driven transport.

Recent theoretical studies on the nonlinear response of spin and orbital degrees of freedom have discovered spin and orbital analogs of the photocurrent, with potential for characterizing topological materials and for applications. In this paper, we develop a general theory for calculating spin and orbital currents in semiconductors and study the properties of optical responses in the Bernevig-Hughes-Zhang and Luttinger models, where nonlinear orbital responses and a topological phase transition occur. We study the evolution of optical responses at the topological phase transition and how they manifest. In addition, we find that the relaxation time dependence of the orbital conductivity is somewhat distinct from that of the photocurrent. The theory is straightforwardly applicable to complex models of real materials, allowing quantitative predictions of the nonlinear responses of orbital and spin.

We introduce a novel coupling scheme for maximizing the synchronization of Kuramoto oscillator networks under a fixed coupling budget. We show that by scaling the interaction strength between oscillators according to their frequency detuning, synchronization is enhanced. The coupling scheme induces a change in criticality, driving the system from a continuous phase transition to an explosive transition by changing a single parameter. Our work offers a general route to efficient synchronization in engineered networks and provides insight into the critical behavior of the Kuramoto model.

While spatial phonon coherence manifested through band folding is believed to be a key factor governing the anomalous thermal conductivity of periodic structures, we investigate phonon transport from the perspective of temporal coherence. Using mode-resolved analyses, we quantify temporal coherent contributions and elucidate the interplay between phonon coherence time and lifetime in heat conduction of graphene/hexagonal boron nitride superlattices. We find that intrinsic coherence of folded phonon modes dominates the enhancement in ultrashort-period superlattices. In contrast, Wigner transport equation yields only a minor effect of band folding on thermal conductivity. The predictions in temperature dependence of models with and without temporal coherence provide a falsifiable experimental signature of this effect. Temporal coherence therefore constitutes a previously overlooked but fundamental channel for heat conduction, extending the conventional picture of spatially coherent transport and deepening the understanding of phonon dynamics in superlattices.

Using a field theory equivalent to a lattice version of the Poland-Scheraga (PS) model, the phase diagram for a long DNA molecule is derived in closed form. A one-loop renormalization group calculation for the generalized PS model with excluded volume interactions shows that there are two stable fixed points. At both fixed points, the excluded volume effect plays a role. At the fixed point reached when the original excluded volume effect is weak, the phase transition is continuous. At the other fixed point, the phase transition is first order.

In their seminal work, Fermi, Pasta, Ulam and Tsingou explored the connection between statistical mechanics and dynamical properties, such as chaos and ergodicity. Even today, seventy years later, the topic is not fully understood: while most results of statistical mechanics require the ergodic hypothesis to be rigorously proved, there are many indications that these predictions, both in and out of equilibrium, hold even in the absence of a rigorous form of ergodicity. Motivated by the above considerations, in this work we reconsider the point of view that the relevant ingredients for the validity of statistical mechanics are the large number of degrees of freedom and the choice of extensive observables, while the details of the dynamics do not play an essential role. This is the idea behind Khinchin's famous proof of the typicality of macroscopic observables at equilibrium. We extend this perspective to the context of non equilibrium, by investigating the thermalization properties of both harmonic (integrable) and nonharmonic (chaotic) oscillator chains initially prepared in out-of-equilibrium conditions. In integrable systems, thermalization occurs, or not, depending on the observable. In the chaotic regime, instead, thermalization is reached by any observable, although the relaxation timescale might be larger than the observation time.

We compare the spectrum and the localisation properties of the eigenmodes of the Laplacian and the adjacency matrix of 2D random geometric graphs, using numerical diagonalization of these matrices for different system sizes and connectivities. For sufficiently large ensembles of systems, we evaluate the spectrum, the probability distribution of the participation ratio and the relation between participation ratios and eigenvalues. While all eigenmodes of the adjacency matrix are localised for sufficiently large system sizes, the Laplacian matrix always leads to a small proportion of system-spanning modes due to a conservation law, and therefore to power-law tails in the probability distribution of the participation ratio and its relation to the eigenvalues. By disentangling the effects of finite system size, of mean degree, of component size distribution, and of network motifs, we provide a thorough understanding of the data.

Mid-infrared (MIR) emitters are critical components in advanced photonic systems, driving progress in fields such as chemical sensing, environmental monitoring, medical diagnostics, thermal imaging and free-space communications. Conventional MIR emitters based on III-V heterostructures rely on complex epitaxial growth on rigid lattice-matched substrates and suffer from limited integration compatibility with CMOS or flexible platforms. The recent development of novel MIR emitters based on two-dimensional (2D) materials such as black phosphorus (BP) is more suitable for on-chip applications but faces challenges related to stability and emission efficiency. Based on the recently discovered highly efficient photoluminescence of Te, we demonstrate a gate-tunable midinfrared light-emitting diode based on a van der Waals heterojunction formed by multilayer transition metal dichalcogenide (TMD) MoS2 and tellurium (Te). The device emits polarized electroluminescence (EL) centered at 3.5 $μ$m under forward bias at 25 K, and the EL persists up to 80 K with reduced intensity. Gate control of the MoS2 Fermi level modulates the band alignment and injection efficiency, enabling dynamic tuning of the EL intensity. The emission remains spectrally stable under varying bias and gating, indicating robust band-edge recombination. These results establish the Te/TMD heterostructure as a promising platform for integrated polarized mid-infrared optoelectronics.

Explicit expressions are proven for derivatives of the ratio of a determinant or Pfaffian determinant and a Vandermonde determinant. Such ratios appear for example in general group integrals of Harish-Chandra--Itzykson--Zuber type and in expectation values of products of characteristic polynomials in random matrix theory. In the latter case we start from known results for general non-Hermitian and Hermitian ensembles for expectation values without derivatives, at finite matrix size. They are given in terms of the determinant or Pfaffian of the corresponding kernel, for unitary or orthogonal and symplectic ensembles, respectively. Several equivalent expressions are proven for general ratios of determinants, starting from first order derivatives containing the Borel transform of the corresponding matrix or kernel. Higher order derivatives are expressed as sums over partitions containing determinants of derivatives of these, with coefficients given in terms of combinatorial expressions. Our most general result is valid for mixed higher order derivatives of ratios of determinants in several variables. This generalises previous findings, e.g. for mixed moments in specific ensembles of random matrices, relevant in applications to the Riemann $ζ$-function. Applications of our results to several examples are presented, including the complex Ginibre ensemble and the circular unitary ensemble.

In the first part of these short lecture notes, we will present an introduction on (auto-)correlation functions and linear-response functions in the language of a physicist. In particular, the fluctuation-dissipation theorem in classical physics is presented underlining the central role of correlation functions. The fundamental importance of (auto-)correlation functions raises the natural question on how they are characterized in general without referring to the concrete underlying dynamical laws. Perhaps unexpectedly -- despite being elegant and long established in the mathematical literature (Bochner's theorem for correlations; Herglotz-Nevanlinna representations for response) -- this answer is not widely appreciated in physics, partly because the requisite tools lie outside the standard curriculum. In the second part we adopt a more rigorous viewpoint: we state the key structural properties of correlation functions and provide selected derivations of these results. Finally, we return to linear response and discuss general characterization results for response functions.

The low temperature magnetoresistance of graphene functionalized by an array of magnetic Terbium Phthalocyanines molecules is found to exhibit a magnetic field-dependent 1/f noise, along with universal conductance fluctuations (UCFs) typical of a mesoscopic phase-coherent sample. A thorough analysis of the magnetic field, temperature and chemical potential dependence of this 1/f noise and UCFs reveals that long range, 2D Ising spin-glass like, magnetic correlations are induced in graphene through exchange interactions between the magnetic molecules and charge carriers in graphene. These experiments show that graphene functionalized with organic molecules constitutes a versatile platform for the investigation of magnetic phase transitions in two dimensions.

The recent proposal and observation of spin inertia, and the consequent high-frequency spin nutation mode, have raised key questions for our understanding of magnetization dynamics, especially considering its high relevance for magnetic memories and ultrafast switching. Notwithstanding recent progress, a clear identification of spin inertia's physical origin thereby offering predictive power remains to be accomplished. Here, discussing general principles for identifying this physical origin, we examine unquenched orbital angular momentum (OAM) finding it to be a key candidate, despite its typically small value. Treating OAM and spin within a two-sublattice model, we derive the equivalent single-sublattice framework for magnetization dynamics making appropriate approximations. The latter naturally manifests the spin inertia term and parameter, which are otherwise introduced phenomenologically. The inertia parameter evaluated within our model is found to be in good agreement with its experimentally observed value in cobalt. We further delineate key experimental signatures that could verify or rule out the unquenched OAM as the origin of the observed high-frequency mode, and avoid a spurious optical mode in a two-sublattice ferromagnet from being identified as nutation. Our analysis offers a potential link between the recently emerged fields of orbitronics and spin inertia, thereby motivating investigations at their intersection.

Recent experiments have revealed heterogeneous dissipation in optically trapped systems, often anticorrelated with local positional fluctuations, exposing a structural gap in the scalar stochastic thermodynamic description. While the conventional scalar framework successfully quantifies dissipation through currents and entropy production rates, it does not reveal the underlying vectorial force geometry that shapes spatial dissipation patterns. Here, we bridge this gap by identifying force geometry as an organizing principle for nonequilibrium thermodynamics and introducing force alignment as a geometric determinant of irreversibility. We show that entropy production depends not only on force magnitudes but also on the relative orientation between deterministic driving forces and entropic gradients, vanishing only under exact anti-alignment with matched magnitudes. We formalize this geometric alignment through a time-dependent force-correlation coefficient, quantifying the relative orientation between the forces. This yields an instantaneous geometric lower bound on entropy production that remains valid even when force magnitudes are matched. For overdamped dynamics, perfect anti-alignment defines a thermodynamic stall where net transport vanishes and the lower bound on entropy production is saturated. This force-level perspective provides a structural explanation for the experimentally observed fluctuation-dissipation anticorrelation and nonuniform dissipation. We construct geometric control charts for both constant dragging and sinusoidal driving protocols, explicitly locating experimental operating points within this force-space representation. Together, these results position force geometry as a unifying structural perspective on irreversibility, spanning active biological systems, microrheology, and naturally extending to underdamped dynamics.

The self-assembling behaviour of ionic liquids in aqueous solution is important for understanding their physicochemical properties and for their industrial applications. While the influence of ionic liquids on surfactant micellization has been widely studied, much less attention has been given to how surfactants affect the aggregation of ionic liquids, particularly when the surfactant concentration is below its critical micelle concentration (CMC). In this work, we examine the effect of the anionic surfactant sodium dodecyl sulfate (SDS), introduced at sub-CMC concentration, on the micellization of 1-methyl-3-octylimidazolium chloride in aqueous solution maintained above the IL CMC, using surface tension measurements, theoretical analysis, and coarse-grained molecular dynamics simulations. We find that at low SDS concentrations (approx 2 mM), SDS inserts smoothly into the pre-existing IL micelles, producing stable mixed micelles with favourable IL-SDS interactions. When the SDS concentration approaches (approx 4 mM), the micelles exhibit distinct changes in their internal dynamics, reflected in deviations in the thermodynamic parameters. Beyond this point, as more SDS is added, the system reorganizes and forms stable mixed micelles again, now containing a higher fraction of SDS but still enriched in IL. The synergistic behaviour is quantified using Clint and Rubingh's models, and simulations supports the structural transitions, showing variations in micelle size, aggregation number, and radial distribution functions. This work demonstrates that SDS acts as an effective modulator of IL aggregation, providing mechanistic insight into IL-surfactant co-assembly.

Artificial spin ices have transcended their origins in frustrated rare-earth pyrochlores to become a versatile platform for engineering exotic states of matter. Across diverse implementations, from nanomagnets and superconducting vortices to colloids, quantum annealers, liquid crystals, and metamaterials, they are unified by the ice rule, which often leads to degeneracy and constrained disorder by enforcing minimization of the local topological charge. Here, we report the first realization of an "anti-spin ice" in which not only the ice rule does not hold, but its opposite is true as the system seeks to maximize, rather than minimize, spin ice charges. Using fast-rotating, in-plane magnetic fields to generate isotropic attraction between colloidal particles, we invert the conventional paradigm of repulsive interactions in colloidal spin ices. Combining experiments and simulations across standard square and honeycomb lattices as well as novel pentaheptite geometries, we establish rules for order and disorder in the anti-spin ice. With the pentaheptite lattice, we demonstrate that the anti-spin ice system can also exhibit frustration, but of a new kind. This topological charge frustration arises from the lattice connectivity, where networks of unequal, odd-sided polygons suppress charge crystallization at high interaction strength.

We investigate heat conduction and energy relaxation in an InAs semiconductor nanowire using a hybrid semiconductor-superconductor architecture. Local electronic temperatures are measured with an in-situ grown quantum dot thermometer, while controlled Joule heating is applied at different locations along the wire to probe temperature gradients at sub-kelvin temperatures. With a onedimensional heat transport model, we calculate an electron-phonon heat flow that scales as Q_{e-ph} \propto T^2.6, which is in close agreement with the T^3 dependence predicted for a clean one-dimensional electron gas coupled to a phonon bath. We further estimate a characteristic length l_{eq} = 370 nm, beyond this length scale, phonon-mediated heat transport dominates over heat conduction in our nanowire. Our results provide a quantitative measure of energy relaxation mechanisms in a onedimensional semiconductor and provide a framework for studying heat flow in low-dimensional nanostructures.

Recent work investigated graphene's hydrogenation with independent control of the electric field, E, and charge density, n, in the crystal and showed that the process is controlled by n. Here, we demonstrate layer-selective conductor-insulator transitions in twisted bilayer graphene, driven by hydrogenation at fixed n under strong E. This process is accompanied by proton transport through the bilayer, enabling several parallel and configurable logic gates in the devices. Selectivity arises because the large twist angle decouples the two layers' electronic systems, enabling independent control of their charge densities. Polarisation by the field then induces a charge imbalance at fixed total n, triggering hydrogenation when one of the layers' charge densities reaches the threshold for monolayer hydrogenation. Our results introduce a new type of electrode-electrolyte interface in which electrochemical processes are controlled with two decoupled 2D electron gases, opening new design opportunities for energy and information processing devices.

Multipoles provide a systematic framework for describing the electronic structures of quantum materials from a symmetry perspective. Thermodynamic multipole moments in crystalline solids exhibit direct microscopic connections to certain allowed physical responses beyond symmetry; however, such relations have thus far been limited to dissipationless responses in equilibrium insulating systems. Here, this framework is extended at a heuristic level by focusing on the Fermi-surface contributions to thermodynamic multipole moments. These contributions establish direct relations to dissipative transport responses characteristic of metals, including charge and spin conductivities. A key consequence is that the conductivities exhibit extrema, typically maxima, at chemical potentials where the corresponding Fermi-surface contributions to the multipoles vanish, specifically, the electric quadrupole for charge conductivity and the magnetic octupole for spin conductivity. These findings uncover a previously overlooked aspect of thermodynamic multipole moments, opening a new perspective on dissipative transport in metallic systems.

First passage phenomena arise across physics, biology, and finance when stochastic processes first reach a threshold, triggering downstream events. Examples include the irreversible exit from a domain, a biochemical reaction, a financial selloff. While typical formulations involve diffusive motion, many stochastic processes are better described as velocity jump processes, characterized by persistent motion interrupted by stochastic velocity changes. Despite their ubiquity, first-passage properties of velocity jump processes remain underdeveloped in higher dimensions, especially under directional bias. We present a general framework to estimate the mean first passage time (MFPT) and higher moments of the survival probability for fixed-speed velocity jump processes where possible reorientations range from strong alignment to full angular anisotropy. For low Knudsen numbers, when the mean free path is small compared to the distance to the target, we derive a universal form for the MFPT in which two bias functions encode broad classes of angular distributions, including von Mises-Fisher, wrapped Cauchy, and elliptical families. In the narrow capture limit of a vanishingly small target, directional persistence induces anomalous scaling, including regimes where the MFPT remains finite whereas standard diffusion would predict divergence. Finally, we obtain a Langevin representation that accurately reproduces first-passage statistics. Analytical predictions are confirmed by numerical simulations.

Quantum Hall edge excitations, whose low-energy behavior admits a chiral conformal-field-theory description, are a promising platform for engineered dynamical experiments, including analog-spacetime proposals. However, establishing their edge dynamics in realistic electrostatic landscapes is essential for controlled dynamical experiments and has remained experimentally challenging. Here we report spatiotemporal imaging of gate-controlled multipath dynamics of edge excitations in a $ν= 1/3$ fractional quantum Hall device using stroboscopic time-resolved photoluminescence microscopy and spectroscopy with $\sim$100-ps resolution. By tuning a control-gate-defined potential landscape, we observe switching between mesa-defined and gate-defined trajectories and identify an intermediate regime in which a single launched excitation accesses multiple pathways. Time-resolved measurements at downstream locations reveal gate-dependent arrival times and pronounced temporal broadening, showing that the propagation dynamics are strongly modified by the local confinement and become increasingly dispersive in a multipath landscape. We further observe a long-range transverse optical response extending tens of micrometers into the bulk and persisting over distances exceeding 200 $μ$m downstream, consistent with the near-field component of an edge magnetoplasmon. These results establish direct experimental access to controllable multipath edge dynamics in the fractional quantum Hall regime and suggest a platform for engineered nonequilibrium and interference-based experiments, as well as future analog-spacetime studies in quantum Hall edge systems.

The quasi-one-dimensional material Bi4I4 hosts two crystallographically similar polymorphs that realize distinct topological insulating phases separated by a first-order structural transition near room temperature. This transition occurs without a change in space group, arising instead from a subtle rearrangement of chain stacking registry. Polarization-resolved Raman spectroscopy directly resolves this structural-topological phase transition through abrupt, hysteretic modifications of the phonon spectrum. Angle-dependent measurements establish the symmetry of the dominant Raman-active modes and require a complex Raman tensor formalism to account for absorption-induced phase effects. Across the transition, selected phonon modes exhibit discontinuous, reversible shifts in frequency, linewidth, and relative intensity despite the absence of a space-group change. Density functional theory calculations reproduce the direction of the observed phonon renormalizations and confirm their sensitivity to stacking-dependent force constants. These results demonstrate that polarization-resolved Raman spectroscopy can detect subtle stacking-driven structural rearrangements that underlie topological band character, even when global crystallographic symmetry remains unchanged. The obtained results provide valuable insights into the interplay among lattice dynamics, structural distortions, and topological properties in this class of low-dimensional materials, with strong potential for unique functionalities.

We study localization in a quasiperiodic spinful antiferromagnetic Hubbard ring within a self-consistent Hartree-Fock framework, emphasizing the interplay of quasiperiodicity, staggered Zeeman-field-induced antiferromagnetic order, and electron correlations. Localization properties are characterized through inverse participation ratios, normalized participation ratios, and multifractality, and are consistently supported by a broad class of real-space mean-field observables, including double occupancy, density fluctuations, local entropy, spin-density-wave (SDW) order, and other related correlation measures. We uncover a pronounced nonmonotonic evolution of localization with interaction strength, featuring an intermediate regime marked by enhanced localization, strong spatial inhomogeneity, and magnetic ordering, followed by a re-entrant tendency toward delocalization at stronger interaction regime. Phase diagrams constructed from complementary localization and mean-field indicators reveal extended, localized, and critical regimes governed by the interplay of quasiperiodicity and interactions. Furthermore, real-time wave-packet dynamics of eigenstates provide direct evidence of ballistic spreading, confinement, and re-entrant transport, in agreement with the underlying spectral characteristics. These results establish a unified framework where diverse mean-field observables and dynamical probes consistently capture correlation-driven localization phenomena in quasiperiodic systems.

We investigate localization in a quasiperiodically engineered diamond lattice with strand-dependent Aubry-André-Harper onsite modulations, highlighting the decisive roles of the modulation ratio $s$ and the averaged potential on the middle strand. The upper strand hosts the primary potential $λ$, the lower strand carries a weaker modulation $λ/s$, and the middle strand follows their average, generating a correlated quasiperiodic landscape across each plaquette. By tuning $λ$ for selected values of $s$, we probe spectral and eigenstate properties via the inverse participation ratio (IPR), normalized participation ratio (NPR), and fractal dimension $D_2$. We uncover a pronounced re-entrant localization behavior, where eigenstates repeatedly switch between extended and localized regimes, which persists only within a finite range of $s$ and crucially relies on the averaged potential construction. This unconventional sequence arises from the interplay of $s$, the correlated potential, and the intrinsic diamond geometry, producing a highly nontrivial interference landscape. Our results reveal localization physics beyond the standard Aubry-André paradigm, further supported by the evolution of extended states, system-size scaling of $\langle \mathrm{NPR} \rangle$ and $\langle D_2 \rangle$, and dynamical signatures from the time-dependent root-mean-square displacement, confirming the robustness of the re-entrant transitions.

In this work, we examine microwave responses of the Pauli spin blockade (PSB) leakage current through a p-type silicon double quantum dot. We observe more than the expected two resonance lines with the main resonance line exhibits both positive and negative peaks as a function of the magnetic field, corresponding to enhancement and suppression of the PSB leakage current, respectively. We attribute the observed spectra to the interplay between two spin rotation mechanisms: spin-orbit-mediated electric dipole spin resonance (EDSR) and multilevel Landau-Zener (MLLZ) interference, both of which are present in electrically driven devices with strong spin-orbit coupling (and enhanced in the vicinity of orbital level crossings). A numerical simulation taking into account both mechanisms show agreement with the experimental results. While these unconventional spectral behaviours can be readily suppressed away from the orbital level crossing or in devices with weak spin-orbit coupling, our study showcases the potential complexity of spin-rotating mechanisms for electrically driven spin qubits.

The subject of the present paper is a detailed theoretical investigation of the energy spectrum and bandgaps, as well as collective properties and linear response, in $d$-wave altermagnets in the presence of an off-resonance optical dressing field. We consider the altermagnets with both $d_{x^2-y^2}$ and $d_{xy}$ pairing symmetries and focus on anisotropic dressing fields applied to an anisotropic and non-linear electron Hamiltonian. We have uncovered several crucial properties of the resulting electron-dressed state; specifically, we found that a finite bandgap is opened by linearly polarized irradiation, a phenomenon not observed in Dirac materials. A number of crucial properties of the electron dressed states in the presence of the linearly polarized light can be uncovered only in the second-order perturbation expansion, which is often omitted. We demonstrate that introducing an anisotropic driving field leads to several subtle yet important changes in the Edelstein susceptibilities of altermagents, enabling the fine-tuning of their spin polarizations. All these results must be in high demand due to the rapidly developing fields of spintronics and device physics.

We introduce a universal criterion for criticality in mean-field rotor Hamiltonians based on the geometric structure of the constant-energy shell. Rather than characterizing the onset of a phase transition through the conventional thermodynamic singularities alone, we show that the relevant information is already encoded in the way the geometry of the shell reorganizes along distinguished collective directions. For a broad class of finite-dimensional trigonometric mean-field interactions, the trace of the Weingarten operator (representing the principal curvatures) admits a universal collective expansion in terms of the order-parameter amplitudes. This expansion defines an energy-dependent quadratic form whose eigenmodes identify the geometrically unstable channels of the system. Criticality is then associated with the vanishing of the corresponding curvature coefficients, yielding a direct geometric selection principle for the modes that become unstable at the transition. In this way, the phase transition in mean-field systems (usually of first- or second-order) is reformulated as a geometric instability phenomenon intrinsic to the microcanonical energy shell. The resulting framework is geometrically universal within the class considered, independent of model-specific details except for a finite set of collective couplings. Moreover, our approach recovers the known critical channels in standard mean-field rotor models while extending naturally to multimode and spectrally coupled cases. These results support a view in which critical behavior can be understood as reorganizations of energy-shell geometry triggered by a collective restructuring of the underlying energy-shell geometry.

We investigate the behavior of the length of the longest weakly increasing subsequences (weak LIS) of $n$-step random walks with nonzero integer increments $k = \pm 1, \pm 2, \dots$ given by a zero-mean, symmetric heavy tailed mass distribution proportional to $|k|^{-1-α}$ for several values of the real parameter $α> 0$ together with that of the simple random walk ($k=\pm 1$), to which the $n$-step heavy tailed random walks tends when $α> (1+o(1))\log_{2}{n}$. By means of exploratory fits, weighted nonlinear least squares, and ANOVA model comparison, we found that the sample average length $\langle{L_{n}}\rangle$ scales like $\langle{L_{n}}\rangle \sim \sqrt{n}\log{n}$ when the distribution of increments has finite variance ($α> 2$) and $\langle{L_{n}}\rangle \sim n^θ$ with a varying exponent $θ> 0.5$ when the variance is infinite ($α\leq 2$). Distributional diagnostics indicate that the bulk of the $L_{n}$ distribution is very well-approximated by a lognormal model, though systematic deviations are observed in the tails. Our results corroborate and expand upon previous results for the LIS of other types of heavy-tailed random walks and raise a conjecture as to whether the distribution of $L_{n}$ is given, or can be effectively described, by a lognormal distribution.

We identify a new class of non-Hermitian causal processes that produce statistically significant temporal correlations invisible to conventional spectral methods. Using a generative model with a strictly causal memory kernel, we demonstrate that time-asymmetric stochastic processes naturally yield sharp transitions at characteristic scales that appear as localized structures in similarity space but leave no trace in power spectra. The model predicts an asymmetric transition profile with orientation-dependent asymmetry parameter $A(θ)=A_0\cos(θ+δ)$ and achieves quantitative agreement ($χ^2/\mathrm{dof}=0.50$, $p=0.86$) with high-precision counting experiments exhibiting $p<10^{-15}$ significance. These results establish a fundamental limitation of spectral analysis for non-Hermitian, non-stationary processes and provide experimentally testable signatures of causal memory in open quantum systems.

A binary fluid mixture in contact with lateral particle reservoirs is considered. By imposing different particle concentrations in these reservoirs, the system can be maintained under controlled non-equilibrium conditions. Previous stochastic approaches have revealed an unexpected property of the system's state trajectory, namely that it remains time-reversible even when the system is driven out of equilibrium. In the absence of relevant experimental evidence, we employ microscopic molecular dynamics simulations to assess the validity of this surprising result. Remarkably, the simulation results unambiguously confirm the prediction of the stochastic analysis.

We present the observation of electrically tunable quantum coupling of orbital states in individual InAs/InGaAs quantum dot molecules emitting in the telecom O-band (~1300 nm). By tuning the static electric field along the growth axis of the QD-molecule, we observe pronounced anticrossings between excitonic transitions and determine the dependence of the interdot electron tunnel coupling on the interdot separation. As the electric field applied along the growth axis of the QD-molecules increases, positively charged exciton complexes sequentially emerge in the time-integrated emission spectra due to electron escape from the system while holes remain trapped. Moreover, for strong pumping, biexciton emission from the O-band molecules is identified. We demonstrate single-photon emission from the InAs/InGaAs QD-molecule emitting around 1300 nm with a g(2)(0) = 0.017(2) and explore the impact of tuning orbital coupling on the second-order correlation function.

Scattering-type scanning near-field optical microscopy (s-SNOM) is a powerful tool for investigating polaritonic modes. However, an inherent limitation of this technique is that excitation and detection occur at the same location. This constraint makes it challenging to resolve excitations with complex spatial structures, which can exhibit delicate dependence on the in-coupling conditions. Here, we present a strategy to overcome this limitation by introducing an auxiliary cavity, which serves as a stationary near-field excitation source. This configuration allows the s-SNOM tip to act solely as a detector, and decouples excitation from detection. We apply this approach to whispering gallery modes (WGMs) of hyperbolic phonon-polaritons in hexagonal boron nitride resonators. Through spatially resolved near-field maps we directly observe subwavelength polaritonic WGMs with large and discrete azimuthal momentum ($k_φ/ k_0$ up to 15). This allows us to map the frequency and angular behavior of the modes. Notably, we observe dynamic tuning of the effective refractive index by the WGMs to preserve consistent azimuthal momentum \(k_φ\) under varying excitation conditions. Numerical simulations support the experimental observations and confirm the observation of hyperbolic WGMs. This approach enables direct visualization of previously hidden mode structures in hyperbolic cavities and opens new pathways for momentum-controlled polaritonic device engineering.

Nanoscale thermal transport is governed by the phonon Boltzmann transport equation (BTE). However, simulating the sub-continuum dynamics remains computationally prohibitive due to the high dimensionality of the phase space and the intrinsic nonlinearity of the scattering collision operator. Traditional numerical solvers and standard physics-informed neural networks (PINNs) inherently struggle with these integro-differential equations due to deterministic quadrature limitations, artificial thermalization introduced by the relaxation time approximation (RTA), and multiscale spectral bias. This work introduces a multiscale auxiliary physics-informed neural network (MTNet) to solve the generalized equation of phonon radiative transfer (GEPRT). By leveraging an auxiliary formulation, this mesh-free framework recasts the GEPRT into a fully differential system, enabling the analytical evaluation of scattering operators via automatic differentiation and facilitating scalable multi-GPU parallelization. To circumvent optimization stiffness, the architecture employs a decoupled, shallow neural network explicitly constrained by radiative equilibrium. MTNet is validated by simulating steady-state cross-plane transport in a silicon thin film, successfully capturing ballistic-diffusive regimes and characteristic boundary slips across extreme temperature gradients ($ΔT = 100$ K) beyond the standard linearization approach. Furthermore, we show that our framework successfully solves a geometric inverse problem in a slab geometry, retrieving the unknown slab thickness based only on interface temperature constraints in the mesoscopic regime. Ultimately, MTNet establishes a robust, fully differentiable foundation for predicting high-fidelity kinetic transport and extracting material properties in next-generation nanostructures.

Filled rubber-like materials are widely used in engineering applications and are well known to exhibit the Mullins effect. In this work, an established local constitutive model from the literature is extended to a non-local setting to resolve the mesh dependence inherent to the local approach. Non-local effects are incorporated using two separate approaches: (i) a Helmholtz-type equation governing a non-local soft volume fraction, and (ii) a Laplacian term introduced directly into the soft volume fraction local evolution law. In both formulations, an additional governing partial differential equation arises and is solved numerically in Abaqus using an analogy with the heat equation. The two approaches yield different results, leaving the choice between them to be guided by experimental findings. The details of the implementation, along with the code developed in this work are also provided.

Separating electrons into emergent fractional quasiparticles is a hallmark of exotic quantum phases of matter with strong interactions. Understanding under which circumstances fractionalized excitations appear is a major conceptual challenge and can help realize long sought-after states, such as quantum spin liquids. Here, we identify a distinct mechanism for fractionalization. Starting from the plaquette-ordered ground state of an SU(4) symmetric t-J model at quarter filling on frustrated triangular lattices, we reveal a compelling interplay between order and fractionalization as a function of doping. For hole doping, we find that the kinetic frustration can be relieved by fractionalizing the holes into fermionic spinons and bosonic holons: the holons minimize their kinetic energy when the spinons form a spinon Fermi surface. We support this mechanism analytically in the large-N limit as well as numerically by simulating the SU(4) case with matrix product states on cylinder geometries and with variational Monte Carlo methods on system sizes up to 40x40. Conversely, electron doping drives the system into a ferromagnetic phase, akin to Nagaoka's theorem. We discuss possible experimental realizations in moiré heterostructures as well as ultracold atoms, and propose dynamical probes to search for key characteristics of the fractionalized quasiparticles.

Multicomponent gel systems have garnered much interest due to their compelling mechanical properties in the past decade. Yet, some mechanisms associated with multicomponent gels, such as sequential gelation, have been explored primarily in the context of chemical nonreversible polymeric and protein gels than in physical reversible colloidal ones. In this study, we use mesoscale simulation techniques to model the sequential gelation of two-component colloidal systems whose components' interspecies and intraspecies electrostatic interactions can be modified independently. We show that by simply leveraging temporal control and interspecies interactions, we can construct markedly different networks; from double networks to mixed and core-shell composite structures of varying coarseness and heterogeneity natures. These findings present a compelling case for further exploration of multicomponent colloidal systems.

Floquet engineering provides a powerful pathway for creating non-equilibrium phases of matter with tailored electronic structures and properties through time-periodic driving. As the original theoretical prototype, graphene established the framework in which the Floquet topological insulator with light-induced anomalous Hall effect was proposed. However, the defining spectroscopic signature of Floquet engineering in graphene--light-induced hybridization (avoided-crossing) gap at Floquet band crossings, has remained experimentally elusive. Here, we report direct observation of Floquet-induced hybridization gap in monolayer graphene under resonant driving by a strong light field. Time- and angle-resolved photoemission spectroscopy reveals gap opening at Floquet band crossings, accompanied by coherent Floquet sidebands. The gap exhibits pronounced momentum anisotropy, featuring two Dirac nodes protected by the spatiotemporal symmetry and tunable by light polarization. These results provide long-sought experimental demonstration of Floquet band engineering in graphene, opening up opportunities for light-field engineered quantum phases in graphene and related materials.

Floquet engineering provides an emerging pathway for tailoring the electronic states of quantum materials through time-periodic drive. A critical step along this direction is achieving light-induced modifications of the dynamical electronic structure, such as avoided-crossing gap at the Floquet Brillouin zone boundary, via efficient coupling of electrons with the coherent light-field. Here, we report robust Floquet-induced gap in bulk graphite that persists despite the presence of interlayer coupling and photo-excitation. Using time- and angle-resolved photoemission spectroscopy with intense mid-infrared pumping, we directly reveal Floquet-induced gaps at resonance points both in the valence and conduction bands, accompanied by coherent Floquet sidebands. The gap and sidebands coexist with photo-excited carriers, yet their distinct timescales allow us to disentangle their origins. Our demonstration of robust Floquet-induced gaps establishes graphite as a platform for coherent manipulation of Dirac fermions and realization of light-engineered quantum phases.

We study the number entropy and quasiparticle width in one-dimensional quasiperiodic many-body localized (MBL) systems and observe slow dynamics that have previously been investigated in detail only in random systems. In contrast, quasiperiodic systems exhibit more structured growth of both observables. We identify the modulation of the Rabi oscillation amplitude of single-particle hoppings as the mechanism underlying the slow growth even deep in the MBL regime. This quantum amplitude modulation and associated beats arise from the interaction between single-particle hopping processes at different positions in the chain. Interestingly, this mechanism is not weakened by increasing the distance between particles and is generic to many-body quantum systems. We develop an analytical model based on the aforementioned mechanism that explains the observed dynamics at all accessible timescales and provides a microscopic picture of the slow dynamics in the MBL regime. Our results are consistent with the stability of the MBL phase in the thermodynamic limit.

