arXiv Daily Digest - 2026-04-17

Despite recent advances in state space models (SSMs) such as Mamba across various sequence domains, research on their standalone capacity for time series classification (TSC) has remained limited. We propose MambaSL, a framework that minimally redesigns the selective SSM and projection layers of a single-layer Mamba, guided by four TSC-specific hypotheses. To address benchmarking limitations -- restricted configurations, partial University of East Anglia (UEA) dataset coverage, and insufficiently reproducible setups -- we re-evaluate 20 strong baselines across all 30 UEA datasets under a unified protocol. As a result, MambaSL achieves state-of-the-art performance with statistically significant average improvements, while ensuring reproducibility via public checkpoints for all evaluated models. Together with visualizations, these results demonstrate the potential of Mamba-based architectures as a TSC backbone.

Recent work has shown that diffusion models trained with the denoising score matching (DSM) objective often violate the Fokker--Planck (FP) equation that governs the evolution of the true data density. Directly penalizing these deviations in the objective function reduces their magnitude but introduces a significant computational overhead. It is also observed that enforcing strict adherence to the FP equation does not necessarily lead to improvements in the quality of the generated samples, as often the best results are obtained with weaker FP regularization. In this paper, we investigate whether simpler penalty terms can provide similar benefits. We empirically analyze several lightweight regularizers, study their effect on FP residuals and generation quality, and show that the benefits of FP regularization are available at substantially lower computational cost. Our code is available at https://github.com/OnnoNiemann/fp_diffusion_analysis.

Oil and gas drilling operations generate extensive time-series data from surface sensors, yet accurate real-time prediction of critical downhole metrics remains challenging due to the scarcity of labelled downhole measurements. This systematic mapping study reviews thirteen papers published between 2015 and 2025 to assess the potential of Masked Autoencoder Foundation Models (MAEFMs) for predicting downhole metrics from surface drilling data. The review identifies eight commonly collected surface metrics and seven target downhole metrics. Current approaches predominantly employ neural network architectures such as artificial neural networks (ANNs) and long short-term memory (LSTM) networks, yet no studies have explored MAEFMs despite their demonstrated effectiveness in time-series modeling. MAEFMs offer distinct advantages through self-supervised pre-training on abundant unlabeled data, enabling multi-task prediction and improved generalization across wells. This research establishes that MAEFMs represent a technically feasible but unexplored opportunity for drilling analytics, recommending future empirical validation of their performance against existing models and exploration of their broader applicability in oil and gas operations.

Post-training quantization (PTQ) assumes that a well-converged model is a quantization-ready model. We show this assumption fails in a structured, measurable, and previously uncharacterized way. Using a calibration-free per-group INT4 probe applied to all 154 publicly available Pythia-160m training checkpoints, we identify a three-phase divergence structure: a rapid-learning phase where both FP32 perplexity and quantization robustness improve together, a meta-stable plateau lasting roughly 70,000 steps where FP32 perplexity stagnates but INT4 gap remains bounded, and an explosive divergence phase where the INT4 gap compounds from 11% to 517% while FP32 perplexity barely moves. Critically, this divergence begins not when the learning rate starts decaying, but precisely when FP32 perplexity converges a finer-grained onset predictor that implies post-convergence weight updates, rather than decay magnitude alone, are the proximate cause. We further show that INT8 quantization is entirely immune throughout all three phases, constraining the mechanism to the coarseness of the 16-level INT4 grid specifically, and rule out weight outlier accumulation as the mechanism via direct kurtosis measurement. Finally, we conduct a controlled fork experiment from the pre-divergence checkpoint comparing three learning rate schedules (cosine continuation, SGDR warm restarts, and our proposed Oscillatory Lock-In) across nine independent runs. SGDR uniformly accelerates divergence (0/9 pairwise wins against cosine), while OLI's settled cool phases reduce the INT4 gap by 2.2 percentage points on average (t = -5.46, p < 0.0001), demonstrating that schedule amplitude calibration, not oscillation alone, determines whether perturbation helps or hurts. Our code, probe implementation, and all 154-checkpoint audit results are released publicly.

Machine unlearning aims to remove targeted knowledge from a trained model without the cost of retraining from scratch. In class unlearning, however, reducing accuracy on forget classes does not necessarily imply true forgetting: forgotten information can remain encoded in internal representations, and apparent forgetting may arise from classifier-head suppression rather than representational removal. We show that existing class-unlearning methods often exhibit weak or negative selectivity, preserve forget-class structure in deep representations, or rely heavily on final-layer bias shifts. We then introduce DAMP (Depth-Aware Modulation by Projection), a one-shot, closed-form weight-surgery method that removes forget-specific directions from a pretrained network without gradient-based optimization. At each stage, DAMP computes class prototypes in the input space of the next learnable operator, extracts forget directions as residuals relative to retain-class prototypes, and applies a projection-based update to reduce downstream sensitivity to those directions. To preserve utility, DAMP uses a parameter-free depth-aware scaling rule derived from probe separability, applying smaller edits in early layers and larger edits in deeper layers. The method naturally extends to multi-class forgetting through low-rank subspace removal. Across MNIST, CIFAR-10, CIFAR-100, and Tiny ImageNet, and across convolutional and transformer architectures, DAMP more closely resembles the retraining gold standard than some of the prior methods, improving selective forgetting while better preserving retain-class performance and reducing residual forget-class structure in deep layers.

LLMs are proving to be adept at machine translation although due to their generative nature they may at times overgenerate in various ways. These overgenerations are different from the neurobabble seen in NMT and range from LLM self-explanations, to risky confabulations, to appropriate explanations, where the LLM is able to act as a human translator would, enabling greater comprehension for the target audience. Detecting and determining the exact nature of the overgenerations is a challenging task. We detail different strategies we have explored for our work in a commercial setting, and present our results.

Large Language Models (LLMs) incur significant computational and memory costs when processing long prompts, as full self-attention scales quadratically with input length. Token compression aims to address this challenge by reducing the number of tokens representing inputs. However, existing prompt-compression approaches primarily operate in token space and overlook inefficiencies in the latent embedding space. In this paper, we propose K-Token Merging, a latent-space compression framework that merges each contiguous block of K token embeddings into a single embedding via a lightweight encoder. The compressed sequence is processed by a LoRA-adapted LLM, while generation remains in the original vocabulary. Experiments on structural reasoning (Textualized Tree), sentiment classification (Amazon Reviews), and code editing (CommitPackFT) show that K-Token Merging lies on the Pareto frontier of performance vs. compression, achieving up to 75% input length reduction with minimal performance degradation.

Large language models have demonstrated strong performance on general-purpose programming tasks, yet their ability to generate executable algorithmic trading strategies remains underexplored. Unlike standard code benchmarks, trading-strategy generation requires simultaneous mastery of domain-specific financial logic, knowledge of a specialized API, and the ability to produce code that is not only syntactically correct but also leads to actual trades on historical data. In this work, we present QuantCode-Bench, a benchmark for the systematic evaluation of modern LLMs in generating strategies for the Backtrader framework from textual descriptions in English. The benchmark contains 400 tasks of varying difficulty collected from Reddit, TradingView, StackExchange, GitHub, and synthetic sources. Evaluation is conducted through a multi-stage pipeline that checks syntactic correctness, successful backtest execution, the presence of trades, and semantic alignment with the task description using an LLM judge. We compare state-of-the-art models in two settings: single-turn, where the strategy must be generated correctly on the first attempt, and agentic multi-turn, where the model receives iterative feedback and may repair its errors. We analyze the failure modes across different stages of the pipeline and show that the main limitations of current models are not related to syntax, but rather to the correct operationalization of trading logic, proper API usage, and adherence to task semantics. These findings suggest that trading strategy generation constitutes a distinct class of domain-specific code generation tasks in which success requires not only technical correctness, but also alignment between natural-language descriptions, financial logic, and the observable behavior of the strategy on data.

As reinforcement Learning with Verifiable Rewards (RLVR) has become the dominant paradigm for scaling reasoning capabilities in LLMs, a new failure mode emerges: LLMs gaming verifiers. We study this phenomenon on inductive reasoning tasks, where models must induce and output logical rules. We find that RLVR-trained models systematically abandon rule induction. Instead of learning generalizable patterns (e.g., ``trains carrying red cars go east''), they enumerate instance-level labels, producing outputs that pass verifiers without capturing the relational patterns required by the task. We show that this behavior is not a failure of understanding but a form of reward hacking: imperfect verifiers that check only extensional correctness admit false positives. To detect such shortcuts, we introduce Isomorphic Perturbation Testing (IPT), which evaluates a single model output under both extensional and isomorphic verification, where the latter enforces invariance under logically isomorphic tasks. While genuine rule induction remains invariant, shortcut strategies fail. We find that shortcut behavior is specific to RLVR-trained reasoning models (e.g., GPT-5, Olmo3) and absent in non-RLVR models (e.g., GPT-4o, GPT-4.5, Ministral). Moreover, shortcut prevalence increases with task complexity and inference-time compute. In controlled training experiments, extensional verification directly induces shortcut strategies, while isomorphic verification eliminates them. These results show that RLVR can incentivize reward hacking not only through overt manipulation but also by exploiting what the verifier fails to enforce.

Reinforcement learning has emerged as an effective paradigm for training large language models to perform search-augmented reasoning. However, existing approaches rely on trajectory-level rewards that cannot distinguish precise search queries from vague or redundant ones within a rollout group, and collapse to a near-zero gradient signal whenever every sampled trajectory fails. In this paper, we propose IG-Search, a reinforcement learning framework that introduces a step-level reward based on Information Gain (IG). For each search step, IG measures how much the retrieved documents improve the model's confidence in the gold answer relative to a counterfactual baseline of random documents, thereby reflecting the effectiveness of the underlying search query. This signal is fed back to the corresponding search-query tokens via per-token advantage modulation in GRPO, enabling fine-grained, step-level credit assignment within a rollout. Unlike prior step-level methods that require either externally annotated intermediate supervision or shared environment states across trajectories, IG-Search derives its signals from the policy's own generation probabilities, requiring no intermediate annotations beyond standard question-answer pairs. Experiments on seven single-hop and multi-hop QA benchmarks demonstrate that IG-Search achieves an average EM of 0.430 with Qwen2.5-3B, outperforming the strongest trajectory-level baseline (MR-Search) by 1.6 points and the step-level method GiGPO by 0.9 points on average across benchmarks, with particularly pronounced gains on multi-hop reasoning tasks. Despite introducing a dense step-level signal, IG-Search adds only ~6.4% to per-step training wall-clock time over the trajectory-level baseline and leaves inference latency unchanged, while still providing a meaningful gradient signal even when every sampled trajectory answers incorrectly.

The rigorous evaluation of the novelty of a scientific paper is, even for human scientists, a challenging task. With the increasing interest in AI scientists and AI involvement in scientific idea generation and paper writing, it also becomes increasingly important that this task be automatable and reliable, lest both human attention and compute tokens be wasted on ideas that have already been explored. Due to the challenge of quantifying ground-truth novelty, however, existing novelty metrics for scientific papers generally validate their results against noisy, confounded signals such as citation counts or peer review scores. These proxies can conflate novelty with impact, quality, or reviewer preference, which in turn makes it harder to assess how well a given metric actually evaluates novelty. We therefore propose an axiomatic benchmark for scientific novelty metrics. We first define a set of axioms that a well-behaved novelty metric should satisfy, grounded in human scientific norms and practice, then evaluate existing metrics across ten tasks spanning three domains of AI research. Our results reveal that no existing metric satisfies all axioms consistently, and that metrics fail on systematically different axioms, reflecting their underlying architectures. Additionally, we show that combining metrics of complementary architectures leads to consistent improvements on the benchmark, with per-axiom weighting achieving 90.1% versus 71.5% for the best individual metric, suggesting that developing architecturally diverse metrics is a promising direction for future work. We release the benchmark code as supplementary material to encourage the development of more robust scientific literature novelty metrics.

This work simulates the developmental process of cortical neurogenesis, initiating from a single stem cell and governed by gene regulatory rules derived from mouse single-cell transcriptomic data. The developmental process spontaneously generates a heterogeneous population of 5,000 cells, yet yields only 85 mature neurons - merely 1.7% of the total population. These 85 neurons form a densely interconnected core of 200,400 synapses, corresponding to an average degree of 4,715 per neuron. At iteration zero, this minimal circuit performs at chance level on MNIST. However, after a single epoch of standard training, accuracy surges to over 90% - a gain exceeding 80 percentage points - with typical runs falling in the 89-94% range depending on developmental stochasticity. The identical circuit, without any architectural modification or data augmentation, achieves 40.53% on CIFAR-10 after one epoch. These findings demonstrate that developmental rules sculpt a domain-general topological substrate exceptionally amenable to rapid learning, suggesting that biological developmental processes inherently encode powerful structural priors for efficient computation.

We introduce DiscoTrace, a method to identify the rhetorical strategies that answerers use when responding to information-seeking questions. DiscoTrace represents answers as a sequence of question-related discourse acts paired with interpretations of the original question, annotated on top of rhetorical structure theory parses. Applying DiscoTrace to answers from nine different human communities reveals that communities have diverse preferences for answer construction. In contrast, LLMs do not exhibit rhetorical diversity in their answers, even when prompted to mimic specific human community answering guidelines. LLMs also systematically opt for breadth, addressing interpretations of questions that human answerers choose not to address. Our findings can guide the development of pragmatic LLM answerers that consider a range of strategies informed by context in QA.

Although modern, AI-centric datacenters heavily rely on SmartNICs, existing devices impose a hard trade-off. Commercial SmartNICs provide high bandwidth and easy software integration, but offer limited support for customization and data processing offload. In contrast, research SmartNICs often suffer from low bandwidth, limited functionality, and poor software compatibility -- to the point that many are not actual NICs in a technical sense. This gap can be closed by treating the NIC datapath as a first-class stream computation substrate with shared hardware/software abstractions for a tight co-design of infrastructure and applications. To demonstrate this, we introduce SCENIC, an open-source datacenter SmartNIC. SCENIC implements a 200G network datapath over offloaded TCP/IP and RDMA stacks, as well as a fallback path for processing arbitrary network traffic. On top of the network logic, SCENIC combines on-datapath Stream Compute Units (SCUs) for data processing and embedded ARM cores for flexible control path manipulation with direct access to GPUs and SSDs. SCENIC is fully integrated with the OS, exposing native Linux network and RDMA verb interfaces, making the programmable datapath transparent to existing applications while enabling control of, e.g., user-defined offloads and programmable congestion control. SCENIC's performance matches commercial platforms, and we show its versatility through several use cases such as offloaded collective communication and network-to-GPU hash-based data partitioning.

Continuous glucose monitoring (CGM) is central to diabetes care, but explaining CGM patterns clearly and empathetically remains time-intensive. Evidence for retrieval-grounded large language model (LLM) systems in CGM-informed counseling remains limited. To evaluate whether a retrieval-grounded LLM-based conversational agent (CA) could support patient understanding of CGM data and preparation for routine diabetes consultations. We developed a retrieval-grounded LLM-based CA for CGM interpretation and diabetes counseling support. The system generated plain-language responses while avoiding individualized therapeutic advice. Twelve CGM-informed cases were constructed from publicly available datasets. Between Oct 2025 and Feb 2026, 6 senior UK diabetes clinicians each reviewed 2 assigned cases and answered 24 questions. In a blinded multi-rater evaluation, each CA-generated and clinician-authored response was independently rated by 3 clinicians on 6 quality dimensions. Safety flags and perceived source labels were also recorded. Primary analyses used linear mixed-effects models. A total of 288 unique responses (144 CA and 144 clinician) generated 864 ratings. The CA received higher quality scores than clinician responses (mean 4.37 vs 3.58), with an estimated mean difference of 0.782 points (95% CI 0.692-0.872; P<.001). The largest differences were for empathy (1.062, 95% CI 0.948-1.177) and actionability (0.992, 95% CI 0.877-1.106). Safety flag distributions were similar, with major concerns rare in both groups (3/432, 0.7% each). Retrieval-grounded LLM systems may have value as adjunct tools for CGM review, patient education, and preconsultation preparation. However, these findings do not support autonomous therapeutic decision-making or unsupervised real-world use.

Sequential associative memories (SAMs) are difficult to build and maintain in real-world streaming environments, where observations arrive incrementally over time, have imbalanced sampling, and non-stationary temporal dynamics. Vector Symbolic Architectures (VSAs) provide a biologically-inspired framework for building SAMs. Entities and attributes are encoded as quasi-orthogonal hyperdimensional vectors and processed with well defined algebraic operations. Despite this rich framework, most VSA systems rely on simple additive updates, where repeated observations reinforce existing information even when no new information is introduced. In non-stationary environments, this leads to the persistence of stale information after the underlying system changes. In this work, we introduce the Sequential Relevance Memory Unit (SRMU), a domain- and cleanup-agnostic update rule for VSA-based SAMs. The SRMU combines temporal decay with a relevance gating mechanism. Unlike prior approaches that solely rely on cleanup, the SRMU regulates memory formation by filtering redundant, conflicting, and stale information before storage. We evaluate the SRMU on streaming state-tracking tasks that isolate non-uniform sampling and non-stationary temporal dynamics. Our results show that the SRMU increases memory similarity by $12.6\%$ and reduces cumulative memory magnitude by $53.5\%$. This shows that the SRMU produces more stable memory growth and stronger alignment with the ground-truth state.

Most existing Byzantine-robust federated learning (FL) methods suffer from slow and unstable convergence. Moreover, when handling a substantial proportion of colluded malicious clients, achieving robustness typically entails compromising model utility. To address these issues, this work introduces FedIDM, which employs distribution matching to construct trustworthy condensed data for identifying and filtering abnormal clients. FedIDM consists of two main components: (1) attack-tolerant condensed data generation, and (2) robust aggregation with negative contribution-based rejection. These components exclude local updates that (1) deviate from the update direction derived from condensed data, or (2) cause a significant loss on the condensed dataset. Comprehensive evaluations on three benchmark datasets demonstrate that FedIDM achieves fast and stable convergence while maintaining acceptable model utility, under multiple state-of-the-art Byzantine attacks involving a large number of malicious clients.