We conducted replica exchange Monte Carlo simulations to investigate the phase diagram of identical hard rhombi systems in two dimensions. The rhombi shape varies from nearly square-like, as their minor angle a approaches 90 degrees, to needle-like, as it approaches 0 degrees. For angles near 90 degrees, we observe an isotropic fluid, a rhombatic fluid, a rotator phase, and a columnar space-filling structure with increasing density. Conversely, as a approaches 0 degrees, the results resemble the needle limit. Even for angles as small as a = 20 degrees, we still obtain isotropic, nematic, and rhombatic fluids before reaching a rhombic solid, but the nematic phase gains importance with decreasing a. At a approximately 60 degrees, aperiodic space-filling structures with long-range six-fold orientational symmetry dominate over periodic candidates such as the rhombic and rhombille. This aperiodic solid undergoes a melting process leading to a phase with quasi-long-range six-fold orientational symmetry, a hexatic fluid, before reaching the isotropic phase.

Microscopic insights into nonlinear interactions are essential for advancing polaritonic devices. Existing studies often rely on phenomenological models that overlook important many-body processes. Based on a material-specific and predictive approach, we investigate monolayer and homobilayer MoS$_2$ embedded in a Fabry-Pérot cavity to characterize the exchange, saturation, and dipole-dipole contributions to polariton-polariton interactions in these technologically promising materials. A key finding is that the exchange interaction induces asymmetric energy shifts of the lower and upper polariton branches in a detuned cavity, a behavior driven by the difference in their excitonic character. Furthermore, we demonstrate that temperature and electron-photon coupling determine the energy renormalization through the equilibrium polariton distribution. In homobilayers, the dipole-dipole interaction is mediated by the interlayer character, enabling electrical control and facilitating the electric-field-induced closing of anti-crossings due to dipolar-interaction shifts. The gained insights on polariton-polariton interactions are important for the development of ultra-compact polaritonic circuitry.

Disordered fiber networks exhibit a floppy to rigid mechanical phase transition as a function of connectivity. Sub-isostatically connected networks can undergo this transition via straining. Critical exponents governing this transition have been estimated theoretically and by numerical simulations of various types of networks. In this study, we present improved results, achieved through a combination of refined numerical simulations, larger system sizes and incorporation of theoretical predictions for better post-simulation analysis. We also report the evolution of the critical strain and critical exponents as the network is sheared while being subjected to non-volume-preserving uniaxial deformations.

The explicit expression of ergotropy (a.k.a. available energy) of a classical system is known for the case when the system phase space density is continuous and with no plateaus. Here we provide the general expression of ergotropy that applies without those limitations. It easily follows upon casting the ergotropy problem as a function rearrangement problem. This leads to the notion of "ergotropic rearangement" which generalises that of "symmetric decreasing rearrangement" (an advanced topic of measure theory). We apply it to investigate the fate of classical ergotropy in the thermodynamic limit, and find that any density of the form $ρ=f(H_0)$ is asymptotically passive, where $H_0$ is the system Hamiltonian and $f$ a generic function.

Artificial spin ice provides a versatile setting for emergent gauge fields and magnetic monopole excitations. Here we propose a driven-dissipative polariton realization of artificial spin ice, in which the circular polarization of each link mode plays the role of an Ising degree of freedom, while an auxiliary lossy vertex mode dynamically enforces a local ice-rule constraint. Adiabatic elimination of the vertex mode yields an effective spin-ice penalty, favoring the two-in two-out manifold in the steady state. We show that local polarization flips generate monopole-antimonopole defects, and that sequential flips transport these defects across the lattice while defining a Dirac string. In an extended spin-ice geometry, the vertex charges and their dynamics can be directly reconstructed from polarization-resolved real-space imaging. Our results establish polariton lattices as a controllable photonic platform for creating, manipulating, and observing emergent gauge charges in nonequilibrium spin-ice systems.

Chemical waves of CO oxidation on platinum surfaces exhibit complex spatio-temporal self-oscillations, yet the local electronic mechanisms driving their propagation remain poorly understood under operando conditions. In this work, we combine operando scanning electron microscopy with frequency-modulated Kelvin probe force microscopy (FM-KPFM) to simultaneously map secondary electron contrast and local work-function variations during CO oxidation on Pt. By utilizing the KPFM tip as a localized sensor, we provide the first work-function-resolved imaging of reaction fronts, enabling an unambiguous physical assignment of CO- and oxygen-covered states. Our results demonstrate that the spillover process of chemical wave-the transition and expansion of adsorbate phases-is characterized by a pronounced temporal asymmetry and spatial heterogeneity transition thresholds. KPFM identifies a rapid onset of oxygen coverage followed by a gradual, diffuse relaxation back to the CO-covered state, indicative of relaxation-type oscillations even at low pressures (10^-2 Pa). Correlative reaction-diffusion simulations reproduce this wave morphology, confirming that the high-resolution work-function signal provides unique insights into the internal structure and kinetic heterogeneity of the working catalyst surface.

In one-dimensional (1D) electron systems, the Fermi liquid state breaks down due either to electron interactions, which results in a Tomonaga-Luttinger liquid (TLL) state, or to Peierls instability, which leads to an insulating charge-density-wave (CDW) phase. In general, these two phenomena are mutually exclusive, and their coexistence remains elusive in real materials. Here, we report the discovery of a new quasi-1D material, Cs$_{1-δ}$Cr$_3$S$_3$, which unexpectedly exhibits coexistence of the antithetical CDW and TLL states. The CDW state is evidenced by the intra-unit-cell dimerization, and the opening of an optical band gap of $\sim$250 meV. Meanwhile, TLL behaviour is unambiguously demonstrated by the measurements of electrical transport and angle-resolved photoemission spectroscopy, which reveal a power-law scaling with temperature, bias voltage and electron energy. Band structure calculations reveal isolated, linearly dispersive, 1D bands around the Fermi level. For the dimerized CDW phase, the 1D Fermi-surface sheets located at the boundary of the Brillouin zone are gapped from intra-unit-cell bond symmetry breaking. Experimentally, subtle Cs vacancies shift the Fermi level into the linearly dispersive valence band, enabling the observation of TLL behaviour without interrupting the CDW order. This work establishes Cs$_{1-δ}$Cr$_3$S$_3$ as a rare material platform in which the antagonistic Fermi-liquid instabilities coexist and intertwine, opening new avenues for studying emergent quantum phenomena in 1D systems.

Under a thermodynamic gradient, for example, the concentration or temperature gradients, the colloidal particles immersed in the solvent can exhibit a directional migration along or against the gradient -- phoresis, a cross transport effect. When the solvent is an odd fluid, where the time-reversal and parity symmetries are broken microscopically, the odd transport phenomenon is allowed. This means an odd phoresis may appear: the colloidal particle migrates perpendicularly to the thermodynamic gradient. Here, we realize the odd diffusiophoresis and odd thermophoresis for a colloidal particle immersed in a two-dimensional odd fluid by performing mesoscale fluid simulations. We further provide the flow field driven by the diffusiophoretic force, which is quantitatively consistent with the numerical solutions of the corresponding odd fluid dynamics equations.

The wave-particle duality of massive macromolecules -- such as the fullerene C$_{60}$ -- is a well-established quantum phenomenon. However, whether the quantum behavior of large organic molecules actively dictates the macroscopic structure and function of synthetic materials remains unknown. In organic semiconductors, crystal polymorphism fundamentally determines optoelectronic performance, yet classical thermodynamic models consistently fail to resolve the microscopic origins of phase selection. This includes the long-standing anomaly of divergent polymorph formation under identical thermodynamic parameters across different reactor scales. Here we show that the polymorphism of copper phthalocyanine (CuPc, 576 Da) -- a planar macromolecule comparable in mass to C$_{60}$ -- is governed by quantum coherence during atmospheric-pressure organic vapor phase deposition (OVPD). We establish the Dissipative structure field-Induced Multipartite Entanglement (DIME) framework, which integrates ambient blackbody radiation, molecular de Broglie wavelengths, and frontier orbital directionality to model field-driven quantum entanglement. We demonstrate that exceptionally weak environmental decoherence at room temperature preserves the coherence of molecular matter waves, enabling the self-assembly of ultralong ($>1$ cm) single-crystalline $η$-CuPc nanowires. By leveraging the DIME framework to manipulate environmental decoherence, we rationally designed an OVPD reactor to synthesize a previously undiscovered polymorph, designated $ω$-CuPc. Our findings reveal that multipartite quantum entanglement acts as the decisive regulator of organic crystal assembly, opening a deterministic, quantum-level pathway for engineering organic semiconductor polymorphism.

The dependence of surface-enhanced Raman scattering (SERS) spectra on the precise axial position of the laser focus relative to a solid nanostructured substrate has received little to no attention in the literature. Here we show this dependence is both real and physically meaningful. Through vertical (Z-axis) scans varying the distance between the laser focus and a planar SERS substrate, we find that the SERS signal intensity follows a Lorentzian axial profile that peaks consistently above the physical sample surface. More significantly, the relative intensities of different spectral regions, i.e. SERS bands and background, vary non-monotonically and non-uniformly along the Z axis, meaning that band intensity ratios are focus-dependent. Finite-Difference Time-Domain (FDTD) simulations attribute these effects to plasmonic near-field responses specific to the focused and defocused beam interacting with the nanostructured metal surface. These findings reveal a previously overlooked source of spectral distortion in solid-substrate SERS measurements, with direct implications for the design and interpretation of quantitative assays based on band intensity ratios.

Coherent acoustic phonons, whose damping sets the upper bound of quality factors in acoustic resonators, play a critical role in advanced telecommunication and quantum information technologies. Yet, probing their decay in the GHz regime remains challenging using conventional surface-based techniques. Dark-field X-ray microscopy (DFXM) offers a solution by enabling through-depth, non-destructive and full-field imaging of strain fields and dislocations inside bulk materials with high spatial and angular resolution. We previously used kinematic diffraction theory to describe DFXM signals based on how the Bragg peak shifts due to the strain wave, allowing us to reconstruct the frequency spectrum of coherent phonons as a function of depth through the sample. The approach of tracking the Bragg peak shifts to study phonon dynamics, however, places an upper-bound to the highest phonon frequency that can be studied, determined by the spatial resolution of the measurement. In this work, we discuss how coherent phonon dynamics can be studied with DFXM from time-dependent intensity oscillation sidebands. This approach simultaneously allows studying coherent phonon dynamics in real and reciprocal space, overcoming frequency resolution limits imposed by the real-space resolution of Bragg-peak tracking. Using Takagi-Taupin dynamical diffraction formalism, we establish the spatial and reciprocal space resolution achievable for studying the coherent phonon dynamics and evaluate conditions for observing long-lived intensity oscillations. We close by proposing experimental strategies to optimize excitation bandwidths and reciprocal-space selectivity. The formalism in the paper enables the design of DFXM experiments for quantitative, frequency-resolved measurements of acoustic phonon decay and phonon-defect interactions in bulk crystalline materials.

We extend ferromagnetic resonance (FMR) modulation theory to describe systems in which bulk and boundary states of topological materials coexist, with both appearing at the same energy. As an application of the formulation, we investigate the enhancement of the Gilbert damping constant on the $(110)$ surface of a $d$-wave superconductor where nodal quasiparticles coexist with edge states, which are one-dimensional boundary states, known as surface zero-energy Andreev bound states. We find two characteristic features: a pronounced edge-to-edge excitation peak near zero energy, and an additional edge-to-bulk excitation peak at the superconducting gap energy. We also observe power-law decay at low temperatures and exponential decay at intermediate temperatures in the low-energy regime. These features demonstrate the comparable contributions of the bulk and boundary states to the FMR response. Our theory provides a broadly applicable framework for the analysis of topological materials.

Power-law distributions are widely observed in complex systems, yet establishing their thermodynamic consistency remains a theoretical challenge. In this paper, we present a thermodynamic framework for power-law statistics based on the \textit{renormalized entropy} $s_{2-q}$. Derived from the asymptotic scaling of the combinatorial $q$-factorial, this quantity yields a stable thermodynamic limit, remaining finite ($O(N^0)$) for systems with strong correlations. Furthermore, we clarify the physical origin of the nonlinearity parameter $q$ through the concept of \textit{Varentropy} (Variance of Entropy). By unifying the macroscopic variational principle with the microscopic Superstatistics framework, we derive the relation $|q-1| \simeq 1/C$, where $C$ is the heat capacity of the reservoir. This result suggests that power-law statistics provides a thermodynamic description of finite systems, where the finite heat capacity of the heat bath necessitates a generalization beyond the standard Boltzmann-Gibbs limit ($C \to \infty$).

We study the scaling properties of long-range loop-erased random walks (LR-LERW), where the underlying random walker performs Lévy-flight-like jumps with a power-law step-length distribution $P(\mathbf{r})\sim |\mathbf{r}|^{-(d+σ)}$. Using extensive Monte Carlo simulations, we measure the scaling relation $N \sim R^{d_N}$ between the loop-erased step number $N$ and the spatial extent $R$, and determine the geometric exponent $d_N$ for various values of $σ$ in spatial dimensions $d = 1, 2,$ and $3$, as well as at the marginal point $σ= 2$ in $d=4$ and $5$. We observe a continuous crossover from long-range (LR) to short-range (SR) behavior as $σ$ increases. Below the upper critical dimension $d<d_c=4$, for $σ< d/2$, loop erasure is asymptotically irrelevant and $d_N=σ$, consistent with Lévy-flight scaling. For $d/2 < σ< 2$, loop erasure becomes relevant and $d_N$ varies continuously toward the SR-LERW value. At the marginal points with $σ=d/2$ or $σ=2$, clear logarithmic corrections are observed. At and above the upper critical dimension, $d \geq 4$, the scaling at $σ=2$ is found to be $N \sim R^2/\ln R$, consistent with that of the corresponding Lévy flight. Our results provide a systematic numerical determination of $d_N(σ)$ for the LR-LERW across dimensions, and are consistent with $σ_* = 2$ as the boundary between LR and SR critical behaviors recently established in a broad variety of statistical models.

Frequency combs are a spectrum of equally spaced frequency components with very high time-frequency accuracy, which have been widely used in the optical and microwave frequency ranges. We propose the realization of a frequency comb operating at the terahertz regime in terms of the nonlinear dynamics of electric-polarization waves, or ferrons as their quanta, in the ferroelectric materials. The efficiency of the frequency comb of the electric-polarization waves is exactly proportional to the static electric polarization carried by the ferron modes, which thereby offers new opportunities for the direct observation and application of the intrinsic properties of ferrons.

We prove an exact finite-volume symmetry formula for two-point functions in the periodic $N$-state superintegrable chiral Potts spin chain. We show that, for every chain length $L$ and every simultaneous eigenvector of the Hamiltonian and the one-site translation operator, the correlations satisfy $\langle Z_0^r Z_R^{\dagger r}\rangle^*=\langle Z_0^r Z_{L-R}^{\dagger r}\rangle$ for $1\leqslant r\leqslant N-1$. Hence, whenever $L$ is even, the midpoint correlation $\langle Z_0^r Z_{L/2}^{\dagger r}\rangle$ is real. Then we generalise the three-state chain case to arbitrary $N$ and to every translation eigensector. This resolves a conjecture of Fabricius and McCoy.

The bulk-edge correspondence principle, a cornerstone of topological physics, ensures that first-order topological systems host robust chiral edge states in two dimension. This was later extended to higher-order phases, where second-order topological insulators exhibit localized, topologically protected corner states. While the transition between these distinct phases has been demonstrated in periodic systems, its existence in disordered platforms remains an open question. Here, we demonstrate a controllable topological phase transition between a second-order topological phase and a first-order topological phase in a two-dimensional photonic alloy. By tuning the magnetic doping concentration - implemented by attaching permanent magnets randomly to nonmagnetized yttrium iron garnet rods in an alternately magnetized honeycomb lattice with C3 rotational symmetry - we flexibly control the system's topology. At zero doping, we observe higher-order corner states, confirmed by a trivial Chern number and non-zero bulk polarizations of 1/3. As doping concentration increases, these corner states progressively merge with the bulk states, culminating in the closure of the bulk transmission gap. After the bulk transmission gap reopens with further increased doping, the system transitions to a first-order topological phase, characterized by a nontrivial Chern number of -1 and the emergence of a chiral edge state. This transition is reversible, providing a highly tunable and experimentally simple platform for flexibly switching between localized corner states and delocalized chiral edge states within a single photonic system.

This review describes the development and applications of multi-chain coarse-grained simulations for entangled polymer dynamics. The mean-field tube model has long served as the standard paradigm for describing the many-body entanglement problem as the motion of a single chain in a static field; it faces intrinsic limitations when addressing spatial correlations, fluctuations, and complex topological rearrangements. To overcome these limitations, "multi-chain" approaches -- specifically the primitive chain network and multi-chain slip-spring models -- were developed. These simulations explicitly resolve the force balance and topological coupling between multiple chains in three-dimensional space. This review covers the primitive chain network model, which emphasizes real-space force balance, and the multi-chain slip-spring model, which is derived from a well-defined free-energy functional. Linear and nonlinear rheology predictions are discussed, along with molecular mechanisms such as constraint release and stretch/orientation-induced reductions in friction. Extensions to branched polymers, wall-slip phenomena, and network polymers are also mentioned.

Wave-like dark matter described by a high-occupancy self-gravitating bosonic field provides a microscopic setting in which both amplitude and phase are dynamical. We study a one-dimensional Gross--Pitaevskii--Poisson toy model and ask which coarse-grained variable, if any, can be meaningfully compared with the 1+1-dimensional Kardar--Parisi--Zhang (KPZ) fixed point. We show that the relevant field is not the raw microscopic phase but a branch-resolved coarse-grained phase built from the sound sector. Above the Jeans scale and below the microscopic cutoff, self-gravity acts as a weak deformation of local sound dynamics. In this window the exact linear modes admit a local sound form, and a weakly nonlinear projection yields a nonvanishing same-chirality Burgers self-coupling. Under one-branch dominance together with a local Markov closure, the dominant branch reduces conditionally to a KPZ-type equation. We also formulate a dictionary from microscopic initial data to the canonical curved, flat, and stationary KPZ benchmarks. Our results do not establish KPZ universality for self-gravitating bosonic dark matter, but they identify the proper comparison field and the controlled regime in which an exact fixed-point test can be posed.

We present a computational framework for characterizing the molecular self-organization of cocoa butter (Theobroma cacao) during dark chocolate tempering through the lens of Topological Data Analysis (TDA). A physics-inspired particle simulation models N=100 triglyceride molecules across the full temperature range 15--60 degrees C, spanning all six crystalline polymorphs of cocoa butter (Forms I--VI) as well as the melt and superheating regimes. At each temperature tick, we construct a Vietoris-Rips filtration and compute the persistent homology groups H0 (connected components), H1 (independent cycles), and H2 (3D voids). The resulting persistence diagrams are analyzed via persistent entropy E = -sum_i p_i log2(p_i), where p_i = l_i / sum_j l_j and l_i = death_i - birth_i denotes feature lifetime; essential classes are assigned death = m+1 (m = eps_max) following the standard persistent entropy convention (Rucco 2026, arXiv:2602.09058). Our results demonstrate that Form V (the optimal tempering polymorph, 29.5--34 degrees C) is characterized by a distinctive topological signature: a local minimum in the H0 persistent entropy (E0 = 5.74 +/- 0.04 bits), a pronounced depression in the first Betti number beta_1 (1562 +/- 35), and a global minimum in the H2 entropy (E2 = 12.29 +/- 0.25 bits) reflecting coherent inter-bilayer lamellar cavities. Via Theorem 1 and Corollary 1 of Rucco (2026), persistent entropy is proven to separate the ordered and disordered phases by an asymptotically non-vanishing gap whenever a phase transition induces the creation or destruction of topological mass at macroscopic scales -- a condition we verify empirically across all eight cocoa butter regimes. These findings suggest that TDA-based metrics could serve as non-invasive quality indicators for industrial chocolate tempering processes.

We report the thermal conductances of molecular junctions made of self-assembled monolayers of benzo-bis(imidazole) molecules, without side groups or functionalized with two phenylamine side groups. In the two cases, when the molecules are connected to the electrodes by thiol anchoring groups in the meta-position, the thermal conductance is decreased compared to the same molecules connected in the para-position (ca. 16-29 nW/K and ca. 37-40 nW/K, respectively) in agreement with the theoretically predicted phonon interference effect in molecular junctions. Upon protonation, the thermal conductances of the meta-connected molecular junction increase by about 50% (reversible behavior upon deprotonation). The fact that only the thermal conductance of the meta-connected molecular junction is sensitive to the protonation/deprotonation is tentatively related to modifications of the structural organization of the molecules in the monolayer, which modifies the thermal conductance at the molecule/electrode interfaces. The electrical conductance is lower for the meta-connected molecule than for the para-connected one, due to destructive quantum interferences, as expected and reported for other molecular junctions. The conductance further decreases (reversibly) upon protonation. The energy position of the molecular orbital involved in the electron transport is not modified by the protonation and the decrease in current is related to changes in the molecule organization in the monolayer, which modulate the electronic coupling energy at the molecule/electrode interfaces.

We investigate the roles of spin and packing fraction on the dynamics of cratering when a solid projectile impacts a granular bed at different incident angles. For that, we carried out DEM (discrete element method) computations in which we varied the magnitude and direction of the projectile spin, the impact velocity, the bed packing fraction, and the incident angle. For a given incident velocity, we found that the projectile can rebound for small angles, or be completely or partially buried for larger angles, and that when buried it can sometimes migrate large horizontal distances depending on the incident angle. We also found that increasing the packing fraction strengthens rebounds, and that the initial spin, depending on its direction and orientation, induces rebound, burying, or transverse deviations. The crater morphology also changes with the varying parameters, acquiring circular, elliptical, goutte-like, tadpole-like, and transitional shapes, correlating well with the projectile behavior. Finally, we propose diagrams organizing and classifying the dynamics observed. Our results shed new light on the different shapes of craters found in nature and the fate of the impacting material.

Controlling collective electronic phases in low-dimensional materials is a central challenge for developing technologies based on charge-density waves. Here, we report that perpendicular electric and magnetic fields can be used to tune charge-density-wave transport in the quasi-two-dimensional material 1T-TaS2. Using h-BN-encapsulated thin-film heterostructures with both top-gate and bottom-gate configurations, we find that electrical gating produces a non-monotonic shift in the depinning threshold, a behavior distinct from that of quasi-one-dimensional charge-density-wave systems. We further show that a perpendicular magnetic field increases the threshold voltage for domain depinning and can drive the nearly commensurate-to-incommensurate charge-density-wave phase transition, demonstrating magnetic control over a two-dimensional electron-lattice condensate. The obtained results shed light on mechanisms governing charge-density-wave domain dynamics and reveal combined electrical and magnetic-field control as a strategy for engineering low-power-dissipation devices and electronics for extreme environments.

We study charge transport across a junction between a normal metal and a Rashba metal in the presence of a Zeeman field applied to the spin--orbit coupled region. While an out-of-plane Zeeman field does not generate a transverse response in a homogeneous Rashba system, we show that such a junction exhibits a finite transverse conductivity that is inherently nonreciprocal, i.e., it depends on the direction of the applied bias. We demonstrate that this effect originates from the breaking of the $k_y \to -k_y$ symmetry of the Hamiltonian in the presence of the Zeeman field, which prevents cancellation of transverse current contributions from opposite transverse momenta. We further show that evanescent modes in the spin--orbit coupled region play a crucial role by carrying a finite spin polarization that gives rise to a transverse current localized near the junction. The transverse conductivity exhibits a peak at an energy scale set by the Zeeman field, displays distinct behavior for opposite bias directions, and shows spatial dependence governed by the nature of the contributing modes. We also identify bound states at the junction for attractive barrier strengths, which enhance conductivity when their energies lie near the transport window. Our results reveal a mechanism for nonreciprocal transverse charge transport in Rashba systems without requiring in-plane magnetic fields or ferromagnetic contacts, and should be experimentally accessible in semiconductor heterostructures.

Accurate phase estimation at the edge of data segments is crucial for EEG applications such as EEG-TMS in offline and real-time data analysis. Our research evaluates the phase estimation performance of four commonly used methods (Phastimate, SSPE, ETP, and PhastPadding) for accuracy and systemic biases, using data from young and elderly healthy controls and chronic stroke participants. To address the identified limitations of the established methods, we introduce Phase Estimation with Autoregressive Padding (PEAP), a method that prevents strong bandpass filtering-induced artifacts. Contrary to the established methods, PEAP does not show significant biases and improves accuracy by 3.2 to 9.2% for the continuous phase estimation. Our offline analysis demonstrates how established methods are systematically biased towards some estimates and how they induce phase shifts. We also show that differences between methods do not vary between clinical and control populations, supporting their translatability. This work indicates that systematic biases in established phase estimation methods may compromise the validity and comparability of phase-dependent findings. PEAP addresses these limitations and thus offers a more reliable and more accurate alternative method.

Background and Objective: Accurate surrogate modeling of knee joint contact mechanics is important for reconstructing stress distributions and identifying risk-relevant regions, yet the relative suitability of different modeling paradigms under practically relevant input-limited conditions remains unclear. Methods: Nine male soccer players performed 90° change-of-direction trials. Finite element simulations driven by subject-specific joint posture and reaction forces were converted into graph-structured samples. Five surrogate architectures representing local diffusion, history-context enhancement, hierarchical multi-scale modeling, explicit global interaction, and local-global hybridization were compared using three-fold cross-subject validation under full, pose-corrupted, load-corrupted, and minimal-input conditions. Performance was evaluated using full-field error, high-stress error, high-risk region overlap, and hotspot localization metrics. Results: The hybrid model achieved the best overall performance under full inputs and remained the most robust under pose- and load-corrupted conditions. Under minimal inputs, no single model dominated all metrics: the history-context model yielded lower overall and high-stress errors, the hybrid model better preserved high-risk region reconstruction, and the hierarchical model showed an advantage in hotspot localization. Conclusion: Evaluation of surrogate models for knee joint contact mechanics should shift from accuracy comparisons under ideal inputs to a comprehensive assessment of the preservation of risk-relevant information under realistic input constraints. Although the local-global hybrid model showed the best overall robustness, the optimal model under minimal-input conditions remained task-dependent.

Effectively modeling irregularly sampled longitudinal data is essential for understanding disease progression and improving risk prediction. We propose a two-view mixture model that integrates static baseline covariates and longitudinal biomarker trajectories within a unified probabilistic clustering framework. Temporal patterns are modeled using Neural Ordinary Differential Equations. Model training uses an EM algorithm with a sparsity-inducing log-penalty for interpretable subgroup discovery. Application of the model to an Irish cohort of ANCA-associated vasculitis patients reveals subgroups with heterogeneous serum creatinine trajectories and variation in end-stage kidney disease outcomes.

Parkinsons disease (PD) alters cortical neural dynamics, yet reliable non-invasive electrophysiological biomarkers remain elusive. This study examined whether interpretable EEG features capturing complementary aspects of neural dynamics can discriminate Parkinsonian neural states. A comprehensive set of interpretable features was extracted and grouped into Standard descriptors (spectral power, phase synchronization, time-domain statistics) and Dynamical descriptors (aperiodic activity, cross-frequency coupling, scale-free dynamics, neuronal avalanche statistics, and instantaneous frequency measures). A multi-head attention transformer classifier was trained using strict LOSO validation. Group-level comparisons were performed to identify electrophysiological differences associated with disease and medication state. Standard feature sets achieved strongest performance in discriminating medication states (PDoff vs PDon), whereas Dynamical performed competitively in contrasts between PD patients and healthy controls. Random feature ablation analyses indicated that Dynamical descriptors provide complementary information distributed across features while correlation analysis revealed low redundancy within both feature sets. Group-level comparisons revealed medication-sensitive reductions in delta power and voltage variance, modulation of neuronal avalanche statistics, persistent increases in theta phase synchronization in PD patients, and disease-related alterations in cross-frequency interactions. Traditional spectral and synchronization features primarily reflect medication-related neural modulation, whereas dynamical descriptors reveal broader alterations in cortical network organization associated with disease but also with medication. These findings support multivariate EEG representations as a promising framework for developing non-invasive biomarkers of PD.

The evolutionary and ecological dynamics of tumors under immune responses and therapeutic interventions pose major challenges to long-term treatment success. Although treatment may initially achieve short-term disease control, resistant cancer cell subpopulations often arise, leading to relapse with more aggressive and treatment-resistant forms of the disease. Here, we develop and analyze mathematical models describing the interactions among effector cells, chemo-resistant tumor cells, and immuno-resistant tumor cells under distinct immune-evasion strategies. The models incorporate competition and cooperation between resistant and sensitive tumor subpopulations. We identify threshold conditions governing tumor persistence, elimination, and phenotype dominance under varying therapeutic intensities. These findings provide a theoretical framework for designing targeted and combination therapies and offer insights into strategies for mitigating the treatment resistance.

Motivated by classical communications engineering, early works in molecular communication (MC) largely adopted established modeling and signal processing concepts from wireless electromagnetic communication systems. In the context of the human cardiovascular system (CVS), MC channel models evolved from simple unbounded and single-duct environments mimicking individual blood vessels to complex vessel network (VN) topologies, generally at the expense of analytical tractability. Up until now, this has largely prohibited rigorous communication-theoretic analysis of large-scale VNs. In this work, we leverage a recently established closed-form analytical channel model for VNs, named mixture of inverse Gaussians for hemodynamic transport (MIGHT), to conduct the first systematic communication-theoretic study of MC in complex, large-scale VNs. Based on MIGHT, we derive a Poisson channel noise model and unveil structural analogies between multipath wireless communications (MWC) and advective-diffusive MC in VNs. In particular, we establish classical MWC metrics, namely the root mean squared (RMS) delay spread, the mean excess delay, and the coherence bandwidth, for MC in VNs and derive closed-form expressions for the channel frequency response and power delay profile (PDP). Building on this characterization, we propose a VN-adapted, coherent decision-feedback (DF) detector and show how the derived multipath metrics can inform the choice of critical system parameters like the symbol duration, the sampling time, and the memory length. Additionally, we evaluate the detector's performance in different VNs exhibiting inter-symbol interference (ISI). Together, these contributions open the door to a systematic, MWC-inspired MC system design for large-scale VNs.

This work presents the Parallelized Hierarchical Connectome (PHC), a general framework that upgrades temporal-only State-Space Models (SSMs) into spatiotemporal recurrent networks. Conventional SSMs achieve high-speed sequence processing through parallel scans, yet are limited to temporal recurrence without lateral or feedback interactions within a single timestep. PHC maps the diagonal SSM core to a shared Neuron Layer and inter-neuronal communication to a shared Synapse Layer, where neurons are partitioned into hierarchical regions governed by the connectome topology. A Multi-Transmission Loop enables intra-slice spatial recurrence, allowing signals to propagate across the hierarchical connectome within each temporal window while preserving O(logT) parallelism. This framework enables integration of neuro-physical priors typically intractable for standard SSMs, including adaptive leaky integrate-and-fire dynamics, Dale's Law, short-term plasticity, and reward-modulated spike-timing-dependent plasticity. The framework is instantiated as PHCSSM, the first model to unify recurrent spiking neural network dynamics with diagonal SSM parallelism while enforcing all five biological constraints and learnable lateral connections within a fully parallelizable training pipeline. Empirical results on physiological benchmarks from the UEA multivariate time-series archive demonstrate that PHCSSM achieves performance competitive with state-of-the-art SSMs while reducing parameter complexity from Theta(D^2 L) for L-layer stacked architectures to Theta(D^2). These findings suggest that biologically grounded inductive biases offer a principled route to parameter-efficient sequence modeling, opening diagonal SSMs to spatiotemporal recurrence and enabling fully parallelizable recurrent spiking neural network training.

When competing species grow into new territory, the population is dominated by descendants of successful ancestors at the expansion front. Successful ancestry depends on both the reproductive advantage (fitness), as well as ability and opportunity to colonize new domains. We present a model that integrates both elements by coupling the classic description of one-dimensional competition (Fisher equation) to the minimal model of front shape (KPZ equation). Macroscopic manifestations of these equations are distinct growth morphologies controlled by expansion rates, competitive abilities, or spatial anisotropy. In some cases the ability to expand in space may overcome reproductive advantage in colonizing new territory. When new traits appear with accumulating mutations, we find that variations in fitness in range expansion may be described by the Tracy--Widom distribution.

Boolean networks are dynamical models of disease development in which the activation levels of genes are represented by binary variables. Given a Boolean network, controls represent mutations or medical treatments that fix the activation levels of selected genes so that all states in every attractor (i.e., long-term recurrent states) satisfy a desired phenotype. Our goal is to enumerate all minimal controls, identifying critical gene subsets in disease development and therapy. This problem has an inherent bilevel integer programming structure and is computationally challenging. We propose an infeasibility-based Benders decomposition, a logic-based Benders framework for bilevel integer programs with multiple subproblems. In our application, each subproblem finds a forbidden attractor of a given length and yields a problem-specific feasibility cut. We also propose an auxiliary IP called subspace separation that finds a Boolean subspace that includes multiple forbidden attractors and thereby strengthens the cut. Numerical experiments show that the resulting algorithms are much more scalable than state-of-the-art methods and that subspace separation substantially improves performance.

Prion-like propagation of misfolded proteins is a key mechanism underlying the progression of neurodegenerative diseases such as Alzheimer's disease. In previous work, we introduced the HeMiTo framework, describing these prion-like dynamics for a class of heterodimer models in terms of three phases: the healthy (He), mixed (Mi), and toxic (To) phases. While the He-phase was characterised analytically, the Mi-phase was described numerically and the To-phase was inferred from linear stability arguments. In this work, we provide a complete analytical characterisation of the Mi- and To-phases for our class of heterodimer models. We derive exact inner solutions governing the Mi-phase and match them with outer solutions from the He-phase, explaining the concave-like behaviour of the healthy species and establishing explicit conditions for exponential growth of the toxic species with a mechanistically interpretable growth rate. Furthermore, we formalise a quasi steady-state reduction near the toxic steady state and show that the dynamics reduce to a logistic growth equation, linking exponential growth to saturation. Together, these results provide a unified and mechanistic description of prion-like dynamics across all phases of disease progression and establish a foundation for predictive modelling of biomarker trajectories.

Stochastic reaction networks are mathematical models with a wide range of applications in biochemistry, ecology, and epidemiology, and are often complex to analyze. Except for some special cases, it is generally difficult to predict how the abundances of all considered species evolve over time. A possible approach to address this issue is to develop tools to compare the model under study with a similar one whose behavior is better understood. The main contribution of our work is to provide direct and computable conditions that can be used to ensure the existence of an ordered coupling between two stochastic reaction networks and to identify which parameter changes in a given model lead to an increase or decrease in the count of certain species. We also make available an algorithm that implements our theory, and we illustrate it with several applications.

Natural genomes sometimes encode two different proteins in staggered reading frames of the same DNA sequence. Despite the prevalence of these 'overlapping genes' across the tree of life, it remains unknown whether arbitrary protein pairs can overlap, to what extent such overlaps are feasible, or what design principles govern them. Here, we study compatibility, frustration, and connectivity in the fitness landscape of overlapping genes. We computationally design sequences de novo that satisfy the dual functional constraints of two distinct protein families. The joint fitness landscape, inferred via Potts models from multiple sequence alignments, reveals a fundamental trade-off between the two proteins and provides a simple criterion for when overlap is feasible. We find widespread compatibility between protein families, with one class of reading frames markedly more permissible than others. By exploring alternative genetic codes, we find that the natural genetic code is uniquely well-suited to support overlapping genes. Constructing mutational paths between sequences, we find that sequence-diverse overlapped genes can be connected via a network of near-neutral mutations. Overall, our results suggest that protein fitness landscapes are sufficiently flexible so as to accommodate the stringent, orthogonal requirements of overlapping genes.

Mammalian sleep is characterized by multiple alternations between episodes of rapid-eye-movement sleep (REMS) and non-REM sleep (NREMS). While the mechanisms governing the timing of these ultradian NREMS-REMS cycles remain poorly understood, the phenomenon of REMS pressure, namely a drive for REMS that builds up between REMS episodes, is thought to be a contributing factor. Prior analyses of NREMS-REMS cycles in mice has suggested that time in NREMS is a primary contributor to REMS pressure. Building on that finding, we previously introduced a REMS propensity measure defined as the probability to enter REMS before the accumulation of an additional amount of NREMS. Analyzing mouse ultradian cycle data, we showed that REMS propensity at REMS onset was positively correlated with REMS bout duration and with the probability of the occurrence of a REMS bout followed by a short inter-REMS interval, called a sequential REMS cycle. In this paper, we extend our analyses of REMS propensity to human and rat ultradian NREMS-REMS cycle data. We show that, as in mice, human and rat sleep contain both short NREMS-REMS sequential cycles and longer single NREMS-REMS cycles, though there are some differences in the relative distributions of cycle durations. Although rodents exhibit polyphasic sleep in contrast with the consolidated sleep of humans, the calculated REMS propensity measures in all three species show similar profiles as functions of time spent in NREMS: specifically, REMS propensity increases with time spent in NREMS until it reaches a peak value, and then it decays with additional time in NREMS. Positive correlations of REMS propensity at REMS onset with REMS bout duration were present in both human and rat data as in mouse data, suggesting that time spent in NREMS also influences REMS duration in these species.

Foraging is a central decision-making behavior performed by all animals, essential to garnishing enough energy for an organism to survive. Similarly, mating is crucial for evolutionary continuity and offspring production. Mate choice is one of the central tenets of sexual selection, driving major evolutionary processes, and can be regarded as a decision-making process between potential mating partners. Often researchers have used coarse-grained models to describe macroscopic phenomenology pertaining to mate choice without detailed quantitative mechanisms of how animals use individual and environmental signals to guide their mating decisions. In this letter, we show that mate choice can be cast as a foraging problem, and we present an analytically tractable optimal foraging-inspired mechanistic theory of decision-making underlying mate choice. We begin from the premise that deciding upon which partner with which to mate is at its core a stochastic decision-making process. Agents adopt a variety of decision strategies, tuned by decision thresholds for leaving or committing to a mate. We find that sensitive leaving thresholds are favored independently of signal availability in the population. By contrast, optimal thresholds for committing to a mate depend upon signal availability in the population, with signal-rich populations generally favoring less eager strategies compared to signal-poor populations.

Low-intensity transcranial focused ultrasound (tFUS) is rapidly emerging as a transformative non-invasive brain stimulation (NIBS) modality characterized by high spatial resolution and ability to target deep brain circuits. Unlike electromagnetic techniques such as transcranial magnetic stimulation and transcranial direct current stimulation, which are constrained by centimeter-scale resolution and a depth-focality tradeoff, tFUS leverages mechanical pressure waves to modulate both superficial cortical and deep subcortical structures with millimeter precision. This article discusses recent scientific observations and engineering breakthroughs in the advancement of tFUS for next-generation ultrasonic brain-computer interfaces (uBCIs) and human-machine interfaces. These advancements move beyond open-loop systems and demonstrate closed-loop architectures that incorporate real-time electrophysiological feedback to optimize cognitive variables such as attention, learning, trust, and cooperation in various applications. Other advances in the development of ultrasound sensors for sonomyography to decode muscle activation and functional ultrasound to monitor hemodynamic brain activity are discussed as potential elements in bidirectional uBCIs. Together, these advances position ultrasound as a foundational technology for the development of intelligent, adaptive, and bidirectional neural interfaces that will seamlessly integrate human cognition with next-generation automation and robotic systems.

In cases of prevalent diseases and disorders, such as Prenatal Alcohol Exposure (PAE), multi-site data collection allows for increased study samples. However, multi-site studies introduce additional variability through heterogeneous collection materials, such as scanner and acquisition protocols, which confound with biologically relevant signals. Neuroscientists often utilize statistical methods on image-derived metrics, such as volume of regions of interest, after all image processing to minimize site-related variance. HACA3, a deep learning harmonization method, offers an opportunity to harmonize image signals prior to metric quantification; however, HACA3 has not yet been validated in a pediatric cohort. In this work, we investigate HACA3's ability to remove site-related variance and preserve biologically relevant signal compared to a statistical method, neuroCombat, and pair HACA3 processing with neuroCombat to evaluate the efficacy of multiple harmonization methods in a pediatric (age 7 to 21) population across three unique scanners with controls and cases of PAE with downstream MaCRUISE volume metrics. We find that HACA3 qualitatively improves inter-site contrast variations, but statistical methods reduce greater site-related variance within the MaCRUISE volume metrics following an ANCOVA test, and HACA3 relies on follow-up statistical methods to approach maximal biological preservation in this context.

The HEALthy Brain and Childhood Development (HBCD) Study is an ongoing longitudinal initiative to understand population-level brain maturation; however, large-scale studies must overcome site-related variance and preserve biologically relevant signal. In addition to diffusion-weighted magnetic resonance imaging images, the HBCD dataset offers analysis-ready derivatives for scientists to conduct their analysis, including scalar diffusion tensor (DTI) metrics in a predetermined set of bundles. The purpose of this study is to characterize HBCD-specific site effects in diffusion MRI data, which have not been systematically reported. In this work, we investigate the sensitivity of HBCD bundle metrics to scanner model-related variance and address these variations with ComBat-GAM harmonization within the current HBCD data release 1.1 across six scanner models. Following ComBat-GAM, we observe zero statistically significant differences between the distributions from any scanner model following FDR correction and reduce Cohen's f effect sizes across all metrics. Our work underscores the importance of rigorous harmonization efforts in large-scale studies, and we encourage future investigations of HBCD data to control for these effects.

The modern deep learning field is a scale-centric one. Larger models have been shown to consistently perform better than smaller models of similar architecture. In many sub-domains of biomedical research, however, the model scaling is bottlenecked by the amount of available training data, and the high cost associated with generating and validating additional high quality data. Despite the practical hurdle, the majority of the ongoing research still focuses on building bigger foundation models, whereas the alternative of improving the ability of small models has been under-explored. Here we experiment with building models with 10-30M parameters, tiny by modern standards, to perform the single-cell segmentation task. An important design choice is the incorporation of a recursive structure into the model's forward computation graph, leading to a more parameter-efficient architecture. We found that for the single-cell segmentation, on multiple benchmarks, our small model, UCell, matches the performance of models 10-20 times its size, and with a similar generalizability to unseen out-of-domain data. More importantly, we found that ucell can be trained from scratch using only a set of microscopy imaging data, without relying on massive pretraining on natural images, and therefore decouples the model building from any external commercial interests. Finally, we examined and confirmed the adaptability of ucell by performing a wide range of one-shot and few-shot fine tuning experiments on a diverse set of small datasets. Implementation is available at https://github.com/jiyuuchc/ucell

We present a unified theoretical model relating stochastic microscopic epidemic dynamics with macroscopic non-linear population behavior. Utilizing the Doi-Peliti formalism, we model the pathogen as a gauge mediator field coupled to susceptible and infected host populations, and introduce a Reactive Immunity Field capable of spontaneous symmetry breaking. We demonstrate that the naive epidemic vacuum is destabilized by radiative loop corrections via the Coleman-Weinberg mechanism, generating a dynamic herd immunity threshold. By extracting the classical saddle-point limit of the Effective Action, we derive the macroscopic reaction-diffusion equations governing the host population. We show that integrating out the gauge mediator inherently generates a thermodynamic Free Energy dependent on the square of the susceptible density. This non-linearity produces a macroscopic spatial ``Fear Drift'' proportional to the magnitude of the immunity field, and a cubic shielding penalty in the effective reproductive number ($R_{eff}$). In this work, we establish a mapping between fundamental field-theoretic mechanisms and specific terms in the macroscopic behavioral equations. We demonstrate that Debye screening is physically executed by the spatial cross-diffusion fluxes driving host evacuation. Simultaneously, vacuum polarization manifests as a non-linear cubic penalty ($-S^3 I$) in the dressed reaction rate that dynamically suppresses the effective reproductive number. As a validation of our model, we apply the formalism to high-resolution spatiotemporal COVID-19 data from Germany.

This study examines the evolution of Intelligent and Secure Smart Hospital Ecosystems using a Scoping Review with Bibliometric Analysis (ScoRBA) to map research patterns, identify gaps, and derive policy implications. Analyzing 891 journal articles from Scopus (2006-2025) through co-occurrence analysis, network visualization, overlay analysis, and the Enhanced Strategic Diagram (ESD), the study applies the PAGER framework to link Patterns, Advances, Gaps, Research directions, and Evidence-based policy implications. Findings reveal three interrelated clusters: AI-driven intelligent healthcare systems, decentralized privacy-preserving digital health ecosystems, and scalable cloud-edge infrastructures, showing a convergence toward integrated ecosystem architectures where intelligence, trust, and infrastructure reinforce each other. Despite progress in AI, blockchain, and cloud computing, gaps remain in interoperability, real-world implementation, governance, and cross-layer integration. Emerging themes such as explainable AI, federated learning, and privacy mechanisms highlight areas needing further research. Policy-relevant recommendations focus on coordinated governance, scalable infrastructure, and secure data ecosystems, particularly for developing country contexts. The study bridges bibliometric evidence with actionable policies, supporting informed decision-making in smart hospital development.

Mathematical models of natural and man-made systems often have many adjustable parameters that must be estimated from multiple, potentially conflicting datasets. Rather than reporting a single best-fit parameter vector, it is often more informative to generate an ensemble of parameter sets that collectively map out the trade-offs among competing objectives. This paper presents ParetoEnsembles.jl, an open-source Julia package that generates such ensembles using Pareto Optimal Ensemble Techniques (POETs), a simulated-annealing-based algorithm that requires no gradient information. The implementation corrects the original dominance relation from weak to strict Pareto dominance, reduces the per-iteration ranking cost from $O(n^2 m)$ to $O(nm)$ through an incremental update scheme, and adds multi-chain parallel execution for improved front coverage. We demonstrate the package on a cell-free gene expression model fitted to experimental data and a blood coagulation cascade model with ten estimated rate constants and three objectives. A controlled synthetic-data study reveals parameter identifiability structure, with individual rate constants off by several-fold yet model predictions accurate to 7%. A five-replicate coverage analysis confirms that timing features are reliably covered while peak amplitude is systematically overconfident. Validation against published experimental thrombin generation data demonstrates that the ensemble predicts held-out conditions to within 10% despite inherent model approximation error. By making ensemble generation lightweight and accessible, ParetoEnsembles.jl aims to lower the barrier to routine uncertainty characterization in mechanistic modeling.

We propose new analytical tools for describing growth-rate distributions generated by stationary time-series. Our analysis shows how deviations from normality are not pathological behaviour, as suggested by some traditional views, but instead can be accounted for by clean and general statistical considerations. In contrast, strict normality is the effect of specific modelling choices. Systems characterized by stationary Gamma or heavy-tailed abundance distributions produce log-growth-rate distributions well described by a generalized logistic distribution, which can describe tent-shaped or nearly normal datasets and serves as a useful null model for these observables. These results prove that, for large enough time lags, in practice, growth-rate distributions cease to be time-dependent and exhibit finite variance. Based on this analysis, we identify some key stylized macroecological patterns and specific stochastic differential equations capable of reproducing them. A pragmatic workflow for heuristic selection between these models is then introduced. This approach is particularly useful for systems with limited data-tracking quality, where applying sophisticated inference methods is challenging.

Brain connectomics is still largely dominated by pairwise-based models, such as graphs, which cannot represent circulatory or higher-order functional interactions. In this paper, we propose a multimodal framework based on Topological Signal Processing (TSP) that models the brain as a higher-order topological domain and treats functional interactions as discrete vector fields. We integrate diffusion MRI and resting-state fMRI to learn subject-specific brain cell complexes, where statistically validated structural connectivity defines a sparse scaffold and phase-coupling functional edge signals drive the inference of higher-order interactions (HOIs). Using Hodge-theoretic tools, spectral filtering, and sparse signal representations, our framework disentangles brain connectivity into divergence (source-sink organization), gradient (potential-driven coordination), and curl (circulatory HOIs), enabling the characterization of temporal dynamics through the lens of discrete vector calculus. Across 100 healthy young adults from Human Connectome Project, node-based HOIs are highly individualized, yet robust mesoscale structure emerges under functional-system aggregation. We identify a distributed default mode network-centered gradient backbone and limbic-centered rotational flows; divergence polarization and curl profiles defining circulation regimes with insightful occupancy and dwell-time statistics. These topological signatures yield significant brain-behavior associations, revealing a relevant higher-order organization intrinsic to edge-based models. By making divergence, circulation, and recurrent mesoscale coordination directly measurable, this work enables a principled and interpretable topological phenotyping of brain function.

Causal inference in brain networks has traditionally relied on regression-based models such as Granger causality, structural equation modeling, and dynamic causal modeling. While effective for identifying directed associations, these methods remain descriptive and acyclic, leaving open the fundamental question of intervention: what would the causal organization become if a pathway were disrupted or externally modulated? We introduce a unified framework for counterfactual causal analysis that models both pathological disruptions and therapeutic interventions as an energy-perturbation problem on network flows. Grounded in Hodge theory, directed communication is decomposed into dissipative and persistent (harmonic) components, enabling systematic analysis of how causal organization reconfigures under hypothetical perturbations. This formulation provides a principled foundation for quantifying network resilience, compensation, and control in complex brain systems.

FcsIT is a platform-independent, open-source tool for calculating the correlation and fitting fluorescence correlation spectroscopy data. The software is written in Python and uses a powerful Dear PyGUI engine for its interface. It provides reading and correlating the TTTR data, as well as TCSPC filtering of the photon time-trace data. The circular-block bootstrap method applied to the calculation of correlation data and its variance results in data quality comparable to that obtained with commercially available software. An intuitive fitting interface provides efficient analysis of large datasets and includes nine predefined mathematical models for fitting correlation curves. Moreover, it allows users to add their own models in a user-friendly manner. Validation of the FcsIT tool against simulated FCS data and real FCS experiments confirms its usability and potential appeal to a wide variety of FCS users.

Reservoirs, typically implemented as recurrent neural networks with fixed random connection weights, can be combined with a simple trained readout layer to perform a wide range of computational tasks. However, increasing the magnitude of reservoir connection weights to exploit nonlinear dynamics can cause the network to develop strong spontaneous activity that drives neurons into saturation, dramatically degrading performance. In this work, we investigate two distinct countermeasures against such runaway excitation. The first approach introduces a subtle non-homogeneous structure into the matrix of connection weigths $w_{ij}$, without altering the overall probability distribution $p(w)$. We identify several favorable structuring principles, such as creating a small subset of neurons with weaker-than-average input connections. Even if the rest of the reservoir falls into runaway saturating behavior, this weakly coupled subset remains in a mildly nonlinear regime whose dynamics can still be exploited by the readout layer. The second approach implements a form of automatic gain control, in which a dedicated control unit dynamically regulates the reservoir's average global activation toward an optimal setpoint. Although the control unit modulates the excitability of the reservoir only via a global gain factor, this mechanism substantially enlarges the dynamical regime favorable for computation and renders performance largely independent of the underlying connection statistics.

Flux sampling is an analysis that, based on a distribution, picks randomly an efficient number of points from the solution space of a metabolic model. Unlike most constraint-based analyses, flux sampling does not require an objective function to optimize, allowing for the exploration of the whole spectrum of the phenotypes a species can exhibit. However, sampling can also be restricted to a subspace where a chosen objective reaches at least a specified fraction of its optimum. This targeted approach adds value when investigating phenotypes that are optimal for a specific function. Contrary to Flux Balance Analysis, which returns a single solution, sampling leverages statistical power to uncover phenotypes that otherwise would be masked. This can be especially useful when changing the conditions (medium) in which a species lives. Here, we highlight some state-of-the-art methods for applying flux sampling at Genome-Scale Metabolic Models in different scenarios, and we showcase flux sampling applications

The spread of infectious disease and the evolution of antigenically distinct strains are often modeled separately, despite strong feedbacks mediated by host immune memory and heterogeneous contacts. To tackle this challenging problem, we introduce a coevolutionary framework in which transmission occurs on a metapopulation network while mutational exploration of strain space follows a mutation network. In this multiscale model, cross-immunity is encoded by similarity in the latent diffusion geometry of the strain network, so that nearby strains confer partial immune protection. We first identify an effective critical region that controls the transition between extinction, recurrent outbreak episodes, and long-lived endemic persistence, thus characterizing the resulting strain-turnover dynamics. We then derive a replicator-mutator-like equation for strain composition and an explicit dynamical evolutionary landscape induced by the coupling of mutation and transmission. Finally, allowing host heterogeneity to modulate the local mutation structure, we show that spreading across demes can effectively connect otherwise disconnected components of strain space, increasing long-term endemic diversity while producing a non-monotonic change in overall prevalence. Together, our results isolate minimal mechanisms by which immune-mediated competition and network structure can shape antigenic diversification.

Anthropogenic activity threatens biodiversity through climate change, habitat fragmentation, and increasing frequency and scale of disturbance. Various theoretical studies have sought to shed light on how these factors could promote or hinder the coexistence of species. However, our understanding of the relative importance of, and interactions between, these factors remains limited. In this study, we employ a theoretical approach integrating three commonly cited coexistence mechanisms -- the competition-colonisation trade-off, the intermediate disturbance hypothesis, and spatial heterogeneity -- into a unified model. We implement a novel method to integrate habitat autocorrelation into a system of differential equations, to create a simple and flexible model that can be used to investigate coexistence of multiple species arranged in a competitive hierarchy under different disturbance and habitat structure scenarios. Using this model, we find that considering interactions between different mechanisms is crucial for explaining the coexistence of species. Biodiversity patterns alternative to the uni-peak curve predicted by the intermediate disturbance hypothesis (e.g., bimodal) emerge along disturbance gradients as habitat fragmentation increases. Furthermore, habitat loss outweighs habitat autocorrelation effects in highly disturbed scenarios, yet autocorrelation can shape species coexistence under low disturbance. These findings underscore the need to integrate spatial and temporal mechanisms in biodiversity management.

Surveillance of diseases in a pandemic is an important part of public health policy. Diagnostic testing at the individual level is often infeasible due to resource constraints. To circumvent these constraints, group testing can be applied. The economic cost evaluation from the payer's perspective typically focuses only on deterministic costs which overlooks the substantial economic impact of productivity losses resulting from quarantine and workplace disruptions. The objective of this article is to develop a mathematical model for a retrospective economic evaluation of group testing that incorporates both deterministic costs and income-based economic loss. Group testing algorithms are revisited and simulated at optimized pool sizes to determine the required number of tests. Income data from the German Socio-Economic Panel are integrated into a mathematical model to capture the economic loss. Afterward, hybrid Monte Carlo experiments are conducted by evaluating the economic cost in the Coronavirus disease 2019 pandemic in Germany. Monte Carlo experiments show that the optimal choice of group testing algorithms changes substantially when income-based economic losses are included. Evaluations considering only deterministic costs systematically underestimate the total economic cost. Algorithms with a longer quarantine duration are less attractive than shorter quarantine duration if income-based economic loss is accounted for. The findings show that current evaluations underestimate the true economic cost. Group testing algorithms with shorter duration and fewer stages are preferred, even when they require a larger number of tests. These results underscore the importance of incorporating income-based economic loss into a mathematical model.

The propagation of light through a turbulent flow field around an aircraft results in optical distortions commonly known as aero-optic effects. The development of methods to mitigate these effects requires large amounts of realistic aero-optic data. However, methods for obtaining this data, including experiment, computational fluid dynamics, and simple phase screen algorithms (e.g., boiling flow), each have significant drawbacks such as high cost, high computation, limited quantity, and/or inaccurate statistics. More recently, data-driven algorithms have been proposed that are computationally efficient and can synthesize aero-optic data to match the statistics of measured data, but these approaches still have drawbacks including limited quality, inaccurate statistics, and the use of complicated algorithms. In this paper, we introduce ReVAR (Re-whitened Vector AutoRegression), a data-driven algorithm for generating synthetic aero-optic data that matches the statistics of measured data. A key contribution in this algorithm is Long-Range AutoRegression, a linear predictive model that combines a standard autoregression with a set of low-pass filters of the data to fit both short-range and long-range temporal statistics. ReVAR uses Long-Range AR together with a spatial re-whitening step to convert measured aero-optic data to temporally and spatially un-correlated white noise. ReVAR can then generate synthetic aero-optic data by reversing this process using white noise input. Using two measured turbulent boundary layer data sets, we demonstrate that ReVAR better matches the measured data's temporal power spectrum and other key metrics than do two conventional phase screen generation methods and an existing single time-lag autoregressive model.

Orthogonal time frequency space (OTFS) modulation offers superior robustness to high-mobility channels compared to conventional orthogonal frequency-division multiplexing (OFDM) waveforms. However, its explicit delay-Doppler (DD) domain representation incurs substantial signal processing complexity, especially with increased DD domain grid sizes. To address this challenge, we present a scalable, real-time Zak-OTFS receiver architecture on GPUs through hardware--algorithm co-design that exploits DD-domain channel sparsity. Our design leverages compact matrix operations for key processing stages, a branchless iterative equalizer, and a structured sparse channel matrix of the DD domain channel matrix to significantly reduce computational and memory overhead. These optimizations enable low-latency processing that consistently meets the 99.9-th percentile real-time processing deadline. The proposed system achieves up to 906.52 Mbps throughput with a DD grid size of (16384,32) using 16QAM modulation over 245.76 MHz bandwidth. Extensive evaluations under a Vehicular-A channel model demonstrate strong scalability and robust performance across CPU (Intel Xeon) and multiple GPU platforms (NVIDIA Jetson Orin, RTX 6000 Ada, A100, and H200), highlighting the effectiveness of compute-aware Zak-OTFS receiver design for next-generation (NextG) high-mobility communication systems.

Fluid antenna systems (FAS) provide extra position agile spatial diversity for integrated sensing and communication (ISAC), by jointly optimizing the port selection and precoding. However, this optimization is challenging in air ground networks due to the intricate dual objective Pareto frontier, complex self-interference, and prohibitive channel state information overhead. To overcome these bottlenecks, this work proposes a novel grey box multi objective Bayesian optimization framework to address the joint design of discrete port selection and ISAC precoding. Unlike black box methods, this architecture explicitly leverages known physical system models to learn unknown channel constituents, dramatically reducing sample complexity. To navigate high dimensional combinatorial spaces, an adaptive trust region mechanism powered by expected hypervolume improvement (EHI) acquisition is implemented. Furthermore, the framework incorporates a spatio-temporal tracking strategy to handle the continuous mobility of users and targets, robustly capturing the drifting optimum in time varying environments. Simulations demonstrate that this framework achieves significantly faster convergence and discovers superior Pareto optimal configurations, validating its efficiency for dynamic real time FAS-ISAC deployments.

Millimeter wave (mmWave) technology at 28 GHz is vital for beyond-5G systems, but indoor deployment remains challenging due to limited statistical evidence on propagation. This study investigates path loss, material penetration, and coverage enhancement using TMYTEK-based measurements. Statistical tests and confidence interval analysis show that path loss aligns with free-space theory, with an exponent of n = 2.07 plus or minus 0.073 (p = 0.385), confirming the suitability of classical models. Material analysis reveals significant variation: desk dividers introduce 3.4 dB more attenuation than display boards (95 percent CI: 1.81 to 4.98 dB, p less than 0.01), contradicting thickness-based assumptions. Reflector optimisation yields a significant mean gain of 2.17 plus or minus 2.33 dB (p less than 0.05), enhancing coverage. The results provide new empirical benchmarks and practical design insights for reliable indoor mmWave deployment.

This paper investigates the passive enhancement of MIMO spectral efficiency through boundary engineering in a simplified two dimensional indoor proof of principle model. The propagation channel is constructed from the electromagnetic Green's function of a room with boundaries modeled as free space, drywall, perfect electric conductor (PEC), or binary gratings. Within this framework, grating coated walls enrich the non line of sight (NLoS) multipath field, reduce channel correlation, and enhance spatial multiplexing over a broad range of receiver locations. Comparisons with the drywall and PEC reference cases further reveal that the observed capacity enhancement arises not from diffraction alone, but from the combined effects of effective wall reflectivity, which confines and reradiates energy within the room, and diffraction induced angular redistribution, which enriches the channel eigenstructure.

Continuous aperture arrays (CAPAs) have emerged as a promising physical-layer paradigm for sixth generation (6G) systems, offering spatial degrees of freedom beyond those of conventional discrete antenna arrays. This paper investigates the interaction between the CAPA receive architecture and low-cost 1-bit analog-to-digital converters (ADCs), which impose a severe nonlinear distortion penalty in conventional discrete systems. For Rayleigh fading, we derive a moment matching approximation (MMA)-based closed-form symbol error probability (SEP) approximation based on Gamma moment-matching of the spatial eigenvalue distribution, and show that CAPAs incur a diversity-order penalty governed by Jensen's inequality on the mode eigenvalues. For line-of-sight (LoS) propagation, we prove that CAPA achieves exactly the unquantized additive white Gaussian noise (AWGN) performance bound under perfect spatial and phase alignment, completely eliminating the 1-bit penalty that forces discrete systems to double their antenna count. Monte Carlo simulations under Rayleigh, Rician, and LoS conditions validate all analytical results.

Millimeter-wave and terahertz communications face a fundamental challenge: overcoming severe path loss without sacrificing spectral efficiency. Pinching antenna systems (PASS) address this by bringing radiators physically close to users, yet existing frameworks treat the waveguide as a mere transmission line, overlooking its inherent multi-mode capabilities and the critical role of polarization. This paper develops the first polarization-aware, full-wave electromagnetic model for multi-mode PASS (MMPASS), capturing spatial radiation patterns, modal polarization states, and polarization matching efficiency from first principles. Leveraging this physically grounded model, we reveal fundamental trade-offs among waveguide attenuation, atmospheric absorption, and geometric spreading, yielding closed-form solutions for optimal PA placement and orientation in single-user scenarios. Extending to multi-user settings, we propose a modular optimization framework that integrates fractional programming with closed-form polarization updates, scaling gracefully to arbitrary numbers of waveguides, PAs, and users. Numerical results show that MMPASS achieves up to a 167% increase in spectral efficiency compared with single-mode PASS. Moreover, when comparing MMPASS with its polarization-ignorant counterpart, polarization awareness alone improves the sum rate by up to 23%. By bridging rigorous electromagnetic theory with scalable optimization, MMPASS establishes a physically complete and practically viable foundation for future high-frequency wireless networks.

In urban environments, vehicle-to-everything (V2X) communications require accurate wireless channel characterization. This requirement is particularly critical at street-canyon intersections, where building blockage and rich multipath propagation can severely degrade link reliability. Due to its unique environmental layout, the channel characteristics in urban canyon are influenced by building distribution. However, this feature has not been well captured in existing channel models. In this paper, we propose an environment-related statistical channel model based on 5.8~GHz channel measurements. We construct a composite environmental factor to characterize environmental differences in intersections. Then, the factor is incorporated into 3GPP path-loss model and further linked to small-scale channel parameters. Finally, accuracy of the proposed model is validated using second-order channel statistics. The results show that the proposed model can effectively characterize propagation properties of urban street-canyon intersection channels with different building conditions. The proposed model provides a physically interpretable and statistically effective framework for channel simulation and performance evaluation in urban vehicular scenarios.

The deployment of multiple transmitters (TXs) in integrated sensing and communication (ISAC) networks necessitates efficient resource sharing to overcome the limitations of orthogonal allocation. While conventional interleaved (CI) pilots combined with non-orthogonal multiple access (NOMA) improve spectral efficiency (SE), they inherently compromise sensing resolution due to spectral sparsity, rendering the CI nulling (CIN) extension a strictly limited remedy. This paper proposes a phase-shifted (PS) pilot design and its novel PS nulling (PSN) variant to integrate a communication TX (CTX) over the PS-ISAC framework. The PSN variant strategically punctures sensing signals at CTX pilot locations to preserve initial channel estimates, enabling a dense data overlay. To resolve the resulting multi-TX interference, joint iterative interference cancellation (IIC) is adapted for non-nulling configurations and sequential IIC is adapted for nulling variants, optimizing for both detection robustness and convergence speed. Simulation results across varying STX densities and modulation orders demonstrate that the phase-shifted frameworks maintain sensing integrity while explicitly reducing receiver-side computational complexities by $18.8\%$ and $21.0\%$ against their respective interleaved baselines.

To meet the requirements for high data rates and ubiquitous connectivity in 6G networks, higher frequencies and larger array apertures are employed to enhance spatial resolution and spectral efficiency. This evolution leads to an expansion of the near-field region, where spherical-wave focusing can significantly enhance received power. However, the pervasive presence of obstacles in near-field environments makes communication in obstructed scenarios a critical challenge, particularly for sensitive high-frequency links with high penetration losses. In this paper, we propose a new waveform, termed the near-field Airy beam, which is tailored to the amplitude and phase characteristics of obstructed near-field channels. By integrating non-uniform amplitude response with non-linear phase profile, the proposed Airy beam forms specific curved trajectories, energy distributions, and focal points, enabling energy concentration at the user even after circumventing obstacles. An Airy beamforming algorithm is also developed for hybrid beamformer architectures. Considering practical conditions with unknown obstacle and user locations, we design an Airy beam codebook and a low-overhead hierarchical search scheme to identify the optimal user-aligned beam. Simulation results demonstrate that in obstructed environments, the near-field Airy beam achieves a received power gain of over 3 dB compared to conventional waveforms like focused and curved beams, closely approaching the theoretical upper bound. Across the mmWave to THz bands and various obstacle dimensions, the proposed beam training scheme consistently outperforms traditional methods in terms of spectral efficiency while maintaining a comparable training overhead.

Signal gating combined with local-oscillator-frequency switching enables bandwidth scaling of offline coherent reception without costly receiver parallelization. We experimentally verify this concept at symbol rates of up to 288 GBaud.

Uncrewed Aerial Vehicle (UAV) networks require accurate Air-to-Air (A2A) channel models, but most existing work focuses on Air-to-Ground links and leaves the sub-6 GHz A2A channel poorly characterized. We present preliminary 3.4 GHz A2A channel measurements collected with a lightweight, reconfigurable, open-source channel sounder built from USRP B210 software-defined radios and a high-precision GNSS-disciplined oscillator mounted on two UAVs. Measurements were conducted at the AERPAW Lake Wheeler testbed using a spherical flight trajectory around a second drone to capture channel behavior over varying altitudes, elevation angles, and relative headings. From these data, we analyze fundamental channel properties, extract channel impulse responses, model fading behavior as a function of link geometry, and characterize fading statistics including RMS delay spread. The resulting dataset and analysis provide a more realistic basis for the design, emulation, and evaluation of physical-layer and MAC protocols for next-generation UAV communication networks.

In Earth remote sensing, spatial-frequency domain visibility samples are inversely transformed into spatial-domain brightness temperature (BT) images through the signal processing pipeline of synthetic aperture interferometric radiometers (SAIR). However, L-band radio-frequency interference (RFI) contaminates the measured visibilities and severely degrades BT image quality, thereby impairing geophysical parameter retrieval. To address this issue, we propose VFDM, a Visibility-Function Diffusion Model based on Denoising Diffusion Probabilistic Models (DDPM), to mitigate RFI in the spatial-frequency domain while preserving fine-scale structures consistent with natural scene statistics. Furthermore, we construct a comprehensive dataset comprising more than ten thousand pairs of RFI-free natural scene visibility sample sets and their corresponding simulated contaminated counterparts, categorized by varying RFI intensities, numbers, and distributions. Finally, comprehensive experiments on both simulated and real-world data demonstrate the effectiveness and robustness of the proposed VFDM-based approach.

This paper revisits linear precoding, namely match-filter (MF) and zero-forcing (ZF), in a semantic multiple-input multiple-output (MIMO) system empowered by generative AI. The aim is to examine whether interference, channel state information (CSI) accuracy, and scalability limitations in conventional MIMO systems remain critical. Theoretical analysis, which is based on the generative inference model and Lipschitz continuous assumptions, reveals reduced sensitivity to interference and channel imperfections, as well as performance inferiority in high-SINR regimes compared to conventional MIMO systems. Simulation results validate the analysis and show that MF achieves semantic performance comparable to ZF under both perfect and imperfect CSI. These findings suggest that semantic MIMO relaxes the needs for aggressive interference mitigation and highly accurate CSI, while improving scalability with reduced computational and implementation complexity.

Future 6G networks will host massive numbers of embodied intelligent agents, which require real-time channel awareness over continuous-space for autonomous decision-making. By pre-obtaining location-specific channel state information (CSI), channel map can be served as a foundational world model for embodied intelligence to achieve wireless channel perception. However, acquiring CSI via measurements is costly, so in practice only sparse observations are available, leaving agents blind to channel conditions at unvisited locations. Meanwhile, purely model-driven channel maps can provide dense CSI but often yields unsatisfactory accuracy and robustness, while purely data-driven interpolation from sparse measurements is computationally prohibitive for real-time updates. To address these challenges, this paper proposes a data-model co-driven (DMcD) framework that performs a two-stage interpolation toward a space-time continuous channel map, First, a hybrid ray tracing and geometry-based channel model (H-RT/GBSM) is developed to capture dynamic scatterers, providing dense, time-variant channel properties that match measurement statistics as a physically consistent prior. Then, an inductive edge-conditioned graph neural network (InductE-GNN) fuses the prior with sparse measurements to perform real-time spatial interpolation, enabling rapid online adaptation without retraining, ensuring the synchronization with the dynamic physical reality. Evaluations with measured datasets show that the proposed DMcD framework significantly outperforms data-only and model-only baselines, providing accurate and queryable channel information for embodied intelligent agents.