We propose a novel amortized optimization method for predicting optimal transport (OT) plans across multiple pairs of measures by leveraging Kantorovich potentials derived from sliced OT. We introduce two amortization strategies: regression-based amortization (RA-OT) and objective-based amortization (OA-OT). In RA-OT, we formulate a functional regression model that treats Kantorovich potentials from the original OT problem as responses and those obtained from sliced OT as predictors, and estimate these models via least-squares methods. In OA-OT, we estimate the parameters of the functional model by optimizing the Kantorovich dual objective. In both approaches, the predicted OT plan is subsequently recovered from the estimated potentials. As amortized OT methods, both RA-OT and OA-OT enable efficient solutions to repeated OT problems across different measure pairs by reusing information learned from prior instances to rapidly approximate new solutions. Moreover, by exploiting the structure provided by sliced OT, the proposed models are more parsimonious, independent of specific structures of the measures, such as the number of atoms in the discrete case, while achieving high accuracy. We demonstrate the effectiveness of our approaches on tasks including MNIST digit transport, color transfer, supply-demand transportation on spherical data, and mini-batch OT conditional flow matching.

Vector Symbolic Architectures (VSAs) provide a well-defined algebraic framework for compositional representations in hyperdimensional spaces. We introduce HyperSpace, an open-source framework that decomposes VSA systems into modular operators for encoding, binding, bundling, similarity, cleanup, and regression. Using HyperSpace, we analyze and benchmark two representative VSA backends: Holographic Reduced Representations (HRR) and Fourier Holographic Reduced Representations (FHRR). Although FHRR provides lower theoretical complexity for individual operations, HyperSpaces modularity reveals that similarity and cleanup dominate runtime in spatial domains. As a result, HRR and FHRR exhibit comparable end-to-end performance. Differences in memory footprint introduce additional deployment trade-offs where HRR requires approximately half the memory of FHRR vectors. By enabling modular, system-level evaluation, HyperSpace reveals practical trade-offs in VSA pipelines that are not apparent from theoretical or operator-level comparisons alone.

Despite the rapid advancement of Large Language Models (LLMs), uncertainty quantification in LLM generation is a persistent challenge. Although recent approaches have achieved strong performance by restricting LLMs to produce short or constrained answer sets, many real-world applications require long-form and free-form text generation. A key difficulty in this setting is that LLMs often produce responses that are semantically coherent yet factually inaccurate, while the underlying semantics are multifaceted and the linguistic structure is complex. To tackle this challenge, this paper introduces Interrogative Uncertainty Quantification (IUQ), a novel framework that leverages inter-sample consistency and intra-sample faithfulness to quantify the uncertainty in long-form LLM outputs. By utilizing an interrogate-then-respond paradigm, our method provides reliable measures of claim-level uncertainty and the model's faithfulness. Experimental results across diverse model families and model sizes demonstrate the superior performance of IUQ over two widely used long-form generation datasets. The code is available at https://github.com/louisfanhz/IUQ.

Feature selection is a classical problem in statistics and machine learning, and it continues to remain an extremely challenging problem especially in the context of unknown non-linear relationships with dependent features. On the other hand, Shapley values are a classic solution concept from cooperative game theory that is widely used for feature attribution in general non-linear models with highly-dependent features. However, Shapley values are not naturally suited for feature selection since they tend to capture both direct effects from each feature to the response and indirect effects through other features. In this paper, we combine the advantages of Shapley values and adapt them to feature selection by proposing \emph{MinShap}, a modification of the Shapley value framework along with a suite of other related algorithms. In particular for MinShap, instead of taking the average marginal contributions over permutations of features, considers the minimum marginal contribution across permutations. We provide a theoretical foundation motivated by the faithfulness assumption in DAG (directed acyclic graphical models), a guarantee for the Type I error of MinShap, and show through numerical simulations and real data experiments that MinShap tends to outperform state-of-the-art feature selection algorithms such as LOCO, GCM and Lasso in terms of both accuracy and stability. We also introduce a suite of algorithms related to MinShap by using the multiple testing/p-value perspective that improves performance in lower-sample settings and provide supporting theoretical guarantees.

Learning-to-Rank (LTR) is a supervised machine learning approach that constructs models specifically designed to order a set of items or documents based on their relevance or importance to a given query or context. Despite significant success in real-world information retrieval systems, current LTR methods rely on one prefix ranking metric (e.g., such as Normalized Discounted Cumulative Gain (NDCG) or Mean Average Precision (MAP)) for optimizing the ranking objective function. Such metric-dependent setting limits LTR methods from two perspectives: (1) non-differentiable problem: directly optimizing ranking functions over a given ranking metric is inherently non-smooth, making the training process unstable and inefficient; (2) limited ranking utility: optimizing over one single metric makes it difficult to generalize well to other ranking metrics of interest. To address the above issues, we propose a novel listwise LTR framework for efficient and generalizable ranking purpose. Specifically, we propose a new differentiable ranking loss that combines a smooth approximation to the ranking operator with the average mean square loss per query. Then, we adapt gradient-boosting machines to minimize our proposed loss with respect to each list, a novel contribution. Finally, extensive experimental results confirm that our method outperforms the current state-of-the-art in information retrieval measures with similar efficiency.

This beta technical report asks how reusable experience should be represented so that it can function as effective test-time control and as a substrate for iterative evolution. We study this question in 4.590 controlled trials across 45 scientific code-solving scenarios. We find that documentation-oriented Skill packages provide unstable control: their useful signal is sparse, and expanding a compact experience object into a fuller documentation package often fails to help and can degrade the overall average. We further show that representation itself is a first-order factor. A compact Gene representation yields the strongest overall average, remains competitive under substantial structural perturbations, and outperforms matched-budget Skill fragments, while reattaching documentation-oriented material usually weakens rather than improves it. Beyond one-shot control, we show that Gene is also a better carrier for iterative experience accumulation: attached failure history is more effective in Gene than in Skill or freeform text, editable structure matters beyond content alone, and failure information is most useful when distilled into compact warnings rather than naively appended. On CritPt, gene-evolved systems improve over their paired base models from 9.1% to 18.57% and from 17.7% to 27.14%. These results suggest that the core problem in experience reuse is not how to supply more experience, but how to encode experience as a compact, control-oriented, evolution-ready object.

Echocardiography is a widely used modality for cardiac assessment due to its non-invasive and cost-effective nature, but the sparse and heterogeneous spatiotemporal views of the heart pose distinct challenges. Existing masked autoencoder (MAE) approaches typically process images or short clips independently, failing to capture the inherent multi-view structure required for coherent cardiac representation. We introduce Latent Attention Masked Autoencoder (LAMAE), a foundation model architecture tailored to the multi-view nature of medical imaging. LAMAE augments the standard MAE with a latent attention module that enables information exchange across frames and views directly in latent space. This allows the model to aggregate variable-length sequences and distinct views, reconstructing a holistic representation of cardiac function from partial observations. We pretrain LAMAE on MIMIC-IV-ECHO, a large-scale, uncurated dataset reflecting real-world clinical variability. To the best of our knowledge, we present the first results for predicting ICD-10 codes from MIMIC-IV-ECHO videos. Furthermore, we empirically demonstrate that representations learned from adult data transfer effectively to pediatric cohorts despite substantial anatomical differences. These results provide evidence that incorporating structural priors, such as multi-view attention, yields significantly more robust and transferable representations.

Mobile agents powered by vision-language models have demonstrated impressive capabilities in automating mobile tasks, with recent leading models achieving a marked performance leap, e.g., nearly 70% success on AndroidWorld. However, these systems keep their training data closed and remain opaque about their task and trajectory synthesis recipes. We present OpenMobile, an open-source framework that synthesizes high-quality task instructions and agent trajectories, with two key components: (1) The first is a scalable task synthesis pipeline that constructs a global environment memory from exploration, then leverages it to generate diverse and grounded instructions. and (2) a policy-switching strategy for trajectory rollout. By alternating between learner and expert models, it captures essential error-recovery data often missing in standard imitation learning. Agents trained on our data achieve competitive results across three dynamic mobile agent benchmarks: notably, our fine-tuned Qwen2.5-VL and Qwen3-VL reach 51.7% and 64.7% on AndroidWorld, far surpassing existing open-data approaches. Furthermore, we conduct transparent analyses on the overlap between our synthetic instructions and benchmark test sets, and verify that performance gains stem from broad functionality coverage rather than benchmark overfitting. We release data and code at https://njucckevin.github.io/openmobile/ to bridge the data gap and facilitate broader mobile agent research.

This paper introduces the first \emph{self-evolving} logic synthesis framework, which leverages Large Language Model (LLM) agents to autonomously improve the source code of \textsc{ABC}, the widely adopted logic synthesis system. Our framework operates on the \emph{entire integrated ABC codebase}, and the output repository preserves its single-binary execution model and command interface. In the initial evolution cycle, we bootstrap the system using existing prior open-source synthesis components, covering flow tuning, logic minimization, and technology mapping, but without manually injecting new heuristics. On top of this foundation, a team of LLM-based agents iteratively rewrites and evolves specific sub-components of ABC following our ``programming guidance`` prompts under a unified correctness and QoR-driven evaluation loop. Each evolution cycle proposes code modifications, compiles the integrated binary, validates correctness, and evaluates quality-of-results (QoR) on \emph{multi-suite benchmarks including ISCAS~85/89/99, VTR, EPFL, and IWLS~2005}. Through continuous feedback, the system discovers optimizations beyond human-designed heuristics, effectively \emph{learning new synthesis strategies} that enhance QoR. We detail the architecture of this self-improving system, its integration with \textsc{ABC}, and results demonstrating that the framework can autonomously and progressively improve EDA tool at full million-line scale.

Rapid advances in Generative AI are giving rise to increasingly sophisticated Multi-Agent AI (MAAI) systems. While AI fairness has been extensively studied in traditional predictive scenarios, its examination in MAAI remains nascent and fragmented. This scoping review critically synthesizes existing research on fairness in MAAI systems. Through a qualitative content analysis of 23 selected studies, we identify five archetypal approaches. Our findings reveal that fairness in MAAI systems is often addressed superficially, lacks robust normative foundations, and frequently overlooks the complex dynamics introduced by agent autonomy and system-level interactions. We argue that fairness must be embedded structurally throughout the development lifecycle of MAAI, rather than appended as a post-hoc consideration. Meaningful evaluation requires explicit human oversight, normative clarity, and a precise articulation of fairness objectives and beneficiaries. This review provides a foundation for advancing fairness research in MAAI systems by highlighting critical gaps, exposing prevailing limitations, and suggesting pathways.

To navigate a space, the brain makes an internal representation of the environment using different cells such as place cells, grid cells, head direction cells, border cells, and speed cells. All these cells, along with sensory inputs, enable an organism to explore the space around it. Inspired by these biological principles, we developed NEATNC, a Neuro-Evolution of Augmenting Topology guided Navigation Cells. The goal of the paper is to improve NEAT algorithm performance in path planning in dynamic environments using spatial cognitive cells. This approach uses navigation cells as inputs and evolves recurrent neural networks, representing the hippocampus part of the brain. The performance of the proposed algorithm is evaluated in different static and dynamic scenarios. This study highlights NEAT's adaptability to complex and different environments, showcasing the utility of biological theories. This suggests that our approach is well-suited for real-time dynamic path planning for robotics and games.

Open-weight Small Language Models(SLMs) can provide faster local inference at lower financial cost, but may not achieve the same performance level as commercial Large Language Models (LLMs) that are orders of magnitudes larger. Consequently, many of the latest applications of LLMs, such as software engineering agents, tend to be evaluated on larger models only, leaving the issue of improving the cost-benefit trade-off of such applications neglected. This paper proposes Atropos, a predictive early-termination analysis and hotswap technique that aims to improve the cost-benefit trade-off for LLM-based agents that use self-consistency. The core component of ATROPOS is a predictive model based on structural properties of LLM inferences: after merging multiple agentic inference paths into a graph representation, ATROPOS uses Graph Convolutional Network (GCN) to predict whether an ongoing inference will eventually succeed or not. If an agentic task instance running on the source LLM is predicted to fail, ATROPOS subsequently performs hotswapping, i.e., migrating the on-going inference context onto the more capable target LLM: this is feasible because LLM contexts are stateless. An empirical evaluation of ATROPOS using three recent LLM-based agents shows that ATROPOS can predict early termination of eventually failing inferences with the accuracy of 0.85 at the midpoint of the inference. Hotswapping LLMs for such inferences can convert up to 27.57% of them to be successful. Consequently, ATROPOS achieves 74.35% of the performance of closed LLMs with as low as only 23.9% of the cost.

Graph Neural Networks (GNNs) conventionally rely on standard Laplacian or adjacency matrices for structural message passing. In this work, we substitute the traditional Laplacian with a Doubly Stochastic graph Matrix (DSM), derived from the inverse of the modified Laplacian, to naturally encode continuous multi-hop proximity and strict local centrality. To overcome the intractable $O(n^3)$ complexity of exact matrix inversion, we first utilize a truncated Neumann series to scalably approximate the DSM, which serves as the foundation for our proposed DsmNet. Furthermore, because algebraic truncation inherently causes probability mass leakage, we introduce DsmNet-compensate. This variant features a mathematically rigorous Residual Mass Compensation mechanism that analytically re-injects the truncated tail mass into self-loops, strictly restoring row-stochasticity and structural dominance. Extensive theoretical and empirical analyses demonstrate that our decoupled architectures operate efficiently in $O(K|E|)$ time and effectively mitigate over-smoothing by bounding Dirichlet energy decay, providing robust empirical validation on homophilic benchmarks. Finally, we establish the theoretical boundaries of the DSM on heterophilic topologies and demonstrate its versatility as a continuous structural encoding for Graph Transformers.

Pareto optimization via evolutionary multi-objective algorithms has been shown to efficiently solve constrained monotone submodular functions. Traditionally when solving multiple problems, the algorithm is run for each problem separately. We introduce multitasking formulations of these problems that are an effective way to solve multiple related problems with a single run. In our setting the given problems share a monotone submodular function $f$ but have different knapsack constraints. We examine the case where elements within a constraint have the same cost and show that our multitasking formulations result in small Pareto fronts. This allows the population to share solutions between all problems leading to significant improvements compared to running several classical approaches independently. Using rigorous runtime analysis, we analyze the expected time until the introduced multitasking approaches obtain a $(1-1/e)$-approximation for each of the given problems. Our experimental investigations for the maximum coverage problem give further insight into the dynamics behind how the approach works and doesn't work in practice for problems where elements within a constraint also have varied costs.

Gradient inversion attacks threaten client privacy in federated learning by reconstructing training samples from clients' shared gradients. Gradients aggregate contributions from multiple records and existing attacks may fail to disentangle them, yielding incorrect reconstructions with no intrinsic way to certify success. In vision and language, attackers may fall back on human inspection to judge reconstruction plausibility, but this is far less feasible for numerical tabular records, fueling the impression that tabular data is less vulnerable. We challenge this perception by proposing a verifiable gradient inversion attack (VGIA) that provides an explicit certificate of correctness for reconstructed samples. Our method adopts a geometric view of ReLU leakage: the activation boundary of a fully connected layer defines a hyperplane in input space. VGIA introduces an algebraic, subspace-based verification test that detects when a hyperplane-delimited region contains exactly one record. Once isolation is certified, VGIA recovers the corresponding feature vector analytically and reconstructs the target via a lightweight optimization step. Experiments on tabular benchmarks with large batch sizes demonstrate exact record and target recovery in regimes where existing state-of-the-art attacks either fail or cannot assess reconstruction fidelity. Compared to prior geometric approaches, VGIA allocates hyperplane queries more effectively, yielding faster reconstructions with fewer attack rounds.

The increasing integration of artificial intelligence (AI) in everyday life brings with it new challenges and questions for regarding how humans interact with autonomous agents. Multi-agent experiments, where humans and AI act together, can offer important opportunities to study social decision making, but there is a lack of accessible tooling available to researchers to run such experiments. We introduce two tools designed to reduce these barriers. The first, CoGrid, is a multi-agent grid-based simulation library with dual NumPy and JAX backends. The second, Multi-User Gymnasium (MUG), translates such simulation environments directly into interactive web-based experiments. MUG supports interactions with arbitrary numbers of humans and AI, utilizing either server-authoritative or peer-to-peer networking with rollback netcode to account for latency. Together, these tools can enable researchers to deploy studies of human-AI interaction, facilitating inquiry into core questions of psychology, cognition, and decision making and their relationship to human-AI interaction. Both tools are open source and available to the broader research community. Documentation and source code is available at {cogrid, multi-user-gymnasium}.readthedocs.io. This paper details the functionality of these tools and presents several case studies to illustrate their utility in human-AI multi-agent experimentation.

Code optimization remains a core objective in software development, yet modern compilers struggle to navigate the enormous optimization spaces. While recent research has looked into employing large language models (LLMs) to optimize source code directly, these techniques can introduce semantic errors and miss fine-grained compiler-level optimization opportunities. We present HintPilot, which bridges LLM-based reasoning with traditional compiler infrastructures via synthesizing compiler hints, annotations that steer compiler behavior. HintPilot employs retrieval-augmented synthesis over compiler documentation and applies profiling-guided iterative refinement to synthesize semantics-preserving and effective hints. Upon PolyBench and HumanEval-CPP benchmarks, HintPilot achieves up to 6.88x geometric mean speedup over -Ofast while preserving program correctness.

Prefill-decode (PD) disaggregation has become the standard architecture for large-scale LLM serving, but in practice its deployment boundary is still determined by KVCache transfer. In conventional dense-attention models, prefill generates huge KVCache traffics that keep prefill and decode tightly coupled within a single high-bandwidth network domain, limiting heterogeneous deployment and resource elasticity. Recent hybrid-attention architectures substantially reduce KVCache size, making cross-cluster KVCache transport increasingly plausible. However, smaller KVCache alone does not make heterogeneous cross-datacenter PD serving practical: real workloads remain bursty, request lengths are highly skewed, prefix caches are unevenly distributed, and inter-cluster bandwidth fluctuates. A naive design that fully externalizes prefill can therefore still suffer from congestion, unstable queueing, and poor utilization. We present Prefill-as-a-Service (PrfaaS), a cross-datacenter serving architecture that selectively offloads long-context prefill to standalone, compute-dense prefill clusters and transfers the resulting KVCache over commodity Ethernet to local PD clusters for decode. Rather than treating reduced KVCache as sufficient, PrfaaS combines model-side KV efficiency with system-side selective offloading, bandwidth-aware scheduling, and cache-aware request placement. This design removes the requirement that heterogeneous accelerators share the same low-latency RDMA fabric, enabling independent scaling of prefill and decode capacity across loosely coupled clusters. In a case study using an internal 1T-parameter hybrid model, a PrfaaS-augmented heterogeneous deployment achieves 54% and 32% higher serving throughput than homogeneous PD and naive heterogeneous baselines, respectively, while consuming only modest cross-datacenter bandwidth.

The evaluation of fairness in machine learning systems has become a central concern in high-stakes applications, including biometric recognition, healthcare decision-making, and automated risk assessment. Existing approaches typically rely on a small number of fairness metrics to assess model behaviour across group partitions, implicitly assuming that these metrics provide consistent and reliable conclusions. However, different fairness metrics capture distinct statistical properties of model performance and may therefore produce conflicting assessments when applied to the same system. In this work, we investigate the consistency of fairness evaluation by conducting a systematic multi-metric analysis of demographic bias in machine learning models. Using face recognition as a controlled experimental setting, we evaluate model performance across multiple group partitions under a range of commonly used fairness metrics, including error-rate disparities and performance-based measures. Our results demonstrate that fairness assessments can vary significantly depending on the choice of metrics, leading to contradictory conclusions regarding model bias. To quantify this phenomenon, we introduce the Fairness Disagreement Index (FDI), a measure designed to capture the degree of inconsistency across fairness metrics. We further show that disagreement remains high across thresholds and model configurations. These findings highlight a critical limitation in current fairness evaluation practices and suggest that single-metric reporting is insufficient for reliable bias assessment.

Recent advancements in LLM agents are gradually shifting from reactive, text-based paradigms toward proactive, multimodal interaction. However, existing benchmarks primarily focus on reactive responses, overlooking the complexities of proactive intervention and monitoring. To bridge this gap, we introduce ProVoice-Bench, the first evaluation framework specifically designed for proactive voice agents, featuring four novel tasks. By leveraging a multi-stage data synthesis pipeline, we curate 1,182 high-quality samples for rigorous testing. Our evaluation of state-of-the-art Multimodal LLMs reveals a significant performance gap, particularly regarding over-triggering and reasoning capabilities. These findings highlight the limitations of current models and offer a roadmap for developing more natural, context-aware proactive agents.

Recent advances in LLM based agent systems have shown promise in tackling complex, long horizon tasks. However, existing agent protocols (e.g., A2A and MCP) under specify cross entity lifecycle and context management, version tracking, and evolution safe update interfaces, which encourages monolithic compositions and brittle glue code. We introduce \textbf{\textsc{Autogenesis Protocol (AGP)}}, a self evolution protocol that decouples what evolves from how evolution occurs. Its Resource Substrate Protocol Layer (RSPL) models prompts, agents, tools, environments, and memory as protocol registered resources\footnote{Unless otherwise specified, resources refer to instances of the five RSPL entity types: \emph{prompt}, \emph{agent}, \emph{tool}, \emph{environment}, \emph{memory} with agent \emph{outputs}.} with explicit state, lifecycle, and versioned interfaces. Its Self Evolution Protocol Layer (SEPL) specifies a closed loop operator interface for proposing, assessing, and committing improvements with auditable lineage and rollback. Building on \textbf{\textsc{AGP}}, we present \textbf{\textsc{Autogenesis System (AGS)}}, a self-evolving multi-agent system that dynamically instantiates, retrieves, and refines protocol-registered resources during execution. We evaluate \textbf{\textsc{AGS}} on multiple challenging benchmarks that require long horizon planning and tool use across heterogeneous resources. The results demonstrate consistent improvements over strong baselines, supporting the effectiveness of agent resource management and closed loop self evolution.

Increasing demand for computational power has led cloud providers to employ multi-NUMA servers and offer multi-NUMA virtual machines to their customers. However, multi-NUMA VMs introduce additional complexity to scheduling algorithms. Beyond merely selecting a host for a VM, the scheduler has to map virtual NUMA topology onto the physical NUMA topology of the server to ensure optimal VM performance and minimize interference with co-located VMs. Under these constraints, maximizing the number of allocated multi-NUMA VMs on a host becomes a combinatorial optimization problem. In this paper, we derive closed-form expressions to compute the maximum number of VMs for a given flavor that can be additionally allocated onto a physical server. We consider nontrivial scenarios of mapping 2- and 4-NUMA symmetric VMs to 4- and 8-NUMA physical topologies. Our results have broad applicability, ranging from real-time dashboards (displaying available cluster capacity per VM flavor) to optimization tools for large-scale cloud resource reorganization.

In nature, estimating the location of a molecule source in turbulent airflow is a central, and yet highly challenging problem for mate search and foraging. Recently, it has also received increasing attention in synthetic molecular communication (SMC), e.g., for leakage detection. One important aspect of source localization is to estimate the distance to the molecule source, e.g., to determine whether it is worth to travel to a potential mating partner or food source, or to decide whether a leak is close enough for inspection. In this study, based on realistic simulations, we show that the diversity induced by molecule mixtures can aid source localization. In particular, when different molecule types in a mixture are subject to atmospheric degradation with different degradation rates, the relative abundance of the different species observed at the receiver enables low-complexity estimation of the source distance. Furthermore, this feature can be combined with already established concentration-based and temporal features of observed molecular signals to further increase estimation accuracy. Thereby, we show that molecule degradation diversity of molecule mixtures can help to realize one of the important envisioned SMC applications, namely source localization, even in turbulent airflow, opening new opportunities for the exploitation of SMC to solve real-world problems.

Cost-aware routing dynamically dispatches user queries to models of varying capability to balance performance and inference cost. However, the routing strategy introduces a new security concern that adversaries may manipulate the router to consistently select expensive high-capability models. Existing routing attacks depend on either white-box access or heuristic prompts, rendering them ineffective in real-world black-box scenarios. In this work, we propose R$^2$A, which aims to mislead black-box LLM routers to expensive models via adversarial suffix optimization. Specifically, R$^2$A deploys a hybrid ensemble surrogate router to mimic the black-box router. A suffix optimization algorithm is further adapted for the ensemble-based surrogate. Extensive experiments on multiple open-source and commercial routing systems demonstrate that {R$^2$A} significantly increases the routing rate to expensive models on queries of different distributions. Code and examples: https://github.com/thcxiker/R2A-Attack.

Generative AI is changing how research software is developed, but rapid AI-assisted development can weaken continuity, traceability, and methodological clarity. SHAPR (Solo, Human-centred, AI-assisted PRactice) was proposed as a framework for structuring AI-assisted research software development. This paper presents a documented case of applying SHAPR to the development of a modular share trading system. From the outset, the project adopted a SHAPR-informed working configuration that shaped how interaction, implementation, and documentation were organised. Across iterative development cycles, the project generated a structured evidence base including reflection notes, development cycle review notes, source-of-truth documents, contracts, quick captures, workflow notes, and evolving code artefacts. The case showed that continuous documentation updates, supported by quick capture and AI-assisted refinement, helped maintain organised and usable project knowledge throughout development. Five recurring lessons were identified: contracts stabilised AI-assisted coding, a maintained source-of-truth layer improved coherence, cycle-boundary snapshots strengthened continuity, code and documentation co-evolved through quick capture and iterative refinement, and environment setup itself contributed to knowledge generation. The case also illustrates a practical SHAPR operating configuration in which a ChatGPT Project and cycle-specific chats supported interaction, reasoning, summarisation, and coding collaboration, PyCharm supported artefact implementation, and Obsidian supported external working memory, structured documentation, reflection, continuity, and repository-oriented note organisation, while remaining consistent with SHAPR's tool-agnostic principle. The paper contributes practical guidance and good practices for researchers conducting AI-assisted research software development.

EEG foundation models (FMs) achieve strong cross-subject and cross-task generalization but impose substantial computational and memory costs that hinder deployment on embedded BCI systems. Knowledge distillation is a natural solution; however, conventional methods fail for EEG FMs because task-relevant semantics are often distributed across intermediate layers, and aggressive dimensionality reduction can distort oscillatory structure via representational collapse and aliasing. To address these challenges, we propose DLink (Distilling Layer-wise and Dominant Knowledge), a unified framework for transferring knowledge from large EEG FMs to compact students with three key innovations: (1) a dynamic Router that adaptively aggregates teacher layers to capture dominant intermediate representations; (2) an EEG MiC student with a Mimic-then-Compress pipeline, which inherits high-dimensional teacher features and then applies structured spatio-temporal compression to avoid a heavy classification head; and (3) spectral distillation that aligns teacher-student representations in the frequency domain to regularize compression and mitigate aliasing and temporal jitter. Experiments on four EEG benchmarks show that DLink enables compact students to outperform lightweight baselines while approaching fully fine-tuned FM performance at substantially lower model size and inference cost.

When do transformers commit to a decision, and what prevents them from correcting it? We introduce \textbf{prolepsis}: a transformer commits early, task-specific attention heads sustain the commitment, and no layer corrects it. Replicating \citeauthor{lindsey2025biology}'s (\citeyear{lindsey2025biology}) planning-site finding on open models (Gemma~2 2B, Llama~3.2 1B), we ask five questions. (Q1)~Planning is invisible to six residual-stream methods; CLTs are necessary. (Q2)~The planning-site spike replicates with identical geometry. (Q3)~Specific attention heads route the decision to the output, filling a gap flagged as invisible to attribution graphs. (Q4)~Search requires ${\leq}16$ layers; commitment requires more. (Q5)~Factual recall shows the same motif at a different network depth, with zero overlap between recurring planning heads and the factual top-10. Prolepsis is architectural: the template is shared, the routing substrates differ. All experiments run on a single consumer GPU (16\,GB VRAM).

Flow matching retains the generation quality of diffusion models while enabling substantially faster inference, making it a compelling paradigm for generative modeling. However, when applied to language modeling, it exhibits fundamental limitations in representing complex latent distributions with irregular geometries, such as anisotropy and multimodality. To address these challenges, we propose a mixture-of-experts flow matching (MoE-FM) framework, which captures complex global transport geometries in latent space by decomposing them into locally specialized vector fields. Building on MoE-FM, we develop a non-autoregressive (NAR) language modeling approach, named YAN, instantiated with both Transformer and Mamba architectures. Across multiple downstream tasks, YAN achieves generation quality on par with both autoregressive (AR) and diffusion-based NAR language models, while requiring as few as three sampling steps. This yields a $40\times$ speedup over AR baselines and up to a $10^3\times$ speedup over diffusion language models, demonstrating substantial efficiency advantages for language modeling.

LLM-based RTL code generation methods increasingly target both functional correctness and PPA quality, yet existing approaches universally decouple the two objectives, optimizing PPA only after correctness is fully achieved. Whether through sequential multi-agent pipelines, evolutionary search with binary correctness gates, or hierarchical reward dependencies, partially correct but architecturally promising candidates are systematically discarded. Moreover, existing methods reduce the multi-objective PPA space to a single scalar fitness, obscuring the trade-offs among area, delay, and power. To address these limitations, we propose COEVO, a co-evolutionary framework that unifies correctness and PPA optimization within a single evolutionary loop. COEVO formulates correctness as a continuous co-optimization dimension alongside area, delay, and power, enabled by an enhanced testbench that provides fine-grained scoring and detailed diagnostic feedback. An adaptive correctness gate with annealing allows PPA-promising but partially correct candidates to guide the search toward jointly optimal solutions. To preserve the full PPA trade-off structure, COEVO employs four-dimensional Pareto-based non-dominated sorting with configurable intra-level sorting, replacing scalar fitness without manual weight tuning. Evaluated on VerilogEval 2.0 and RTLLM 2.0, COEVO achieves 97.5\% and 94.5\% Pass@1 with GPT-5.4-mini, surpassing all agentic baselines across four LLM backbones, while attaining the best PPA on 43 out of 49 synthesizable RTLLM designs.

As a current trend in Artificial Intelligence (AI), large foundation models are increasingly employed as the core of AI services. However, even after training, serving such models at scale remains a challenging task due to their heavy resource footprints, particularly in terms of GPU memory. While recent works revealed unique characteristics of systems serving foundation models that distinguish them from traditional distributed computing systems, there is still a lack of fundamental understanding of the underlying system management problems. This work aims at addressing this gap by extracting a novel problem of "server chain composition" via block placement and cache allocation for serving chainstructured jobs with large memory footprints, which models a fundamental problem in serving large foundation models through pipeline parallelism. After showing the NP-hardness of the optimal solution, the focus is turned to developing scalable algorithms with guaranteed performance under state-of-the-art load balancing. Application of the proposed solution to a distributed large language model (LLM) serving system shows significant reduction of response times compared to state-of-the-art solutions.

As power systems transition toward renewable-rich and inverter-dominated operations, accurate time-domain dynamic analysis becomes increasingly critical. Such analysis supports key operational tasks, including transient stability assessment, dynamic security analysis, contingency screening, and post-fault trajectory evaluation. In practice, these tasks may operate under several challenges, including unknown and time-varying system parameters, privacy constraints on data sharing, and the need for fast online inference. Existing learning-based approaches are typically trained for individual systems and therefore lack generalization across operating conditions and physical parameters. Hence, this paper proposes LArge Scale Small ODE (LASS)-ODE-Power, a learning framework for general-purpose time-domain prediction. The proposed approach leverages large-scale pretraining on more than 40 GB of DAE or ordinary differential-equation (ODE) trajectories to learn transferable representations. The resulting model supports trajectory prediction from short measurement prefixes across diverse dynamic regimes, including electromechanical and inverter-driven systems. Hence, the model can be directly used without data sharing in a zero-shot setting. In addition, the proposed architecture incorporates parallel and linearized computation to achieve fast inference. Moreover, to enhance task-specific performance in power systems, a specialized fine-tuning strategy is developed based on approximately 1 GB of heterogeneous power-system dynamic data. Extensive experiments over diverse power-system simulation scenarios demonstrate that LASS-ODE-Power consistently outperforms existing learning-based models in trajectory prediction accuracy with efficient inference.

Artificial General Intelligence (AGI) is increasingly being discussed not only as a tool, but also as a potential subject with personal and therefore moral status. In our opinion, the currently dominant alignment strategies, which focus on human control and containment of AI, therefore fall short. Building on Turing's analogy of "child machines", we are developing a vision of the possibility of autonomy-supporting parenting of AI, in which human control over a developing AGI is gradually reduced, allowing AI to become an independent, autonomous subject. Rather than viewing AGI, as is currently prevalent, as a dangerous creature that needs to be locked up and controlled, we should approach potential AGI with respect for a possible developing subject on the one hand, and with full confidence in our human capabilities on the other. Such a perspective opens up the possibility of cooperative coexistence and co-evolution between humans and AGIs. The relationship between humans and AGIs will thus have to be newly determined, which will change our self-image as humans. It will be crucial that humans not only claim control over potential AGIs, but also engage with AGIs through surprise, creativity, and other specifically human qualities, thereby offering them motivating incentives for cooperation.

Recent advances in large language models (LLMs) have sparked growing interest in automatic RTL optimization for better performance, power, and area (PPA). However, existing methods are still far from realistic RTL optimization. Their evaluation settings are often unrealistic: they are tested on manually degraded, small-scale RTL designs and rely on weak open-source tools. Their optimization methods are also limited, relying on coarse design-level feedback and simple pre-defined rewriting rules. To address these limitations, we present Dr. RTL, an agentic framework for RTL timing optimization in a realistic evaluation environment, with continual self-improvement through reusable optimization skills. We establish a realistic evaluation setting with more challenging RTL designs and an industrial EDA workflow. Within this setting, Dr. RTL performs closed-loop optimization through a multi-agent framework for critical-path analysis, parallel RTL rewriting, and tool-based evaluation. We further introduce group-relative skill learning, which compares parallel RTL rewrites and distills the optimization experience into an interpretable skill library. Currently, this library contains 47 pattern--strategy entries for cross-design reuse to improve PPA and accelerate convergence, and it can continue evolving over time. Evaluated on 20 real-world RTL designs, Dr. RTL achieves average WNS/TNS improvements of 21\%/17\% with a 6\% area reduction over the industry-leading commercial synthesis tool.