The Chinese remainder theorem (CRT) provides an efficient way to reconstruct an integer from its remainders modulo several integer moduli, and has been widely applied in signal processing and information theory. Its multidimensional extension (MD-CRT) generalizes this principle to integer vectors and integer matrix moduli, enabling reconstruction in multidimensional signal processing scenarios. However, since matrices are generally non-commutative, the multidimensional extension introduces new theoretical and algorithmic challenges. When all matrix moduli are diagonal, the system is equivalent to applying the one-dimensional CRT independently along each dimension. This work first investigates whether non-diagonal (non-separable) moduli offer fundamental advantages over traditional diagonal ones. We show that under the same determinant constraint, non-diagonal matrices do not increase the dynamic range but yield more balanced and better-conditioned sampling patterns. More importantly, they generate lattices with longer shortest vectors, leading to higher robustness to vector remainder errors, compared to diagonal ones. To further improve the robustness, we develop a multi-stage robust MD-CRT framework that improves the robustness level without reducing the dynamic range. Due to the multidimensional nature and modulo matrix forms, it is challenging and not straightforward to extend the existing one-dimensional multi-stage robust CRT. In this paper, we obtain a new condition for matrix moduli, which can be easily checked, such that a multi-stage robust MD-CRT can be implemented. Both theoretical analysis and simulation results demonstrate that the proposed multi-stage robust MD-CRT achieves stronger error tolerance and more reliable reconstruction under erroneous vector remainders than that of single-stage robust MD-CRT.

With the rapid expansion of low-altitude economy (LAE) services and the growing demand for integrated sensing and communication (ISAC) in air-ground networks, reliable direction-of-arrival (DOA) estimation has become essential for both directional communication and sensing functions. DOA underpins beam alignment, spatial-reuse scheduling, and ISAC-critical tasks such as airspace situational awareness and multi-target monitoring. Hybrid analog-digital (HAD) architectures have emerged as a practical solution for large-aperture directional operation under stringent radio frequency (RF), analog-to-digital converter (ADC), and size, weight, and power (SWaP) constraints. However, HAD compresses antenna-domain observations through analog combining, fundamentally reshaping the measurement model and introducing new algorithmic and system-level challenges for DOA estimation. This article first reviews the principles and representative architectures of HAD, highlighting their advantages for scalable beam-centric and ISAC-oriented operation in LAE scenarios. We then provide a structured overview of HAD-enabled DOA estimation methodologies, including spatial covariance matrix (SCM) reconstruction, multi-combiner scan-based acquisition, and pilot-aided estimation, along with key design tradeoffs. Finally, we discuss open challenges and outline reliability-driven research directions toward robust, deployable HAD-enabled DOA solutions for practical ISAC-enabled low-altitude environments.

We develop a unified Fisher-information framework for localization in environments with both Line-of-Sight (LOS) and Non-Line-of-Sight (NLOS) paths, focusing on diffraction-dominated NLOS propagation characteristic of Outdoor-to-Indoor (O2I) signal propagation. The model couples anchor geometry with a physically grounded path-loss law that is continuous across the LOS/NLOS boundary and serves as an optimization objective for our optimal anchor placement problem. As the first step, we analyze single-target anchor placement and derive the classical A-, D-, and E-optimality criteria. Under a specific path-loss assumption, these criteria collapse to a polygon-closure condition in the complex plane: A-, D-, and E-optimal designs coincide, yielding necessary and sufficient conditions for optimal placement. Next, we extend the notion of optimal anchor placement with respect to a single target to optimality over a feasible region (multi-target setting) using a general formulation that explicitly includes a realistic path loss model. This is achieved by recasting the anchor placement as a combinatorial anchor-selection problem with provable guarantees. Next, we specify E- and D-optimal objectives over multiple targets in a predefined feasible target region and show that E-optimality straddles A-optimality (within a constant factor), while D-optimality provides looser bounds. These insights yield two practical algorithms, both mixed-integer second-order cone programs (MISOCP) with exact E-optimal and exact D-optimal objectives that produce robust, region-wide designs under mixed LOS/NLOS conditions.

The assessment of unwanted radiated emissions from Active Antenna Systems (AAS) has become a critical issue in adjacent-band coexistence scenarios. In this paper, we establish the existence of a deterministic spatial upper bound on the radiated power of active antenna arrays. We show that the maximum radiated power always occurs in the boresight direction, irrespective of frequency or signal nature (useful signal, nonlinear distortion, or noise), or instantaneous beamforming configuration, thereby defining a conservative spatial upper bound whose angular envelope is solely determined by the elementary radiating building block of the antenna architecture, i.e., the element or sub-array radiation pattern. Starting from a two-element array with third-order nonlinearities, we derive the spatial envelope and extend the result to realistic AAS architectures. The theoretical findings are validated by over-the-air (OTA) measurements performed on a 3.5 GHz Massive Multiple-Input Multiple-Output (MIMO) antenna. The proposed approach offers a simple, robust, and measurement-oriented methodology for coexistence assessments involving beamformed radio systems.

We report the design and experimental and simulated performance for a 2050 nm band fiber amplifier with high optical-optical slope efficiency and low ion pairing, using a novel high performance single clad Ho-doped fiber from the Naval Research Laboratory (NRL). We measure an optical-optical slope efficiency of 57% using 1 mW input signal power and 1860 nm pumping which we believe is the highest slope efficiency obtained to date for a single clad single stage copumped HDFA. A new method for non-destructive measurement of the ion pairing coefficient in Ho-doped fibers is introduced and validated. Using this method, we link our 57% slope efficiency to a low ion pairing coefficient of 4% in the NRL Ho-doped fiber as derived from our experimental data. We present an overview and survey of the ion pairing results for Ho-doped fiber amplifiers and lasers reported so far in the literature.

In this work, we consider end-to-end calibration of an integrated sensing and communication (ISAC) base station (BS) under gain-phase and antenna displacement impairments without collecting signals from predefined positions (labeled data). We consider a BS with two impaired uniform linear arrays used for simultaneous multi-target sensing and communication with a user equipment (UE) leveraging orthogonal frequency-division multiplexing signals. The main contribution is the design of a framework that can compensate for the impairments without labeled data and considering coherent receive signals. We harness a differentiable precoder based on the maximum array response in an angular direction at the transmitter and the orthogonal matching pursuit (OMP) algorithm at the sensing receiver. We propose an ISAC loss as a combination of sensing and communication losses that provides a trade-off between the two functionalities. We compare two sensing objective alternatives: (i) maximize the maximum response of the angle-delay map of the targets or (ii) minimize the norm of the residual signal at the output of the OMP algorithm after all estimated targets have been removed. The communication objective maximizes the energy of the received signal at the UE. Additionally, our framework leverages an approximation of the channel gradient that avoids the impractical knowledge of the gradient of the channel. Our results show that the proposed method performs closely to using labeled data and knowledge of the channel gradient in terms of sensing position estimation and communication symbol error rate. When comparing the two sensing losses, minimizing the norm of the OMP residual yields significantly better sensing position estimation with slightly increased complexity.

Learning the graph Laplacian from observed data is one of the most investigated and fundamental tasks in Graph Signal Processing (GSP). Different variants of the Laplacian, such as the combinatorial, signless or signed Laplacians have been considered depending on the type of features to be extracted from the data. The main contribution of this paper is the introduction of a parametric Laplacian, called the deformed Laplacian, defined as a quadratic matrix polynomial that provides a parametric dictionary for graph signal processing. The deformed Laplacian can be interpreted as the generator of a parametric linear reaction-diffusion dynamics on graphs, capturing the interplay between diffusive coupling and nodal reaction effects. It is a parametric polynomial matrix that enables the design of novel topological operators tailored to both the underlying graph structure and the observed signals. Interestingly, we show that several Laplacian variants proposed in the literature arise as special cases of the deformed Laplacian. We then develop a method to jointly learn the deformed Laplacian and the graph signals from data, showing how its use improves signal representation across a broad class of graphs compared to standard Laplacian forms. Through extensive numerical experiments on both synthetic and real-world datasets, including financial and communication networks, we assess the benefits of the proposed method in terms of graph signal reconstruction error and sparsity of the representation.

This paper presents a phase-difference-based scheme for three-dimensional (3D) line-of-sight (LoS) user localization using a uniform planar array (UPA), applicable to both near-field and far-field regimes under the exact spherical-wave model. Unlike the previously studied two-dimensional (2D) uniform linear array (ULA) case, the 3D UPA case requires jointly exploiting the two array axes in order to recover the user's range, azimuth, and zenith angle. Adjacent-antenna phase-differences are first estimated from uplink pilots and then summed along the array axes to obtain unwrapped phase-differences between widely separated antenna elements. These summed phase-differences enable the construction of multiple three-equation systems whose solutions yield the user's range, azimuth, and zenith angle. We quantify the number of such equation systems, provide a representative closed-form estimator that uses only three phase-difference sums, and propose an all-data nonlinear least-squares estimator that exploits all available sums. Numerical results show that the least-squares estimator, when initialized by the closed-form estimate, achieves Cramér--Rao bound accuracy. Moreover, unlike state-of-the-art baseline schemes, whose performance depends on well-tuned hyperparameters, the proposed estimators are hyperparameter-free.

High-Resolution three-dimensional (3D) radio maps (RMs) provide rich information about the radio landscape that is essential to a myriad of wireless applications in the future wireless networks. Although deep learning (DL) methods have shown their effectiveness in RM construction, existing approaches require massive high-resolution 3D RM samples in the training dataset, the acquisition of which is labor-intensive and time-consuming in practice. In this paper, our goal is to devise a data-friendly high-resolution 3D RM construction solution via training over a hybrid dataset, wherein the RMs associated with a small fraction of environment maps (EMs) are of high-resolution, while those corresponding to the majority of EMs are of low-resolution. To this end, we propose a Data-Friendly 3D Radio Map Estimator (DF-3DRME), which comprises two processing stages. Specifically, in the first stage, we leverage the abundant low-resolution 3D RM samples to train a neural network, termed the LR-Net, for predicting the low-resolution 3D RM from the input EM, which provides a coarse characterization of the spatial radio propagation. In the second stage, we employ an advanced super-resolution network, termed the SR-Net, to upscale the predicted low-resolution 3D RM to its high-resolution counterpart. Unlike the LR-Net, the SR-Net can be effectively trained with only the limited high-resolution 3D RM samples available in the hybrid dataset. Experimental results demonstrate that the proposed framework achieves compelling reconstruction performance with only 4% of the EMs in the dataset having high-resolution 3D RM labels, which significantly reduces data acquisition overhead and facilitates practical deployment.

Semantic communications (SemCom) is a promising task-oriented paradigm in which semantic features exhibit non-uniform importance. Consequently, unequal error protection (UEP), which allocates resources based on semantic importance, plays a pivotal role in maximizing system utility. However, most existing schemes adopt passive importance evaluation, which neither proactively reshapes the importance distribution nor explores its impact on UEP performance. In this paper, we propose a novel importance-ordered semantic feature restructuring (ISFR) scheme that proactively enforces a descending importance hierarchy and jointly optimizes multi-dimensional resources to improve system utility. Specifically, modules with decreasing retention probabilities and increasing distortion levels are employed, which drive the model to concentrate key semantics into front-end features and thus strengthen importance differentiation. Moreover, a joint optimization problem that jointly optimizes channel matching, feature selection, modulation schemes, and power allocation is formulated to minimize the importance-weighted total semantic distortion. To solve this non-convex problem, a hierarchical decoupling strategy is proposed, which decomposes it into four tractable subproblems. This approach leverages the ordered prior to drastically prune the search space for feature selection and modulation, while integrating greedy-based channel matching and convex power allocation. Simulation results demonstrate that the proposed ISFR scheme outperforms traditional uniform importance-based schemes under harsh channel conditions and limited resources, validating the significant robustness improvement enabled by the concentration of key semantic information.

This paper investigates the performance of a pinching-antenna (PA) system with a signal waveguide and multiple pinching antennas to serve users distributed across multiple rooms. The performance of the system is evaluated through a comparative analysis under both orthogonal multiple access (OMA) and non-orthogonal multiple access (NOMA) schemes. Specifically, this paper derives closed-form expressions for the outage probability (OP) and ergodic rate (ER) in each scheme. Furthermore, asymptotic analyses are conducted to characterize the system behavior in the high signal-to-noise ratio (SNR) regime. Extensive Monte Carlo simulations are utilized to validate the accuracy of the analytical derivations. The comparative results can be summarized as follows: 1) in the downlink fixed-rate scenario, whether OMA or NOMA achieves better outage performance depends on system parameters, such as the number of users and power allocation coefficients; 2) in the uplink fixed-rate scenario, the outage performance of NOMA is inferior to that of OMA in the high-SNR regime, and the decay rate of the OP for NOMA users depends on the rate thresholds; and 3) for both uplink and downlink adaptive-rate scenarios, the rate performance comparison of the two schemes depends on system parameters in the low-SNR regime, whereas OMA generally outperforms NOMA in the high-SNR regime.

Imaging is a crucial sensing function that finds wide applications in environmental reconstruction, autonomous driving, etc. However, the signal processing methods for existing radio imaging techniques, such as millimeter wave (mmWave) imaging, require high-resolution range estimation enabled by Gigahertz-level or even Terahertz-level bandwidth, and cannot be applied in 6G integrated sensing and communication (ISAC) network with Megahertz-level bandwidth. This paper proposes two novel high-resolution radio imaging schemes that can work on the 6G signals with limited bandwidth - bandwidth-independent synthetic aperture radar (BI-SAR), where the movable base station (BS) revolves along the static targets by 360 degrees; as well as bandwidth-independent inverse synthetic aperture radar (BI-ISAR), where the BS is static and the targets revolve along an axis by 360 degrees. Different from conventional SAR and ISAR counterparts that rely on range estimation, our proposed imaging schemes solely utilize Doppler information to perform imaging without any range information. The main technical challenge of our schemes lies in the anisotropic scattering functions over different directions, which hinder the coherent synthesis of the backscattered signals from all directions. We design an iterative adaptive approach-based Doppler association (IAA-DA) algorithm to tackle the above issue. Moreover, we also derive the imaging resolution to characterize the reconstruction quality. Real-world experiments are provided to show the feasibility and the effectiveness of our proposed 6G imaging schemes.

Pinching-antenna (PA) has recently attracted considerable research attention in wireless systems, realized by attaching small dielectric particles along a waveguide. Building upon which, the segmented waveguide-enabled pinching-antenna system (SWAN) has been proposed to mitigate the inter-antenna radiation problem in uplink transmissions of conventional PA systems. In this work, SWAN-assisted integrated sensing and communication (ISAC) is investigated, where a base station (BS) equipped with SWAN provides downlink communications for multiple communication users (CUs) and performs sensing for multiple targets. The dual-functional signals transmitted by the BS are radiated by the SWAN, and the echo signals reflected by the targets are captured by the SWAN and relayed to the BS for estimating the locations of the targets. We formulate a Cramér-Rao lower bound (CRLB) minimization problem to evaluate the performance of the ISAC system, where the CRLB of the location estimation is minimized under communication rate constraints. To jointly optimize the beamforming and the PA positions of the SWAN, we develop a Riemannian manifold optimization (RMO) method, where each variable is constrained on its corresponding Riemannian manifold, and a Riemannian product manifold (RPM) is constructed as the solution space. A penalty method combined with Riemannian Broyden-Fletcher-Goldfarb-Shanno (RBFGS) algorithm is applied to obtain a feasible solution. Simulation results show that the proposed SWAN-assisted ISAC system yields superior CRLB performance for target localization compared with existing schemes including the multi-waveguide-enabled pinching-antenna-assisted ISAC systems.

Reconfigurable intelligent surfaces (RISs) are often assumed to allow continuous phase control over all elements, leading to hardware cost that scales with the number of elements. Treating the phase of each element as a discrete variable is essential for improving cost effectiveness toward ubiquitous RIS deployment. However, the resulting discrete optimization problem is inherently difficult to solve. To address this challenge, this letter proposes a two-dimensional line-control method to reduce the degrees of freedom of the phase variables. The formulation yields a fourth-order objective function and is not directly compatible with physical optimizers such as coherent Ising machines and quantum annealers, which are designed for quadratic interactions. Conventional methods for reducing the order of the objective function with additional auxiliary variables increase the number of variables and require additional penalty parameters, limiting scalability. We therefore propose a two-step optimization method that transforms the fourth-order objective into two successive quadratic optimization problems. For a RIS with 5,476 elements, the required number of discrete variables is reduced from 11,100 to 5,476. Experiments using a real coherent Ising machine demonstrated that the proposed approach solved the discrete-phase optimization problem with 5,476 elements, while limiting the beamforming-gain loss to 2 dB compared with the full continuous-control case.

Unlike fixed-position arrays with static observation entropy, the scalable fluid antenna system (S-FAS) can dynamically adjust its aperture to form different observation spaces with configuration-dependent entropy budgets. This reconfigurability requires an information-theoretic framework beyond traditional algebraic identifiability analysis. This paper establishes an observation entropy framework for S-FAS, which unifies the derivation of identifiability limits, the diagnosis of processing bottlenecks, and system design optimization. For an S-FAS with mutual coupling suppression, we derive a complete capacity hierarchy among compressed, extended, and jointly stacked configurations. The entropy framework reveals that sequential two-stage processing suffers from an information bottleneck that restricts achievable capacity, while the noise entropy ratio can be used to distinguish fundamental performance limits from algorithmic deficiencies. A joint MUSIC algorithm is proposed to approach the theoretical joint capacity bound. Extensive Monte Carlo simulations, validated by both algebraic and information-theoretic criteria, verify the derived capacity hierarchy and identifiability boundaries.

The coexistence of heterogeneous cellular standards (2G-5G) in shared spectrum demands sophisticated RF source separation techniques, yet no public dataset exists for data-driven research on this problem. We present RFSS (RF Signal Source Separation), an open-source dataset of 100,000 multi-source RF signal samples generated with full 3GPP standards compliance. The dataset covers GSM (TS 45.004), UMTS (TS 25.211), LTE (TS 36.211), and 5G NR (TS 38.211), with 2-4 simultaneous sources per sample plus 4,000 single-source reference samples, at 30.72 MHz sample rate. Each sample passes through independent 3GPP TDL multipath fading channels and realistic hardware impairments: carrier frequency offset, I/Q imbalance, phase noise, DC offset, and PA nonlinearity (Rapp model). Two mixing modes are provided: co-channel (all sources at baseband) and adjacent-channel (each source frequency-shifted to its standard-specific carrier). The dataset totals 103 GB in HDF5 format with a 70/15/15 train/validation/test split. We benchmark five methods: FastICA, Frobenius-norm NMF, Conv-TasNet, DPRNN, and a CNN-LSTM baseline, evaluated using permutation-invariant SI-SINR (PI-SI-SINR). Conv-TasNet achieves -21.18 dB PI-SI-SINR on 2-source mixtures versus -34.91 dB for ICA, a 13.7 dB improvement. On co-channel mixtures, Conv-TasNet reaches -12.34 dB versus -28.04 dB for ICA and -16.19 dB for NMF. The dataset and evaluation code are publicly released at submission time.

mmWave radar enables human sensing in non-visual scenarios-e.g., through clothing or certain types of walls-where traditional cameras fail due to occlusion or privacy limitations. However, robust anomaly detection with mmWave remains challenging, as signal reflections are influenced by material properties, clutter, and multipath interference, producing complex, non-Gaussian distortions. Existing methods lack contextual awareness and misclassify benign signal variations as anomalies. We present mmAnomaly, a multi-modal anomaly detection framework that combines mmWave radar with RGBD input to incorporate visual context. Our system extracts semantic cues-such as scene geometry and material properties-using a fast ResNet-based classifier, and uses a conditional latent diffusion model to synthesize the expected mmWave spectrum for the given visual context. A dual-input comparison module then identifies spatial deviations between real and generated spectra to localize anomalies. We evaluate mmAnomaly on two multi-modal datasets across three applications: concealed weapon localization, through-wall intruder localization, and through-wall fall localization. The system achieves up to 94% F1 score and sub-meter localization error, demonstrating robust generalization across clothing, occlusions, and cluttered environments. These results establish mmAnomaly as an accurate and interpretable framework for context-aware anomaly detection in mmWave sensing.

Offline reinforcement learning (RL) is an attractive tool for unmanned aerial vehicle (UAV) systems, where online exploration is costly and raises safety concerns. In terrain-aware UAV relaying, agents may observe high-dimensional inputs such as terrain and land-cover maps, which describe the propagation environment, but complicate offline learning from fixed datasets. This paper investigates the impact of compact state representations on offline RL for UAV relaying. End-to-end service is jointly constrained by UAV--user access links and a base-station--to--UAV backhaul link, yielding feasibility limits driven by user mobility and independent of UAV control. To distinguish feasibility limits from control-induced sub-optimality, a candidate-set feasibility upper bound (CS-FUB) is introduced, which estimates the maximum achievable user coverage over a restricted set of UAV placements. To address high-dimensional terrain context, map-like observations are compressed into low-dimensional latent representations using a variational autoencoder (VAE) and policies are trained via Conservative Q-Learning (CQL). Simulation results show that training CQL directly on raw high-dimensional terrain-context states leads to slow convergence and large feasibility gaps. In contrast, VAE-encoded representations improve learning stability, enable earlier convergence to feasible relay configurations, and reduce sub-optimality relative to physical limits. Comparisons with autoencoder and linear compression baselines further demonstrate the benefit of structured representation learning for effective offline RL in terrain-aware UAV systems.

The system identification capabilities of a novel information-theoretic method are examined here. Specifically, this work uses information-theoretic metrics and vibration-based measurements to enhance damping estimation accuracy in mechanical systems. The method refers to a key limitation in system identification, signal processing, monitoring, and alert systems. These systems integrate various components, including sensors, data acquisition devices, and alert mechanisms. They are designed to operate in an environment to calculate key parameters such as peak accelerations and duration of high acceleration values. The current operational modal identification methods, though, suffer from limitations related to obtaining poor damping estimates due to their empirical nature. This has a significant impact on alert warning systems. This occurs when their duration is misestimated; specifically, when using the vibration amplitudes as an indicator of danger alerts for monitoring systems in damage or anomaly detection scenarios. To this end, approaches based on the Shannon entropy and the Kullback-Leibler divergence concept are proposed. The primary objective is to monitor the vibration levels in near real-time and provide immediate alerts when predefined thresholds are exceeded. In considering the proposed approach, both new real-world data from the multi-axis simulation table at the University of Bath, as well as the benchmark International Association for Structural Control-American Society of Civil Engineers (IASC-ASCE) structural health monitoring problem are considered. Importantly, the approach is shown to select the optimal model, which accurately captures the correct alert duration, providing a powerful tool for system identification and monitoring.

We address the problem of localizing multiple sources in 3D by combining sensor array measurements with camera observations. We propose a fusion framework extending the covariance matrix fitting method with an unbalanced optimal transport regularization term that softly aligns sensor array responses with visual priors while allowing flexibility in mass allocation. To solve the resulting largescale problem, we adopt a greedy coordinate descent algorithm that efficiently updates the transport plan. Its computational efficiency makes full 3D localization feasible in practice. The proposed framework is modular and does not rely on labeled data or training, in contrast with deep learning-based fusion approaches. Although validated here on acoustic arrays, the method is general to arbitrary sensor arrays. Experiments on real data show that the proposed approach improves localization accuracy compared to sensor-only baselines.

This paper studies an unmanned aerial vehicle (UAV) position and attitude sensing problem, where a base station equipped with an antenna array transmits signals to a predetermined potential flight region of a flying UAV, and exploits the reflected echoes for wireless imaging. The UAV is represented by an electromagnetic point cloud in this region that contains its spatial information and electromagnetic properties (EPs), enabling the unified extraction of UAV position, attitude, and shape from the reconstructed point cloud. To accomplish this task, we develop a generative UAV sensing approach. The position and signal-to-noise ratio embedding are adopted to assist the UAV features extraction from the estimated sensing channel under the measurement noise and channel variations. Guided by the obtained features, a conditional diffusion model is utilized to generate the point cloud. The simulation results demonstrate that the reconstructed point clouds via the proposed approach present higher fidelity compared to the competing schemes, thereby enabling a more accurate capture of the UAV attitude and shape information, as well as a more precise position estimation.

Future wireless systems increasingly require predictive and transferable representations that can support multiple physical-layer (PHY) tasks under dynamic environments. However, most existing supervised learning-based methods are designed for a single task, which leads to high adaptation cost. To address this issue, we propose a joint-embedding predictive architecture for multimodal sensing-assisted communications (JEPA-MSAC), a self-supervised multimodal predictive representation learning framework for wireless environments. The proposed framework first maps multimodal sensing and communication measurements into a unified token space, and then pretrains a shared backbone using temporal block-masked JEPA to learn a predictive latent space that captures environment dynamics and cross-modal dependencies. After pretraining, the backbone is frozen and reused as a general future-feature generator, on top of which lightweight task heads are trained for localization, beam prediction, and received signal strength indicator (RSSI) prediction. Extensive experiments show the latent state supports accurate multi-task prediction with low adaptation cost. Additionally, ablation studies reveal its scaling behavior and the impact of key pretraining setups.

RF fingerprinting authenticates satellite transmitters by exploiting hardware-specific signal impairments, yet existing methods operate without theoretical performance guarantees. We derive the Fisher information matrix (FIM) for joint estimation of in-phase/quadrature (IQ) imbalance and power amplifier (PA) nonlinearity parameters, establishing Cramér-Rao bounds (CRBs) whose structure depends on constellation moments. A necessary condition for full IQ identifiability is that the identifiability factor~$β$ exceeds zero; for binary phase-shift keying (BPSK), $β= 0$ yields a rank-deficient FIM, rendering IQ parameters unidentifiable. This provides a plausible theoretical explanation for OrbID's near-random performance (area under the ROC curve, AUC~$= 0.53$) on Orbcomm. From the FIM, we define a discrimination metric that predicts which hardware parameters dominate authentication for a given modulation. For constant-modulus PSK signals, PA nonlinearity features are predicted to dominate while IQ features are ineffective. We validate the framework on 24~Iridium satellites using two recording campaigns, achieving cross-file PA fingerprint correlation $r = 0.999$ and confirming all four CRB predictions. A discrimination-ratio-weighted (DR-weighted) authentication test achieves AUC~$= 0.934$ from six features versus $0.807$ with equal weighting, outperforming machine-learning classifiers (AUC~$\leq 0.69$) on the same data.

This vision paper addresses the research challenges of integrating traditional signal processing with Artificial Intelligence (AI) to enable energy-efficient, programmable, and scalable wireless connectivity infrastructures. While prior studies have primarily focused on high-level concepts, such as the potential role of Large Language Model (LLM) in 6G systems, this work advances the discussion by emphasizing integration challenges and research opportunities at the system level. Specifically, this paper examines the role of compact AI models, including Tiny and Real-time Machine Learning (ML), in enhancing wireless connectivity while adhering to strict constraints on computing resources, adaptability, and reliability. Application examples are provided to illustrate practical considerations and highlight how AI-driven signal processing can support next-generation wireless networks. By combining classical signal processing with lightweight AI methods, this paper outlines a pathway toward efficient and adaptive connectivity solutions for 6G and beyond.

Driven by the MagNet Challenge 2025 (MC2), increased research interest is directed towards modeling transient magnetic fields within ferrite material. An accurate time-resolved and temperature-aware H-field prediction is essential for optimizing magnetic components in applications with quasi-stationary / non-stationary excitation waveforms. Within the scope of this investigation, a selection of model structures with varying degrees of physically motivated structure are compared. Based on a Pareto investigation, a rather black-box gated recurrent unit (GRU) model structure with a graceful initialization setup is found to offer the most attractive model size vs. model accuracy trade-off, while the physics-inspired models performed worse. For a GRU-based model with only 325 parameters, a sequence relative error of 8.02 % and a normalized energy relative error of 1.07 % averaged across five different materials are achieved on unseen test data. With this excellent parameter efficiency, the proposed model won the first place in the performance category of the MC2.

While Third Generation Partnership Project (3GPP) has confirmed orthogonal frequency division multiplexing (OFDM) as the baseline waveform for sixth-generation (6G), its performance is severely compromised in the high-mobility scenarios envisioned for 6G. Building upon the GEARBOX-PHY vision, we present gear-switching OFDM (GS-OFDM): a unified framework in which the base station (BS) adaptively selects among three gears, ranging from legacy OFDM to delay-Doppler domain processing based on the channel mobility conditions experienced by the user equipments (UEs). We illustrate the benefit of adaptive gear switching for communication throughput and, finally, we conclude with an outlook on research challenges and opportunities.

We present a receiver-side framework for identifying amplitude distortions in frequency-selective OFDM channels. The core novelty is the use of the DCT Neuron, a compact adaptive processor based on the discrete cosine transform (DCT), to characterize the channel's nonlinear response, leveraging its properties for highly efficient estimation. Operating directly in the time domain, the method builds an accurate signal model and tracks channel variations adaptively, achieving reliable identification with as few as two OFDM symbols. The learned nonlinear response can then be exploited for predistortion and iterative decoding, enabling low-complexity, real-time adaptive compensation of complex responses in multicarrier systems.

Non-destructive testing using ultrasound is based on the interaction of sound waves with the object being tested and any defects it may contain. The aim is to extract as much information as possible about the object and its defects from the scattered wave field. In this paper, the concept of information in the context of ultrasonic testing is formalized and quantified physically for the first time. To this end, a balance equation for information is derived, analogous to Poynting's theorem for elastic energy. Various examples demonstrate how structural information is generated and annihilated within a component and along which pathways it travels from the defect to the sensor. Subsequently, the significance and potential of this new information concept for practical ultrasonic testing, structural health monitoring, numerical simulation, and machine learning are discussed. Finally, similarities and differences to mathematical Shannon information and statistical Fisher information are highlighted.

Integrated Sensing and Communication (ISAC) is a key enabler of high speed, ultra low latency vehicular communication in 6G. ISAC leverages radar signal processing (RSP) to localize multiple unknown targets amid static clutter by jointly estimating range, azimuth, and Doppler velocity (3D), thereby enabling highly directional beamforming toward intended mobile users. However, the speed and accuracy of RSP significantly impact communication throughput. This work proposes a novel 3D reconfigurable RSP accelerator, implemented on a Zynq Multi processor System on Chip (MPSoC) using a hardware software codesign approach and fixed point optimization. We propose two RSP frameworks: (1) high accuracy and high complexity, and (2) low complexity and low accuracy, along with their respective architectures. Then, we develop an adaptive architecture that dynamically switches between these two frameworks based on the signal to clutter plus noise ratio. This adaptive reconfiguration achieves up to 5.6 times faster RSP compared to state of the art designs. At the system level, the proposed RSP based ISAC delivers a 24% improvement in communication throughput without increasing hardware complexity.

Reconfigurable Intelligent Surfaces (RISs) are emerging as a key enabler of Programmable Wireless Environments for 6G, but their practical integration into operational networks still lacks orchestration mechanisms that can jointly support resource allocation, energy efficiency, and admission control with low online complexity. This paper presents SCROOGE, a physics-aware orchestration framework for multi-user RIS-assisted networks that operates on information generated offline during RIS codebook compilation, namely optimal codebook entries and per-element influence scores. Rather than relying on online optimization or idealized fading-based abstractions, SCROOGE exploits physics-derived descriptors to support low-latency operating-phase decisions that remain compatible with network-level control requirements. Specifically, SCROOGE introduces: i) an influence-aware, tier-consistent resource-allocation mechanism that combines user priority and element importance in the construction of a common RIS configuration; ii) an energy-efficiency mechanism that deactivates globally low-influence elements; and iii) an admission-control mechanism that accepts or rejects candidate users based on tier-aware compatibility with the currently deployed RIS state.

State-of-the-art optical wireless positioning (OWP) commonly reaches centimeter-level accuracy by depending on dense multi-light-emitting diodes (LED) infrastructures, photodiode (PD) arrays, or image-sensor receivers, incurring hardware complexity and deployment cost. This paper introduces a single beam-steered LED, single-PD OWP architecture that achieves three-dimensional (3D) localization without receiver rotation, cameras, or PD arrays; the core idea is to steer the transmitter through K known orientations and exploit the resulting received-signal-strength variations at the PD to estimate LED-to-PD direction and distance. We derive a composite Cramer-Rao lower bound and position-error bound (PEB) for the joint observation model, and cast the steering-pattern design as a genetic algorithm that minimizes the PEB over a 3D testbed. We develop both model-based a constrained nonlinear estimator and closed-form direction estimators: a statistically efficient generalized least squares solution, and a lightweight weighted least squares approximation. Simulations demonstrate centimeter-level accuracy for 3D OWP with a single beam-steered LED and a single PD.

Accurate characterization of Terahertz (THz) band path loss is critical for reliable high-frequency communication, especially in aerial networks where transceivers may operate at different altitudes. Existing THz-band path loss models for aerial networks focus on horizontal or vertical transceiver deployments, and fall short at modeling the random 3D geometry of transceiver locations. To address this limitation, we propose a new analytical THz path loss model that incorporates arbitrary 3D geometry of transceiver locations and frequency-selective absorption, obtained through a two-dimensional regression. We validate our proposed model with the propagation data collected via the Atmospheric Model (am) tool for multiple aerial link types, including drone-to-drone (Dr2Dr), medium-altitude aerial communication (MAAC), high-altitude unmanned aerial vehicles~(UAV)-to-UAV (U2U) links over varying transceiver separation and sub-THz to low-THz spectrum, i.e., 0.1--1~THz. The proposed framework provides a unified and accurate model for analyzing and designing future high-frequency aerial communication systems.

This survey examines intelligent forensics in next-generation mobile networks, arguing that future wireless security must move beyond real-time detection toward accountable post-incident reconstruction. Unlike traditional digital forensics, wireless investigations rely on short-lived, distributed, and heterogeneous evidence, including radio waveforms, channel measurements, device-side artifacts, and network telemetry, affected by calibration, timing uncertainty, privacy constraints, and adversarial manipulation. To address this limitation, this paper develops an evidence-centric framework that treats wireless measurements as first-class forensic artifacts and organizes the field through a unified taxonomy spanning physical-layer, device-layer, network-layer, and cross-layer forensics. We further systematize the forensic workflow into readiness and preservation-by-design, acquisition, correlation and analysis, and reporting and reproducibility, while comparing the complementary roles of traditional methods and artificial intelligence-assisted techniques. Subsequently, we review major application areas, including anomaly discovery, attribution, provenance and localization, authenticity verification, and timeline reconstruction. Finally, we identify key open challenges, including domain shift, resource-aware evidence capture, and the benefits and admissibility risks of generative evidence. Overall, this paper positions wireless forensics as a foundational capability for trustworthy, auditable, and reproducible security in next-generation wireless systems. Readers can understand and streamline wireless forensics processes for specific applications, such as low-altitude wireless networks, vehicular communications, and edge general intelligence.