Covert channels (CCs) in wireless chips pose a serious security threat, as they enable the exfiltration of sensitive information from the chip to an external attacker. In this work, we propose an AI-based defense mechanism deployed at the RF receiver, where the model directly monitors raw I/Q samples to detect, in real time, the presence of a CC embedded within an otherwise nominal signal. We first compact a state-of-the-art convolutional neural network (CNN), achieving an 80% reduction in parameters, which is an essential requirement for efficient edge deployment. When evaluated on the open-source hardware Trojan (HT)-based CC dataset, the compacted CNN attains an average accuracy of 90.28% for CC detection and 86.50% for identifying the underlying HT, with results averaged across SNR values above 1 dB. For practical communication scenarios where SNR > 20 dB, the model achieves over 97% accuracy for both tasks. These results correspond to a minimal performance degradation of less than 2% compared to the baseline model. The compacted CNN is further benchmarked against alternative classifiers, demonstrating an excellent accuracy-model size trade-off. Finally, we design a lightweight CNN hardware accelerator and demonstrate it on an FPGA, achieving very low resource utilization and an efficiency of 107 GOPs/W. Being the first AI hardware accelerator proposed specifically for CC detection, we compare it against state-of-the-art AI accelerators for RF signal classification tasks such as modulation recognition, showing superior performance.

As companies enter the race for agentic AI adoption, fears surface around agentic autonomy and its subsequent risks. These fears compound as companies scale their agentic AI adoption with low-code applications, without a comparable scaling in their governance processes and expertise resulting in a phenomenon known as "Agent Sprawl". While shadow AI tools can help with agentic discovery and identification, few observability tools offer insights into the agents' configuration and settings or the decision-making process during agent-to-agent communication and orchestration. This paper explores AI governance professionals' concerns in enterprise settings, while offering design-time and runtime explainability techniques as suggested by AI governance experts for addressing those fears. Finally, we provide a preliminary prototype of an Agentic AI Card that can help companies feel at ease deploying agents at scale.

Building on recent advances in AI, hybrid decision making (HDM) holds the promise of improving human decision quality and reducing cognitive load. We work in the context of learning to guide (LtG), a recently proposed HDM framework in which the human is always responsible for the final decision: rather than suggesting decisions, in LtG the AI supplies (textual) guidance useful for facilitating decision making. One limiting factor of existing approaches is that their guidance compounds information about all possible outcomes, and as a result it can be difficult to digest. We address this issue by introducing ConfGuide, a novel LtG approach that generates more succinct and targeted guidance. To this end, it employs conformal risk control to select a set of outcomes, ensuring a cap on the false negative rate. We demonstrate our approach on a real-world multi-label medical diagnosis task. Our empirical evaluation highlights the promise of ConfGuide.

You are a robot and you live in a Markov decision process (MDP) with a finite or an infinite number of transitions from state-action to next states. You got brains and so you plan before you act. Luckily, your roboparents equipped you with a generative model to do some Monte-Carlo planning. The world is waiting for you and you have no time to waste. You want your planning to be efficient. Sample-efficient. Indeed, you want to exploit the possible structure of the MDP by exploring only a subset of states reachable by following near-optimal policies. You want guarantees on sample complexity that depend on a measure of the quantity of near-optimal states. You want something, that is an extension of Monte-Carlo sampling (for estimating an expectation) to problems that alternate maximization (over actions) and expectation (over next states). But you do not want to StOP with exponential running time, you want something simple to implement and computationally efficient. You want it all and you want it now. You want TrailBlazer.

Hate, derogatory, and offensive speech remains a persistent challenge in online platforms and public discourse. While automated detection systems are widely used, most focus on censorship or removal, raising concerns for transparency and freedom of expression, and limiting opportunities to explain why content is harmful. To address these issues, explanatory approaches have emerged as a promising solution, aiming to make hate speech detection more transparent, accountable, and informative. In this paper, we present a hybrid approach that combines Large Language Models (LLMs) with three newly created and curated vocabularies to detect and explain hate speech in English, French, and Greek. Our system captures both inherently derogatory expressions tied to identity characteristics and direct group-targeted content through two complementary pipelines: one that detects and disambiguates problematic terms using the curated vocabularies, and one that leverages LLMs as context-aware evaluators of group-targeting content. The outputs are fused into grounded explanations that clarify why content is flagged. Human evaluation shows that our hybrid approach is accurate, with high-quality explanations, outperforming LLM-only baselines.

Frontier model developers aim to train models continually to possess emergent, diverse capabilities. To extend capabilities, the current pre-training and post-training paradigm requires manually starting training runs with static datasets or reward functions every time. Addressing this limitation, our work pursues the insight that open-endedness (via the coevolution of models and tasks) can discover models with increasingly novel skills in a single run. We introduce a new model development framework that extends coevolution to large language model (LLM) discovery, open-ended \textit{Assessment Coevolving with Diverse Capabilities} (AC/DC). AC/DC evolves both LLMs via model merging and natural language tasks via synthetic data generation. AC/DC discovers growing archives of LLMs that surpass the capabilities of larger LLMs while taking up less GPU memory. In particular, our LLM populations achieve a broader Coverage of expertise than other curated models or baselines on downstream benchmarks, without \textit{any} explicit benchmark optimization. Furthermore, AC/DC improves Coverage over time, continually innovates on tasks and models, and improves performance in multi-agent best-of-N selection. Our findings highlight the potential of coevolution as a means of discovering broader sets of capabilities from base LLMs. Overall, AC/DC brings us one step closer to a profoundly new paradigm of LLM development, where continual improvements to the diversity of model capabilities can be accelerated by leveraging existing models as stepping stones to increasingly powerful models.

Retrieval-Augmented Generation (RAG) extends Large Vision-Language Models (LVLMs) with external visual knowledge. However, existing visual RAG systems typically rely on generic retrieval signals that overlook the fine-grained visual semantics essential for complex reasoning. To address this limitation, we propose UniDoc-RL, a unified reinforcement learning framework in which an LVLM agent jointly performs retrieval, reranking, active visual perception, and reasoning. UniDoc-RL formulates visual information acquisition as a sequential decision-making problem with a hierarchical action space. Specifically, it progressively refines visual evidence from coarse-grained document retrieval to fine-grained image selection and active region cropping, allowing the model to suppress irrelevant content and attend to information-dense regions. For effective end-to-end training, we introduce a dense multi-reward scheme that provides task-aware supervision for each action. Based on Group Relative Policy Optimization (GRPO), UniDoc-RL aligns agent behavior with multiple objectives without relying on a separate value network. To support this training paradigm, we curate a comprehensive dataset of high-quality reasoning trajectories with fine-grained action annotations. Experiments on three benchmarks demonstrate that UniDoc-RL consistently surpasses state-of-the-art baselines, yielding up to 17.7% gains over prior RL-based methods.

Contextual bandit algorithms suffer from high regret during cold-start, when the learner has insufficient data to distinguish good arms from bad. We propose augmenting Disjoint LinUCB with LLM pseudo-observations: after each round, a large language model predicts counterfactual rewards for the unplayed arms, and these predictions are injected into the learner as weighted pseudo-observations. The injection weight is controlled by a calibration-gated decay schedule that tracks the LLM's prediction accuracy on played arms via an exponential moving average; high calibration error suppresses the LLM's influence, while accurate predictions receive higher weight during the critical early rounds. We evaluate on two contextual bandit environments - UCI Mushroom (2-arm, asymmetric rewards) and MIND-small (5-arm news recommendation) - and find that when equipped with a task-specific prompt, LLM pseudo-observations reduce cumulative regret by 19% on MIND relative to pure LinUCB. However, generic counterfactual prompt framing increases regret on both environments, demonstrating that prompt design is the dominant factor, more important than the choice of decay schedule or calibration gating parameters. We analyze the failure modes of calibration gating on domains with small prediction errors and provide a theoretical motivation for the bias-variance trade-off governing pseudo-observation weight.

Network security is a critical concern in the digital landscape of today, with users demanding secure browsing experiences and protection of their personal data. This study explores the dynamic integration of Machine Learning (ML) algorithms with Software-Defined Networking (SDN) controllers to enhance network security through adaptive decision mechanisms. The proposed approach enables the system to dynamically choose the most suitable ML algorithm based on the characteristics of the observed network traffic. This work examines the role of Intrusion Detection Systems (IDS) as a fundamental component of secure communication networks and discusses the limitations of SDN-based attack detection mechanisms. The proposed framework uses adaptive model selection to maintain reliable intrusion detection under varying network conditions. The study highlights the importance of analyzing traffic-type-based metrics to define effective classification rules and enhance the performance of ML models. Additionally, it addresses the risks of overfitting and underfitting, underscoring the critical role of hyperparameter tuning in optimizing model accuracy and generalization. The central contribution of this work is an automated mechanism that adaptively selects the most suitable ML algorithm according to real-time network conditions, prioritizing detection robustness and operational feasibility within SDN environments.

Training GUI agents with traditional centralized methods faces significant cost and scalability challenges. Federated learning (FL) offers a promising solution, yet its potential is hindered by the lack of benchmarks that capture real-world, cross-platform heterogeneity. To bridge this gap, we introduce FedGUI, the first comprehensive benchmark for developing and evaluating federated GUI agents across mobile, web, and desktop platforms. FedGUI provides a suite of six curated datasets to systematically study four crucial types of heterogeneity: cross-platform, cross-device, cross-OS, and cross-source. Extensive experiments reveal several key insights: First, we show that cross-platform collaboration improves performance, extending prior mobile-only federated learning to diverse GUI environments; Second, we demonstrate the presence of distinct heterogeneity dimensions and identify platform and OS as the most influential factors. FedGUI provides a vital foundation for the community to build more scalable and privacy-preserving GUI agents for real-world deployment. Our code and data are publicly available at https://github.com/wwh0411/FedGUI..

Tool learning with foundation models aims to endow AI systems with the ability to invoke external resources -- such as APIs, computational utilities, and specialized models -- to solve complex tasks beyond the reach of standalone language generation. While recent advances in Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs) have expanded their reasoning and perception capabilities, existing tool-use methods are predominantly limited to text-only inputs and closed-world settings. Consequently, they struggle to interpret multimodal user instructions and cannot generalize to tools unseen during training. In this work, we introduce RaTA-Tool, a novel framework for open-world multimodal tool selection. Rather than learning direct mappings from user queries to fixed tool identifiers, our approach enables an MLLM to convert a multimodal query into a structured task description and subsequently retrieve the most appropriate tool by matching this representation against semantically rich, machine-readable tool descriptions. This retrieval-based formulation naturally supports extensibility to new tools without retraining. To further improve alignment between task descriptions and tool selection, we incorporate a preference-based optimization stage using Direct Preference Optimization (DPO). To support research in this setting, we also introduce the first dataset for open-world multimodal tool use, featuring standardized tool descriptions derived from Hugging Face model cards. Extensive experiments demonstrate that our approach significantly improves tool-selection performance, particularly in open-world, multimodal scenarios.

In this paper, we proposed Bayesian Tucker decomposition (BTuD) in which residual is supposed to obey Gaussian distribution analogous to linear regression. Although we have proposed an algorithm to perform the proposed BTuD, the conventional higher-order orthogonal iteration can generate Tucker decomposition consistent with the present implementation. Using the proposed BTuD, we can perform unsupervised feature selection successfully applied to various synthetic datasets, global coupled maps with randomized coupling strength, and gene expression profiles. Thus we can conclude that our newly proposed unsupervised feature selection method is promising. In addition to this, BTuD based unsupervised FE is expected to coincide with TD based unsupervised FE that were previously proposed and successfully applied to a wide range of problems.

Android apps are built on APIs that abstract core Android system functionalities. These APIs are officially documented in multiple files distributed with the Android source code or SDK, which we collectively refer to as Android API Lists (AALs). Prior Android research has relied on specific AALs, often treating them as interchangeable ground truth. However, recent studies suggest that different AALs can lead to substantially different research outcomes, raising concerns about the validity and reproducibility of Android API-based analyses. To address this issue, we present the first in-depth empirical study of four official AALs that are widely used in prior work. We systematically characterize their contents and analyze their evolution across Android releases. We then perform a fine-grained comparison of the APIs recorded in each AAL to uncover their underlying API inclusion policies and inconsistencies. To assess the practical impact of these differences, we further examine API availability on nine Android devices, including both stock Android and vendor-customized systems. Finally, we analyze API usage in 17,759 real-world Android apps (including open-source apps, commercial apps, and malware) to quantify how the choice of AAL affects empirical Android research. Our results reveal that official AALs are neither stable nor mutually consistent, and that discrepancies among them can substantially influence research conclusions. We also observe that vendor-customized APIs are actively used by normal apps, yet remain largely overlooked by existing studies. Based on these findings, we discuss their implications for Android API-based research and provide actionable suggestions to help researchers select and interpret AALs more reliably.

Communicating complex system designs or scientific processes through text alone is inefficient and prone to ambiguity. A system that automatically generates scientific architecture diagrams from text with high semantic fidelity can be useful in multiple applications like enterprise architecture visualization, AI-driven software design, and educational content creation. Hence, in this paper, we focus on leveraging language models to perform semantic understanding of the input text description to generate intermediate code that can be processed to generate high-fidelity architecture diagrams. Unfortunately, no clean large-scale open-access dataset exists, implying lack of any effective open models for this task. Hence, we contribute a comprehensive dataset, \system, comprising scientific architecture images, their corresponding textual descriptions, and associated DOT code representations. Leveraging this resource, we fine-tune a suite of small language models, and also perform in-context learning using GPT-4o. Through extensive experimentation, we show that \system{} models significantly outperform existing baseline models like DiagramAgent and perform at par with in-context learning-based generations from GPT-4o. We make the code, data and models publicly available.

Automatic evaluation metrics are essential for building multilingual translation systems. The common practice of evaluating these systems is averaging metric scores across languages, yet this is suspicious since metrics may suffer from cross-lingual scoring bias, where translations of equal quality receive different scores across languages. This problem has not been systematically studied because no benchmark exists that provides parallel-quality instances across languages, and expert annotation is not realistic. In this work, we propose XQ-MEval, a semi-automatically built dataset covering nine translation directions, to benchmark translation metrics. Specifically, we inject MQM-defined errors into gold translations automatically, filter them by native speakers for reliability, and merge errors to generate pseudo translations with controllable quality. These pseudo translations are then paired with corresponding sources and references to form triplets used in assessing the qualities of translation metrics. Using XQ-MEval, our experiments on nine representative metrics reveal the inconsistency between averaging and human judgment and provide the first empirical evidence of cross-lingual scoring bias. Finally, we propose a normalization strategy derived from XQ-MEval that aligns score distributions across languages, improving the fairness and reliability of multilingual metric evaluation.

End-to-end spoken dialogue models have garnered significant attention because they offer a higher potential ceiling in expressiveness and perceptual ability than cascaded systems. However, the intelligence and expressiveness of current open-source spoken dialogue models often remain below expectations. Motivated by the success of online reinforcement learning(RL) in other domains, one might attempt to directly apply preference optimization to spoken dialogue models, yet this transfer is non-trivial. We analyze these obstacles from the perspectives of reward modeling and rollout sampling, focusing on how sparse preference supervision interacts with dense speech generation under shared-parameter updates. Based on the analysis, we propose a modality-aware adaptive post-training recipe that makes RL practical for spoken dialogue: it constrains preference updates to the semantic channel and improves acoustic behavior via explicit anchoring, while dynamically regulating their mixture from rollout statistics to avoid unreliable preference gradients. We evaluate the method across multiple spoken dialogue benchmarks and representative architectures, and observe consistent improvements in semantic quality and speech expressiveness.

Concatenating quantum error correction codes scales error correction capability by driving logical error rates down double-exponentially across levels. However, the noise structure shifts under concatenation, making it hard to choose an optimal code sequence. We automate this choice by estimating the effective noise channel after each level and selecting the next code accordingly. In particular, we use learning-based methods to tailor small, non-additive encoders when the noise exhibits sufficient structure, then switch to standard codes once the noise is nearly uniform. In simulations, this level-wise adaptation achieves a target logical error rate with far fewer qubits than concatenating stabilizer codes alone--reducing qubit counts by up to two orders of magnitude for strongly structured noise. Therefore, this hybrid, learning-based strategy offers a promising tool for early fault-tolerant quantum computing.

Information Extraction aims to distill structured, decision-relevant information from unstructured text, serving as a foundation for downstream understanding and reasoning. However, it is traditionally treated merely as a terminal objective: once extracted, the resulting structure is often consumed in isolation rather than maintained and reused during multi-step inference. Moving beyond this, we propose \textit{IE-as-Cache}, a framework that repurposes IE as a cognitive cache to enhance agentic reasoning. Drawing inspiration from hierarchical computer memory, our approach combines query-driven extraction with cache-aware reasoning to dynamically maintain compact intermediate information and filter noise. Experiments on challenging benchmarks across diverse LLMs demonstrate significant improvements in reasoning accuracy, indicating that IE can be effectively repurposed as a reusable cognitive resource and offering a promising direction for future research on downstream uses of IE.

Many CAD learning pipelines discretize Boundary Representations (B-Reps) into triangle meshes, discarding analytic surface structure and topological adjacency and thereby weakening consistent instance-level analysis. We present STEP-Parts, a deterministic CAD-to-supervision toolchain that extracts geometric instance partitions directly from raw STEP B-Reps and transfers them to tessellated carriers through retained source-face correspondence, yielding instance labels and metadata for downstream learning and evaluation. The construction merges adjacent B-Rep faces only when they share the same analytic primitive type and satisfy a near-tangent continuity criterion. On ABC, same-primitive dihedral angles are strongly bimodal, yielding a threshold-insensitive low-angle regime for part extraction. Because the partition is defined on intrinsic B-Rep topology rather than on a particular triangulation, the resulting boundaries remain stable under changes in tessellation. Applied to the DeepCAD subset of ABC, the pipeline processes approximately 180{,}000 models in under six hours on a consumer CPU. We release code and precomputed labels, and show that STEP-Parts serves both as a tessellation-robust geometric reference and as a useful supervision source in two downstream probes: an implicit reconstruction--segmentation network and a dataset-level point-based backbone.

Recently, sparse autoencoders (SAEs) have emerged as a promising technique for interpreting activations in foundation models by disentangling features into a sparse set of concepts. However, identifying the optimal level of sparsity for each neuron remains challenging in practice: excessive sparsity can lead to poor reconstruction, whereas insufficient sparsity may harm interpretability. While existing activation functions such as ReLU and TopK provide certain sparsity guarantees, they typically require additional sparsity regularization or cherry-picked hyperparameters. We show in this paper that dynamically sparse attention mechanisms using sparsemax can bridge this trade-off, due to their ability to determine the activation numbers in a data-dependent manner. Specifically, we first explore a new class of SAEs based on the cross-attention architecture with the latent features as queries and the learnable dictionary as the key and value matrices. To encourage sparse pattern learning, we employ a sparsemax-based attention strategy that automatically infers a sparse set of elements according to the complexity of each neuron, resulting in a more flexible and general activation function. Through comprehensive evaluation and visualization, we show that our approach successfully achieves lower reconstruction loss while producing high-quality concepts, particularly in top-n classification tasks.

Reinforcement Learning (RL) has emerged as a critical driver for enhancing the reasoning capabilities of Large Language Models (LLMs). While recent advancements have focused on reward engineering or data synthesis, few studies exploit the model's intrinsic representation characteristics to guide the training process. In this paper, we first observe the presence of high-magnitude activations within the query and key vectors when processing long contexts. Drawing inspiration from model quantization -- which establishes the criticality of such high-magnitude activations -- and the insight that long-context reasoning inherently exhibits a sparse structure, we hypothesize that these weights serve as the pivotal drivers for effective model optimization. Based on this insight, we propose LongAct, a strategy that shifts from uniform to saliency-guided sparse updates. By selectively updating only the weights associated with these significant activations, LongAct achieves an approximate 8% improvement on LongBench v2 and enhances generalization on the RULER benchmark. Furthermore, our method exhibits remarkable universality, consistently boosting performance across diverse RL algorithms such as GRPO and DAPO. Extensive ablation studies suggest that focusing on these salient features is key to unlocking long-context potential.

Achieving seamless, human-like interaction remains a key challenge for full-duplex spoken dialogue models (SDMs). Reinforcement learning (RL) has substantially enhanced text- and vision-language models, while well-designed reward signals are crucial for the performance of RL. We consider RL a promising strategy to address the key challenge for SDMs. However, a fundamental barrier persists: prevailing automated metrics for assessing interaction quality rely on superficial proxies, such as behavioral statistics or timing-prediction accuracy, failing to provide reliable reward signals for RL. On the other hand, human evaluations, despite their richness, remain costly, inconsistent, and difficult to scale. We tackle this critical barrier by proposing a Dual-Axis Generative Reward Model, which is trained to understand complex interaction dynamics using a detailed taxonomy and an annotated dataset, produces a single score and, crucially, provides separate evaluations for semantic quality and interaction timing. Such dual outputs furnish precise diagnostic feedback for SDMs and deliver a dependable, instructive reward signal suitable for online reinforcement learning. Our model achieves state-of-the-art performance on interaction-quality assessment across a wide spectrum of datasets, spanning synthetic dialogues and complex real-world interactions.

We study downlink beam and rate adaptation in a multi-user mmWave MISO system where multiple base stations (BSs), each using analog beamforming from finite codebooks, serve multiple single-antenna user equipments (UEs) with a unique beam per UE and discrete data transmission rates. BSs learn about transmission success based on ACK/NACK feedback. To encode service goals, we introduce a satisficing throughput threshold $τ_r$ and cast joint beam and rate adaptation as a combinatorial semi-bandit over beam-rate tuples. Within this framework, we propose SAT-CTS, a lightweight, threshold-aware policy that blends conservative confidence estimates with posterior sampling, steering learning toward meeting $τ_r$ rather than merely maximizing. Our main theoretical contribution provides the first finite-time regret bounds for combinatorial semi-bandits with satisficing objective: when $τ_r$ is realizable, we upper bound the cumulative satisficing regret to the target with a time-independent constant, and when $τ_r$ is non-realizable, we show that SAT-CTS incurs only a finite expected transient outside committed CTS rounds, after which its regret is governed by the sum of the regret contributions of restarted CTS rounds, yielding an $O((\log T)^2)$ standard regret bound. On the practical side, we evaluate the performance via cumulative satisficing regret to $τ_r$ alongside standard regret and fairness. Experiments with time-varying sparse multipath channels show that SAT-CTS consistently reduces satisficing regret and maintains competitive standard regret, while achieving favorable average throughput and fairness across users, indicating that feedback-efficient learning can equitably allocate beams and rates to meet QoS targets without channel state knowledge.

Online hate speech and abusive language pose a growing challenge for content moderation, especially in multilingual settings and for low-resource languages such as Lithuanian. This paper investigates to what extent modern multilingual sentence embedding models can support accurate hate speech detection in Lithuanian, Russian, and English, and how their performance depends on downstream modeling choices and feature dimensionality. We introduce LtHate, a new Lithuanian hate speech corpus derived from news portals and social networks, and benchmark six modern multilingual encoders (potion, gemma, bge, snow, jina, e5) on LtHate, RuToxic, and EnSuperset using a unified Python pipeline. For each embedding, we train both a one class HBOS anomaly detector and a two class CatBoost classifier, with and without principal component analysis (PCA) compression to 64-dimensional feature vectors. Across all datasets, two class supervised models consistently and substantially outperform one class anomaly detection, with the best configurations achieving up to 80.96% accuracy and AUC ROC of 0.887 in Lithuanian (jina), 92.19% accuracy and AUC ROC of 0.978 in Russian (e5), and 77.21% accuracy and AUC ROC of 0.859 in English (e5 with PCA). PCA compression preserves almost all discriminative power in the supervised setting, while showing some negative impact for the unsupervised anomaly detection case. These results demonstrate how modern multilingual sentence embeddings combined with gradient boosted decision trees provide robust soft-computing solutions for multilingual hate speech detection applications.

The SARS-CoV-2 RNA pseudoknot is a promising target for antiviral intervention, as it regulates the efficiency of $-$1 programmed ribosomal frameshifting ($-$1 PRF), a mechanism that is essential for viral protein synthesis. The pseudoknot represents a viral RNA sequence composed of helical stems that adopts two long-lived topologies, threaded and unthreaded. Ligand-induced distortion of this fold is thought to underlie the susceptibility of $-$1 PRF to small-molecule inhibitors. Resolving these distortions from unbiased molecular dynamics (MD) requires collective variables (CVs) that isolate the slowest dynamic modes of the RNA--ligand system from the high-frequency fluctuations. Here, we use spectral map (SM), a thermodynamics-driven machine-learning method, to learn such CVs directly from MD trajectories of the SARS-CoV-2 RNA pseudoknot in complex with the $-$1 PRF inhibitor merafloxacin and two related analogs. We examine both threaded and unthreaded pseudoknot topologies and consider the neutral and ionized ligand forms relevant at physiological pH. Free-energy landscapes show that ligand-induced destabilization is topology-selective: merafloxacin and its analogs destabilize the S2 stem in the threaded pseudoknot, whereas in the unthreaded pseudoknot, destabilization shifts to the S1 and S3 stems. We find that the zwitterionic form of merafloxacin uniquely imposes slow dynamics on the otherwise featureless unthreaded pseudoknot. Furthermore, the neutral and zwitterionic forms of merafloxacin differ qualitatively in their mechanisms within the same RNA topology. Overall, these results clarify how pseudoknot topology, ligand type, and protonation state shape the slow conformational dynamics of viral RNA and establish physiological protonation as an essential factor for modeling RNA-targeted drug action.

Intelligent embodied agents should not simply follow instructions, as real-world environments often involve unexpected conditions and exceptions. However, existing methods usually focus on directly executing instructions, without considering whether the target objects can actually be manipulated, meaning they fail to assess available affordances. To address this limitation, we introduce DynAfford, a benchmark that evaluates embodied agents in dynamic environments where object affordances may change over time and are not specified in the instruction. DynAfford requires agents to perceive object states, infer implicit preconditions, and adapt their actions accordingly. To enable this capability, we introduce ADAPT, a plug-and-play module that augments existing planners with explicit affordance reasoning. Experiments demonstrate that incorporating ADAPT significantly improves robustness and task success across both seen and unseen environments. We also show that a domain-adapted, LoRA-finetuned vision-language model used as the affordance inference backend outperforms a commercial LLM (GPT-4o), highlighting the importance of task-aligned affordance grounding.

Large language models have advanced rapidly, from pattern recognition to emerging forms of reasoning, yet they remain confined to linguistic simulation rather than grounded understanding. They can produce fluent outputs that resemble reflection, but lack temporal continuity, causal feedback, and anchoring in real-world interaction. This paper proposes a complementary approach in which reasoning is treated as a relational process distributed between human and model rather than an internal capability of either. Building on recent work on "System-2" learning, we relocate reflective reasoning to the interaction layer. Instead of engineering reasoning solely within models, we frame it as a cognitive protocol that can be structured, measured, and governed using existing systems. This perspective emphasizes collaborative intelligence, combining human judgment and contextual understanding with machine speed, memory, and associative capacity. We introduce "The Architect's Pen" as a practical method. Like an architect who thinks through drawing, the human uses the model as an external medium for structured reflection. By embedding phases of articulation, critique, and revision into human-AI interaction, the dialogue itself becomes a reasoning loop: human abstraction -> model articulation -> human reflection. This reframes the question from whether the model can think to whether the human-AI system can reason. The framework enables auditable reasoning traces and supports alignment with emerging governance standards, including the EU AI Act and ISO/IEC 42001. It provides a practical path toward more transparent, controllable, and accountable AI use without requiring new model architectures.

We present an initial investigation into Agentic Retrieval-Augmented Generation (RAG) for Ukrainian, conducted within the UNLP 2026 Shared Task on Multi-Domain Document Understanding. Our system combines two-stage retrieval (BGE-M3 with BGE reranking) with a lightweight agentic layer performing query rephrasing and answer-retry loops on top of Qwen2.5-3B-Instruct. Our analysis reveals that retrieval quality is the primary bottleneck: agentic retry mechanisms improve answer accuracy but the overall score remains constrained by document and page identification. We discuss practical limitations of offline agentic pipelines and outline directions for combining stronger retrieval with more advanced agentic reasoning for Ukrainian.

We propose a new perspective on policy optimization: rather than reweighting all samples by their importance ratios, an optimizer should select which samples are trustworthy enough to drive a policy update. Building on this view, we introduce Rejection-Gated Policy Optimization (RGPO), which replaces the importance sampling ratio r_theta = pi_theta / pi_old with a smooth, differentiable acceptance gate alpha_theta(s, a) = g(r_theta(s, a)) in the range [0, 1]. Unlike prior work that applies rejection sampling as a data-level heuristic before training, RGPO elevates rejection to an optimization principle: the gate participates directly in gradient computation and is implicitly updated alongside the policy. RGPO provides a unified framework: the policy gradients of TRPO, PPO, and REINFORCE all correspond to specific choices of the effective gradient weight w(r) = g'(r) * r. We prove that RGPO guarantees finite, bounded gradient variance even when importance sampling ratios are heavy-tailed (where IS variance diverges). We further show that RGPO incurs only a bounded, controllable bias and provides an approximate monotonic policy improvement guarantee analogous to TRPO. RGPO matches PPO in computational cost, requires no second-order optimization, and extends naturally to RLHF-style preference alignment. In online preference fine-tuning of Qwen2.5-1.5B-Instruct on Anthropic HH-RLHF (n = 3 seeds), RGPO uses a dual-ratio gate that anchors learning to both the previous policy and the reference model, achieving a Pareto-dominant outcome: the highest reward among online RL methods (+14.8% vs. PPO-RLHF) and the lowest KL divergence to the reference model (-16.0% vs. PPO-RLHF, -53.1% vs. GRPO).

Evaluating medical AI systems using expert clinician panels is costly and slow, motivating the use of large language models (LLMs) as alternative adjudicators. Here, we evaluate an LLM jury composed of three frontier AI models scoring 3333 diagnoses on 300 real-world middle-income country (MIC) hospital cases. Model performance was benchmarked against expert clinician panel and independent human re-scoring panel evaluations. Both LLM and clinician-generated diagnoses are scored across four dimensions: diagnosis, differential diagnosis, clinical reasoning and negative treatment risk. For each of these, we assess scoring difference, inter-rater agreement, scoring stability, severe safety errors and the effect of post-hoc calibration. We find that: (i) the uncalibrated LLM jury scores are systematically lower than clinician panels scores; (ii) the LLM Jury preserves ordinal agreement and exhibits better concordance with the primary expert panels than the human expert re-score panels do; (iii) the probability of severe errors is lower in \lj models compared to the human expert re-score panels; (iv) the LLM Jury shows excellent agreement with primary expert panels' rankings. We find that the LLM jury combined with AI model diagnoses can be used to identify ward diagnoses at high risk of error, enabling targeted expert review and improved panel efficiency; (v) LLM jury models show no self-preference bias. They did not score diagnoses generated by their own underlying model or models from the same vendor more (or less) favourably than those generated by other models. Finally, we demonstrate that LLM jury calibration using isotonic regression improves alignment with human expert panel evaluations. Together, these results provide compelling evidence that a calibrated, multi-model LLM jury can serve as a trustworthy and reliable proxy for expert clinician evaluation in medical AI benchmarking.

While Chain-of-thought (CoT) reasoning enables LLMs to solve challenging reasoning problems, as KV cache grows linearly with the number of generated tokens, CoT reasoning faces scaling issues in terms of speed and memory usage. In this work, we propose MemoSight (Memory-Foresight-based reasoning), a unified framework that integrates both context compression and multi-token prediction to mitigate the efficiency issues while maintaining CoT reasoning performance. Our framework adopts the same minimalist design for both context compression and multi-token prediction via special tokens and their corresponding position layout tailored to each token type. Comprehensive experiments on four reasoning benchmarks demonstrate that MemoSight reduces the KV cache footprint by up to 66% and accelerates inference by 1.56x, while outperforming existing CoT compression methods.

Recent advances in vision language models (VLMs) offer reasoning capabilities, yet how these unfold and integrate visual and textual information remains unclear. We analyze reasoning dynamics in 18 VLMs covering instruction-tuned and reasoning-trained models from two different model families. We track confidence over Chain-of-Thought (CoT), measure the corrective effect of reasoning, and evaluate the contribution of intermediate reasoning steps. We find that models are prone to answer inertia, in which early commitments to a prediction are reinforced, rather than revised during reasoning steps. While reasoning-trained models show stronger corrective behavior, their gains depend on modality conditions, from text-dominant to vision-only settings. Using controlled interventions with misleading textual cues, we show that models are consistently influenced by these cues even when visual evidence is sufficient, and assess whether this influence is recoverable from CoT. Although this influence can appear in the CoT, its detectability varies across models and depends on what is being monitored. Reasoning-trained models are more likely to explicitly refer to the cues, but their longer and fluent CoTs can still appear visually grounded while actually following textual cues, obscuring modality reliance. In contrast, instruction-tuned models refer to the cues less explicitly, but their shorter traces reveal inconsistencies with the visual input. Taken together, these findings indicate that CoT provides only a partial view of how different modalities drive VLM decisions, with important implications for the transparency and safety of multimodal systems.

In data-sensitive domains such as healthcare, cross-silo federated learning (CFL) allows organizations to collaboratively train AI models without sharing raw data. However, practical CFL deployments are inherently coopetitive, in which organizations cooperate during model training while competing in downstream markets. In such settings, training contributions, including data volume, quality, and diversity, can improve the global model yet inadvertently strengthen rivals. This dilemma is amplified by non-IID data, which leads to asymmetric learning gains and undermines sustained participation. While existing competition-aware CFL and incentive-design approaches reward organizations based on marginal training contributions, they fail to account for the costs of strengthening competitors. In this paper, we introduce CoCoGen+, a coopetition-compatible data generation and incentivization framework that jointly models non-IID data and inter-organizational competition while endogenizing GenAI-based synthetic data generation as a strategic decision. Specifically, CoCoGen+ formulates each training round as a weighted potential game, where organizations strategically decide how much synthetic data to generate by balancing learning performance gains against computational costs and competition-caused utility losses. We then provide a tractable equilibrium characterization and derive implementable generation strategies to maximize social welfare. To promote long-term collaboration, we integrate a payoff redistribution-based incentive mechanism to compensate organizations for their contributions and competition-caused utility degradation. Experiments on varying learning tasks validate the feasibility of CoCoGen+. The results show how non-IID data, competition intensity, and incentives shape organizational strategies and social welfare, while CoCoGen+ outperforms baselines in efficiency.

Autoregressive decoding in Large Language Models (LLMs) generates one token per step, causing high inference latency. Speculative decoding (SD) mitigates this through a guess-and-verify strategy, but existing training-free variants face trade-offs: retrieval-based drafts break when no exact match exists, while logits-based drafts lack structural guidance. We propose $\textbf{RACER}$ ($\textbf{R}$etrieval-$\textbf{A}$ugmented $\textbf{C}$ont$\textbf{e}$xtual $\textbf{R}$apid Speculative Decoding), a lightweight and training-free method that integrates retrieved exact patterns with logit-driven future cues. This unification supplies both reliable anchors and flexible extrapolation, yielding richer speculative drafts. Experiments on Spec-Bench, HumanEval, and MGSM-ZH demonstrate that RACER consistently accelerates inference, achieving more than $2\times$ speedup over autoregressive decoding, and outperforms prior training-free methods, offering a scalable, plug-and-play solution for efficient LLM decoding. Our source code is available at $\href{https://github.com/hkr04/RACER}{https://github.com/hkr04/RACER}$.