Low Earth orbit (LEO) satellite downlinks are fundamentally limited by severe channel correlation: the line-of-sight (LoS)-dominant propagation and high orbital altitude confine users to a narrow angular region, rendering the multiuser channel matrix ill-conditioned. This paper provides a rigorous characterization of this limitation by exploiting the Vandermonde structure of the channel. Specifically, we link the minimum eigenvalue of the channel Gram matrix to user crowding through a balls-and-bins abstraction, and derive asymptotic sum rate scaling laws for both uniform linear arrays and uniform planar arrays. Our analysis reveals a sharp density threshold beyond which zero-forcing (ZF) precoding provably fails. To overcome this spatial multiplexing breakdown, we propose space-time adaptive beamforming (STAB), which exploits user-dependent residual Doppler shifts as an additional discrimination dimension. By constructing a time-extended channel in the joint space-Doppler domain, STAB restores a non-vanishing sum rate in regimes where purely spatial ZF collapses. We further develop a space-Doppler user selection (SDS) algorithm that leverages both spatial and Doppler separability for scheduling. Numerical results corroborate the analytical predictions and demonstrate that STAB with SDS achieves substantial sum rate gains over conventional methods in dense LEO downlink scenarios.

Tacho-less rotational speed estimation is critical for vibration-based prognostics and health management (PHM) of rotating machinery, yet traditional methods--such as time-domain periodicity, cepstrum, and harmonic comb matching--struggle under noise, non-stationarity, and inharmonic interference. Probabilistic tracking offers a principled way to fuse multiple estimators, but a major challenge is that heterogeneous estimators produce evidence on incompatible axes and scales. We address this with ARC (Alignment-based RPM Estimation with Curvature-adaptive Tracking) by unifying the observation representation. Each estimator outputs a one-dimensional evidence curve on its native axis, which is mapped onto a shared RPM grid and converted into a comparable grid-based log-likelihood via robust standardization and a Gibbs-form energy shaping. Standard recursive filtering with fixed-variance motion priors can fail under multi-modal or ambiguous evidence. To overcome this, ARC introduces a curvature-informed, state-dependent motion prior, where the transition variance is derived from the local discrete Hessian of the previous log-posterior. This design enforces smooth tracking around confident modes while preserving competing hypotheses, such as octave alternatives. Experiments on synthetic stress tests and real vibration-table data demonstrate stable, physically plausible trajectories with interpretable uncertainty, and ablations confirm that these gains arise from uncertainty-aware temporal propagation rather than per-frame peak selection or ad hoc rules.

Cell search engineers face significant challenge in reducing computation time to meet the requirements for fast initial access and radio link recovery. Since the majority of cell search time is consumed by Primary Synchronization Signal (PSS) detection, reducing the computational burden of this step is critical for shortening the overall procedure. This paper proposes a novel timing offset estimation scheme designed to accelerate 5G cell search. Leveraging the 5G Synchronization Signal Block (SSB) structure, the proposed scheme employs a two-step estimation process using dual-rate sampling. This approach effectively reduces the PSS detection search space without compromising the performance of subsequent processes. Performance evaluations in practical system and channel environments demonstrate that the proposed scheme reduces the cell search procedure time by 68\% compared to the baseline, while maintaining Physical Broadcast CHannel (PBCH) decoding performance.

Imagine advanced humanoid robots, powered by multimodal large language models (MLLMs), coordinating missions across industries like warehouse logistics, manufacturing, and safety rescue. While individual robots show local autonomy, realistic tasks demand coordination among multiple agents sharing vast streams of sensor data. Communication is indispensable, yet transmitting comprehensive data can overwhelm networks, especially when a system-level orchestrator or cloud-based MLLM fuses multimodal inputs for route planning or anomaly detection. These tasks are often initiated by high-level natural language instructions. This intent serves as a filter for resource optimization: by understanding the goal via MLLMs, the system can selectively activate relevant sensing modalities, dynamically allocate bandwidth, and determine computation placement. Thus, R2X is fundamentally an intent-to-resource orchestration problem where sensing, communication, and computation are jointly optimized to maximize task-level success under resource constraints. This survey examines how integrated design paves the way for multi-robot coordination under MLLM guidance. We review state-of-the-art sensing modalities, communication strategies, and computing approaches, highlighting how reasoning is split between on-device models and powerful edge/cloud servers. We present four end-to-end demonstrations (sense -> communicate -> compute -> act): (i) digital-twin warehouse navigation with predictive link context, (ii) mobility-driven proactive MCS control, (iii) a FollowMe robot with a semantic-sensing switch, and (iv) real-hardware open-vocabulary trash sorting via edge-assisted MLLM grounding. We emphasize system-level metrics -- payload, latency, and success -- to show why R2X orchestration outperforms purely on-device baselines.

In this work, a self-attention based conditional generative adversarial network (SA-cGAN) framework for the sixth generation (6G) semantic communication system is proposed, explicitly designed to balance the trade-off between distortion criticality and information representability under varying channel conditions. The proposed SA-cGAN model continuously learns compact semantic representations by jointly considering semantic importance, reconstruction distortion, and channel quality, enabling adaptive selection of semantic tokens for transmission. A knowledge graph is integrated to preserve contextual relationships and enhance semantic robustness, particularly in low signal-to-noise ratio (SNR) regimes. The resulting optimization framework incorporates continuous relaxation, submodular semantic selection, and principled constraint handling, allowing efficient semantic resource allocation under bandwidth and multi-constraint conditions. Simulation results show that, although SA-cGAN achieves modest syntactic bilingual evaluation understudy scores at low SNR to approximately 0.72 at 20 dB, it significantly outperforms conventional and JSCC-based schemes in semantic metrics, with semantic similarity, semantic accuracy, and semantic completeness consistently improving above 0.90 with SNR. Additionally, the model exhibits adaptive compression behavior, aggressively reducing redundant content while preserving critical semantic information to maintain fidelity. The convergence of training loss further validates stable and efficient learning of semantic representations. Overall, the results confirm that the proposed SA-cGAN model effectively captures distortion-invariant semantic representations and dynamically adapts transmitted content based on distortion criticality and information representability for meaning-centric communication in future 6G networks.

This paper presents a dual-domain low-complexity expectation propagation (EP) detection framework for affine frequency division multiplexing (AFDM) systems. By analyzing the structural properties of the effective channel matrices in both the time and affine frequency (AF) domains, our key observation is the domain-specific quasi-banded sparsity patterns, including AF-domain sparsity under frequency-selective channels and time-domain sparsity under doubly-selective channels. Based on these observations, we develop an AF-domain EP (EP-AF) detector for frequency-selective channels and a time-domain EP (EP-T) detector for doubly-selective channels, respectively. By performing iterative inference in the time domain using the Gaussian approximation, the proposed EP-T detector avoids inverting the dense channel matrix in the AF domain. Furthermore, the proposed EP-AF and EP-T detectors leverage the aforementioned quasi-banded sparsity of the AF domain and time domain channel matrices, respectively, to reduce the complexity of matrix inversion from cubic to linear order. Simulation results demonstrate that the proposed low-complexity EP-AF detector achieves nearly identical error rate performance to its conventional counterpart, while the proposed low-complexity EP-T detector offers an attractive trade-off between detection performance and complexity.

The long-term forecasting of electricity demand has been a prevalent research topic, primarily because of its economic and strategic relevance. Several machine learning as well as deep learning techniques have been developed in parallel with the growing complexity of the peak demand, planning for generation facilities and transmission augmentation in future. Most of these proposed techniques work on short-term forecasting as long-term forecasting is considerably more challenging due to unpredictable and unforeseeable variables that may arise in the future. This paper proposes a Temporal Fusion Transformer based deep learning approach for long term forecasting of peak power demand. The dataset used in this paper consists of peak power demand in India for a period of 6 years and the prediction was done for a period of 1 year. Our proposed model was compared with other popular forecasting models and it performed considerably better in benchmarks and was also more accurate in modelling the variance in the power demand.

The convergence of communication, sensing, and Artificial Intelligence (AI) in the Radio Access Network (RAN) offers compelling economic advantages through shared spectrum and infrastructure. How can inference and sensing be integrated in the RAN infrastructure at a system level? Current abstractions in O-RAN and 3GPP lack the interfaces and capabilities to support (i) a dynamic life cycle for inference and Integrated Sensing and Communication (ISAC) algorithms, whose requirements and sensing targets may change over time and across sites; (ii) pipelines for AI-driven ISAC, which need complex data flows, training, and testing; (iii) dynamic device and stack configuration to balance trade-offs between connectivity, sensing, and inference services. This paper analyzes the role of a programmable, software-driven, open RAN in enabling the intelligent edge for 5G and 6G systems. We identify real-time user-plane data exposure, open interfaces for plug-and-play inference and ISAC models, closed-loop control, and AI pipelines as elements that evolutions of the O-RAN architecture can uniquely provide. Specifically, we describe how dApps - a real-time, user-plane extension of O-RAN - and a hierarchy of controllers enable real-time AI inference and ISAC. Experimental results on an open-source RAN testbed demonstrate the value of exposing I/Q samples and real-time RAN telemetry to dApps for sensing applications.

While battery aging is commonly studied at the cell-level, evaluating aging and performance within battery modules remains a critical challenge. Testing cells within fully assembled modules requires hardware solutions to access cell-level information without compromising module integrity. In this paper, we design and develop a hardware testing platform to monitor and control the internal cells of battery modules contained in the Audi e-tron battery pack. The testing is performed across all 36 modules of the pack. The platform integrates voltage sensors, balancing circuitry, and a micro-controller to enable safe, simultaneous cell screening without disassembling the modules. Using the proposed testing platform, cell voltage imbalances within each module are constrained to a defined reference value, and cell signals can be safely accessed, enabling accurate and non-invasive cell-level state-of-health assessments. On a broader scale, our solution allows for the quantification of internal heterogeneity within modules, providing valuable insights for both first- and second-life applications and supporting efficient battery pack maintenance and repurposing.

Sensing and communication are fundamental enablers of next-generation networks. While communication technologies have advanced significantly, sensing remains limited to conventional parameter estimation and is far from fully explored. Motivated by these limitations, we propose semantic sensing (SemS), a novel framework that shifts the design objective from reconstruction fidelity to semantic effective recognition. Specifically, we mathematically formulate the interaction between transmit waveforms and semantic entities, thereby establishing SemS as a semantics-oriented transceiver design. Within this architecture, we leverage the information bottleneck (IB) principle as a theoretical criterion to derive a unified objective, guiding the sensing pipeline to maximize task-relevant information extraction. To practically solve this optimization problem, we develop a deep learning (DL)-based framework that jointly designs transmit waveform parameters and receiver representations. The framework is implemented in an orthogonal frequency division multiplexing (OFDM) system, featuring a shared semantic encoder that employs a Gumbel-Softmax-based pilot selector to discretely mask task-irrelevant resources. At the receiver, we design distinct decoding architectures tailored to specific sensing objectives, comprising a 2D residual network (ResNet)-based classifier for target recognition and a correlation-driven 1D regression network for high-precision delay estimation. Numerical results demonstrate that the proposed semantic pilot design achieves superior classification accuracy and ranging precision compared to reconstruction-based baselines, particularly under constrained resource budgets.

Fluid reconfigurable intelligent surfaces (FRIS) extend conventional RIS architectures by enabling physical reconfiguration of element positions, thereby introducing a fundamentally new degree of freedom for controlling spatial correlation and improving link reliability. Despite this promise, rigorous performance analysis of FRIS-assisted wireless systems has remained challenging, as exact statistical analyses of the end-to-end cascaded channels have been unavailable. This paper addresses this gap by providing the first exact closed-form characterization of the end-to-end cascaded channel gain in FRIS-aided systems under general spatial correlation. By exploiting the spectral structure of the FRIS-induced correlation matrix, we show that the channel gain statistics can be represented as a finite linear combination of K-distributions. This unified formulation naturally captures fully correlated, effectively decorrelated, and intrinsically uncorrelated operating regimes as special cases. Building on the derived channel statistics, we further obtain exact closed-form expressions for the outage probability and ergodic capacity. We also conduct an outage-based asymptotic analysis, which reveals the true diversity order of the system. Numerical results corroborate the proposed analytical framework via Monte Carlo simulations, benchmark its accuracy against state-of-the-art approximation-based approaches, and demonstrate that fluidic reconfiguration can yield tangible reliability gains by reshaping the spatial correlation structure.

Near-field propagation in extremely large aperture arrays requires joint angle-range estimation. In hybrid architectures, only $N_\mathrm{RF}\ll M$ compressed snapshots are available per slot, making the $N_\mathrm{RF}\times N_\mathrm{RF}$ compressed sample covariance the natural sufficient statistic. We propose the Curvature-Learning KL (CL-KL) estimator, which grids only the angle dimension and \emph{learns the per-angle inverse range} directly from the compressed covariance via KL divergence minimisation. CL-KL uses a $Q_θ$-element dictionary instead of the $Q_θQ_r$ atoms of 2-D polar gridding, eliminating the range-dimension dictionary coherence that plagues polar codebooks in the strong near-field regime, and operates entirely on the compressed covariance for full compatibility with hybrid front-ends. At $N_\mathrm{MC}=400$ ($f_c=28$~GHz, $M=64$, $N_\mathrm{RF}=8$, $N=64$, $d=3$, $r\in[0.05,1.0]\,r_\mathrm{RD}$), CL-KL achieves the lowest channel NMSE among all six evaluated methods -- including four full-array baselines using $64\times$ more data -- at $\mathrm{SNR}\in\{-5,0,+5,+10\}$~dB. Running in approximately 70~ms per trial (vs.\ 5~ms for the compressed-domain peer P-SOMP), CL-KL's dominant cost is the $N_\mathrm{RF}{\times}N_\mathrm{RF}$ inversion rather than $M$: measured runtime stays near 70~ms across $M\in\{32,64,128,256\}$, making it aperture-scalable for XL-MIMO deployments. CL-KL is further validated against a derived compressed-domain Cramér-Rao bound and confirmed robust to non-Gaussian (QPSK) source distributions, with a maximum NMSE gap below 0.6~dB.

The number of spatial degrees of freedom (NDoF) and channel strength in antenna systems are examined within a geometric framework. Starting from a correlation-operator representation of the channel between transmitter and receiver regions, we analyze the associated eigenspectrum and relate the NDoF to its spectral transition (corner). We compare the spectrum-based effective NDoF and effective rank metrics, clarifying their behavior for both idealized and realistic eigenvalue distributions. In parallel, we develop geometry-based asymptotic estimates in terms of mutual shadow (view) measures and coupling strength. Specifically, we show that while the projected length or area predicts the number of usable modes in two- and three-dimensional settings, the coupling strength determines the average eigenvalue level. Canonical configurations of parallel lines and regions are used to derive closed-form asymptotic expressions for the effective NDoF, revealing significant deviations from the spectral corner in closely spaced configurations. The results illustrate that these are physically grounded. The proposed theory and techniques are computationally efficient and form a toolbox for estimating the modal richness in near-field channels, with implications for array design, inverse problems, and high-capacity communication systems.

This paper introduces a methodology to calibrate Radio-Frequency Digital Twins (RF-DTs) for Integrated Sensing and Communication (ISAC) in dynamic wireless environments. The approach leverages high-resolution ray tracing in combination with wideband channel sounding to ensure consistency between simulated and measured propagation. The methodology is validated in urban scenarios featuring both mono-static and bi-static configurations, as well as moving user platforms and vehicles. Results show that the calibrated RF-DT reproduces key propagation effects, including multipath evolution, dynamic scatterers, and Doppler-induced signatures, with close agreement to measurements. These findings confirm that accurate geometry, material modeling, antenna patterns, and diffuse scattering are essential for realistic high-frequency ISAC simulation. By bridging the gap between simulation and measurement, the proposed calibration framework provides a scalable tool for developing and evaluating ISAC algorithms in complex, time-varying environments envisioned for 6G.

This paper investigates energy efficiency (EE) optimization for an uplink multiuser system assisted by a movable-element reconfigurable intelligent surface (ME-RIS) and a base station equipped with movable antennas (MA-BS). We jointly optimize the uplink postcoder vectors, user transmit powers, RIS phase shift, and the positions of both the BS antennas and RIS elements to maximize the system EE. The resulting non-convex fractional problem is solved using an alternating optimization (AO) framework, where subproblems are handled via Dinkelbach's method combined with successive convex approximation (SCA). Simulation results show that the proposed scheme achieves significant EE gains over fixed-antenna BS and fixed-element RIS benchmarks.

This dataset provides a large collection of 10,915 synthetic hyperspectral image cubes paired with pixel-level vegetation trait maps, designed to support research in radiative transfer emulation, vegetation trait retrieval, and uncertainty quantification. Each hyperspectral cube contains 211 bands spanning 400--2500 nm at 10 nm resolution and a fixed spatial layout of 64 \times 64 pixels, offering continuous simulated surface reflectance spectra suitable for emulator development and machine-learning tasks requiring high spectral detail. Vegetation traits were derived by inverting Sentinel-2 Level-2A surface reflectance using a PROSAIL-based lookup-table approach, followed by forward PROSAIL simulations to generate hyperspectral reflectance under physically consistent canopy and illumination conditions. The dataset covers four ecologically diverse regions -- East Africa, Northern France, Eastern India, and Southern Spain -- and includes 5th and 95th percentile uncertainty maps as well as Sentinel-2 scene classification layers. This resource enables benchmarking of inversion methods, development of fast radiative transfer emulators, and studies of spectral--biophysical relationships under controlled yet realistic environmental variability.

Integrated Sensing and Communications (ISAC) is regarded as a key element of the beyond-fifth-generation (5G) and sixth-generation (6G) systems, raising the question of whether current 5G New Radio (NR) signal structures can meet the sensing accuracy requirements specified by the Third Generation Partnership Project (3GPP). This paper addresses this issue by analyzing the fundamental limits of range and velocity estimation through the Cramér-Rao lower bound (CRLB) for a monostatic unmanned aerial vehicle (UAV) sensing use case currently under consideration in the 3GPP standardization process. The study focuses on standardized signals and also evaluates the potential performance gains achievable with reference signals specifically designed for sensing purposes. The compact CRLB expressions derived in this work highlight the fundamental trade-offs between estimation accuracy and system parameters. The results further indicate that information from multiple slots must be exploited in the estimation process to attain the performance targets defined by the 3GPP. As a result, the 5G NR positioning reference signal (PRS), whose patterns may be suboptimal for velocity estimation when using single-slot resources, becomes suitable when multislot estimation is employed. Finally, we propose a two-step iterative range and radial-velocity estimator that attains the CRLB over a significantly wider range of distances than conventional maximum-likelihood (ML) estimators, for which the well-known threshold effect severely limits the distance range over which the accuracy requirements imposed by the 3GPP are satisfied.

Providing guaranteed quality of service for cell-edge users remains a longstanding challenge in wireless networks. While coordinated interference management was proposed decades ago, its potential has been limited by computational complexity and backhaul resource constraints. Distributed user scheduling and coordinated beamforming (D-USCB) offers a scalable solution but faces practical challenges in acquiring inter-cell channel state information (CSI), as base stations (BSs) are often restricted to signal strength measurements, and high-dimensional CSI exchange incurs substantial overhead. Inspired by integrated sensing and communication (ISAC), this paper proposes a sensing-assisted D-USCB (SD-USCB) framework to maximize the network throughput of multi-cell mmWave networks. Firstly, the framework leverages channel knowledge maps (CKMs) that map user locations to CSI estimates, where user locations are proactively sensed via ISAC echoes. Secondly, we employ a signal-to-average-leakage-plus-interference-plus-noise ratio (SALINR) metric for distributed ISAC beamforming optimization, in which BSs simultaneously communicate with users and sense their locations. These two components jointly enable distributed coordinated transmission with only user location information exchanged among BSs, thereby substantially reducing backhaul overhead. In addition, we devise efficient distributed user scheduling and ISAC beamforming algorithms to jointly optimize communication and sensing performance. Extensive numerical results demonstrate significant improvements in network throughput, validating the efficacy of the proposed framework.

Distributed scheduling is essential for open radio access network (O-RAN) employing advanced physical-layer techniques such as multi-user MIMO (MU-MIMO), carrier aggregation (CA), and joint transmission (JT). This work investigates the multi-component-carrier (multi-CC) resource block group (RBG) scheduling in MU-MIMO O-RAN with both JT and non-JT users. We formulate a scheduling optimization problem to maximize throughput subject to user-specific quality of service (QoS) requirements while ensuring consistent allocations across cooperating O-RAN radio units (O-RUs) required by JT transmission. The strong variable coupling, non-convexity, and combinatorial complexity make the problem highly challenging. To tackle this, we extend the eigen-based zero-forcing transceiver design to JT users and leverage massive MIMO asymptotic properties to derive a tractable, separable rate approximation. Building on this, we develop two solutions: a centralized block coordinate descent benchmark and a distributed scheduler aligned with the O-RAN architecture. The proposed distributed scheme achieves near-centralized performance with only one round of lightweight coordination among cells, significantly reducing complexity and delay. Extensive simulations validate that our distributed scheduler achieves high scalability, fast convergence, and better QoS satisfaction rate in large-scale MU-MIMO networks.

Environment-aware 6G wireless networks demand the deep integration of multimodal and wireless data. However, most existing datasets are confined to 2D terrestrial far-field scenarios, lacking the 3D spatial context and near-field characteristics crucial for low-altitude extremely large-scale multiple-input multiple-output (XL-MIMO) systems. To bridge this gap, this letter introduces Multimodal-NF, a large-scale dataset and specialized generation framework. Operating in the upper midband, it synchronizes high-fidelity near-field channel state information (CSI) and precise wireless labels (e.g., Top-5 beam indices, LoS/NLoS) with comprehensive sensory modalities (RGB images, LiDAR point clouds, and GPS). Crucially, these multimodal priors provide spatial semantics that help reduce the near-field search space and thereby lower the overhead of wireless sensing and communication tasks. Finally, we validate the dataset through representative case studies, demonstrating its utility and effectiveness. The open-source generator and dataset are available at https://lmyxxn.github.io/6GXLMIMODatasets/.

We investigate a novel integrated sensing and communication (ISAC) system enabled by pinching antennas (PAs), which are dynamically activated along a dielectric waveguide. Unlike prior designs, the PAs are organized into multiple clusters of movable antennas. The movement of the antennas within each cluster enables transmit beamforming, while the spatial separation of different clusters allows the system to illuminate the target from multiple angular perspectives.

Carrier frequency offset estimation (CFOE) is a critical stage in modern coherent optical communication systems. Although conventional all-digital techniques perform reliably in typical fiber-optic communication links, CFOE often becomes a major bottleneck in low-symbol-rate scenarios with large carrier CFOs (approaching the signal bandwidth) and severe additive noise levels. These conditions are particularly prevalent in links between optical ground stations (OGSs) and low Earth orbit (LEO) satellites, where Doppler-induced frequency shifts of several gigahertz and atmospheric attenuation significantly degrade CFOE performance and can render traditional methods ineffective. In this paper, we propose a robust non-data-aided (NDA) scheme designed for wide-range CFOE. Such a coarse CFOE (C-CFOE) algorithm partially compensates for the CFO, enabling the operation of a subsequent fine CFOE algorithm. By applying low-complexity operations to the spectrum of the received signal, we recast the frequency estimation task as a segmented linear regression (SLR) problem. Numerical simulations in stress-test scenarios involving large CFOs, low SNR, and low symbol rates show that the proposed approach achieves good estimation accuracy and robust convergence. Experimental offline validation further confirms the practical feasibility of the method.

A multiple-input multiple-output (MIMO) system operating at terahertz (THz) frequencies and consisting of a transmitter, Alice, that encodes secret keys using Gaussian-modulated coherent states, which are communicated to a legitimate receiver, Bob, under the assistance of a reconfigurable intelligent surface (RIS) is considered in this paper. The composite wireless channel comprising the direct Alice-to-Bob signal propagation path and the RIS-enabled reflected one is modeled as a passive linear Gaussian quantum channel, allowing for a unitary dilation that preserves the canonical commutation relations. The security of the considered RIS-empowered MIMO system is analyzed under collective Gaussian entangling attacks, according to which an eavesdropper, Eve, is assumed to have access to environmental modes associated with specific propagation segments. We also study, as a benchmark, the case where Eve has access to the purification of the overall channel. The legitimate receiver, Bob, is designed to deploy homodyne detection and reverse reconciliation for key extraction. Novel expressions for the achievable secret key rate (SKR) of the system are derived for both the considered eavesdropping scenarios. Furthermore, an optimization framework is developed to determine the optimal RIS phase configuration matrix that maximizes the SKR performance. The resulting optimization problem is efficiently solved using particle swarm optimization. Numerical results are presented to demonstrate the system's performance with respect to various free parameters. It is showcased that the considered RIS plays a crucial role in enhancing the SKR of the system as well as in extending the secure communication range. This establishes RIS-assisted THz MIMO CV-QKD as a promising solution for next generation secure wireless networks.

The stringent link budget, caused by long propagation distances and payload constraints, poses a fundamental bottleneck for single-satellite transmission. Although LEO mega-constellations make multi-satellite cooperative transmission (MSCT), such as distributed precoding (DP), increasingly feasible, its cooperative gains critically rely on stringent time-frequency-phase synchronization (TFP-Sync), which is difficult to maintain under rapid channel variation and feedback latency. To address this issue, this paper proposes a joint CSI acquisition, feedback, and phase-level synchronization (JCAFPS) framework for MSCT. Specifically, to enable reliable, overhead-efficient CSI acquisition, we design a beam-domain adjustable phase-shift tracking reference signal (TRS) transmission scheme, along with criteria for the TRS and CSI-feedback periods. Then, exploiting deterministic orbital motion and dominant LoS propagation, we establish a polynomial model for the temporal evolution of delay and Doppler shift, and derive an OFDM-based multi-satellite signal model under non-ideal synchronization. The analysis reveals that, unlike the single-satellite case, the composite multi-satellite channel exhibits nonlinear time-frequency-varying phase behavior, necessitating symbol- and subcarrier-wise phase precompensation for coherent transmission. Based on these results, we develop a practical closed-loop realization integrating single-TRS-based channel parameter estimation, multi-TRS-based channel prediction, predictive CSI feedback, and user-specific TFP precompensation. Numerical results demonstrate that the proposed framework achieves accurate CSI acquisition and precise TFP-Sync, enabling DP-based dual-satellite cooperative transmission to approach the theoretical 6 dB power gain over single-satellite transmission, while remaining robust under extended prediction durations and enlarged TRS periods.

In this paper, a rate-splitting multiple access (RSMA) based joint optimization framework for the space-air-ground integrated network (SAGIN) is proposed, where the satellite and base stations employ uniform planar array (UPA) antennas for signal transmission, and unmanned aerial vehicles (UAVs) relay the satellite signals. Earth stations (ESs) and user equipments (UEs) receive signals from satellite and base stations (BSs), respectively, resulting in mutual interference. We first model the channels and signals in this scenario and analyse the interference at BSs and UEs. Then, We formulate a joint optimization problem aimed at maximizing the weighted sum-rate, involving beamforming, RIS-UAV deployment and phase shifts, and rate splitting. However, this problem is highly non-convex. To tackle this challenge, we apply a block coordinate descent (BCD) approach to decompose the problem and employ the weighted minimum mean square error (WMMSE) method to transform the non-convex objective function. For the rate-splitting sub-problem, a greedy algorithm is proposed and a successive convex approximation (SCA) algorithm is used for beamforming. Besides, the alternating direction method of multipliers (ADMM) algorithm is employed for the RIS phase-shift problem with unit-modulus constraints, and an exhaustive search method is adopted for the complex UAV positioning and orientation. Simulation results validate that the proposed algorithm achieves superior performance in terms of user weighted sum-rate.

Achieving coherent integration in distributed Internet of Things (IoT) sensing networks requires precise synchronization to jointly compensate clock offsets and radio-frequency (RF) phase errors. Conventional two-step protocols suffer from time-phase coupling, where residual timing offsets degrade phase coherence. This paper proposes a generalized hyper-plane regression (GHR) framework for joint calibration by transforming coupled spatiotemporal phase evolution into a unified regression model, enabling effective parameter decoupling. To support resource-constrained IoT edge nodes, a feature-level distributed architecture is developed. By adopting a linear frequency-modulated (LFM) waveform, the model order is reduced, yielding linear computational complexity. In addition, a unidirectional feature transmission mechanism eliminates the communication overhead of bidirectional timestamp exchange, making the approach suitable for resource-constrained IoT networks. Simulation results demonstrate reliable picosecond-level synchronization accuracy under severe noise across kilometer-scale distributed IoT sensing networks.

Edge deployment of large language models (LLMs) can reduce latency for interactive services, but mobility introduces service interruptions when an user equipment (UE) hands over between base stations (BSs). To promptly resume decoding, the target-side edge server must recover the UE context state, which can be provisioned either by token forwarding followed by prefill computation or by direct key-value (KV) cache transmission over backhaul. This paper proposes a unified handover (HO) design that jointly selects the prefill length and schedules backhaul KV cache delivery to minimize the worst-user LLM HO delay for multiple UEs. The resulting scheme admits a tractable step-wise solution with explicit feasibility conditions and a constructive rate-scheduling policy. Simulations show that the proposed method consistently outperforms baselines across a wide range of backhaul capacities, prefill speeds, and context sizes, providing practical guidelines for mobility-aware Edge LLM token streaming.

The deployment of ultra-dense networks (UDNs), particularly cell-free massive MIMO (CF-mMIMO), is mainly hindered by costly and capacity-limited fronthaul links. This work proposes a two-tiered optimization framework for cost-effective hybrid fronthaul planning, comprising a Near-Optimal Fronthaul Association and Configuration (NOFAC) algorithm in the first tier and an Integer Linear Program (ILP) in the second, integrating fiber optics, millimeter-wave (mmWave), and free-space optics (FSO) technologies. The proposed framework accommodates various functional split (FS) options (7.2x and 8), decentralized processing levels, and network configurations. We introduce the hierarchical scheme (HS) as a resilient, cost-effective fronthaul solution for CF-mMIMO and compare its performance with radio-stripes (RS)-enabled CF-mMIMO, validating both across diverse dense topologies within the open radio access network (O-RAN) architecture. Results show that the proposed framework achieves better cost-efficiency and higher capacity compared to traditional benchmark schemes such as all-fiber fronthaul network. Our key findings reveal fiber dominance in highly decentralized deployments, mmWave suitability in moderately centralized scenarios, and FSO complements both by bridging deployment gaps. Additionally, FS7.2x consistently outperforms FS8, offering greater capacity at lower cost, affirming its role as the preferred O-RAN functional split. Most importantly, our study underscores the importance of hybrid fronthaul effective planning for UDNs in minimizing infrastructural redundancy, and ensuring scalability to meet current and future traffic demands.

We revisit the recently proposed ``self-Wiener" (SW) filtering method for robust deconvolution, and generalize it to the classical denoising problem. The resulting estimator, termed generalized SW (GSW) filtering, retains the nonlinear shrinkage structure of SW but introduces a tunable threshold parameter. This tunability enables GSW to flexibly adapt to varying signal-to-noise ratio (SNR) regimes by balancing noise suppression and signal preservation. We derive closed-form expressions for its mean-square error (MSE) performance in both low- and high-SNR regimes, and demonstrate that GSW closely approximates the oracle MMSE at high SNR while maintaining strong robustness at low SNR. Simulation results validate the analytical findings, showing that GSW consistently achieves favorable denoising performance across a wide range of SNRs. Its analytical tractability, parameter flexibility, and close connection to the optimal Wiener filter structure make it a promising tool for practical applications including compressive sensing, sparse signal recovery, and domain-specific shrinkage in wavelet, Fourier, and potentially learned orthonormal representations.

We study joint compression and detection in distributed sensing systems motivated by emerging applications such as IoT-based localization. Two spatially separated sensors observe noisy signals and can exchange only a $k$-bit message over a reliable one-way low-rate link. One sensor compresses its observation into a $k$-bit description to help the other decide whether their observations share a common underlying signal or are statistically independent. We propose a simple extremum-based strategy, in which the encoder sends the index of its largest sample and the decoder performs a scalar threshold test. We derive exact nonasymptotic false-alarm and misdetection probabilities and validate the analysis with representative simulations.

To advance integrated sensing and communications (ISAC) in sixth-generation (6G) extremely large-scale multiple-input multiple-output (XL-MIMO) networks, a low-complexity compressed sensing (CS)-based dictionary design is proposed for wideband near-field (WB-NF) target localization. Currently, the massive signal dimensions in the WB-NF regime impose severe computational burdens and high spatial-frequency coherence on conventional grid-based algorithms. Furthermore, a unified framework exploiting both wideband (WB) and near-field (NF) effects is lacking, and the analytical conditions for simplifying this model into decoupled approximations remain uncharacterized. To address these challenges, the proposed algorithm mathematically decouples the mutual coherence function and introduces a novel angle-distance sampling grid with customized distance adjustments, drastically reducing dictionary dimensions while ensuring low coherence. To isolate the individual WB and NF impacts, two coherence-based metrics are formulated to establish the effective boundaries of the narrowband near-field (NB-NF) and wideband far-field (WB-FF) regions, where respective multiple signal classification (MUSIC) algorithms are utilized. Simulations demonstrate that the CS-based method achieves robust performance across the entire regime, and the established boundaries provide crucial theoretical guidelines for WB and NF effect decoupling.

This paper investigates the physical layer security of a non-orthogonal multiple access (NOMA) system assisted by a tertiary-mode simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS), which can perform transmission, reflection, and jamming simultaneously. The system comprises a base station (BS) serving two users located on opposite sides of the STAR-RIS, assuming perfect channel state information (CSI) at the transmitter. To enhance secrecy performance, a subset of STAR-RIS elements is adaptively configured for jamming. A penalty-based alternating optimization algorithm is developed to jointly optimize the BS's active beamforming and the STAR-RIS's passive beamforming and mode selection. Simulation results demonstrate that the proposed design substantially improves the achievable sum rate and secrecy performance compared to conventional RIS-assisted and no-RIS benchmarks, highlighting the potential of tertiary-mode STAR-RIS for secure and efficient next-generation wireless communications.