Recent advances in Deep Learning (DL) have strengthened data-driven System Identification (SysID), with Neural and Fuzzy Ordinary Differential Equation (NODE/FODE) models achieving high accuracy in nonlinear dynamic modeling. Yet, system states in these frameworks are often reconstructed without clear physical meaning, and input contributions to the state derivatives remain difficult to interpret. To address these limitations, we propose Explainable FODE (xFODE), an interpretable SysID framework with integrated DL-based training. In xFODE, we define states in an incremental form to provide them with physical meanings. We employ fuzzy additive models to approximate the state derivative, thereby enhancing interpretability per input. To provide further interpretability, Partitioning Strategies (PSs) are developed, enabling the training of fuzzy additive models with explainability. By structuring the antecedent space during training so that only two consecutive rules are activated for any given input, PSs not only yield lower complexity for local inference but also enhance the interpretability of the antecedent space. To train xFODE, we present a DL framework with parameterized membership function learning that supports end-to-end optimization. Across benchmark SysID datasets, xFODE matches the accuracy of NODE, FODE, and NLARX models while providing interpretable insights.

Rapid urbanization and continuous population growth have made municipal solid waste management increasingly challenging. These challenges highlight the need for smarter and automated waste management solutions. This paper presents the design and evaluation of an integrated waste management framework that combines two connected systems, a robotic waste segregation module and an optimized bio-digestor. The robotic waste segregation system uses a MyCobot 280 Jetson Nano robotic arm along with YOLOv8 object detection and robot operating system (ROS)-based path planning to identify and sort waste in real time. It classifies waste into four different categories with high precision, reducing the need for manual intervention. After segregation, the biodegradable waste is transferred to a bio-digestor system equipped with multiple sensors. These sensors continuously monitor key parameters, including temperature, pH, pressure, and motor revolutions per minute. The Particle Swarm Optimization (PSO) algorithm, combined with a regression model, is used to dynamically adjust system parameters. This intelligent optimization approach ensures stable operation and maximizes digestion efficiency under varying environmental conditions. System testing under dynamic conditions demonstrates a sorting accuracy of 98% along with highly efficient biological conversion. The proposed framework offers a scalable, intelligent, and practical solution for modern waste management, making it suitable for both residential and industrial applications.

Large language models are increasingly integrated into decision-making in areas such as healthcare, law, finance, engineering, and government. Yet they share a critical limitation: they produce fluent outputs even when their internal reasoning has drifted. A confident answer can conceal uncertainty, speculation, or inconsistency, and small changes in phrasing can lead to different conclusions. This makes LLMs useful assistants but unreliable partners in high-stakes contexts. Humans exhibit a similar weakness, often mistaking fluency for reliability. When a model responds smoothly, users tend to trust it, even when both model and user are drifting together. This paper is the first in a five-paper research series on stabilising human-AI reasoning. The series proposes a two-layer approach: Parts II-IV introduce human-side mechanisms such as uncertainty cues, conflict surfacing, and auditable reasoning traces, while Part V develops a model-side Epistemic Control Loop (ECL) that detects instability and modulates generation accordingly. Together, these layers form a missing operational substrate for governance by increasing signal-to-noise at the point of use. Stabilising interaction makes uncertainty and drift visible before enforcement is applied, enabling more precise capability governance. This aligns with emerging compliance expectations, including the EU AI Act and ISO/IEC 42001, by making reasoning processes traceable under real conditions of use. The central claim is that fluency is not reliability. Without structures that stabilise both human and model reasoning, AI cannot be trusted or governed where it matters most.

Recent advances in Deep Learning (DL) have boosted data-driven System Identification (SysID), but reliable use requires Uncertainty Quantification (UQ) alongside accurate predictions. Although UQ-capable models such as Fuzzy ODE (FODE) can produce Prediction Intervals (PIs), they offer limited interpretability. We introduce Explainable Type-2 Fuzzy Additive ODEs for UQ (xFODE+), an interpretable SysID model which produces PIs alongside point predictions while retaining physically meaningful incremental states. xFODE+ implements each fuzzy additive model with Interval Type-2 Fuzzy Logic Systems (IT2-FLSs) and constraints membership functions to the activation of two neighboring rules, limiting overlap and keeping inference locally transparent. The type-reduced sets produced by the IT2-FLSs are aggregated to construct the state update together with the PIs. The model is trained in a DL framework via a composite loss that jointly optimizes prediction accuracy and PI quality. Results on benchmark SysID datasets show that xFODE+ matches FODE in PI quality and achieves comparable accuracy, while providing interpretability.

Nonlinear system identification must balance physical interpretability with model flexibility. Classical methods yield structured, control-relevant models but rely on rigid parametric forms that often miss complex nonlinearities, whereas Neural ODEs are expressive yet largely black-box. Physics-Informed Neural Networks (PINNs) sit between these extremes, but inverse PINNs typically assume a known governing equation with fixed coefficients, leading to identifiability failures when the true dynamics are unknown or state-dependent. We propose \textbf{SOLIS}, which models unknown dynamics via a \emph{state-conditioned second-order surrogate model} and recasts identification as learning a Quasi-Linear Parameter-Varying (Quasi-LPV) representation, recovering interpretable natural frequency, damping, and gain without presupposing a global equation. SOLIS decouples trajectory reconstruction from parameter estimation and stabilizes training with a cyclic curriculum and \textbf{Local Physics Hints} windowed ridge-regression anchors that mitigate optimization collapse. Experiments on benchmarks show accurate parameter-manifold recovery and coherent physical rollouts from sparse data, including regimes where standard inverse methods fail.

Generative Retrieval (GR) offers a promising paradigm for recommendation through next-token prediction (NTP). However, scaling it to large-scale industrial systems introduces three challenges: (i) within a single request, the identical model inputs may produce inconsistent outputs due to the pagination request mechanism; (ii) the prohibitive cost of encoding long user behavior sequences with multi-token item representations based on semantic IDs, and (iii) aligning the generative policy with nuanced user preference signals. We present GenRec, a preference-oriented generative framework deployed on the JD App that addresses above challenges within a single decoder-only architecture. For training objective, we propose Page-wise NTP task, which supervises over an entire interaction page rather than each interacted item individually, providing denser gradient signal and resolving the one-to-many ambiguity of point-wise training. On the prefilling side, an asymmetric linear Token Merger compresses multi-token Semantic IDs in the prompt while preserving full-resolution decoding, reducing input length by ~2X with negligible accuracy loss. To further align outputs with user satisfaction, we introduce GRPO-SR, a reinforcement learning method that pairs Group Relative Policy Optimization with NLL regularization for training stability, and employs Hybrid Rewards combining a dense reward model with a relevance gate to mitigate reward hacking. In month-long online A/B tests serving production traffic, GenRec achieves 9.5% improvement in click count and 8.7% in transaction count over the existing pipeline.

Does reinforcement learning genuinely expand what LLM agents can do, or merely make them more reliable? For static reasoning, recent work answers the second: base and RL pass@k curves converge at large k. We ask whether this holds for agentic tool use, where T rounds of interaction enable compositional strategies that re-sampling cannot recover. We introduce PASS@(k,T), a two-dimensional metric that jointly varies sampling budget k and interaction depth T, separating capability expansion from efficiency improvement. Our main finding is that, contrary to the static-reasoning result, tool-use RL genuinely enlarges the capability boundary: the RL agent's pass-curve pulls above the base model's and the gap widens at large k rather than converging. The expansion is specific to compositional, sequential information gathering; on simpler tasks RL behaves as prior work predicts. Under matched training data, supervised fine-tuning regresses the boundary on the same compositional tasks, isolating self-directed exploration as the causal factor. Mechanism analysis shows RL reweights the base strategy distribution toward the subset whose downstream reasoning more often yields a correct answer, with the improvement concentrated on how the agent integrates retrieved information. These results reconcile optimistic and pessimistic readings of RL for LLMs: both are correct, on different task types.

We study the tail behavior of regret in stochastic multi-armed bandits for algorithms that are asymptotically optimal in expectation. While minimizing expected regret is the classical objective, recent work shows that even such algorithms can exhibit heavy regret tails, incurring large regret with non-negligible probability. Existing sharp characterizations of regret tails are largely restricted to parametric settings, such as single-parameter exponential families. In this work, we extend the $\KLinf$-UCB algorithm of to a broad nonparametric class of reward distributions satisfying mild assumptions, and establish its asymptotic optimality in expectation. We then analyze the tail behavior of its regret and derive a novel upper bound on the regret tail probability. As special cases, our results recover regret-tail guarantees for both bounded-support and heavy-tailed (moment-bounded) bandit models. Moreover, for the special case of finitely-supported reward distributions, our upper bound matches the known lower bound exactly. Our results thus provide a unified and tight characterization of regret tails for asymptotically optimal KL-based UCB algorithms, going beyond parametric models.

Local loss-landscape stabilization under sample growth is typically measured either pointwise or through isotropic averaging in the full parameter space. Despite practical value, both choices probe directions that contribute little to the dominant local deformation of strongly anisotropic neural landscapes. We recast stabilization as an observational problem and introduce a unified family of criteria parameterized by an aggregation order and a probing distribution; within this family we propose a curvature-aligned criterion $Δ_2^{(D)}$ that probes the loss increment field in the top-$D$ eigenspace of the empirical Hessian near a trained solution. Solely from a local quadratic model, we prove that $Δ_2^{(D)}$ preserves the $O(k^{-2})$ mean-squared rate of the full-space criterion while replacing ambient-dimension curvature dependence with dependence on the subspace dimension $D$; a corollary gives a closed-form spectral expression and a proposition identifies the top-$D$ eigenspace as extremal within the eigenspace-aligned family. We also derive scalable estimators based on Hessian-vector products, subspace Monte Carlo, and a closed-form Gaussian-moment proxy. On a decoder-only transformer, a curvature-aligned probe occupying a tiny fraction of parameter space already reproduces the full-space mean-squared signal to within numerical noise throughout the validated local regime, and the closed-form estimator is orders of magnitude faster than direct Monte Carlo after subspace construction.

Vibe coding inherently assumes iterative refinement of LLM-generated code through feedback loops. While effective for conventional software tasks, its reliability in runtime-adaptive systems is unclear -- especially when generated code is not manually inspected. This paper studies feedback-based automated verification of LLM-generated adaptation managers in Collective Adaptive Systems (CAS). We focus on the key challenges of verification in the loop: how to detect failures of generated code at runtime and how to report them precisely enough for an LLM to fix them. We combine the adaptation loop with a vibe-coding feedback loop where correctness is checked against (i) generic architectural constraints and (ii) functional constraints formalized in Functional Constraints Logic (FCL), a novel first-order temporal logic over potentially finite traces. Conducting the Dragon Hunt CAS case study, we show that fine-grained constraint violations provide actionable feedback that typically yields a valid adaptation manager within a few iterations, while simple coarse metric-based feedback often stalls. Our findings suggest that feedback precision is the dominant factor for reliable vibe coding in systems designed by domain experts with no programming skills, thereby obviating the need for human code inspection.

Vision-Language Models (VLMs) have demonstrated significant potential in medical image analysis, yet their application in intraoral photography remains largely underexplored due to the lack of fine-grained, annotated datasets and comprehensive benchmarks. To address this, we present MetaDent, a comprehensive resource that includes (1) a novel and large-scale dentistry image dataset collected from clinical, public, and web sources; (2) a semi-structured annotation framework designed to capture the hierarchical and clinically nuanced nature of dental photography; and (3) comprehensive benchmark suites for evaluating state-of-the-art VLMs on clinical image understanding. Our labeling approach combines a high-level image summary with point-by-point, free-text descriptions of abnormalities. This method enables rich, scalable, and task-agnostic representations. We curated 60,669 dental images from diverse sources and annotated a representative subset of 2,588 images using this meta-labeling scheme. Leveraging Large Language Models (LLMs), we derive standardized benchmarks: approximately 15K Visual Question Answering (VQA) pairs and an 18-class multi-label classification dataset, which we validated with human review and error analysis to justify that the LLM-driven transition reliably preserves fidelity and semantic accuracy. We then evaluate state-of-the-art VLMs across VQA, classification, and image captioning tasks. Quantitative results reveal that even the most advanced models struggle with a fine-grained understanding of intraoral scenes, achieving moderate accuracy and producing inconsistent or incomplete descriptions in image captioning. We publicly release our dataset, annotations, and tools to foster reproducible research and accelerate the development of vision-language systems for dental applications.

Large Language Models (LLMs) are increasingly exposed to adaptive jailbreaking, particularly in high-stakes Chemical, Biological, Radiological, and Nuclear (CBRN) domains. Although streaming probes enable real-time monitoring, they still make systematic errors. We identify a core issue: existing methods often rely on a few high-scoring tokens, leading to false alarms when sensitive CBRN terms appear in benign contexts. To address this, we introduce a streaming probing objective that requires multiple evidence tokens to consistently support a prediction, rather than relying on isolated spikes. This encourages more robust detection based on aggregated signals instead of single-token cues. At a fixed 1% false-positive rate, our method improves the true-positive rate by 35.55% relative to strong streaming baselines. We further observe substantial gains in AUROC, even when starting from near-saturated baseline performance (AUROC = 97.40%). We also show that probing Attention or MLP activations consistently outperforms residual-stream features. Finally, even when adversarial fine-tuning enables novel character-level ciphers, harmful intent remains detectable: probes developed for the base LLMs can be applied ``plug-and-play'' to these obfuscated attacks, achieving an AUROC of over 98.85%.

Constrained decoding has been widely adopted for structured generation with large language models (LLMs), ensuring that outputs satisfy predefined formats such as JSON and XML. However, existing approaches largely treat schemas as purely structural constraints and overlook the possibility that their linguistic formulation may affect model behavior. In this work, we study how instruction placement influences model performance in structured generation and show that merely changing the wording of schema keys, without modifying the prompt or model parameters, can significantly alter model performance under constrained decoding. Based on this observation, we propose to reinterpret structured generation as a multi-channel instruction problem, where instructions can be conveyed explicitly through prompts and implicitly through schema keys during decoding. To the best of our knowledge, this is the first work to systematically study how schema key formulation acts as an implicit instruction channel and affects model performance under constrained decoding. Experiments on multiple mathematical reasoning benchmarks show that different model families exhibit distinct sensitivities to these instruction channels: Qwen models consistently benefit from schema-level instructions, while LLaMA models rely more heavily on prompt-level guidance. We further observe non-additive interaction effects between instruction channels, showing that combining multiple channels does not always lead to further improvement. These findings suggest that schema design not only determines output structure, but also carries instruction signals, offering a new perspective on structured generation in LLMs.

We study bandit best-arm identification with arbitrary and potentially adversarial rewards. A simple random uniform learner obtains the optimal rate of error in the adversarial scenario. However, this type of strategy is suboptimal when the rewards are sampled stochastically. Therefore, we ask: Can we design a learner that performs optimally in both the stochastic and adversarial problems while not being aware of the nature of the rewards? First, we show that designing such a learner is impossible in general. In particular, to be robust to adversarial rewards, we can only guarantee optimal rates of error on a subset of the stochastic problems. We give a lower bound that characterizes the optimal rate in stochastic problems if the strategy is constrained to be robust to adversarial rewards. Finally, we design a simple parameter-free algorithm and show that its probability of error matches (up to log factors) the lower bound in stochastic problems, and it is also robust to adversarial ones.

As agent systems move into increasingly diverse execution settings, trajectory-level safety evaluation and diagnosis require benchmarks that evolve with them. ATBench is a diverse and realistic agent trajectory benchmark for safety evaluation and diagnosis. This report presents ATBench-Claw and ATBench-CodeX, two domain-customized extensions that carry ATBench into the OpenClaw and OpenAI Codex / Codex-runtime settings. The key adaptation mechanism is to analyze each new setting, customize the three-dimensional Safety Taxonomy over risk source, failure mode, and real-world harm, and then use that customized taxonomy to define the benchmark specification consumed by the shared ATBench construction pipeline. This extensibility matters because agent frameworks remain relatively stable at the architectural level even as their concrete execution settings, tool ecosystems, and product capabilities evolve quickly. Concretely, ATBench-Claw targets OpenClaw-sensitive execution chains over tools, skills, sessions, and external actions, while ATBench-CodeX targets trajectories in the OpenAI Codex / Codex-runtime setting over repositories, shells, patches, dependencies, approvals, and runtime policy boundaries. Our emphasis therefore falls on taxonomy customization, domain-specific risk coverage, and benchmark design under a shared ATBench generation framework.

Understanding climate change requires reasoning over complex causal networks. Yet, existing causal discovery datasets predominantly capture explicit, direct causal relations. We introduce ClimateCause, a manually expert-annotated dataset of higher-order causal structures from science-for-policy climate reports, including implicit and nested causality. Cause-effect expressions are normalized and disentangled into individual causal relations to facilitate graph construction, with unique annotations for cause-effect correlation, relation type, and spatiotemporal context. We further demonstrate ClimateCause's value for quantifying readability based on the semantic complexity of causal graphs underlying a statement. Finally, large language model benchmarking on correlation inference and causal chain reasoning highlights the latter as a key challenge.

Test-time compute scaling, the practice of spending extra computation during inference via repeated sampling, search, or extended reasoning, has become a powerful lever for improving large language model performance. Yet deploying these techniques under finite inference budgets requires a decision that current systems largely ignore: which inputs deserve more compute, and which can be answered cheaply? We formalize this as a constrained optimization problem (maximize expected accuracy subject to an average compute budget) and solve it with a two-stage Solve-then-Learn pipeline. In the solve stage, Lagrangian relaxation decomposes the global constraint into per-instance sub-problems, each admitting a closed-form oracle action that optimally prices accuracy against cost. We prove that the induced cost is monotone in the dual variable, enabling exact budget targeting via binary search. In the learn stage, a lightweight classifier is trained to predict oracle actions from cheap input features, amortizing the allocation rule for real-time deployment. We establish that the task-level regret of the learned policy is bounded by its imitation error times the worst-case per-instance gap, yielding a clean reduction from constrained inference to supervised classification. Experiments on MATH and GSM8K with three LLMs (DeepSeek-V3, GPT-4o-mini, Qwen2.5-7B) show that our method consistently outperforms uniform and heuristic allocation baselines, achieving up to 12.8% relative accuracy improvement on MATH under matched budget constraints, while closely tracking the Lagrangian oracle upper bound with over 91% imitation accuracy.

Purpose: Adaptive skip modules can improve medical image segmentation, but searching for them is computationally costly. Implantable Adaptive Cells (IACs) are compact NAS modules inserted into U-Net skip connections, reducing the search space compared with full-network NAS. However, the original IAC framework still requires a 200-epoch differentiable search for each backbone and dataset. Methods: We analyzed the temporal behavior of operations and edges within IAC cells during differentiable search on public medical image segmentation benchmarks. We found that operations selected in the final discrete cell typically emerge among the strongest candidates early in training, and their architecture parameters stabilize well before the final epoch. Based on this, we propose a Jensen--Shannon-divergence-based stability criterion that tracks per-edge operation-importance distributions and progressively prunes low-importance operations during search. The accelerated framework is called IAC-LTH. Results: Across four public benchmarks (ACDC, BraTS, KiTS, AMOS), several 2-D U-Net backbones, and a 2-D nnU-Net pipeline, IAC-LTH discovers IAC cells whose patient-level segmentation performance matches and sometimes slightly exceeds that of cells found by the original full-length search, while reducing wall-clock NAS cost by 3.7x to 16x across datasets and backbones. These results are consistent across architectures, benchmarks, and both non-augmented and augmented training settings, while preserving the gains of IAC-equipped U-Nets over strong attention-based and dense-skip baselines. Conclusion: Competitive IAC architectures can be identified from early-stabilizing operations without running the full search, making adaptive skip-module design more practical for medical image segmentation under realistic computational constraints.

Large Reasoning Models (LRMs) achieve strong performance on complex tasks through extended chains of thought but suffer from high inference latency due to autoregressive reasoning. Recent work explores using Small Reasoning Models (SRMs) to accelerate LRM inference. In this paper, we systematically characterize the capability boundaries of SRMs and identify three common types of reasoning risks: (1) path divergence, where SRMs lack the strategic ability to construct an initial plan, causing reasoning to deviate from the most probable path; (2) cognitive overload, where SRMs fail to solve particularly difficult steps; and (3) recovery inability, where SRMs lack robust self-reflection and error correction mechanisms. To address these challenges, we propose TrigReason, a trigger-based collaborative reasoning framework that replaces continuous polling with selective intervention. TrigReason delegates most reasoning to the SRM and activates LRM intervention only when necessary-during initial strategic planning (strategic priming trigger), upon detecting extraordinary overconfidence (cognitive offload trigger), or when reasoning falls into unproductive loops (intervention request trigger). The evaluation results on AIME24, AIME25, and GPQA-D indicate that TrigReason matches the accuracy of full LRMs and SpecReason, while offloading 1.70x - 4.79x more reasoning steps to SRMs. Under edge-cloud conditions, TrigReason reduces latency by 43.9\% and API cost by 73.3\%. Our code is available at \href{https://github.com/QQQ-yi/TrigReason}{https://github.com/QQQ-yi/TrigReason}

Retail theft costs the global economy over \$100 billion annually, yet existing AI-based detection systems require expensive custom model training on proprietary datasets and charge \$200-500/month per store. We present Paza, a zero-shot retail theft detection framework that achieves practical concealment detection without training any model. Our approach orchestrates multiple existing models in a layered pipeline - cheap object detection and pose estimation running continuously, with an expensive vision-language model (VLM) invoked only when behavioral pre-filters trigger. A multi-signal suspicion pre-filter (requiring dwell time plus at least one behavioral signal) reduces VLM invocations by 240x compared to per-frame analysis, bounding calls to <=10/minute and enabling a single GPU to serve 10-20 stores. The architecture is model-agnostic: the VLM component accepts any OpenAI-compatible endpoint, enabling operators to swap between models such as Gemma 4, Qwen3.5-Omni, GPT-4o, or future releases without code changes - ensuring the system improves as the VLM landscape evolves. We evaluate the VLM component on the DCSASS synthesized shoplifting dataset (169 clips, controlled environment), achieving 89.5% precision and 92.8% specificity at 59.3% recall zero-shot - where the recall gap is attributable to sparse frame sampling in offline evaluation rather than VLM reasoning failures, as precision and specificity are the operationally critical metrics determining false alarm rates. We present a detailed cost model showing viability at \$50-100/month per store (3-10x cheaper than commercial alternatives), and introduce a privacy-preserving design that obfuscates faces in the detection pipeline. The source code is available at https://github.com/xHaileab/Paza-AI.

Behavioral Profile (BP) annotation is difficult to automate because it requires simultaneous coding across multiple linguistic dimensions. We treat BP annotation as a bundle of annotation skills rather than a single task and evaluate LLM-assisted BP annotation from this perspective. Using 3,134 concordance lines of 30 Chinese metaphorical color-term derivatives and a 14-feature BP schema, we implement a skill-file-driven pipeline in which each feature is externally defined through schema files, decision rules, and examples. Two human annotators completed a two-round schema-only protocol on a 300-instance validation subset, enabling BP skills to be classified as directly operable, recoverable under focused re-annotation, or structurally underspecified. GPT-5.4 and three locally deployable open-source models were then evaluated under the same setup. Results show that BP annotation is highly heterogeneous at the skill level: 5 skills are directly operable, 4 are recoverable after focused re-annotation, and 5 remain structurally underspecified. GPT-5.4 executes the retained skills with substantial reliability (accuracy = 0.678, \k{appa} = 0.665, weighted F1 = 0.695), but this feasibility is selective rather than global. Human and GPT difficulty profiles are strongly aligned at the skill level (r = 0.881), but not at the instance level (r = 0.016) or lexical-item level (r = -0.142), a pattern we describe as shared taxonomy, independent execution. Pairwise agreement further suggests that GPT is better understood as an independent third skill voice than as a direct human substitute. Open-source failures are concentrated in schema-to-skill execution problems. These findings suggest that automatic annotation should be evaluated in terms of skill feasibility rather than task-level automation.

Current single-cell foundation model benchmarks universally extract final layer embeddings, assuming these represent optimal feature spaces. We systematically evaluate layer-wise representations from scFoundation (100M parameters) and Tahoe-X1 (1.3B parameters) across trajectory inference and perturbation response prediction. Our analysis reveals that optimal layers are task-dependent (trajectory peaks at 60% depth, 31% above final layers) and context-dependent (perturbation optima shift 0-96% across T cell activation states). Notably, first-layer embeddings outperform all deeper layers in quiescent cells, challenging assumptions about hierarchical feature abstraction. These findings demonstrate that "where" to extract features matters as much as "what" the model learns, necessitating systematic layer evaluation tailored to biological task and cellular context rather than defaulting to final-layer embeddings.

Evaluating large language models (LLMs) for clinical documentation tasks such as SOAP note generation remains challenging. Unlike standard summarization, these tasks require clinical abstraction, normalization of colloquial language, and medically grounded inference. However, prevailing evaluation methods including automated metrics and LLM as judge frameworks rely on lexical faithfulness, often labeling any information not explicitly present in the transcript as hallucination. We show that such approaches systematically misclassify clinically valid outputs as errors, inflating hallucination rates and distorting model assessment. Our analysis reveals that many flagged hallucinations correspond to legitimate clinical transformations, including synonym mapping, abstraction of examination findings, diagnostic inference, and guideline consistent care planning. By aligning evaluation criteria with clinical reasoning through calibrated prompting and retrieval grounded in medical ontologies we observe a significant shift in outcomes. Under a lexical evaluation regime, the mean hallucination rate is 35%, heavily penalizing valid reasoning. With inference aware evaluation, this drops to 9%, with remaining cases reflecting genuine safety concerns. These findings suggest that current evaluation practices over penalize valid clinical reasoning and may measure artifacts of evaluation design rather than true errors, underscoring the need for clinically informed evaluation in high context domains like medicine.

Educational assistants should spend more computation only when the task needs it. This paper rewrites our earlier draft around the system that was actually implemented and archived in the repository: a sample-level 1B to 7B cascade for the shared-8 EduBench benchmark. The final system, Pangu-ACE, uses a 1B tutor-router to produce a draft answer plus routing signals, then either accepts the draft or escalates the sample to a 7B specialist prompt. We also correct a major offline evaluation bug: earlier summaries over-credited some open-form outputs that only satisfied superficial format checks. After CPU-side rescoring from saved prediction JSONL, the full Chinese test archive (7013 samples) shows that cascade_final improves deterministic quality from 0.457 to 0.538 and format validity from 0.707 to 0.866 over the legacy rule_v2 system while accepting 19.7% of requests directly at 1B. Routing is strongly task dependent: IP is accepted by 1B 78.0% of the time, while QG and EC still escalate almost always. The current archived deployment does not yet show latency gains, so the defensible efficiency story is routing selectivity rather than wall-clock speedup. We also package a reproducible artifact-first paper workflow and clarify the remaining external-baseline gap: GPT-5.4 re-judging is implemented locally, but the configured provider endpoint and key are invalid, so final sampled-baseline alignment with GPT-5.4 remains pending infrastructure repair.

We present Nautilus, a novel tensor compiler that moves toward fully automated math-to-kernel optimization. Nautilus compiles a high-level algebraic specification of tensor operators into efficient tiled GPU kernels. Nautilus's successive lowering design allows high-level optimizations, expression rewrites, and tile optimizations to be jointly applied in a single end-to-end system. Nautilus presents a novel auto-scheduler that discovers sequences of high-level optimizations, while preserving the regular program structure needed by tile optimizers. Nautilus's auto-scheduler captures complex interactions and trade-offs in the high-level optimizations, including aggressive global transformations like advanced reduction fusion. Nautilus is the first end-to-end tensor compiler capable of starting from a math-like description of attention and automatically discovering FlashAttention-3-like kernels, offloading the entire burden of optimization from the programmer to the compiler. Across five transformer-based models and 150 evaluation configurations on NVIDIA GH200 and RTX 5090 GPUs, Nautilus achieves up to 23% higher throughput than state-of-the-art compilers on GH200 and up to 42% on RTX 5090, while matching or exceeding manually written cuDNN kernels on many long-sequence configurations.

Resolving real-world software engineering (SWE) issues with autonomous agents requires complex, long-horizon reasoning. Current pipelines are bottlenecked by unoptimized demonstration data, sparse execution rewards, and computationally prohibitive inference scaling, which collectively exacerbate token bloat, reward hacking, and policy degradation. We present SWE-TRACE (Trajectory Reduction and Agentic Criteria Evaluation), a unified framework optimizing the SWE agent lifecycle across data curation, reinforcement learning (RL), and test-time inference. First, we introduce an LLM multi-task cascading method, utilizing stepwise oracle verification to distill a 60K-instance Supervised Fine-Tuning (SFT) corpus strictly biased toward token-efficient, shortest-path trajectories. Second, to overcome the instability of sparse outcome rewards, we design a MemoryAugmented Agentic RL pipeline featuring a Rubric-Based Process Reward Model (PRM). An auxiliary Rubric-Agent provides dense, fine-grained heuristic feedback on intermediate steps, guiding the model through long-horizon tasks. Finally, we bridge training and inference by repurposing the PRM for heuristic-guided Test-Time Scaling (TTS). By dynamically evaluating and pruning action candidates at each step, SWE-TRACE achieves superior search efficiency without the latency overhead of standard parallel sampling. Extensive experiments on standard SWE benchmarks demonstrate that SWE-TRACE significantly advances the state-of-the-art, maximizing resolution rates while drastically reducing both token consumption and inference latency.

In NLP classification tasks where little labeled data exists, domain fine-tuning of transformer models on unlabeled data is an established approach. In this paper we have two aims. (1) We describe our observations from fine-tuning the Finnish BERT model on Finnish medical text data. (2) We report on our attempts to predict the benefit of domain-specific pre-training of Finnish BERT from observing the geometry of embedding changes due to domain fine-tuning. Our driving motivation is the common\situation in healthcare AI where we might experience long delays in acquiring datasets, especially with respect to labels.

Ad hoc wireless networks exhibit complex, innate and coupled dynamics: node mobility, energy depletion and topology change that are difficult to model analytically. Model-free deep reinforcement learning requires sustained online interaction whereas existing model based approaches use flat state representations that lose per node structure. Therefore we propose G-RSSM, a graph structured recurrent state space model that maintains per node latent states with cross node multi head attention to learn the dynamics jointly from offline trajectories. We apply the proposed method to the downstream task clustering where a cluster head selection policy trains entirely through imagined rollouts in the learned world model. Across 27 evaluation scenarios spanning MANET, VANET, FANET, WSN and tactical networks with N=30 to 1000 nodes, the learned policy maintains high connectivity with only trained for N=50. Herein, we propose the first multi physics graph structured world model applied to combinatorial per node decision making in size agnostic wireless ad hoc networks.

In online clustering problems, there is often a large amount of uncertainty over possible cluster assignments that cannot be resolved until more data are observed. This difficulty is compounded when clusters follow complex distributions, as is the case with text data. Sequential Monte Carlo (SMC) methods give a natural way of representing and updating this uncertainty over time, but have prohibitive memory requirements for large-scale problems. We propose a novel SMC algorithm that decomposes clustering problems into approximately independent subproblems, allowing a more compact representation of the algorithm state. Our approach is motivated by the knowledge base construction problem, and we show that our method is able to accurately and efficiently solve clustering problems in this setting and others where traditional SMC struggles.

We study a classification problem with three key challenges: pervasive informative missingness, the integration of partial prior expert knowledge into the learning process, and the need for interpretable decision rules. We propose a framework that encodes prior knowledge through an expert-guided class-conditional model for one or more classes, and use this model to construct a small set of interpretable goodness-of-fit features. The features quantify how well the observed data agree with the expert model, isolating the contributions of different aspects of the data, including both observed and missing components. These features are combined with a few transparent auxiliary summaries in a simple discriminative classifier, resulting in a decision rule that is easy to inspect and justify. We develop and apply the framework in the context of seismic monitoring used to assess compliance with the Comprehensive Nuclear-Test-Ban Treaty. We show that the method has strong potential as a transparent screening tool, reducing workload for expert analysts. A simulation designed to isolate the contribution of the proposed framework shows that this interpretable expert-guided method can even outperform strong standard machine-learning classifiers, particularly when training samples are small.

Machine unlearning for large language models (LLMs) aims to remove targeted knowledge while preserving general capability. In this paper, we recast LLM unlearning as an asymmetric two-task problem: retention is the primary objective and forgetting is an auxiliary. From this perspective, we propose a retention-prioritized gradient synthesis framework that decouples task-specific gradient extraction from conflict-aware combination. Instantiating the framework, we adapt established PCGrad to resolve gradient conflicts, and introduce SAGO, a novel retention-prioritized gradient synthesis method. Theoretically, both variants ensure non-negative cosine similarity with the retain gradient, while SAGO achieves strictly tighter alignment through constructive sign-constrained synthesis. Empirically, on WMDP Bio/Cyber and RWKU benchmarks, SAGO consistently pushes the Pareto frontier: e.g., on WMDP Bio (SimNPO+GD), recovery of target model MMLU performance progresses from 44.6% (naive) to 94.0% (+PCGrad) and further to 96.0% (+SAGO), while maintaining comparable forgetting strength. Our results show that re-shaping gradient geometry, rather than re-balancing losses, is the key to mitigating unlearning-retention trade-offs.

The rapid integration of large language models (LLMs) into everyday workflows has transformed how individuals perform cognitive tasks such as writing, programming, analysis, and multilingual communication. While prior research has focused on model reliability, hallucination, and user trust calibration, less attention has been given to how LLM usage reshapes users' perceptions of their own capabilities. This paper introduces the LLM fallacy, a cognitive attribution error in which individuals misinterpret LLM-assisted outputs as evidence of their own independent competence, producing a systematic divergence between perceived and actual capability. We argue that the opacity, fluency, and low-friction interaction patterns of LLMs obscure the boundary between human and machine contribution, leading users to infer competence from outputs rather than from the processes that generate them. We situate the LLM fallacy within existing literature on automation bias, cognitive offloading, and human--AI collaboration, while distinguishing it as a form of attributional distortion specific to AI-mediated workflows. We propose a conceptual framework of its underlying mechanisms and a typology of manifestations across computational, linguistic, analytical, and creative domains. Finally, we examine implications for education, hiring, and AI literacy, and outline directions for empirical validation. We also provide a transparent account of human--AI collaborative methodology. This work establishes a foundation for understanding how generative AI systems not only augment cognitive performance but also reshape self-perception and perceived expertise.

Effective abstention (EA), recognizing evidence insufficiency and refraining from answering, is critical for reliable multimodal systems. Yet existing evaluation paradigms for vision-language models (VLMs) and multi-agent systems (MAS) assume answerability, pushing models to always respond. Abstention has been studied in text-only settings but remains underexplored multimodally; current benchmarks either ignore unanswerability or rely on coarse methods that miss realistic failure modes. We introduce MM-AQA, a benchmark that constructs unanswerable instances from answerable ones via transformations along two axes: visual modality dependency and evidence sufficiency. Evaluating three frontier VLMs spanning closed and open-source models and two MAS architectures across 2079 samples, we find: (1) under standard prompting, VLMs rarely abstain; even simple confidence baselines outperform this setup, (2) MAS improves abstention but introduces an accuracy-abstention trade-off, (3) sequential designs match or exceed iterative variants, suggesting the bottleneck is miscalibration rather than reasoning depth, and (4) models abstain when image or text evidence is absent, but attempt reconciliation with degraded or contradictory evidence. Effective multimodal abstention requires abstention-aware training rather than better prompting or more agents.