Microwave-frequency acoustic waves in solids have emerged as a versatile platform for both classical and quantum applications. While phononic integrated devices and circuits are being developed on various material platforms, an ideal phononic integrated circuit (PnIC) platform should simultaneously support low-loss waveguide structures, high-quality-factor resonators, high-performance modulators, and efficient electromechanical transducers. Here, we establish a low-loss gigahertz-frequency PnIC platform based on patterned thin-film silicon nitride (SiN) on lithium niobate (LN) substrate. We develop low-loss PnIC building blocks including waveguides, directional couplers, and high-quality-factor (high-Q) ring resonators. As an application, we demonstrate a 1-GHz phononic oscillator based on a ring resonator, reaching a low phase noise of -159.0 dBc/Hz at a 100-kHz offset frequency. Our low-loss PnICs could meet the requirements in microwave acoustics, quantum phononics, and integrated hybrid systems combining phonons, photons, superconducting qubits, and solid-state defects.

Reconfigurable intelligent surfaces have emerged as a cutting-edge technology for next-generation wireless communications that are capable of reconfiguring the wireless environment using a large number of cost-effective reflecting elements. However, a significant body of prior studies has focused on single-layer surfaces that lack the capability of significantly mitigating inter-user interference. Moreover, previous studies mostly consider far-field operation and neglect working in the near-field region. In this paper, we propose a stacked intelligent metasurfaces (SIM)-assisted near-field multi-user multiple-input-single-output covert communication system. More specifically, we have a multi-antenna base station that is assisted with a SIM to serve multiple single-antenna users in the presence of multiple single-antenna wardens. We aim at optimizing the beamfocusing vectors at the BS and SIM phase shift matrices to maximize the sum covert rate under maximum transmit power budget constraint, quality-of-service (QoS) constraint for all users, and covertness constraint. Since the formulated problem is highly non-convex due to the coupling between the variables, we adopt alternating optimization to tackle it, where we divide the problem into beamfocusing sub-problem and SIM phase shift sub-problem, which are solved alternately until convergence. We leverage successive convex approximation (SCA) to solve the two sub-problems. Additionally, we formulate the SIM phase shift sub-problem using the widely adopted projected gradient ascent (PGA) method for comparison purposes. The conducted simulations reveal that the SCA-based algorithm outperforms the existing PGA-based algorithm as well as other benchmarks in terms of the achieved sum covert rate, demonstrating its consistent performance and robustness under various system parameter configurations.

Movable antennas (MAs) enable the reconfiguration of array geometry within a bounded region to exploit sub-wavelength spatial degrees of freedom in wireless communication and sensing systems. However, most prior research has predominantly focused on the communication and sensing performance, overlooking the mechanical power consumption inherent in antenna movement. To bridge this gap, this paper investigates a velocity profile optimization framework for MA-assisted direction-of-arrival (DoA) estimation, explicitly balancing sensing accuracy with mechanical energy consumption of MAs. We first establish a Newtonian-based mechanical energy model, and formulate a functional optimization problem for sensing energy efficiency (EE) maximization. By applying the calculus of variations, this formulation is transformed into an infinite-dimensional problem defined by the Euler-Lagrange equation. To solve it, we propose a spectral discretization framework based on the Galerkin method, which expands the velocity profile over a sinusoidal basis. In the regime where energy consumption is dominated by linear damping, we prove that the optimal velocity profile follows a closed-form sinusoidal shape. For more general scenarios involving strong nonlinear aerodynamic drag, we leverage the Markov-Lukács theorem to transform the kinematic constraints into strictly convex sum-of-squares (SOS) conditions. Consequently, the infinite-dimensional problem is reformulated as a tractable finite-dimensional nonlinear algebraic system, which is solved by a two-layer algorithm combining Dinkelbach's method with successive convex approximation (SCA). Numerical results demonstrate that our optimized velocity profile significantly outperforms baselines in terms of EE across various system configurations. Insights into the optimized velocity profiles and practical design guidelines are also provided.

Analytical modeling of field-assisted molecular communication under dynamic electric fields is fundamentally challenging due to the coupling between stochastic transport and complex boundary geometries, which renders conventional partial differential equation (PDE) approaches intractable. In this work, we introduce an effective stochastic modeling approach to address this challenge. By leveraging trajectory-reweighting techniques, we derive analytically tractable channel impulse response (CIR) expressions for both fully-absorbing and passive spherical receivers, where the latter serves as an exact theoretical baseline to validate our modeling accuracy. Building upon these models, we establish a dynamic waveform design framework for system optimization. Under a maximum \textit{a posteriori} decision-feedback equalizer (MAP-DFE) framework, we show that the first-slot received probability serves as the primary determinant of the bit error probability (BEP), while inter-symbol interference manifests as higher-order corrections. Exploiting the monotonic response of the fully-absorbing architecture and using the limitations of the passive model to justify this strategic focus, we reformulate BEP minimization into a distance-based optimization problem. We propose a unified, low-complexity Maximize Received Probability (MRP) algorithm, encompassing the Maximize Hitting Probability (MHP) and Maximize Sensing Probability (MSP) methods, to dynamically enhance desired signals and suppress inter-symbol interference. Numerical results validate the accuracy of the proposed modeling approach and demonstrate near-optimal detection performance.

In this work, we propose a zero constellation for binary modulation on conjugate-reciprocal zeros (BMOCZ), called jutted BMOCZ (J-BMOCZ), and study its application to non-coherent orthogonal frequency division multiplexing (OFDM). With J-BMOCZ, we introduce asymmetry to the zero constellation for Huffman BMOCZ, which removes ambiguity at the receiver under a uniform rotation of the zeros. The asymmetry is controlled by the magnitude of "jutted" zeros and enables the receiver to estimate zero rotation using a simple cross-correlation. The proposed method, however, leads to a natural trade-off between asymmetry and zero stability. Accordingly, we introduce a reliability metric to measure the stability of a polynomial's zeros under an additive perturbation of the coefficients, and we apply the metric to optimize the J-BMOCZ zero constellation parameters. We then combine the advantages of J-BMOCZ and Huffman BMOCZ to design a hybrid waveform for OFDM with BMOCZ (OFDM-BMOCZ). The pilot-free waveform enables blind synchronization/detection and has a fixed peak-to-average power ratio that is independent of the message. Finally, we assess the proposed scheme through simulation and demonstrate non-coherent OFDM-BMOCZ using low-cost software-defined radios.

We study Anderson localization in a one-dimensional disordered system with long-range correlated hopping decaying as $1/r^{a}$ with complex hopping amplitudes that break time-reversal symmetry in a tunable fashion by varying their argument. We find analytically a corelation-induced algebraic localization that is robust to a finite strength of the time-reversal-symmetry-breaking parameter, beyond which all states delocalize. This establishes a localization--delocalization transition driven by the interplay between long-ranged correlated hopping and time-reversal symmetry breaking. In addition to obtaining the static localization phase diagram, we also investigate the dynamical phase diagram through the lens of wavepacket spreading. We find that the growth in time of the mean-squared displacement of a wavepacket, which is subdiffusive for the time-reversal symmetric case, becomes diffusive for any finite value of the time-reversal-symmetry-breaking parameter.

Single-photon sources with high single-photon purity and high brightness are key elements of many future quantum technologies. While photon blockade (PB) is widely exploited in the development of such sources, achieving the coexistence of high purity and high brightness remains a long-standing challenge. Here, we identify a novel mechanism for high-brightness PB and demonstrate that near-ideal purity and near-ideal brightness can be simultaneously achieved in an extended nondegenerate two-photon Jaynes-Cummings model with two-body and three-body interactions. This mechanism is underpinned by a distinctive energy-level structure arising from the combined action of the two interactions. The energy levels in the multi-excitation manifold essentially retain a harmonic ladder of degenerate doublets, whereas in the single-excitation subspace the doublet degeneracy is lifted, with a finite splitting between the two levels. Consequently, when one bosonic mode is driven by a coherent continuous-wave pump, the former degeneracy enables the other bosonic mode to exhibit near-perfect PB even in the strong driving regime, while the latter splitting allows the mean photon number of that mode to approach unity. Our proposed scheme overcomes the outstanding challenge and offers a promising pathway toward realizing ideal single-photon sources.

Solving the Elliptic Curve Discrete Logarithm Problem (ECDLP) is critical for evaluating the quantum security of widely deployed elliptic-curve cryptosystems. Consequently, minimizing the number of logical qubits required to execute this algorithm is a key object. In implementations of Shor's algorithm, the space complexity is largely dictated by the modular inversion operation during point addition. Starting from the extended Euclidean algorithm (EEA), we refine the register-sharing method of Proos and Zalka and propose a space-efficient reversible modular inversion algorithm. We use length registers together with location-controlled arithmetic to store the intermediate variables in a compact form throughout the computation. We then optimize the stepwise update rules and give concrete circuit constructions for the resulting controlled arithmetic components. This leads to a modular inversion circuit that uses $3n + 4\lfloor \log_2 n \rfloor + O(1)$ logical qubits and $204n^2\log_2 n + O(n^2)$ Toffoli gates. By inserting this modular inversion component into the controlled affine point-addition circuit, we obtain a space-efficient algorithm for the ECDLP with $5n + 4\lfloor \log_2 n \rfloor + O(1)$ qubits and $O(n^3)$ Toffoli gates. In particular, for a 256-bit prime-field curve, our estimate reduces the logical-qubit count to 1333, compared with 2124 in the previous low-width implementation of Häner et al.

In trapped-ion chains, multipartite GHZ states can be prepared natively with the help of a single bichromatic laser pulse. However, higher-order terms in the expansion in the Lamb-Dicke parameter $η$ limit the GHZ state preparation infidelity for rectangular and bell-like pulses to the order of $η^4$. For tens of ions, the infidelity caused by out-of-Lamb-Dicke effects can reach several percents. We propose an amplitude and phase-modulated pulse shape, an "echoed lemniscate pulse", which cancels this contribution into error in the leading order. For the proposed pulse, the infidelity scales as $η^6$. The improved scaling is achieved because of a special phase trajectory of a collective motional mode following the figure-eight curve (lemniscate). We demonstrate that the lemniscate pulse allows achieving lower infidelity than bell-like pulses, which can be as low as $10^{-4}$ for $20$-ion chains.

We investigate the semiclassical regularity of thermal equilibria in the presence of a harmonic potential at low temperature; that is, we obtain the asymptotic behavior of the Schatten norms of commutators of the one-body operators associated with these equilibria and the position and momentum operators. We also obtain upper bounds in the magnetic field case for the Fock-Darwin Hamiltonian. Our estimates, in particular, allow us to observe several regimes depending on the joint behavior of the Planck constant, the temperature, and the strength of the magnetic field.

Organising the space of entanglement structures of a multipartite quantum system is a much more challenging task than its bipartite version: while the local unitary (LU) orbit of a bipartite pure state can be conveniently characterized by its entanglement spectrum, invariants of multipartite entanglement structures are comparatively difficult to define and work with. The root cause of this difference is that the bipartite problem can be reduced to the analysis of matrix invariants, while its multipartite version is governed by a much richer space of tensor invariants. The present work explores the latter through the lens of so-called trace-invariants, which are in one-to-one correspondence with combinatorial objects known as colored graphs. We first explain why trace-invariant evaluations can serve as labels of LU-orbits of multipartite pure states, how this strategy extends to random states, and how the effect of local operations (LO) can be analyzed through such data. We then focus on entanglement classification within an (infinite-dimensional) subspace of reference states, whose basic building blocks are GHZ states of various dimensions. We show that relatively simple subclasses of trace-invariants are sufficient to separate the LU-orbits of reference states, and enable a complete (resp. an incomplete) characterization of their relations in the LO (resp. LOCC) resource theory of entanglement. Finally, we investigate how a (still infinite) subclass of reference states of local dimension N can be efficiently distinguished at leading and subleading orders in an asymptotic large-N expansion (among themselves, or from Haar-random states). This analysis relies crucially on combinatorial quantities associated to colored graphs, some of which have already played instrumental roles in the recent literature on random tensors. Results of broader relevance are reported along the way.

We study the power spectrum of the velocity field induced during a first-order phase transition occurring in the radiation-dominated era. We focus on the phase of bubble expansion, assuming that it ends with the onset of the sound-wave regime. The main result we present is a refined template for the velocity spectrum at the beginning of the sound-wave phase, which can be used for studying the resulting anisotropic stresses and gravitational wave production. In particular, we find that the breaks in the velocity spectrum are not associated to the bubble size and the sound shell thickness, as previously proposed, but to the position of the discontinuities. This distinction is particularly relevant for supersonic deflagrations, as it implies that the intermediate slope is more pronounced and the two breaks are more separated when the wall velocity approaches the Chapman-Jouget speed, instead of the sound speed. We also show that the asymptotic branches of the velocity power spectrum are determined by the integral over the single-bubble profiles at large scales, and by the discontinuities of the velocity profiles at small scales. Furthermore, we study the dependence of the two breaks and the intermediate slope on the distribution function of the times of bubble nucleation (exponential and simultaneous). All the results presented in this work have been included in the public Python package CosmoGW.

We present a detailed computational and algebraic study of Mutually Unbiased Bases (MUBs) in finite-dimensional Hilbert spaces, with a particular focus on dimensions 2, 3, 4, and the challenging case of 6. Starting from the Hadamard-phase parametrization, we derive explicit analytical conditions for mutual unbiasedness in dimension 4, providing a tractable system of trigonometric constraints on the phase parameters. We then explore a tensor-product construction via Pauli operators, highlighting the algebraic and group-theoretical origin of MUBs in two-qubit systems, and demonstrating how these constructions yield a complete set of 5 MUBs in dimension 4. Extending our approach, we investigate the Fourier-family method in dimension 6, where the absence of a prime-power structure imposes strong rigidity constraints and limits the known constructions to sets of 3 MUBs. We provide a systematic computational framework for testing candidate phase vectors, bridging the gap between analytical insight and numerical exploration. Finally, we generalize the discussion to arbitrary prime-power dimensions, emphasizing the role of finite-field structures, Heisenberg-Weyl operators, and discrete symmetries in generating complete sets of MUBs. Our work offers a transparent, line-by-line verification methodology, highlighting both the geometric and algebraic richness of MUBs, and clarifying why certain dimensions resist full analytical constructions. This study serves as a comprehensive resource for researchers seeking both theoretical understanding and practical construction of MUBs in quantum information science.

We investigate the quantum algorithms for dynamic programming by Ambainis et al. (SODA'19). While giving provable complexity speedups and applicable to a variety of NP-hard problems, these algorithms have a notable drawback: they require a large amount of Quantum Random Access Memory (QRAM), which potentially could be very challenging to implement in a physical quantum computer. In this work, we study how we can improve the space complexity by trading it for time, while still retaining a speedup over the classical algorithms. We show novel quantum time-space tradeoffs, which we obtain by adjusting the parameters of these algorithms and combining them with "quantized" classical strategies.

We quantize the degrees of freedom on a gravitational null ray segment in a fully gauge-invariant manner by using the dressing time as a quantum reference frame (QRF). Our work goes beyond previous models in that the QRF we employ is made out of the gravitational field itself, and accounts for the full group of diffeomorphisms along the ray, not just a locally compact subgroup. The key tool we introduce is covariant normal ordering, a QRF-dependent but background-independent renormalization prescription that restores diffeomorphism covariance at the quantum level. This enables the definition of a quantum dressing map whose image is the algebra of gauge-invariant observables. We find that this algebra carries the structure of a Virasoro crossed product, and that the dressing map induces a deformed product on gauge-fixed operators which can be understood as a quantization of the Dirac bracket, with consequences for the fluctuations of observables. We explain how to cancel anomalies in the physical Hilbert space representation of the gauge-invariant algebra by including a deformation at the classical level, thereby eliminating all spurious degrees of freedom at the quantum level. The physical Hilbert space admits a Page-Wootters reduction map to the perspective of the dressing time, and we show that the dressing time is non-ideal in the sense that its distinct coherent states have non-vanishing overlaps governed by the Teo-Takhtajan energy, i.e. the Kähler potential for Virasoro coadjoint orbits.

Ultra-narrow-linewidth lasers with suppressed high-frequency phase noise are critical for quantum control and precision metrology. While optical phase locking (OPL) is the standard technique for cloning the coherence of such sources, its effectiveness is often limited at high frequencies by feedback latency. We present a robust feedforward architecture that overcomes this limitation by recycling and demodulating the existing master-slave beat signal to drive a single electro-optic modulator for near-instantaneous noise cancellation. This approach eliminates the extraneous sidebands and transmission losses typical of more complex modulators. Through active stabilization of the beat amplitude and demodulation phase, we demonstrate robust suppression exceeding 30 dB from 10 kHz to 10 MHz. This hardware-efficient framework is readily compatible with standard OPL setups, offering a scalable solution for high-fidelity coherent control.

In an effective-field-theory framework for gravity, black-hole quasinormal mode spectra acquire corrections in quadratic-curvature, scalar-tensor extensions of general relativity. Previous calculations of such corrections were limited to moderate spins, since the corresponding background solutions relied on expansions in the spin parameter. Using recently constructed numerical black-hole solutions valid for large spin, we compute the leading-order deviations from general relativity in the scalar quasinormal mode spectrum of rotating black holes in scalar Gauss-Bonnet and dynamical Chern-Simons gravity. We solve the resulting perturbation equations with pseudo-spectral collocation methods, allowing us to determine the quasinormal-mode corrections for dimensionless spins up to $a/M=0.99$, with accuracy better than $\lesssim 10^{-3}$ for the $l=m=0$ mode and $\lesssim 10^{-6}$ for higher multipoles. For spins $a/M>0.9$, the corrections to certain modes can increase by orders of magnitude.

Applications of quantum mechanics have led to many successful predictions and explanations of puzzling phenomena, and we now apply quantum mechanics to gain, process, and communicate information in novel ways. We can understand quantum mechanics by understanding how we have applied it. We should not seek agreement on the nature of the world it represents, because this theory does not itself represent the physical world (though its applications do help us to represent it better). When applied to a quantum state, quantum mechanics yields probabiities for physical events: both state and probability are objective--not because they represent elements of phyiscal reality, but because each exerts norrmative authority over the beliefs of anyone who accepts quantum mechanics and applies it relative to a physical situation they may (but need not) occupy. These events may be described by statements that are meaningful in an appropriate environmental context, and quantum mechanics can help one to say when that is. Measurement creates an appropriate context, so here the Born rule indirectly yields probabilities of measurement outcomes. The quantum state of a system does not "collapse" on measurement: a new state must be assigned relative to a physical situation in which information about the outcome is accessible. Understood this way, there is no measurement problem, and violations of Bell inequalities does not demonstrate "spooky" non-local action. Quantum field theories have no physical ontology of their own: a quantum field is a mathematical object in a model whose application helps us to improve and extend our descriptions of the world in other terms. We cannot realise the scenario of Wigner's friend and its recent extensions: but the data that provide overwhelming evidence for quantum mechanics are objective in the same sense as the relative measurement outcomes described in those scenarios.

Powerful approaches to squeeze the motional state of a harmonic oscillator rely on the stepwise modulation of its resonance frequency. Such protocols can be limited by forces that vary slowly between experimental runs but are constant during a single experimental shot. Such shot-to-shot noise gives rise to a spread in experimental outcomes that masks the uncertainty intrinsic to quantum theory. Taking inspiration from spin-echo protocols, we propose a decoupling technique that, under ideal conditions, perfectly cancels shot-to-shot force noise in squeezing experiments based on parametric modulation. We implement the protocol using an optically levitated nanoparticle, where shot-to-shot force noise arises from slowly varying stray fields acting on the charge carried by the particle. Using our oscillator-echo protocol, we demonstrate shot-to-shot noise suppression to the measurement-backaction limit.

Distributing arbitrary graph states across quantum networks is a central challenge for modular quantum computing and measurement-based quantum communication. I introduce the phase quantum walk (PQW), a discrete-time quantum walk in which the conventional position-permuting shift operator is replaced by a diagonal conditional phase (CZ) gate, enabling distribution of arbitrary graph states, not merely GHZ states, from elementary two-qubit resources. The Byproduct Lemma shows that each walk step teleports edge entanglement with a correctable Pauli byproduct; the Coin Invariance Theorem proves that the optimal fidelity F*(C,E) = F*(H,E) for all unitary coins C and noise channels E, with closed-form expressions F_dep = (1 - 3p/4)^k and F_pd = ((1 + sqrt(1 - p))/2)^k. Analytical correction formulas are derived for tree graphs (general theorem) and ring graphs (C4 case study), with F = 1.0 verified across eight topologies (up to 4096 outcomes). Hardware validation on ibm marrakesh (IBM Heron r2, CZ-native) yields F_cl = 0.924 for |GHZ4> and 0.922 for |L4>, statistically identical, providing the first experimental confirmation that fidelity is independent of graph topology as predicted by the Coin Invariance Theorem.

We study axial gravitational perturbations of a hairy black hole constructed in the framework of gravitational decoupling and investigate the geometric origin of echo-like late-time signals in this spacetime. We derive the odd-parity master equation and the corresponding effective potential, and we compute the quasinormal-mode spectrum by using frequency-domain and time-domain methods. We show that, in a suitable region of parameter space, the axial potential develops a double-peak structure that supports a trapping cavity and gives rise to echo-like late-time waveforms. Rather than imposing near-horizon reflectivity by hand, the delayed pulses therefore arise dynamically from the geometry of the effective potential. We also clarify that the parameter region exhibiting echoes need not coincide with the region in which the weak energy condition is satisfied everywhere outside the event horizon, and this distinction must be taken into account when interpreting the physical status of the solution. Our results provide a useful framework for probing black-hole hair through gravitational-wave ringdown and for exploring possible observational departures from the standard no-hair paradigm.

Q2NS is an open-source quantum network simulator built on ns-3, the de facto standard for classical network simulation. By inheriting ns-3's mature classical stack and event-driven execution model, Q2NS enables faithful co-simulation of quantum-network dynamics and classical signaling, a core requirement for the functioning of any quantum network. Its modular architecture is designed for extensibility, with pluggable quantum-state backends (state-vector, density matrix, stabilizer) and a clean separation between network control and node-level operations. Q2NS comes with a quantum network visualizer Q2NSViz, supporting interactive inspection of both physical- and entanglement-induced connectivity graphs, helping users interpret protocol behavior and entanglement manipulation processes. We present a demonstration of Q2NS, highlighting its ability to capture and simulate the coexistence of quantum and classical communication. The proposed demonstration presents quantum communication scenarios of increasing complexity: from entanglement distribution basics to multipartite graph-state manipulation, complemented by pre-loaded examples in Q2NSViz that require no prior quantum communication or coding experience.

We prove a localized big bang formation result, which does not require proximity of the initial data to any background solution. Suppose that we are given initial data for the Einstein--nonlinear scalar field equations on an open set $U \subset \mathbb{R}^3$. We identify a general condition on the initial data such that if the condition is satisfied in a large enough neighborhood of $x \in U$, then the corresponding maximal globally hyperbolic development has a local quiescent big bang singularity with curvature blow up to the past of $x$. We achieve the localization by introducing a new kind of foliation by spacelike hypersurfaces, given by the level sets of a time function satisfying a certain second order differential equation. This time function allows us to synchronize the singularity while at the same time yielding a symmetric hyperbolic formulation of Einstein's equations. Our new formulation also has two key advantages over previous localized big bang stability results. First, it is independent of the matter model, so it is possible that it could be used to prove big bang formation with matter models different from a scalar field. And second, it allows us to conclude that our solutions induce geometric initial data on the singularity, thus giving a complete description of the asymptotics towards the big bang.

We study the conformal wave equation $\square_g ψ+ \frac{2}{l^2} ψ= 0$ on 4-dimensional Schwarzschild--Anti de Sitter spacetimes under dissipative boundary conditions. We prove boundedness and decay of the non-degenerate energy of $ψ$ at an arbitrary polynomial rate of $(1+v)^{-n}$ provided that we control the (up to) $n$-times $T$-commuted energy. This contrasts with the inverse logarithmic decay obtained under Dirichlet boundary conditions and is in line with the result obtained in the pure Anti-de Sitter case under dissipative boundary conditions. In particular, the decay is not affected by the additional trapping at the photon sphere.

The Quantum Approximate Optimization Algorithm (QAOA) is a leading hybrid heuristic for combinatorial optimization, but efficiently handling hard constraints remains a significant challenge. XY-mixers successfully confine quantum state evolution to a feasible subspace, such as the Hamming-weight-1 sector for one-hot constraints. On the contrary, warm-starting biases the search toward promising regions based on preliminary solutions. Combining these two techniques requires maintaining the essential alignment between the initial state and the mixer Hamiltonian to preserve convergence guarantees. Previous work demonstrated warm-starting with XY-mixers via a biased initial state, but relying only on standard mixer Hamiltonians. Consequently, the initial state is no longer a ground state of the mixer. In this work, we overcome these limitations by formulating a warm-started XY-mixer Hamiltonian for one-hot constraints and proving its ground-state properties. Furthermore, we provide a shallow circuit implementation suitable for NISQ implementations. We embed the warm-starting into a classical heuristic that iteratively updates the bias based on previous samples, called Iterative Warm-Starting (IWS). Extensive numerical simulations on Max-$k$-Cut and Traveling Salesperson Problem instances demonstrate that IWS-QAOA significantly accelerates the solution-finding process, increasing the probability of sampling optimal solutions by orders of magnitude compared to standard XY-QAOA. Finally, we validate our approach on the ibm_boston QPU using hardware-tailored 144-qubit problem instances. By coupling IWS-QAOA with a greedy steepest-descent post-processing strategy to repair infeasible measurements caused by hardware noise, we successfully identify optimal solutions on actual quantum devices.

Quantum information processing, communication, and sensing networks are being developed with various qubit platforms that use different encoding schemes. Connecting quantum network nodes to distribute entanglement requires matching photon qubit basis encoding. In this work, we implement an interconversion protocol which converts photon qubit encoding from the polarization basis to the time-bin basis, transmits the photons through a transport fiber with large fluctuations in polarization, and converts back to polarization encoding for ease of measurement. This interconversion scheme faithfully transmits a polarization Bell state across the transport fiber by converting sources of infidelity to changes in transmission rate. These results illustrate a practical approach for interfacing distinct qubit platforms to enable modular and flexible operation in heterogeneous quantum networks.

The quantum information content of Hawking radiation holds the key to understanding black-hole evaporation and the fate of unitarity. Motivated by recent advances in cold-atom experiments, we develop a lattice-regularization approach aimed at simulating the coarse-grained entanglement scaling of a quantum field in a 1+1D analogue black-hole background. We provide the first concrete demonstration that logarithmic negativity -- an entanglement monotone that typically exhibits a UV-divergent log-scaling for the conformal vacuum -- acquires a UV-finite volume term from the nonlocal correlations seeded by Hawking radiation. We show that this volume term encodes both the number density and spatial distribution of entangled Hawking pairs along the black-hole interior and exterior. We highlight its prospective detection in currently realizable experiments as well as its implications beyond the analogue paradigm, in particular for black-hole thermodynamics.

In this paper, based on noncommutative-geometry-inspired Schwarzschild black hole, we employ a third-order WKB approximation approach to systematically calculate the quasinormal mode frequencies (QNFs), greybody factors (GFs), and absorption cross section (ACS) under massive scalar field perturbations. The results show that the QNFs satisfy Im($ω$)<0, confirming the stability of the black hole under perturbations. Furthermore, increasing the noncommutative parameter $θ$ reduces the absolute values of both the real and imaginary parts of the frequency, while increasing mass $μ$ increases the real part and reduces the imaginary part. The GFs and ACS increase with increasing $θ$ and decrease with increasing $μ$, indicating opposite modulation effects of these two types of parameters. It is worth emphasizing that the QNFs of the extreme black hole approach the corresponding values of the classical Schwarzschild black hole at angular quantum number $\ell=1$ and large $μ$, suggesting that, the effects of mass and noncommutative geometry quantum corrections cancel each other out to some extent. It is hoped that these results provide a viable theoretical basis for both the theoretical and experimental aspects of the perturbative dynamics of black hole.

We provide here a universal approximation theorem with precise quantitative error bounds for noisy quantum neural networks. We focus on applications to Quantitative Finance, where target functions are often given as expectations. We further provide a detailed numerical analysis, testing our results on actual noisy quantum hardware.

We demonstrate that strong-field ionization of atoms in circularly polarized laser fields generates a photoelectron spin texture with toroidal topology in momentum space. Using time-dependent Schrödinger equation simulations, spin-resolved classical-trajectory Monte Carlo calculations, and an extended spin-resolved strong-field approximation including intermediate excitation pathways, we show that the rotation angle of this spin torus provides access to attosecond relative time delays associated with photoelectron wave packets released by tunneling from the counter-rotating and co-rotating \(p\)-orbital channels. When intermediate-state dynamics become significant, the torus develops a clear splitting. These results establish photoelectron spin textures as a complementary source of dynamical information beyond conventional momentum spectroscopy, and identify spin polarization as a robust internal degree of freedom for self-referenced attosecond metrology.

Auxiliary Field Quantum Monte Carlo (AFQMC) has emerged as a powerful framework for treating strongly correlated electronic systems, offering a favorable balance between computational cost and accuracy. In this paper, we present a novel AFQMC method that uses the isometric tensor hypercontraction (ITHC) technique to diagonalize the two-body Coulomb interaction of molecular electronic Hamiltonians by introducing additional fictitious fermionic modes. Our method shows reduced theoretical complexity and better practical performance for both propagation and local energy evaluation compared to the standard AFQMC method. We demonstrate the efficacy of this approach by computing the ground-state energies of a linear $\ce{H10}$-chain and the benzene molecule. Our results show that the extended-basis AFQMC recovers many-body correlations with a precision comparable to that of high-level wavefunction methods such as Coupled Clusters (CC) or Density Matrix Renormalization Group (DMRG), while offering significantly improved scaling.

We investigate the asymptotic symmetries of General Relativity at spatial infinity within the first-order formalism described by the Holst action. Employing the covariant phase space method, we propose a set of relaxed boundary conditions for the co-tetrad and Lorentz connection that admit the full Bondi-Metzner-Sachs (BMS) group, including non-trivial supertranslations, which are typically eliminated in standard treatments. We demonstrate that the logarithmic divergences appearing in the symplectic structure can be removed by imposing specific, symmetry-preserving parity conditions on the asymptotic fields without suppressing the supertranslation sector. A detailed analysis of the conserved charges reveals that the Holst term contributes non-trivially to the charge variations due to the linear growth of Lorentz generators. We show that the naive surface integrals for the Holst charges exhibit linear divergences arising from the rotation of the background tetrad. These divergences are successfully regularized by supplementing the asymptotic symmetry generator with a compensating internal Lorentz gauge transformation defined to preserve the background structure. The resulting charges are manifestly finite and integrable. Crucially, we prove that while the Holst modification shifts the charges associated with Lorentz boosts and rotations, it leaves the supertranslation charges identically invariant. This framework provides a consistent derivation of the full BMS algebra at spatial infinity in terms of Ashtekar-Barbero variables, offering new insights into the role of the Immirzi parameter in classical and quantum gravity.

Topological codes have many desirable properties that allow fault-tolerant quantum computation with relatively low overhead. A core challenge for these codes, however, is to achieve a low-overhead universal gate set with limited connectivity. In this work, we explore a non-Pauli stabilizer code that can be used to complete a universal gate set on topological toric and surface codes in strictly two dimensions. Fault-tolerant syndrome extraction for the non-Pauli code requires mid-circuit $X$ corrections, a key difference to conventional Pauli codes. We construct and benchmark a just-in-time (JIT) matching decoder to reliably decide these corrections. Under a phenomenological error model with equally likely physical and measurement errors, we find a high threshold of $\approx 2.5\,\%$, close to the $\approx 2.9\,\%$ of a decoder with access to the full syndrome history. We also perform a finite-size scaling analysis to estimate how the logical error rate scales below threshold and verify an exponential suppression in both physical error rate and in the system size. A second global decoding step for $Z$ errors is required and the non-Clifford gates in the circuit reduce the threshold from $\approx 2.9\,\%$ to $\approx 1.8\,\%$ with a naive decoder. We show how $Z$ decoding can be improved using knowledge of the $X$ corrections, pushing the threshold to $\approx 2.2\,\%$. Our results suggest non-Clifford logic in 2D codes could perform comparably to 2D quantum memory. Our formalism for efficient benchmarking and decoding directly generalizes to a broader family of CSS codes whose $X$ stabilizers are twisted by diagonal Clifford operators, and spacetime versions thereof, defined by CSS-like circuits enriched by $CCZ$, $CS$, and $T$ gates.

We introduce a novel quantum algorithm for similar subgraph identification in form of an NP-hard cardinality-constrained binary quadratic optimization problem. Given a weighted reference graph with Laplacian $\boldsymbol{B}$, our algorithm determines the subgraph featuring Laplacian $\boldsymbol{B'}$ on the same vertex set, but $x$ out of $N$ inactive edges, minimizing the Frobenius distance $||\boldsymbol{B} - \boldsymbol{B'}||_\mathrm{F}^2$. We represent the $\binom{N}{x}$ graph topologies by an equal-weight superposition in form of a Dicke state, enabling controlled transformations applied to the quantum state associated with the vectorized Laplacian of the reference graph. Combined with amplitude estimation and a minimum finding approach, our algorithm provides a polynomial speed up $\mathcal{O}(\sqrt{N^{x}/x!}N\log\log N)$ compared to $\mathcal{O}(N^{x+1}/x!)$ of classical brute-force search algorithms. We demonstrate the application of our method on standard test cases, which represent electric power grids, by reconstructing $||\boldsymbol{B} -\boldsymbol{B'}||_\mathrm{F}^2$ from measurements and show how our approach can be additionally used to calculate energy functional like quadratic forms of the Laplacians with respect to a given vector.

I take a pragmatist perspective on quantum theory. This is not a view of the world described by quantum theory. In this view quantum theory itself does not describe the physical world, nor our observatons, experiences or opinions of it. Instead, the theory offers reliable advice on when to expect an event of one kind or another, and on how strongly to expect each possible outcome of that event. The actual outcome is a perspectival fact: a fact relative to a physical context of assessment. Measurement outcomes and quantum states are both perspectival. By noticing that each must be relativized to an appropriate physical context one can resolve the measurement problem and the problem of nonlocal action. But if the outcome of a quantum measurement is not an absolute fact, then why shoud the statistics of such outcomes give us any objective reason to accept quantum theory? One can describe extensions of the scenario of Wigner's friend in which a statement expressing the outcome of a quantum measurement would be true relative to one such context but not relative to another. However, physical conditions in our world prevent us from realizing such scenarios. Since the outcome of every actual quantum measurement is certified at what is essentially a single context of assessment, the outcome relative to that context is an objective fact in the only sense that matters for science. We should accept quantum theory because the statistics these outcomes display are just those it leads us to expect.

In their fourth observing run, the LIGO--Virgo--KAGRA gravitational-wave observatories have found hundreds of new signals, but many are contaminated by non-Gaussian transient noise artefacts known as glitches. Left unaddressed, glitches can bias parameter inference and lead to misleading astrophysical conclusions. We present a series of case studies using GPBilby, a parameter estimation tool that employs a time-domain likelihood jointly modelling the astrophysical signal with a physical waveform and non-Gaussian noise with a Gaussian process. We first show that when the detector noise is Gaussian, GPBilby produces results consistent with those obtained with the standard Gaussian-noise likelihood, and then consider events affected by non-Gaussian features. For GW231123, the highest-mass binary black hole candidate observed to date, analyses using IMRPhenomXPHM reveal coherent residual structure that leads to measurable shifts in inferred source parameters. In contrast, analyses employing NRSur7dq4 show no significant excess residual power and remain consistent across likelihood choices. This demonstrates that waveform systematics and flexible noise modelling are intrinsically coupled, as the Gaussian process terms can partially absorb coherent waveform mismatches. For GW191109, we find that evidence for spin misalignment remains robust despite glitches in both LIGO detectors. For GW230630_070659, excluded from GWTC-4.0 owing to poor data quality, we find the data to be consistent with a BBH waveform model, with no additional residual power identified by the Gaussian process component. Overall, these results highlight how GPBilby can be used to perform glitch-robust inference and as a tool to understand waveform modelling systematics.

In this article, we propose a practical construction of Lie group-equivariant and permutation-invariant functions of $N$ variables from the knowledge of a one-particle basis that is stable with respect to the group action. The construction is generic for any linear Lie group and relies on building a matrix constructed from the Lie algebra whose kernel is spanned by a group-equivariant and permutation-invariant basis. In particular, this construction does not require the knowledge of Clebsch--Gordan coefficients and instead directly builds generalized Clebsch--Gordan coefficients. For specific groups such as $SO(3)$ and $SU(2)$, we exploit the Lie algebra structure to simplify the matrix, which then allows us to derive an explicit formula for the exact dimension of the group-equivariant and permutation-invariant space. Numerical simulations are provided to show that the proposed method scales linearly instead of exponentially for existing methods in the literature. We also show that for large values of $N$, the number of rotation-equivariant and permutation-invariant basis functions is of a comparable order as the number of permutation-invariant basis functions, while pre-asymptotically, a large gain can be achieved by explicitly enforcing rotation-equivariance on top of permutation-invariance.

The sensitivity of continuous gravitational-wave searches is strictly limited by non-Gaussian spectral artefacts that accumulate coherent power over long observation baselines. In this paper, we present an unsupervised mitigation framework based on adaptive network coherence analysis. Unlike traditional veto methods that discard entire frequency bands, our pipeline selectively suppresses local artefacts while preserving global potentially astrophysical signals. We validate the method using Advanced LIGO O3 data, analysing the cleaning performance across integration times of 1, 3, and 5 days. For the 5-day dataset, the pipeline identifies and mitigates 89\% and 77\% of the total spectral lines in the Hanford and Livingston detectors, respectively, while effectively preserving the coherent population consistent with astrophysical morphologies. This is achieved while modifying less than 7\% of the analysis bandwidth spanning 20~Hz to 2000~Hz. Rigorous statistical verification demonstrates that the mitigation effectively suppresses the non-Gaussian tail of the noise distribution while strictly preserving the statistical integrity of coherent signal candidates. By recovering detector sensitivity in parameter spaces previously contaminated by the spectral forest, this framework provides a robust preprocessing strategy for all-sky searches.

Non-Hermitian systems driven along slow parametric loops undergo non-adiabatic transitions whose outcome depends sensitively on the driving speed, yet no explicit formula has been available for the critical timescale $T_{\mathrm{cr}}$ at which these transitions develop. Using a $2\times 2$ Hamiltonian with circular parameter trajectories, we derive $T_{\mathrm{cr}} = \mathcal{G}\,\ln(1/|Δ|)$ in closed form for non-encircling loops, phase-shifted loops, offset loops, and loops encircling exceptional points, where $\mathcal{G}$ is a geometry-dependent growth factor and $Δ$ is the instability seed. This formula sharply separates the regime where the system remains in the averagely dominant eigenstate ($T< T_{\mathrm{cr}}$) from the superadiabatic regime where the instantaneous dominant eigenstate takes over ($T> T_{\mathrm{cr}}$), resolving the apparent tension between the previous literature. We identify two competing seeds: a geometric Stokes multiplier and the finite-precision floor. When the geometric seed vanishes, precision alone governs the transition, yielding $T_{\mathrm{cr}} \propto m\lnβ$, linear in the number of precision bits $m$. This provides a purely forward-evolution manifestation of precision-induced irreversibility (PIR)~\cite{PIR}, demonstrating that the fundamental limit identified through echo protocols also controls the outcome of slow non-Hermitian dynamics without requiring time reversal. For PT-symmetric energy spectra, $T_{\mathrm{cr}}$ additionally determines the onset of chirality: the dynamics is non-chiral for $T< T_{\mathrm{cr}}$ and chiral for $T> T_{\mathrm{cr}}$.

The realization of the Quantum Internet promises transformative capabilities in secure communication, distributed quantum computing, and high-precision metrology. However, transitioning from laboratory experiments to a scalable, multi-tenant network utility introduces deep orchestration challenges. Current development is often siloed within physics communities, prioritizing hardware, while the classical networking community lacks architectural models to manage fragile quantum resources. This tutorial bridges this divide by providing a network-centric view of quantum networking. We dismantle idealized assumptions in current simulators to address the "simulation-reality gap," recasting them as explicit control-plane constraints. To bridge this gap, we establish Software-Defined Quantum Networking (SDQN) as a prerequisite for scale, prioritizing a symbiotic, dual-plane architecture where classical control dictates quantum data flow. Specifically, we synthesize reference models for SDQN and the Quantum Network Operating System (QNOS) for hardware abstraction, and adapt a Quantum Network Utility Maximization (Q-NUM) framework as a unifying mathematical lens for engineers to reason about trade-offs between entanglement routing, scheduling, and fidelity. Furthermore, we analyze Distributed Quantum AI (DQAI) over imperfect networks as a case study, illustrating how physical constraints such as probabilistic stragglers and decoherence dictate application-layer viability. Ultimately, this tutorial equips network engineers with the tools required to transition quantum networking from a bespoke physics experiment into a programmable, multi-tenant global infrastructure.

Inspired by the work of McGlynn and Simenel [Phys. Rev. C {\bf 102}, 064614 (2020)], this study investigates the quantum tunneling of two interacting distinguishable particles in two potential wells. We first benchmark the system by reproducing key established results: the exact quantum solution and the spurious self-trapping effect that arises in the real-time mean-field dynamics for strong interactions. To exactly capture the tunneling dynamics, we apply the time-dependent generator coordinate method (TDGCM) to the model. Numerical simulations demonstrate that the TDGCM, by utilizing the real-time mean-field states as generator states, successfully overcomes the self-trapping effect, yielding tunneling dynamics in excellent agreement with the exact solution. Furthermore, we explore the expectation values of the generator coordinates from the correlated TDGCM many-body wave function. While different methods for calculating expectation values show consistent results in some cases, significant discrepancies are observed in others, providing critical insights into the emergence of collective and single-particle behaviors in interacting systems. This work also verifies the TDGCM as a robust framework for describing collective quantum tunneling and opens avenues for its application to more complex and realistic systems.

We present a loop-shaping approach to coherent feedback (CF) control. By formulating the coupling between a quantum system and its environment in terms of the noise power spectrum, our method enables direct manipulation of the effective dissipation coefficients through spectral shaping. A systematic design framework for CF controllers is also developed, in which transfer functions are shaped to realize desired spectral responses. Applying this framework to optomechanical sideband cooling, we demonstrate that suppression of the Stokes process and enhancement of the anti-Stokes process can be simultaneously achieved, enabling ground-state cooling even in the unresolved-sideband regime. This loop-shaping framework provides an intuitive and general foundation for the design of CF controllers and can be extended to a wide class of quantum systems in which interactions with environments are characterized by noise power spectra.

Nonreciprocity, characterized by direction-dependent signal propagation, is fundamental to technologies such as isolators, signal routing, and precision sensing. This letter theoretically demonstrates nonreciprocal phonon transport and the conversion between photon and acoustic phonon signals in coupled optomechanical cavities via phase-dependent driving. It is demonstrated that, in contrast to nonreciprocal phonon transport, which necessitates both dissipation and phase-induced violation of time reversal symmetry, the nonreciprocity in photon-phonon conversion can occur without violating time reversal symmetry. We demonstrate that such nonreciprocity arises due to the path-dependent asymmetry in photon-phonon conversion. Furthermore, we demonstrate that the nonreciprocity of photon-phonon conversion can be further enhanced, achieving isolation levels of up to 40 dB by suitably modifying the phase difference of the driving lasers.

While QKD ensures information-theoretic security at the link level, real-world deployments depend on trusted repeaters, creating potential vulnerabilities. In this paper, we thus introduce a topology-hiding connectivity assurance protocol to enhance trust in quantum key distribution (QKD) network infrastructures. Our protocol allows network providers to jointly prove the existence of a secure connection between endpoints without revealing internal topology details. By extending graph-signature techniques to support multi-graphs and hidden endpoints, we enable zero-knowledge proofs of connectivity that ensure both soundness and topology hiding. We further discuss how our approach can certify, e.g., multiple disjoint paths, supporting multi-path QKD scenarios. This work bridges cryptographic assurance methods with the operational requirements of QKD networks, promoting verifiable and privacy-preserving inter-network connectivity.

We exhibit nontrivial transversal logical multi-controlled-$Z$ gates on $[\![N,Θ(N),\tildeΘ(N)]\!]$ quantum low-density parity-check codes and $[\![N,Θ(N),\tildeΘ(N)]\!]$ quantum locally testable codes with soundness $\tildeΘ(1)$, combining nearly optimal code parameters with fault-tolerant non-Clifford gates for the first time. Remarkably, our proofs are almost entirely algebraic-topological, showing that such presumably intricate logical gates naturally arise as a fundamental topological phenomenon. We develop a general framework for constructing a rich new family of homological invariant forms which we call ''cupcap gates'' that induce transversal logical multi-controlled-$Z$ and, building on insights from [Li et al., arXiv:2603.25831], covering space methods to certify their nontriviality. The claimed almost-good code results follow immediately as examples.

A classical particle under spatial constraints is strictly confined to live on a specific space manifold or path, but this assumption is incompatible with the zero-point fluctuations of a quantum particle. One way to describe quantum mechanics under constraints is the confinement potential approach (CPA). For a non-relativistic particle, the CPA maps the problem onto the solution of a Schrödinger-type equation in an isometrically embedded Riemannian submanifold of Euclidean space while the motion along orthogonal directions are decoupled and spatially confined. This approach respects quantum uncertainty, and one of its key results is the appearance of geometry- and metric-induced potentials that affect the stationary states and the dynamics of the particle. For particles constrained to different spaces, such as structures hosting sharp bents, vertices, wedges, conical apices, tips, or self-intersections, a formalism beyond the CPA is needed. Here, a step towards a CPA extension for irregular spaces is presented. After classifying the possible geometric irregularities concerning the CPA formalism, the presentation is focused on a sharply bent quantum wire modeled as an embedded curve with singular (but absolute integrable) curvature. For a subclass fulfilling the additional requirement that the geometric potential is a distribution of first order, a solution scheme for the confined Schrödinger equation is presented based on singular Sturm-Liouville theory and operator theoretic methods. The analytical considerations and numerical simulations evidence the existence of curvature-induced bound states with non-differentiable wave functions localized around the singular point, with an extension well beyond the singularity. Furthermore, a multitude of scattering states appear that may affect the transport and optical properties of the system.

Secure long-distance communication in quantum key distribution (QKD) networks depends on trusted repeater nodes along the entire transmission path. Consequently, these nodes will be subject to strict auditing and certification in future large-scale QKD deployments. However, trust must also extend to the network operator, who is responsible for fulfilling contractual obligations -- such as ensuring certified devices are used and transmission paths remain disjoint where required. In this work, we present a path validation protocol specifically designed for QKD networks. It enables the receiver to verify compliance with agreed-upon policies. At the same time, the protocol preserves the operator's confidentiality by ensuring that no sensitive information about the network topology is revealed to users. We provide a formal model and a provably secure generic construction of the protocol, along with a concrete instantiation. For long-distance communication involving 100 nodes, the protocol has a computational cost of 1-2.5s depending on the machine, and a communication overhead of less than 70kB - demonstrating the efficiency of our approach.

Traditional diagnostics of black hole phase transitions rely on thermodynamic quantities defined at the event horizon or asymptotic boundary. Here, we demonstrate that the near-singularity geometry offers a sharp, independent probe of both first-order phase transitions and supercritical crossover. For scalarized AdS black holes exhibiting a first-order phase transition, the Kasner exponent $p_t$, which characterizes the approach to the singularity, undergoes a dramatic transformation. On one side of the transition, $p_t$ oscillates strongly with temperature, reflecting violent interior dynamics. On the other side, it becomes a smooth, monotonically varying function. These two distinct behaviors converge as the critical point is approached. Beyond the critical point, in the supercritical region, $p_t(T)$ develops a distinct extremum, defining a ''Kasner crossover line'' that is entirely independent of traditional thermodynamic (Widom line) or dynamic (Frenkel line) criteria. Our work establishes the black hole singularity as a novel class of diagnostics for phase transitions, revealing that a change in the macroscopic thermodynamic state fundamentally reshapes the deepest interior structure of spacetime.

We propose a general scheme to controllably distribute pairwise entanglement in a quantum network of qubits by exploiting environmental ancilla qubits interacting with the network nodes through tunable Hamiltonians. Our approach leverages collision models, in which a quantum syatem interacts sequentially with ancilla units. We explore two distinct scenarios within this framework: one in which the ancilla is reset to its initial coherent state after each interaction (the traditional collision model), and another where the ancilla is not reset but its state is simply carried over to the next interaction, which we dub the repeated interaction model. In both scenarios, we ensure that the system-ancilla correlations are discarded between steps. We also demonstrate how varying the ancilla-system interaction patterns enables selective generation of entanglement between different qubit pairs, including non-neighbouring nodes that do not directly interact. The scheme is analyzed in networks up to three qubits under both collision and repeated interaction dynamics, revealing the genaration of maximally entangled bell pairs even in configurations where the interacting ancilla couples to only a single node. Our results provide a systematic and physically implementable route to entanglement distribution, offering potential applications in quantum communication, metrology and modular quantum computing.

We theoretically investigate the control of the Goos-Hänchen shift (GHS) of a reflected weak probe field in both parity-time ($\mathcal{PT}$)-symmetric and conventional optomechanical systems. The proposed scheme consists of a single optomechanical platform where a passive optical cavity is coupled to an active mechanical resonator, in contrast to standard passive-passive configurations. Analysis of the eigenfrequency spectrum reveals the emergence of an exceptional point under balanced gain-loss conditions at a tunable effective optomechanical coupling strength. Using the transfer-matrix method combined with stationary-phase analysis, we examine the GHS across broken and unbroken $\mathcal{PT}$ phases and compare it with that in the conventional system. The lateral shift exhibits strong phase dependence: it is markedly enhanced in the unbroken regime relative to both the broken phase and the passive configuration. We further show that the GHS can be actively tuned through the cavity detuning and the intracavity medium length. These results provide a controlled means for manipulating beam shifts in optomechanical systems and suggest pathways toward tunable photonic components and precision optical sensing.

Magnetic Resonance Spectroscopy (MRS) offers a unique non-invasive window into metabolic processes, yet its potential remains strictly constrained by severe spectral congestion and intrinsic insensitivity. Traditional pulse sequence design, tethered to human intuition, predominantly targets simple quantum states, thereby overlooking the vast majority of the exponentially scaling operator space which consists of complex spin superpositions. Here, we introduce a spectrum-driven, end-to-end differentiable physical framework that transcends these heuristic limitations. By integrating physical laws with automatic differentiation algorithm, our approach directly navigates the high-dimensional spin dynamics space, bypassing the intractable inverse problem of state preparation. This enables the discovery of non-intuitive, complex mixed states that simultaneously satisfy the dual objectives of selective excitation and interferometric signal enhancement. We validate this paradigm by achieving the robust separation of Glutamate and Glutamine, which is a longstanding neuroimaging challenge, in the human brain at 3T, demonstrating spectral fidelity superior to conventional methods. By unlocking the "dark" informational content of nuclear spin ensembles, our work establishes a generalizable paradigm for goal-driven quantum state engineering in magnetic resonance and beyond.

We investigate the imprints of the Generalized Uncertainty Principle on cosmological scales by using redshift-space distortion measurements in combination with background cosmological data to determine constraints on the deformation parameter $β$. We consider the modified Poisson bracket related to the existence of a minimal length, which leads to a modified Raychaudhuri equation for the standard $Λ$CDM model and gives rise to a phenomenological one-parameter dynamical dark energy scenario. Through this modification, we can reveal the effects of the minimal length on the late-time structure of the universe. We employ the $f$ and $fσ_8$ measurements of the growth rate combined with background data, including cosmic chronometers, baryon acoustic oscillations and Type Ia supernova observations. The observational constraints reveal a systematically negative value for the deformation parameter $β$, with the $Λ$CDM limit lying within the 95\% credible interval. When supernova data are included, the Akaike Information Criterion indicates weak-to-strong support in favour of the GUP-modified model depending on the SNIa catalogue, while the Bayesian evidence suggests a weak preference.

We consider an application of the Multipolar Post Minkowskian formalism to the case of a two-body system in radial fall. We compute, within the post-Newtonian approximation, the associated gravitational waveform reaching the 2.5 Post-Newtonian accuracy level. At this level the presence of a radiation-reaction force manifests, modifying the fall with a corresponding bremsstrahlung radiation. We evaluate then all emissions: energy, angular momentum (vanishing identically) and linear momentum. We also evaluate the (nonlocal) inertial forces contributions appearing (at the next PN order, 4.5PN) in the center-of-mass due to the losses paving the way for future more accurate computations.

In arXiv:2507.07171, we demonstrate that the high-redshift supermassive black holes in the so-called "little red dots" discovered by James Webb Space Telescope (JWST) can be explained by the primordial black hole (PBH) clustering on small scales. In this paper, we present a comprehensive simulation of the successive PBH mergers within a cluster by solving the Smoluchowski coagulation equation. We derive the coagulation kernel considering both cases with and without the effects of mass segregation. Then we employ the Monte Carlo method to solve the equation, implementing the full-conditioning scheme using the discrete inverse transformation method. Our simulations determine the runaway timescales of clusters and the mass population evolution of PBHs across a wide range of cosmic redshifts, depending on the number of PBHs within the cluster and the associated density.

A modification of the Einstein-Hilbert Lagrangian by introducing a coupling between the Weyl tensor and the stress-energy tensor was proposed to explain flat galactic rotation curves without the exotic (non-baryonic) dark matter (DM) [1]. The proposed coupling constant was previously determined by fitting the rotational velocities of the Milky Way and M31 modeled with constant density, yielding the same coupling constant for both [2,3]. In this work, we have modified the formalism for a variable density by modeling the galactic systems with realistic, spherically symmetric and radially varying density profiles for the baryonic matter and this analysis is applied to seven edge-on spiral galaxies of the local cluster [4-10] and the Milky Way.

Quadratic gravity is a well-motivated extension of general relativity~(GR) wherein the Einstein-Hilbert action is augmented by quadratic curvature terms. This theory is equivalent to GR in an effective-field-theory framework, while the two theories are different at the non-perturbative level. As we have recently shown, black holes in quadratic gravity have a rich linear response, including extra scalar, vector, and tensor quasinormal modes that can be excited in physical processes, even when the stationary solution is the same as in GR. Here, by studying the gravitational-wave emission from point particles plunging into a Schwarzschild black hole in quadratic gravity, we show that observable deviations from GR are exponentially suppressed in the GR limit. This provides a nonperturbative realization of the equivalence between quadratic gravity and GR predicted in the effective-field-theory framework.

We propose a privacy-aware hybrid framework for federated medical image classification that combines tensor-network representation learning, MPC-secured aggregation, and post-aggregation quantum refinement. The framework is motivated by two practical constraints in privacy-aware federated learning: MPC can introduce substantial communication overhead, and direct quantum processing of high-dimensional medical images is unrealistic with a small number of qubits. To address both constraints within a single architecture, client-side tensor-network frontends, Matrix Product State (MPS), Tree Tensor Network (TTN), and Multi-scale Entanglement Renormalization Ansatz (MERA), compress local inputs into compact latent representations, after which a Quantum-Enhanced Processor (QEP) refines the aggregated latent feature through quantum-state embedding and observable-based readout. Experiments on PneumoniaMNIST show that the effect of the QEP is frontend-dependent rather than uniform across architectures. In the present setting, the TTN+QEP combination exhibits the most balanced overall profile. The results also suggest that the QEP behaves more stably when the qubit count is sufficiently matched to the latent dimension, while noisy conditions degrade performance relative to the noiseless setting. The MPC benchmark further shows that communication cost is governed primarily by the dimension of the protected latent representation. This indicates that tensor-network compression plays a dual role: it enables small-qubit quantum processing on compressed latent features and reduces the communication overhead associated with secure aggregation. Taken together, these results support a co-design perspective in which representation compression, post-aggregation quantum refinement, and privacy-aware deployment should be optimized jointly.

A fundamental challenge in quantum physics is determining the ground-state properties of many-body systems. Whereas standard approaches, such as variational calculations, consist of writing down a wave function ansatz and minimizing over the possible states expressible by this ansatz, one can alternatively formulate the problem as a noncommutative polynomial optimization problem. This optimization problem can then be addressed using a hierarchy of semidefinite programming relaxations. In contrast to variational calculations, the semidefinite program can provide lower bounds for ground state energies and upper and lower bounds on observable expectation values. However, this approach typically suffers from severe scalability issues, limiting its applicability to small-to-medium-scale systems. In this article, we demonstrate that leveraging the inherent structures of the system can significantly mitigate these scalability challenges and thus allows us to compute meaningful bounds for quantum spin systems on up to $16\times16$ square lattices.

We present a model in which the cosmological constant emerges as a purely geometric effect from the four-dimensional compactification of five-dimensional Einstein-Chern-Simons gravity. The compactification of the extra dimension generates an effective cosmological constant $Λ$ depending on the compactification radius $r_c$, the coupling parameter $l$, and the trace $\tilde{h}$ of the compactified field $h^a$, rather than being introduced as a free parameter. The resulting field equations are structurally equivalent to those of General Relativity with a cosmological constant, so all known vacuum solutions -- Schwarzschild--de Sitter, Kerr--de Sitter, and FLRW spacetimes -- remain valid. As a concrete application, we derive the Kottler (Schwarzschild--de Sitter) black hole solution. We identify two dynamical regimes. In the weak-field regime, $Λ\propto l^{2}\tilde{h}/r_{c}^{3}$, whose sign is controlled by $l^2\tilde{h}$, requiring fine-tuning to reproduce $Λ_{\text{obs}} \approx 10^{-52}\,\text{m}^{-2}$. In the strong-field regime, dependence on $l$ and $\tilde{h}$ cancels algebraically, yielding $Λ\approx 3/(4r_{c}^{2})$ independently of the Chern-Simons coupling. This regime naturally reproduces $Λ_{\rm obs}$ for $r_{c} \approx 0.78\,H_{0}^{-1} \approx 8.2 \times 10^{25}\,\text{m}$, without fine-tuning. The Bekenstein-Hawking entropy of the cosmological horizon gives $S_{\rm cosm} = 4πk_B r_c^2/l_{\rm Pl}^2 \sim 10^{122}\,k_B$, consistent with the Gibbons-Hawking result and admitting a direct geometric interpretation in terms of $r_c$. This framework geometrically reframes the cosmological constant problem: rather than asking why $Λ$ is small, one asks why $r_c$ is large -- a reformulation compatible with a large extra dimension without violating established gravitational tests.

Single-shot measurement learning (SSML) learns a compensation unitary from a one-bit success/failure record and halts after a prescribed run of consecutive successes. We recast SSML as an adaptive estimator on a parameterized sensing manifold and ask what role it can play in quantum-enhanced sensing. First, we show that the terminal run itself furnishes an intrinsic certificate of local alignment: longer terminal runs certify smaller infidelity, and near the optimum this becomes a Fisher-calibrated certificate of parameter error. Second, for compensation-type sensing families, the Bernoulli success/failure record is locally matched to the probe quantum Fisher information (QFI), so SSML preserves the probe's metrological content despite using only one classical bit per copy. In this sense, SSML makes the quantum enhancement carried by the probe operationally available in an online self-terminating protocol. Applied to GHZ/NOON probes of depth $m$, SSML retains the familiar square-root entanglement gain over product probes at fixed total resource, while an ideal multiscale architecture remains compatible with Heisenberg scaling. Monte Carlo simulations of photonic NOON-state phase sensing show the expected near-inverse decay of terminal infidelity with entangled shots, SQL-like total-resource scaling at fixed entanglement depth, the corresponding fixed-resource entanglement gain, the global limitation of a single fringe scale, and the recovery of Heisenberg-compatible behavior under ideal multiscale hand-off. These results identify SSML as a Fisher-preserving, self-certifying estimator layer for quantum-enhanced sensing.

We explore the connection between black hole shadows, thermodynamic phase structure, and microstructure of charged and rotating black holes within General Relativity and Geometrothermodynamics. Focusing on Reissner-Nordström and Kerr solutions, we establish a criterion to select the most suitable Geometrothermodynamic metric for a system, revealing that the first metric from enthalpy and the second from mass correctly reproduce heat capacity singularities. We show that the shadow radius encodes the same phase information as entropy and introduce Shadow-Microstructure diagrams to extract insights into stability and microscopic interaction types directly from observational bounds. Applying this framework to Sagittarius A*, we constrain the macroscopic parameters and the allowed microscopic thermodynamic phases. Our findings indicate that shadow measurements offer a novel probe of thermodynamic and microscopic aspects of black holes, enabling tests of alternative theories of gravity and thermodynamic frameworks.

We study the computational complexity of estimating the normalized trace $2^{-n}Tr[f(A)]$ for a log-local Hamiltonian $A$ acting on $n$ qubits. This problem arises naturally in the DQC1 model, yet its complexity is only understood for a limited class of functions $f(x)$. We show that if $f(x)$ is a continuous function with approximate degree $Ω({\rm poly}(n))$, then estimating $2^{-n}Tr[f(A)]$ up to constant additive error is DQC1-complete, under a technical condition on the polynomial approximation error of $f(x)$. This condition holds for a broad class of functions, including exponentials, trigonometric functions, logarithms, and inverse-type functions. We further prove that when $A$ is sparse, the classical query complexity of this problem is exponential in the approximate degree, assuming a conjectured lower bound for a trace variant of the $k$-Forrelation problem in the DQC1 query model. Together, these results identify the approximate degree as the key parameter governing the complexity of normalized trace estimation: it characterizes both the quantum complexity (via efficient DQC1 algorithms) and, conditionally, the classical hardness, yielding an exponential quantum-classical separation. Our proof develops a unified framework that cleanly combines circuit-to-Hamiltonian constructions, periodic Jacobi operators, and tools from polynomial approximation theory, including the Chebyshev equioscillation theorem.

We investigate vacuum decays in the early Universe in the presence of curvature perturbations. For sufficiently large perturbations associated with over-densities, we find that the bounce solution develops an oscillating middle stage near the bubble wall. For small perturbations, we analytically show within the thin-wall approximation that an over- (under-) density would enhance (suppress) the vacuum decay rate with a smaller (larger) initial bubble radius. By numerically solving for the bounce solutions and evaluating the corresponding Euclidean action, we further confirm this behaviour in thick-wall cases. Our results indicate that over-densities can generically trigger vacuum decay at an earlier moment.

Let $G$ be a strongly regular graph of prime order $p$ with connection degree $k \geq 6$. We prove that the \emph{quantum walk characteristic polynomial} $χ_q(G,λ) \coloneqq \det(λI - U_G)$, where $U_G$ is the coined quantum walk operator on $G$, completely determines $G$ up to isomorphism within the class of strongly regular graphs of the same order. The proof proceeds in three steps. First, we show that $U_G$ block-diagonalizes under the discrete Fourier transform over $\Z_p$, yielding $p$ blocks $U_G^{(j)}$ of size $k \times k$. Second, we prove an explicit formula \[ χ_q\!\bigl(U_G^{(j)}, λ\bigr) = (λ-1)^{(k-2)/2}(λ+1)^{(k-2)/2} \!\left(λ^2 - \tfrac{2\widehat{A}_G(j)}{k}\,λ+ 1\right), \] from which the Fourier coefficient $\widehat{A}_G(j)$ is recovered as the unique real part of an eigenvalue of $U_G^{(j)}$ distinct from $\pm 1$. Third, the inverse discrete Fourier transform recovers the connection set $S$ of $G$, and Turner's theorem (1967) identifies $G$ up to isomorphism. As a consequence, graph isomorphism is decidable in polynomial time within this class using the quantum walk spectrum, without resorting to the general quasi-polynomial algorithm of Babai (2016).

Characterising multi-time quantum processes is essential for analysing temporally correlated noise and for designing effective control and mitigation strategies. A complete operational description through multi-time process tomography requires an informationally complete set of probes, which necessarily includes non-deterministic intermediate operations. On present-day quantum devices, such operations are commonly implemented using mid-circuit measurements and reset, which are technologically limited and can introduce noise and overhead in terms of ancilla requirement. In this work, we study the minimal ancillary dimension required for complete characterisation of multi-time processes. We show that sequential interactions with a single qubit ancilla can generate an informationally complete family of correlated probes for processes of arbitrary length, without requiring mid-circuit measurements or reset. Our result provides a resource-efficient route for complete multi-time process tomography and establishes that one qubit of coherent ancillary memory suffices for full reconstruction of arbitrary multi-time dynamics.

We introduce a twisted fiber-bundle construction of quantum CSS codes over group algebras \(R=\mathbb F_2[G]\), where each base generator carries a generator-dependent \(R\)-linear fiber twist satisfying a flatness condition. This construction extends the untwisted lifted product code, recovered when all twists are identities. We show that invertible twists (satisfying a flatness condition) give a complex chain-isomorphic to the untwisted one, so the resulting binary CSS codes have the same blocklength \(n\) and encoded dimension \(k\). In contrast, singular chain-compatible twists can lower boundary ranks and increase the number of logical qubits. Examples over \(R=\mathbb F_2[D_3]\) show that the twisted fiber bundle code can outperform the corresponding untwisted lifted-product code in \(k\) while keeping the same \(n\) and, in our examples, the same minimum distance \(d\).

Classical shadows (CS) has recently emerged as an important framework to efficiently predict properties of an unknown quantum state. A common strategy in CS protocols is to parametrize the basis in which one measures the state by a random group action; many examples of this have been proposed and studied on a case-by-case basis. In this work, we present a unified theory that allows us to simultaneously understand CS protocols based on sampling from general group representations, extending previous approaches that worked in simplified (multiplicity-free) settings. We identify a class of measurement bases which we call "centralizing bases" that allows us to analytically characterize and invert the measurement channel, minimizing classical post-processing costs. We complement this analysis by deriving general bounds on the sample-complexity necessary to obtain estimates of a given precision. Beyond its unification of previous CS protocols, our method allows us to readily generate new protocols based on other groups, or different representations of previously considered ones. For example, we characterize novel shadow protocols based on sampling from the spin and tensor representations of $\textsf{SU}(2)$, symmetric and orthogonal groups, and the exceptional Lie group $G_2$.

Commutativity gadgets provide a technique for lifting classical reductions between constraint satisfaction problems to quantum-sound reductions between the corresponding nonlocal games. We develop a general framework for commutativity gadgets in the setting of quantum homomorphisms between finite relational structures. Building on the notion of quantum homomorphism spaces, we introduce a uniform notion of commutativity gadget capturing the finite-dimensional quantum, quantum approximate, and commuting-operator models. In the robust setting, we use the weighted-algebra formalism for approximate quantum homomorphisms to capture corresponding notions of robust commutativity gadgets. Our main results characterize both non-robust and robust commutativity gadgets purely in terms of quantum polymorphism spaces: in any model, existence of a commutativity gadget is equivalent to the collapse of the corresponding quantum polymorphisms to classical ones at arity $|A|^2$, and robust gadgets are characterized by stable commutativity of the appropriate weighted polymorphism algebra. We use this characterisation to show relations between the classes of commutativity gadget, notably that existence of a robust commutativity gadget is equivalent to the existence of a corresponding non-robust one. Finally, we prove that quantum polymorphisms of complete graphs $K_n$ have a very special structure, wherein the noncommutative behaviour only comes from the quantum permutation group $S_n^+$. Combining this with techniques from combinatorial group theory, we construct separations between commutativity-gadget classes: we exhibit a relational structure admitting a finite-dimensional commutativity gadget but no quantum approximate gadget, and, conditional on the existence of a non-hyperlinear group, a structure admitting a quantum approximate commutativity gadget but no commuting-operator gadget.

Ultralight bosons can form macroscopic gravitational-atom clouds around rotating black holes via superradiance, sourcing quasi-monochromatic gravitational waves through level transitions and annihilation. Primordial black holes provide a natural setting for such systems in a frequency range relevant for resonant-cavity experiments. We present a unified treatment of gravitational-wave emission from both isolated and binary-perturbed gravitational atoms in this regime. For isolated systems, we derive analytic expressions for the time- and frequency-domain strain from transition and annihilation channels, emphasizing their narrow-band structure. For binaries, we model resonantly driven level transitions using the Landau--Zener formalism and compute the resulting transient signals. We find that, while binary-driven transitions generically yield signals with durations compatible with detector response times, their characteristic strain lies well below the sensitivity of current experiments at astrophysically plausible distances, and event rates further suppress detectability by requiring sources at unrealistically small separations. We quantify the improvements in sensitivity, bandwidth, and response needed to render these signals observable, and identify gravitational-atom systems around primordial black holes as a theoretically well-motivated target for future high-frequency gravitational-wave searches.

The Tolman-Oppenheimer-Volkoff (TOV) equations for a massless fluid take the form of a pair of coupled autonomous first order differential equations, which can be employed in a model for a ``dynamical gravastar'' black hole mimicker. The mimicker has no true horizon, but rather a ``simulated horizon'', outside which the geometry resembles a Schwarzschild black hole, but inside which the $g_{00}$ component of the metric is always positive and becomes exponentially small. Collins has reinterpreted the relevant TOV equations in terms of a two-dimensional flow with a spiral form, and Zöllner and Kämpfer have mapped the simulated horizon to a specific segment of the Collins spiral. We give here results of an asymptotic analysis, relating initial values at the small radius end of this spiral segment to the black hole mimicker mass and other parameters that emerge at the large radius kink in the TOV solution, which corresponds to the simulated horizon.

Resource estimation is a significant challenge in evaluating fault tolerant quantum computers. Existing approaches often rely on either fixed architectural assumptions or coarse analytical models that fail to capture the interaction between hardware constraints and circuit compilation. This challenge is particularly acute for neutral atom quantum computers, where architectural features such as atom movement, measurement zones, and multi-species arrays introduce a broad design space for implementing fault tolerant computation. Addressing the need for a tighter feedback loop between hardware design and practical application development, we present a compilation-driven framework for quantum resource estimation that translates arbitrary quantum circuits into logical primitive operations with known physical resource costs. This framework allows for easily configurable hardware assumptions that enable rapid comparison of different architectural design choices. We apply our approach to two early fault tolerant quantum simulation and optimization workloads, assuming the use of the surface code, revealing several architectural trends. While the production of magic states continues to be the dominant source of overhead for these benchmarks, access to movement can save time on cultivation and important transversal gates. As problem size grows, routing and qubit movement become dominant bottlenecks, highlighting the need for movement-aware compiler optimizations and frugal routing strategies. Finally, our results suggest that neutral atom architectures combining dual-species arrays with controlled qubit movement offer a promising path toward near-term advantage on fault tolerant devices.

General-purpose programmable photonic processors offer a flexible foundation for integrating various functionalities within a single chip. A two-dimensional hexagonal waveguide mesh of Mach Zehnder interferometers has been shown to have great potential in the field of microwave photonics. Additionally, they are a promising platform for the creation of unitary linear transformations, which are key elements in photonic neural networks, In this article, we expand the portfolio of available applications for recirculating bricks mesh architecture to quantum technologies. We will show that a single programmable optical system is capable of performing various functions depending on the requirements. In particular, we will focus in this work on boson sampling, a task that best demonstrates quantum advantage, as well as on tasks that enable the determination of photon indistinguishability, which plays a key role in photonic quantum technologies. We will also show that, in addition to spatial modes, the same optical system can be equally well-suited for work on temporal modes through the implementation of an appropriate number of loops.

We study algorithms for approximating the permanent of a random matrix when the entries are slightly biased away from zero. This question is motivated by the goal of understanding the classical complexity of linear optics and \emph{boson sampling} (Aaronson and Arkhipov '11; Eldar and Mehraban '17). Barvinok's interpolation method enables efficient approximation of the permanent, provided one can establish a sufficiently large zero-free region for the polynomial $\mathrm{per}(zJ + W)$, where $J$ is the all-ones matrix and $W$ is a random matrix with independent mean-zero entries. We show that when the entries of $W$ are standard complex Gaussians, all zeros of the random polynomial $\mathrm{per}(zJ + W)$ lie within a disk of radius $\tilde{O}(n^{-1/3})$, which yields an approximation algorithm when the bias of the entries is $\tildeΩ(n^{-1/3})$. Previously, there were no efficient algorithms at biases smaller than $1/\mathrm{polylog}(n)$, and it was unknown whether there typically exist zeros $z$ with $|z| \ge 1$. As a complementary result, we show that the bulk of the zeros, namely $(1 - ε)n$ of them, have magnitude $Θ(n^{-1/2})$. This prevents our interpolation method from contradicting the conjectured average-case hardness of approximating the permanent. We also establish analogous zero-free regions for the hardcore model on general graphs with complex vertex fugacities. In addition, we prove universality results establishing zero-free regions for random matrices $W$ with i.i.d. subexponential entries.

We construct a broad class of completely positive maps and Go\-rini--Kossakowski--Sudarshan-Lindblad generators for fermionic systems induced by linear transformations of system and environment modes. For these maps, we derive explicit Heisenberg-picture formulas for arbitrary normally ordered monomials in terms of minors of the underlying mode-transformation matrices and environment correlation tensors. We show that for even environment states the linear span of monomials up to any fixed order is invariant, which yields closed equations for low-order moments and makes their computation efficient. We also discuss the relation of this construction to second quantization of non-Hermitian one-particle contractions and extend the formalism to completely positive maps arising from post-selection.

We present an exact solution in the large-$N$ limit of the Lévy Sachdev-Ye-Kitaev (LSYK) model introduced in Ref. [1], wherein the couplings are drawn from a Lévy Stable distribution parameterized by a tail exponent $μ\in [0, 2]$. Starting from the Hamiltonian and its associated partition function, we highlight the key differences from the standard Gaussian SYK model and derive the large-$N$ Schwinger-Dyson equations via a bosonic oscillator representation of the action. These equations are solved both numerically and analytically in the large-$q$ and infrared limits. We subsequently analyze the chaotic properties of the model by computing the Krylov exponent from the large-$q$ Green's function and extracting the Lyapunov exponent from the $4$-point function. The parameter $μ$ continuously interpolates between a free theory at $μ= 0$ and the conventional, maximally chaotic Gaussian SYK model at $μ= 2$, with non-maximal chaos persisting throughout the intermediate regime $0 < μ< 2$. Thermodynamic quantities, including the entropy, free energy, average energy, and specific heat capacity, are computed and compared with their Gaussian SYK counterparts. The interpretations of the thermodynamics are discussed with respect to the holographic dual and non-Fermi liquid theory. Finally, we discuss an alternative representation of the LSYK model based on a distinct decomposition of the Lévy Stable distribution, which establishes a non-trivial connection to Gaussian SYK, and provide supporting analytical and numerical results in the appendices.

Achieving fast, excitation-free quantum control is a vital challenge in modern quantum technologies. In many cases, shortcuts to adiabaticity enable fast adiabatic-like protocols, yet determining control parameters that satisfy practical constraints is often challenging in complex systems. Here, we combine an analytical shortcut to adiabaticity approach with several numerical optimization methods to boost the performance of the protocol. As a proof-of-principle for this hybrid approach, we study a particularly intricate control problem, the separation of two trapped ions. We show that this analytical-numerical approach, along with the physical insight gained through the variety of suboptimal solutions, leads to the exploration of new solutions in a complex landscape that yield improvements of up to 3 orders of magnitude. Moreover, this improvement comes with no additional cost from an experimental point of view.

In the GEVAG (Generalized Entropy Varying-G) framework, any generalization to horizon entropy leads to a varying gravitational "constant" $G_\text{eff}$ that is a function the horizon area. In this work, it is shown that if we promote $G_\text{eff}$ to be valid in the neighborhood of the horizon, then Hawking temperature consists of two terms, the second of which is related to the variation of $G_\text{eff}$. When applied to the logarithmic correction of the entropy, as is common across various quantum gravity approaches, the first term in the Schwarzschild black hole temperature exactly agrees with that obtained from utilizing generalized uncertainty principle (GUP), while the second term improves on the GUP result by driving the Hawking temperature to zero as the black hole approaches a minimum mass. This resolves the inconsistency in the GUP result concerning a nonzero temperature minimum mass remnant. This work also derives simple general formulas for both the thermodynamic energy and the Bekenstein bound for any correction to the area law under the same assumption that $G_\text{eff}$ can be extended off-shell in the horizon neighborhood. The (generalized) Bekenstein bound can be interpreted as a statement regarding the renormalization group scaling dimension of the entropy functional $f(A)$ and the naturalness of the theory.

The origin of cosmic structure is widely regarded as quantum, yet the Universe today appears classical. Standard lore attributes this to a "quantum-to-classical" transition on super-horizon scales during inflation. Gravity plays a central role: super-horizon dynamics squeeze quantum states, while the cosmological horizon enforces a system-environment split, leading to decoherence. But are these mechanisms always sufficient? We revisit this question, identifying assumptions and limitations in conventional arguments. We highlight recent work showing that beyond slow roll, non-linear dynamics of cosmological perturbations can generate non-classical features that may survive in observables. This raises the tantalizing possibility that quantum signatures may persist in cosmic structure. We propose a phase-space analysis based on the Wigner function as a concrete route to identifying and probing such signatures.

An important measure of utility for a quantum code is the identification of which logical operations can be implemented fault-tolerantly on its codespace. We introduce a framework which leverages the automorphism groups of associated classical codes, the choice of logical basis and exploitation of code equivalence to construct all distinct implementable realisations of each valid logical operation for a given $[[n,k,d]]$ code. We establish conjugacy classes and group transversals (unrelated to transversality) as key explanatory concepts. We subsequently motivate and calculate two figures-of-merit that can be optimised with this framework. Our results yield a table of optimal logical operations and their corresponding physical circuits for all small stabiliser codes with $n \leq 7$ and $k \leq 2$, drawn from quantum databases. This exhaustive table of results provides the optimal physical implementations of logical operations which may be advantageous for both magic state cultivation and experimental purposes.

Spontaneous scale symmetry breaking is commonly associated with a flat direction in the action. We show that this need not be so if the dilaton is coupled to a three-form field in a manner compatible with gauge invariance and dilatations. The resulting effective dynamics lifts the flat direction without introducing explicit scale-violating operators. When gravity is included, the corresponding potential takes the form of an exponential plateau.

Pandres has developed a theory which extends the geometrical structure of a real four-dimensional space-time via a field of orthonormal tetrads with an enlarged covariance group. This new group, called the conservation group, contains the group of diffeomorphisms as a proper subgroup. The free-field Lagrangian density involves only the curvature vector which is a vector which measures curvature. When massive objects are present a source term is added to this Lagrangian density. The weak-field approximation implies that gravitational waves travel at the speed of light. Spherically symmetric solutions for both the free field and the field with sources are found. In the free-field case, the field equations require nonzero stress-energy tensors. However, we find that for our model to be an acceptable model, we must have a source term in the Lagrangian. In our framework, we divide up the galaxy into three spherically symmetric regions: a baryonic matter-dominated central bulge, a dark matter-dominated mesosphere and an outside region where neither type dominates. Assuming the density of baryonic matter has a central cusp, we show how to model the bulge. Via an isothermal condition we find a model for the mesosphere and show this model implies flat rotation curves with one free parameter. The outside region is readily modeled via previously published results. The models for the bulge, mesosphere and outside region are combined into one continuous model. Using the radial acceleration relation we then show how a galaxy model may be set up for a rotationally supported galaxy.

We construct a static, spherically symmetric black hole (BH) solution embedded in a cored Plummer dark matter (DM) halo and a Letelier cloud of strings (CoS). Starting from the Plummer-Schwarzschild metric of Senjaya et al.~\cite{Senjaya2026}, we incorporate the string-cloud tension parameter $α$ into the lapse function, obtaining $A(r) = h_{\rm Plummer}(r) - α$. The resulting spacetime admits a single, non-degenerate event horizon (EH) for $α< 1$ and a naked singularity for $α\ge 1$. We determine the photon sphere (PS) and BH shadow radii, compute the weak deflection angle via the Gauss-Bonnet theorem (GBT), and analyze the innermost stable circular orbit (ISCO). Scalar perturbations are studied through the effective potential, greybody factor (GF) bounds obtained from the Boonserm-Visser method, the Hawking emission spectrum, and quasinormal mode (QNM) frequencies computed with the WKB approximation. The thermodynamic analysis covers the Hawking temperature, Bekenstein-Hawking entropy, heat capacity, and Gibbs free energy; the heat capacity is found to be strictly negative for all parameter values, confirming the absence of any Davies-type phase transition. A consistent hierarchy emerges across all six analyses: the CoS tension $α$ governs the leading-order modifications to every observable, while the Plummer halo density $ρ_0$ provides a subdominant, additive correction.

Traditional perturbation theory, based on local analyticity (Taylor expansion), often fails in many-body systems with exponentially small energy gaps and strong interactions. This work presents an alternative methodological framework built on two core principles: (1) starting from the pole expansion of the resolvent to directly capture the global analytic structure, and (2) treating local fluctuations statistically (in the spirit of the eigenstate thermalization hypothesis) to close the mean-field equations. Crucially, we go beyond the mean-field level by deriving an exact recursive re-expansion of the cross-correlated terms, which systematically generates higher-order corrections that control the distribution tails, branch splitting, and fluctuations. The framework is realized through a hierarchical ansatz strategy, solving self-consistent equations with Lorentzian, Gaussian, and hybrid forms to describe the bulk, tail, and full distribution, respectively. This methodology does not rely on weak-coupling assumptions and is applicable to the quantitative analysis of global properties such as entropy production and distribution functions in nonintegrable many-body systems. We detail its mathematical structure, the recursive expansion of fluctuations, conditions of validity, comparison with traditional methods, and provide a general implementation workflow.

Routing in wireless communication networks is shaped by mobility, interference, congestion, and competing service requirements, making route selection a high-dimensional constrained optimization problem rather than a simple shortest-path task. This paper investigates the use of hybrid classical--quantum methods for wireless routing, focusing on the Quantum Approximate Optimization Algorithm (QAOA) and quantum walks as candidate mechanisms for exploring complex routing spaces. The paper examines how wireless routing can be expressed as a constrained graph optimization problem in which routing objectives, flow constraints, connectivity requirements, and interference effects are mapped into quantum-compatible Hamiltonian representations. It then discusses how these approaches can be integrated into a hybrid architecture in which classical systems perform network monitoring, graph construction, pre-processing, and deployment, while quantum subroutines are used for selected optimization components. The analysis shows that the potential value of quantum routing lies primarily in the treatment of difficult combinatorial subproblems rather than end-to-end replacement of classical routing frameworks. The paper also highlights practical limitations arising from state preparation, constraint encoding, oracle construction, hardware noise, limited qubit resources, and hybrid execution overhead. It is argued that any meaningful near-term advantage will depend on careful problem decomposition, compact encoding, and tight classical--quantum integration.

In this study, we propose an efficient quantum multiplication approach based on a QFT-assisted parallelized addition scheme. The multiplication stage is implemented using a structure composed entirely of Toffoli gates, which generate partial products. In the second stage, these partial results are accumulated using a QFT-based adder. Unlike conventional QFT-based arithmetic circuits, the proposed design eliminates the repeated application of QFT and inverse QFT (IQFT) operations during intermediate summation processes. This leads to a significant reduction in the total gate count and circuit complexity, enabling a more resource-efficient implementation. To demonstrate the feasibility of the proposed approach, a quantum circuit that performs the multiplication of two 3-bit numbers is designed. The circuit is tested and validated using IBM quantum simulators. The results indicate that the proposed method provides a more efficient alternative to traditional quantum multiplication techniques in terms of gate cost and circuit depth.

This paper investigates the optical images of neutron stars within the framework of the radiatively inefficient accretion flow model, taking into account a polytropic equation of state. After obtaining the numerical solutions of the neutron star, we solved numerically the geodesic equations together with the radiative transfer equation. We mainly examine the effects of the polytropic index $N$ and the observer inclination angle $θ_o$ on the image morphology. The obtained images are also compared with the shadow of a Schwarzschild black hole. It is shown that, under the assumption that photon trajectories are terminated at the neutron star surface, the image exhibits a bright higher order structure surrounding an inner dark region. As $N$ increases, the size of the higher-order image gradually expands. As $θ_o$ increases, the obscuration of the neutron star silhouette by radiation originating outside the equatorial plane becomes more pronounced. Compared with the black hole shadow obtained under the same parameter configuration, the neutron star exhibits a larger higher order image and a more extended obscured inner dark region, whereas the higher order image of the black hole is more readily distinguishable. These results indicate significant differences in the optical appearance of neutron stars and black holes, and thus provide a theoretical basis for distinguishing between them through high resolution imaging.

For Hamiltonian systems, level statistics provide a faithful diagnostic of quantum chaos. By analogy, the statistics of the Lindbladian spectrum are often used in open quantum systems, and the Grobe-Haake-Sommers conjecture proposes that systems with chaotic classical counterparts should exhibit level repulsion in the Lindbladian spectrum. Here we point out an important flaw in this analogy: Hamiltonian and Lindbladian spectra behave differently and have distinct physical interpretations, and one should therefore not expect the latter to provide a reliable diagnostic. For Lindbladians, the late-time dynamics are not determined by the bulk of the eigenvalues but only by those eigenvalues -- and their corresponding eigenvectors -- with small real parts. Combined with the strong non-normality typical of Lindbladians, this allows situations in which the level statistics can be tuned almost arbitrarily without affecting the dynamics on either short or long time scales. We explicitly demonstrate this phenomenon and provide examples in which Ginibre level repulsion arises while the system dynamics at no time show signatures of chaos. We further relate this mechanism to the emergence of a non-Hermitian skin effect in Liouville space, linking boundary-induced eigenvector localization to the observed spectral instability. Our results show that level statistics cannot universally serve as a reliable diagnostic of quantum chaos in open quantum systems and highlight the need for alternative diagnostics that remain robust in strongly non-normal regimes.

Superconducting quantum processor units (QPUs) are incapable of producing massive datasets for quantum error correction (QEC) because of hardware limitations. Thus, QEC decoders heavily depend on synthetic data from qubit error models. Classic depolarizing error models with polynomial complexity present limited accuracy. Coherent density matrix methods suffer from exponential complexity $\propto O(4^n)$ where $n$ represents the number of qubits. This paper introduces PAEMS: a precise and adaptive qubit error model. Its qubit-wise separation framework, incorporating leakage propagation, captures error evolvements crossing spatial and temporal domains. Utilizing repetition-code experiment datasets, PAEMS effectively identifies the intrinsic qubit errors through an end-to-end optimization pipeline. Experiments on IBM's QPUs have demonstrated a 19.5$\times$, 9.3$\times$, and 5.2$\times$ reduction in timelike, spacelike, and spacetime error correlation, respectively, surpassing all of the previous works. It also outperforms the accuracy of Google's SI1000 error model by 58$\sim$73\% on multiple quantum platforms, including IBM's Brisbane, Sherbrooke, and Torino, as well as China Mobile's Wuyue and QuantumCTek's Tianyan.

Neural network field theory formulates field theory as a statistical ensemble of fields defined by a network architecture and a density on its parameters. We extend the construction to topological settings via the inclusion of discrete parameters that label the topological quantum number. We recover the Berezinskii--Kosterlitz--Thouless transition, including the spin-wave critical line and the proliferation of vortices at high temperatures. We also verify the T-duality of the bosonic string, showing invariance under the exchange of momentum and winding on $S^1$, the transformation of the sigma model couplings according to the Buscher rules on constant toroidal backgrounds, the enhancement of the current algebra at self-dual radius, and non-geometric T-fold transition functions.

Symmetry and entanglement are two fundamental concepts in quantum many-body physics. Their interplay is captured by symmetry-resolved entanglement, which decomposes the total entanglement into contributions from different symmetry sectors. Computing symmetry-resolved entanglement in strongly interacting higher-dimensional quantum systems remains challenging. Here, we introduce a quantum Monte Carlo (QMC) approach for computing symmetry-resolved Rényi entropies (SRRE) in large-scale interacting systems by measuring disorder (symmetry-twisted) operators on replica manifolds and reconstructing SRRE from the corresponding charged moments. We apply this method to the transverse-field Ising model (TFIM) in one and two dimensions. In one dimension, we recover the conformal-field-theory prediction for the logarithmic scaling of the disorder operator and observe the expected approach to entanglement equipartition. In two dimensions, our data provide numerical evidence consistent with entanglement equipartition at the (2+1)D Ising critical point. We further apply the framework to the 1D Heisenberg chain and obtain results consistent with the expected asymptotic scaling and finite-size corrections in the U(1)-resolved sectors. Our work establishes a practical numerical route to symmetry-resolved entanglement in interacting lattice models and provides a useful framework for future studies beyond one dimension.

Models of gravity in warped extra dimensions enjoy invariance under diffeomorphism. We derive the nonlinear transformation rules for the metric perturbations in the unitary gauge. As an off-shell symmetry, the main consequence of diffeomorphism is a set of recursive relations linking consecutive orders in the field expansion of the effective Lagrangian. The physical consequences are briefly explored for the Randall-Sundrum model with hard branes.

To harness the power of Language Models in answering domain specific specialized technical questions, Retrieval Augmented Generation (RAG) is been used widely. In this work, we have developed a Q\&A application inspired by the Retrieval Augmented Generation (RAG), which is comprised of an in-house database indexed on the arXiv articles related to the Electron-Ion Collider (EIC) experiment - one of the largest international scientific collaboration and incorporated an open-source LLaMA model for answer generation. This is an extension to it's proceeding application built on proprietary model and Cloud-hosted external knowledge-base for the EIC experiment. This locally-deployed RAG-system offers a cost-effective, resource-constraint alternative solution to build a RAG-assisted Q\&A application on answering domain-specific queries in the field of experimental nuclear physics. This set-up facilitates data-privacy, avoids sending any pre-publication scientific data and information to public domain. Future improvement will expand the knowledge base to encompass heterogeneous EIC-related publications and reports and upgrade the application pipeline orchestration to the LangGraph framework.

In a class of theories, dark matter is explained by postulating the existence of a `dark sector', which interacts gravitationally with ordinary matter. If this dark sector contains a U(1) symmetry, and a corresponding `dark' photon ($A_{D}$) , it is natural to expect that this particle kineticly mix with the ordinary photon, and hence become a `portal' through which the dark sector can be studied. The strength of the mixing is given by a mixing parameter $(ε)$. This same parameter governs both the production and the decay of the $A_{D}$ back to SM particles, and for values of $ε$ not already excluded, the signal would be a quite small, and quite narrow resonance: If $ε$ is large enough to yield a detectable signal, its decay width will be smaller than the detector resolution, but so large that the decay back to SM particles is prompt. For masses of the dark photon above the reach of Belle II, future high energy e$^+$e$^-$ colliders are ideal for searches for such a signal, due to the low and well-known backgrounds, and the excellent momentum resolution and equally excellent track-finding efficiency of the detectors at such colliders. This contribution will discuss a study investigating the dependency of the limit on the mixing parameter and the mass of the $A_{D}$ using the $A_{D}\rightarrowμ^{+}μ^{-}$ decay mode in the presence of standard model background, using fully simulated signal and background events in the ILD detector at the ILC Higgs factory. In addition, a more general discussion about the capabilities expected for generic detectors at e$^+$e$^-$ colliders operating at other energies will be given.

In particle physics, as in many areas of science, parameter inference relies on simulations to bridge the gap between theory and experiment. Recent developments in simulation-based inference have boosted the sensitivity of analyses; however, biases induced by simulation-data mismodeling can be difficult to control within standard inference pipelines. In this work, we propose a Template-Adapted Mixture Model to confront this problem in the context of signal fraction estimation: inferring the population proportion of signal in a mixed sample of signal and background, both of which follow arbitrarily complex distributions. We harness many biased simulations to perform data-driven estimates of each process distribution in the signal region, substantially reducing the bias on the signal fraction due to the domain shift between simulation and reality. We explore different methodological choices, including model selection, feature representation, and statistical method, and apply them to a Gaussian toy example and to a semi-realistic di-Higgs measurement. We find that the presented methods successfully leverage the biased simulations to provide estimates with well-calibrated uncertainties.

Recent ACT measurements favor a scalar spectral index n_s larger than the Planck value, posing a challenge to many single-field slow-roll inflation models. We show that a smooth, localized step in the inflaton potential can shift the predicted values of n_s and r by displacing the field value at which the CMB pivot scale exits the horizon. This mechanism can move monomial and, in particular, plateau-like attractor models toward the ACT-favored region, whereas the induced shift remains insufficient in natural inflation.

A large solid angle magnetic spectrometer for high luminosity and forward scattering angles was constructed at the Thomas Jefferson National Accelerator Facility. A number of physics experiments have used this spectrometer, and a significant physics program of future experiments has already been approved. A key feature of the spectrometer concept is a horizontal slit opening that allows the beamline to pass through the yoke of the spectrometer magnet. This design enables a short distance between the target and spectrometer, resulting in a 70~msr solid angle acceptance. The residual magnetic-field on the beamline inside the slit is reduced by a two-layer magnetic shielding system, with the external layer comprising a set of iron rings. Two correcting magnets, before and after the dipole, were used to compensate for the transverse component of the fringe field outside of the dipole yoke. The mechanical stability of the tall dipole magnet in close proximity to the target was provided by means of a heavy counterweight.

A search for Higgs boson pair production is presented, targeting final states where one Higgs boson decays to a pair of bottom quarks and the other Higgs boson decays to two W bosons, both of which decay leptonically, to an electron or a muon, and a neutrino. For the first time, the search is conducted with proton-proton collision data from the LHC at $\sqrt{s}$ = 13.6 TeV, recorded with the CMS detector in 2022 and 2023 and corresponding to an integrated luminosity of 62 fb$^{-1}$. The results are consistent with the standard model predictions. An upper limit of 12.0 times the standard model prediction at 95% confidence level is set on the Higgs boson pair production cross section, with an expected limit of 18.5. The results are also used to constrain the strength of the trilinear self-coupling of the Higgs boson, as well as of the quartic coupling between two Higgs bosons and two vector bosons.

We investigate the symmetry structure of the $3+1$ D staggered fermion Hamiltonian and its implications for anomalies. Since the spin and flavor degrees of freedom of Dirac fermions are distributed over the lattice, in addition to the standard on-site mass term, the staggered fermion system also admits one-, two-, and three-link bilinear terms within a unit cube as local, charge conserving mass terms with different spin and flavor dependence. We identify the spin flavor structures of all those bilinear mass terms and determine the symmetries preserved by each of them. Among them, one of the one-link mass terms preserves a larger residual symmetry associated with conserved charges that generate the Onsager algebra. Motivated by this structure, we consider a kink profile of the one-link mass and analyze the resulting domain-wall system. In the low-energy limit, the $3+1$ D bulk becomes gapped, while two-flavor massless Dirac fermions appear as localized modes on the $2+1$ D domain wall. We show that the bulk conserved charges act on the wall as generators of a flavor $\mathrm{SU}(2)$ symmetry, and that no symmetric mass gap is allowed for the boundary theory when this $\mathrm{SU}(2)$ symmetry and space reflection symmetry are both imposed. This realizes the parity anomaly of the boundary theory and shows that the boundary flavor symmetry and anomaly descend from the ultraviolet staggered-fermion Hamiltonian rather than emerging only in the infrared.

We construct the functional integral of Abelian Chern-Simons theory with toral gauge group $\mathbb T=\mathfrak t/Λ\cong U(1)^n$ at level $K$, where $K:Λ\timesΛ\to\mathbb Z$ is an even, integral, nondegenerate symmetric bilinear form, by exact zeta-regularized Gaussian evaluation of the formal quotient integral over connections modulo gauge. For closed $3$-manifolds, this yields a topological invariant; for manifolds with boundary, the relative functional integral produces the canonical boundary state. The resulting theory satisfies the required axioms of a $(2+1)$-dimensional TQFT.

Understanding axion production in the early Universe remains a pivotal challenge, given the axion as a compelling cold dark matter candidate. Conventional misalignment scenarios often overlook the possibility that a large initial axion velocity can fundamentally reshape the subsequent evolution of the axion field. In this letter, we provide a comprehensive analysis of how primordial magnetic fields impact the axion relic abundance. By accounting for the axion coupling to the Chern-Simons term of the hypercharge gauge field, the equation of motion of the axion is recast as a driven oscillator equation. This modification effectively shifts the onset of axion oscillations, leading to a significant reevaluation of the final relic abundance, a novel effect we term the axion helical misalignment mechanism. Furthermore, in the presence of primordial chiral asymmetries, the chiral magnetic effect (CME) emerges as a critical driver of axion dynamics. The interplay between the axion field and the CME not only profoundly influences the evolution of Standard Model chiral fermions but also provides a viable pathway for generating the observed baryon asymmetry of the Universe.

In this kick-off presentation for the "Recent developments in QCD" session at Baryons 2025 I will tie together the recent progress made on the extraction of parton distribution functions (PDFs) in lattice QCD and the long standing efforts in solving the inverse problem in the form of spectral function reconstruction.

Recent measurements of the $t\bar{t}$ cross section, performed both inclusively and differentially by the ATLAS and CMS Collaborations, are reported. In particular, off-shell effects are probed in the $pp\to W^+bW^-\bar{b}$ and $pp\to e^\pmμ^\mp +b\bar{b}$ processes, and modelling aspects of the POWHEG bb4$\ell$ Monte Carlo generator are discussed. Cross section and properties measurements are also performed at the threshold: we review an indirect extraction of the top quark Yukawa coupling, as well as the recent observations by both experiments of an excess of events near the top pair production threshold that is consistent with the formation of quasi-bound states.

In this work, we investigate the worldsheet symmetry of bosonic brane theories and its quantum consistency in the tensionless limit. We find that the residual worldsheet symmetry after specific gauge fixing is generated by a novel algebra, denoted as $g^{(p)}_λ$. To achieve full quantization of the tensionless brane, we introduce a $bc$ ghost system and derive the overall BRST charge. Moreover, we calculate the quantum anomaly of the $g^{(p)}_λ$ algebra for general parameters $p$ and $λ$ in the framework of canonical quantization. After demanding that this quantum anomaly vanishes, we successfully derive the critical dimensions of the bosonic brane theories. Especially, we obtain nontrivial solutions: $p=3$ in $D=4$ spacetime dimensions when $λ=-3$ and $p=6$ in $D=7$ spacetime dimensions when $λ=3$.

We investigate the leading-power fragmentation of triply heavy $Ω$ baryons in high-energy hadronic collisions. Extending our previous work on the $Ω_{3c}$ sector, we release the full OMG3Q1.0 family of collinear fragmentation functions by completing the description of the charm channel and delivering the novel $Ω_{3b}$ functions. These hadron-structure-oriented functions are constructed from improved proxy-model calculations for heavy-quark and gluon fragmentation, matched to a flavor-aware DGLAP evolution based on the HF-NRevo scheme. For phenomenological applications, we employ the (sym)JETHAD multimodular interface to compute and analyze NLL/NLO$^+$ semi-inclusive $Ω_{3Q}$ plus jet distributions at the HL-LHC and FCC. This work consolidates the link between hadron structure, rare baryon production, and resummed QCD at the energy frontier.

We investigate the core dynamics behind exotic matter formation via the TQ4Q1.1 set of collinear fragmentation functions for fully charmed or bottomed tetraquarks in three quantum configurations: scalar ($0^{++}$), axial vector ($1^{+-}$), and tensor ($2^{++}$). We adopt leading-power single-parton fragmentation within a nonrelativistic QCD framework tailored to tetraquark Fock states. Initial-scale inputs are constructed from updated gluon- and heavy-quark channels, and evolved through threshold-consistent DGLAP within HF-NRevo. We present the first systematic propagation of uncertainties from color-composite long-distance matrix elements governing tetraquark hadronization. This study advances the connection between hadronic structure, precision QCD, and exotic matter.

We investigate the high-energy production of tetraquark-jet systems at the LHC and its forthcoming High-Luminosity upgrade. In this review, we examine the leading-power fragmentation of fully heavy tetraquarks ($T_{4Q}$) in hadronic collisions, highlighting their relevance as novel probes of multiquark dynamics in QCD. Our analysis relies on the hadron-structure-oriented TQ4Q1.1 fragmentation functions, built within a nonrelativistic QCD framework that incorporates both gluon- and heavy-quark-initiated channels. Threshold-consistent DGLAP evolution is performed through the HF-NRevo scheme, enabling a unified treatment of mass thresholds and scale variations. We also provide a systematic discussion of uncertainties arising from color-composite long-distance matrix elements (LDMEs) and from perturbative hard- and fragmentation-scale inputs (H- and F-MHOUs). Phenomenological predictions are obtained using the (sym)Jethad framework at NLL/NLO$^+$ accuracy for semi-inclusive tetraquark-jet production at the LHC and beyond. This review connects the emerging spectroscopy of fully heavy exotics with modern fragmentation-based approaches to hadron structure and high-energy QCD.

We examine the Regge theoretical properties for the scaling observed in pp elastic scattering differential cross-sections at the LHC. A positive signature amplitude (i.e. the Pomeron) with scaling properties has been derived. It is found to describe the dip-bump region of momentum transfer at LHC energies in agreement with data. We derive the analytic continuation in the whole plane of the t-channel partial waves of index $l_t$ specific to the Regge formalism. The analytic form of the amplitude exhibits a specific scaling property without singularities, except for a series of poles in the $l_t$ real axis at fractional values.

We study the light asymptotic limit of the one-point torus conformal block in $A_{n-1}$ Toda field theory. Through the AGT correspondence, this problem can be translated into the computation of the instanton partition function of four-dimensional ${\cal N}=2^{\ast}$ $U(n)$ supersymmetric Yang--Mills theory, which we then examine in the limit $b\to 0$ at fixed conformal dimensions. We show that, in this regime, the instanton sum simplifies drastically: for each Young diagram, only boxes with specific arm lengths contribute to the bifundamental factors. Exploiting this property, we derive an explicit representation for the light one-point torus $W_n$ conformal block valid for arbitrary $n\ge 2$. As a consistency check, we specialize our construction to the Liouville case $n=2$ and compare it with the previously known hypergeometric representation of the torus block in the light limit. We also discuss the $W_3$ case and its relation to a known alternative representation obtained by the shadow formalism.

Exclusive semileptonic and nonleptonic $J/ψ$ decays are investigated in the framework of the relativistic quark model based on the quasipotential approach and quantum chromodynamics. The form factors parameterizing the hadronic matrix element of the weak current are calculated with the complete account of the relativistic effects. These form factors are expressed as the overlap integrals of the meson wave functions and are determined in the whole accessible kinematic range. On this basis the semileptonic decay branching fractions are evaluated for decays involving both electrons and muons. The nonleptonic decays are considered in the factorization approximation in the limit for the number of colors $N_c\to \infty$. The obtain branching fractions are found to be of the order $10^{-9}\sim 10^{-12}$. They are compared with the previous theoretical predictions and available experimental upper bounds.

We study the $l^-l^+\to ν\barνZh$ process for future lepton colliders, whose cross section becomes larger than that for $l^-l^+\to Zh$ in the energy region above a few TeV. % We classify the amplitudes into three main groups based on the topology of each Feynman diagram; vector boson scattering, $l^-W^+$ scattering, and $W^-l^+$ scattering, and study the interference patterns among the amplitudes. % We show that subtle gauge cancellation among the amplitudes at high energies in the unitary gauge is absent in the recently proposed Feynman-diagram gauge, and the physical distributions can be interpreted by the contributions from each subgroup. % We also find that the interference patterns in kinematical distributions of the Z boson can be understood by those in the $l^-l^+\to ν\barνZ$ process.

Studying the final state interactions and finding possible bound states is helpful for understanding the strong interactions and comprehending the equation-of-state (EoS) of the nuclear matter. In these proceedings, we present recent femtoscopy results of \pXi{}, \LaLa{}, \pOm{} femtoscopic correlations with high statistics Isobar (Ru+Ru, Zr+Zr) and Au+Au collisions measured by the STAR experiment. For the \pXi{} and \pOm{} pairs, the centrality dependence of source size and the scattering parameters are extracted with the Lednický-Lyuboshitz approach. The results show that there is an attractive interaction in \pXi{} pairs and a bound state in \pOm{} pairs.