Proteins carry out biological functions through the coordinated action of groups of residues organized into structural arrangements. These arrangements, which we refer to as protein units, exist at an intermediate scale, being larger than individual residues yet smaller than entire proteins. A deeper understanding of protein function can be achieved by identifying these units and their associations with function. However, existing approaches either focus on residue-level signals, rely on curated annotations, or segment protein structures without incorporating functional information, thereby limiting interpretable analysis of structure-function relationships. We introduce PUFFIN, a data-driven framework for discovering protein units by jointly learning structural partitioning and functional supervision. PUFFIN represents proteins as residue-level structure graphs and applies a graph neural network with a structure-aware pooling mechanism that partitions each protein into multi-residue units, with functional supervision that shapes the partition. We show that the learned units are structurally coherent, exhibit organized associations with molecular function, and show meaningful correspondence with curated InterPro annotations. Together, these results demonstrate that PUFFIN provides an interpretable framework for analyzing structure-function relationships using learned protein units and their statistical function associations. We made our source code available at https://github.com/boun-tabi-lifelu/puffin.

Interactive Evolutionary Computation (IEC) provides a powerful framework for optimizing subjective criteria such as human preferences and aesthetics, yet it suffers from a fundamental limitation: in high-dimensional generative representations, defining crossover in a semantically consistent manner is difficult, often leading to a mutation-dominated search. In this work, we explicitly define crossover in diffusion models. We propose Diffusion crossover, which formulates evolutionary recombination as step-wise interpolation of noise sequences in the reverse process of Denoising Diffusion Probabilistic Models (DDPMs). By applying spherical linear interpolation (Slerp) to the noise sequences associated with selected parent images, the proposed method generates offspring that inherit characteristics from both parents while preserving the geometric structure of the diffusion process. Furthermore, controlling the time-step range of interpolation enables a principled trade-off between diversity (exploration) and convergence (exploitation). Experimental results using PCA analysis and perceptual similarity metrics (LPIPS) demonstrate that Diffusion crossover produces perceptually smooth and semantically consistent transitions between parent images. Qualitative interactive evolution experiments further confirm that the proposed method effectively supports human-in-the-loop image exploration. These findings suggest a new perspective: diffusion models are not only powerful generators, but also structured evolutionary search spaces in which recombination can be explicitly defined and controlled.

Deploying deep neural networks on edge devices requires balancing accuracy, latency, and resource constraints under realistic execution conditions. To fit models within these constraints, two broad strategies have emerged: static compression techniques such as pruning and quantization, which permanently reduce model size, and dynamic approaches such as early-exit mechanisms, which adapt computational cost at runtime. While both families are widely studied in isolation, they are rarely compared under identical conditions on physical hardware. This paper presents a unified deployment-oriented comparison of static compression and dynamic early-exit mechanisms, evaluated on real edge devices using ONNX based inference pipelines. Our results show that static and dynamic techniques offer fundamentally different trade-offs for edge deployment. While pruning and quantization deliver consistent memory footprint reduction, early-exit mechanisms enable input-adaptive computation savings that static methods cannot match. Their combination proves highly effective, simultaneously reducing inference latency and memory usage with minimal accuracy loss, expanding what is achievable at the edge.

Developing an MR sequence is challenging and remains largely constrained by human intuition. Recently, AI-driven approaches have been proposed; however, most require an initial sequence for parameter optimization or extensive training datasets, limiting their general applicability. In this study, we propose "Sequence Search," an automated sequence design framework based on neural architecture search. The method takes tissue properties, imaging parameters, and design objectives as inputs and generates pulse sequences satisfying the design objectives, without requiring prior knowledge of conventional sequence structures. Sequence Search iteratively generates candidate sequences through neural architecture search and optimizes them via a differentiable Bloch simulator and objective-specific loss functions using gradient-based learning. The framework successfully replicated conventional spin-echo, T2-weighted spin-echo, and inversion recovery sequences. Less intuitive solutions were also discovered, such as three-RF spin-echo-like sequences with reduced RF energy and refocusing phases deviating from the conventional Hahn-echo. This work establishes a generalizable framework for automated MR sequence design, highlighting the potential to explore configurations beyond conventional designs based on human intuition.

Network Digital Twins (NDTs) enable safe what-if analysis for 6G cloud-edge infrastructures, but adoption is often limited by fragmented workflows from telemetry to validation. We present a data-driven NDT framework that extends 6G-TWIN with a scalable pipeline for cloud-edge telemetry aggregation and semantic alignment into unified data models. Our contributions include: (i) scalable cloud-edge telemetry collection, (ii) regime-aware feature engineering capturing the network's scaling behavior, and (iii) a validation methodology based on Sign Agreement and Directional Sensitivity. Evaluated on a Kubernetes-managed cluster, the framework extrapolates performance to unseen high-load regimes. Results show both Deep Neural Network (DNN) and XGBoost achieve high regression accuracy (R2 > 0.99), while the XGBoost model delivers superior directional reliability (Sa > 0.90), making the NDT a trustworthy tool for proactive resource scaling in out-of-distribution scenarios.

Generative Agents, owing to their precise modeling and simulation capabilities of human behavior, have become a pivotal tool in the field of Artificial Intelligence in Education (AIEd) for uncovering complex cognitive processes of learners. However, existing educational agents predominantly rely on static personas to simulate student learning behaviors, neglecting the decisive role of deep cognitive capabilities in learning outcomes during practice interactions. Furthermore, they struggle to characterize the dynamic fluidity of knowledge internalization, transfer, and cognitive state transitions. To overcome this bottleneck, this paper proposes a human-like educational agent capable of simulating student cognitive evolution: CogEvolution. Specifically, we first construct a cognitive depth perceptron based on the Interactive, Constructive, Active, Passive (ICAP) taxonomy from cognitive psychology, achieving precise quantification of learner cognitive engagement. Subsequently, we propose a memory retrieval method based on Item Response Theory (IRT) to simulate the connection and assimilation of new and prior knowledge. Finally, we design a dynamic cognitive update mechanism based on evolutionary algorithms to simulate the real-time integration of student learning behaviors and cognitive evolution processes. Comprehensive evaluations demonstrate that CogEvolution not only significantly outperforms baseline models in behavioral fidelity and learning curve fitting but also uniquely reproduces plausible and robust cognitive evolutionary paths consistent with educational psychology expectations, providing a novel paradigm for constructing highly interpretable educational agents.

Recent progress in Multimodal Large Language Models (MLLMs) has demonstrated remarkable advances in perception and reasoning, suggesting their potential for embodied intelligence. While recent studies have evaluated embodied MLLMs in interactive settings, current benchmarks mainly target capabilities to perceive, understand, and interact with external objects, lacking a systematic evaluation of self-centric intelligence. To address this, we introduce MirrorBench, a simulation-based benchmark inspired by the classical Mirror Self-Recognition (MSR) test in psychology. MirrorBench extends this paradigm to embodied MLLMs through a tiered framework of progressively challenging tasks, assessing agents from basic visual perception to high-level self-representation. Experiments on leading MLLMs show that even at the lowest level, their performance remains substantially inferior to human performance, revealing fundamental limitations in self-referential understanding. Our study bridges psychological paradigms and embodied intelligence, offering a principled framework for evaluating the emergence of general intelligence in large models. Project page: https://fflahm.github.io/mirror-bench-page/.

In continual visual question answering (VQA), existing Continual Learning (CL) methods are mostly built for symmetric, unimodal architectures. However, modern Vision-Language Models (VLMs) violate this assumption, as their trainable components are inherently asymmetric. This structural mismatch renders VLMs highly prone to catastrophic forgetting when learning from continuous data streams. Specifically, the asymmetry causes standard global regularization to favor the massive language decoder during optimization, leaving the smaller but critical visual projection layers highly vulnerable to interference. Consequently, this localized degradation leads to a severe loss of compositional reasoning capabilities. To address this, we propose Asymmetric Information Masking (AIM), which balances stability and plasticity by applying targeted masks based on modality-specific sensitivity. Experiments on VQA v2 and GQA under continual VQA settings show that AIM achieves state-of-the-art performance in both Average Performance (AP) and Average Forgetting (AF), while better preserving generalization to novel skill-concept compositions.

While LLMs have demonstrated remarkable potential in Question Answering (QA), evaluating personalization remains a critical bottleneck. Existing paradigms predominantly rely on lexical-level similarity or manual heuristics, often lacking sufficient data-driven validation. We address this by mining Community-Individual Preference Divergence (CIPD), where individual choices override consensus, to distill six key personalization factors as evaluative dimensions. Accordingly, we introduce CoPA, a benchmark with 1,985 user profiles for fine-grained, factor-level assessment. By quantifying the alignment between model outputs and user-specific cognitive preferences inferred from interaction patterns, CoPA provides a more comprehensive and discriminative standard for evaluating personalized QA than generic metrics. The code is available at https://github.com/bjzgcai/CoPA.

The pre-training and fine-tuning paradigm has become the dominant approach for model adaptation. However, conventional pre-training typically yields models at a fixed scale, whereas practical deployment often requires models of varying sizes, exposing its limitations when target model scales differ from those used during pre-training. To address this, we propose an innovative constraint-based pre-training paradigm that imposes structured constraints during pre-training to disentangle size-agnostic knowledge into reusable weight templates, while assigning size-specific adaptation to lightweight weight scalers, thereby reformulating variable-sized model initialization as a multi-task adaptation problem. Within this paradigm, we further introduce WeiT, which employs Kronecker-based constraints to regularize the pre-training process. Specifically, model parameters are represented as compositions of weight templates via concatenation and weighted aggregation, with adaptive connections governed by lightweight weight scalers whose parameters are learned from limited data. This design enables flexible and efficient construction of model weights across diverse downstream scales. Extensive experiments demonstrate the efficiency and effectiveness of WeiT, achieving state-of-the-art performance in initializing models with varying depths and widths across a broad range of perception and embodied learning tasks, including Image Classification, Image Generation, and Embodied Control. Moreover, its effectiveness generalizes to both Transformer-based and Convolution-based architectures, consistently enabling faster convergence and improved performance even under full training.

Large Language Models (LLMs) exhibit strong mathematical reasoning when trained on high-quality Chain-of-Thought (CoT) that articulates intermediate steps, yet costly CoT curation hinders further progress. While existing remedies such as distillation from stronger LLMs and self-synthesis based on test-time search alleviate this issue, they often suffer from diminishing returns or high computing overhead.In this work, we propose CoTEvol, a genetic evolutionary framework that casts CoT generation as a population-based search over reasoning trajectories.Candidate trajectories are iteratively evolved through reflective global crossover at the trajectory level and local mutation guided by uncertainty at the step level, enabling holistic recombination and fine-grained refinement. Lightweight, task-aware fitness functions are designed to guide the evolutionary process toward accurate and diverse reasoning. Empirically, CoTEvol improves correct-CoT synthesis success by over 30% and enhances structural diversity, with markedly improved efficiency. LLMs trained on these evolutionary CoT data achieve an average gain of 6.6% across eight math benchmarks, outperforming previous distillation and self-synthesis approaches. These results underscore the promise of evolutionary CoT synthesis as a scalable and effective method for mathematical reasoning tasks.

Preventing machine failure is inherently superior to reactive remediation, particularly for critical assets like gas turbines, where early fault detection (FD) is a cornerstone of industrial sustainability. However, modern deep learning-based FD models often face a significant trade-off between architectural complexity and real-time operational constraints, often hindered by a lack of temporal context within restricted vibration signal windows. To address these challenges, this study proposes a Temporal Cross-Modal Knowledge-Distillation Transfer-Learning (TCMKDTL) framework. The framework employs a "privileged" teacher model trained on expansive temporal windows incorporating both past and future signal context to distill latent feature-based knowledge into a compact student model. To mitigate issues of data scarcity and domain shift, the framework leverages robust pre-training on benchmark datasets (such as CWRU) followed by adaptation to target industrial data. Extensive evaluation using experimental and industrial gas turbine (MGT-40) datasets demonstrates that TCMKDTL achieves superior feature separability and diagnostic accuracy compared to conventional pre-trained architectures. Ultimately, this approach enables high-performance, unsupervised anomaly detection suitable for deployment on resource-constrained industrial hardware.

We present a geometric framework for Reinforcement Learning (RL) that views policies as maps into the Wasserstein space of action probabilities. First, we define a Riemannian structure induced by stationary distributions, proving its existence in a general context. We then define the tangent space of policies and characterize the geodesics, specifically addressing the measurability of vector fields mapped from the state space to the tangent space of probability measures over the action space. Next, we formulate a general RL optimization problem and construct a gradient flow using Otto's calculus. We compute the gradient and the Hessian of the energy, providing a formal second-order analysis. Finally, we illustrate the method with numerical examples for low-dimensional problems, computing the gradient directly from our theoretical formalism. For high-dimensional problems, we parameterize the policy using a neural network and optimize it based on an ergodic approximation of the cost.

The communication bottleneck in federated learning (FL) has spurred extensive research into techniques to reduce the volume of data exchanged between client devices and the central parameter server. In this paper, we systematically classify gradient and model compression schemes into three categories based on the type of correlations they exploit: structural, temporal, and spatial. We examine the sources of such correlations, propose quantitative metrics for measuring their magnitude, and reinterpret existing compression methods through this unified correlation-based framework. Our experimental studies demonstrate that the degrees of structural, temporal, and spatial correlations vary significantly depending on task complexity, model architecture, and algorithmic configurations. These findings suggest that algorithm designers should carefully evaluate correlation assumptions under specific deployment scenarios rather than assuming that they are always present. Motivated by these findings, we propose two adaptive compression designs that actively switch between different compression modes based on the measured correlation strength, and we evaluate their performance gains relative to conventional non-adaptive approaches. In summary, our unified taxonomy provides a clean and principled foundation for developing more effective and application-specific compression techniques for FL systems.

Large language models still struggle with faithfulness and hallucinations despite their remarkable reasoning abilities. In Knowledge Graph Question Answering (KGQA), semantic parsing-based approaches address the limitations by understanding constraints in a user's question and converting them into a logical form to execute on a knowledge graph. However, existing KGQA benchmarks and methods are biased toward positive and calculation constraints. Negative constraints are neglected, although they frequently appear in real-world questions. In this paper, we introduce a new task, NEgative-conSTrained (NEST) KGQA, where each question contains at least one negative constraint, and a corresponding dataset, NestKGQA. We also design PyLF, a Python-formatted logical form, since existing logical forms are hardly suitable to express negation clearly while maintaining readability. Furthermore, NEST questions naturally contain multiple constraints. To mitigate their semantic complexity, we present a novel framework named CUCKOO, specialized to multiple-constrained questions and ensuring semantic executability. CUCKOO first generates a constraint-aware logical form draft and performs schema-guided semantic matching. It then selectively applies self-directed refinement only when executing improper logical forms yields an empty result, reducing cost while improving robustness. Experimental results demonstrate that CUCKOO consistently outperforms baselines on both conventional KGQA and NEST-KGQA benchmarks under few-shot settings.

Conventional Graph Contrastive Learning (GCL) on Text-Attributed Graphs (TAGs) relies on blind stochastic augmentations, inadvertently entangling task-relevant signals with noise. We propose SDM-SCR, a robust framework anchored in Approximate Orthogonal Decomposition. First, the Semantic Decoupling Module (SDM) leverages the instruction-following capability of Large Language Models (LLMs) to actively parse raw attributes into asymmetric, task-oriented signal and noise views. This shifts the paradigm from random perturbation to semantic-aware disentanglement. Subsequently, Semantic Consistency Regularization (SCR) exploits the spectral observation that semantic signals are topologically smooth while residual noise is high-frequency. SCR functions as a selective spectral filter, enforcing consistency only on the signal subspace to eliminate LLM hallucinations without over-smoothing. This ``Disentangle-then-Refine'' mechanism ensures rigorous signal purification. Extensive experiments demonstrate that SDM-SCR achieves SOTA performance in accuracy and efficiency.

In many real-world settings, problem instances that need to be solved are quite similar, and knowledge from previous optimization runs can potentially be utilized. We explore this for the Traveling Salesperson problem with time windows (TSPTW), which often arises in settings where the travel-time matrix is fixed but time-window constraints change across related tasks. Existing TSPTW studies, however, have not systematically compared solving such task sequences independently with sequential transfer from previously solved tasks. We address this gap using a multi-task benchmark in which each base instance is expanded into five related tasks under two environments: partial time-window expansion and swap-additive time reassignment. We compare a standard from-scratch protocol with an iterative protocol that initializes each task from the best tour of the previous task, using the popular local search approaches LNS, VNS, and LKH-3 under a common penalized-score objective. Our experimental results show that the iterative protocol is consistently superior in the progressive-relaxation setting and generally competitive under swap-additive changes, with improvements increasing on more difficult instances.

Large-scale renewable energy deployment introduces pronounced volatility into the electricity system, turning grid operation into a complex stochastic optimization problem. Accurate electricity price forecasting (EPF) is essential not only to support operational decisions, such as optimal bidding strategies and balancing power preparation, but also to reduce economic risk and improve market efficiency. Probabilistic forecasts are particularly valuable because they quantify uncertainty stemming from renewable intermittency, market coupling, and regulatory changes, enabling market participants to make informed decisions that minimize losses and optimize expected revenues. However, it remains an open question which models to employ to produce accurate forecasts. Should these be task-specific machine learning (ML) models or Time Series Foundation Models (TSFMs)? In this work, we compare four models for day-ahead probabilistic EPF (PEPF) in European bidding zones: a deterministic NHITS backbone with Quantile-Regression Averaging (NHITS+QRA) and a conditional Normalizing-Flow forecaster (NF) are compared with two TSFMs, namely Moirai and ChronosX. On the one hand, we find that TSFMs outperform task-specific deep learning models trained from scratch in terms of CRPS, Energy Score, and predictive interval calibration across market conditions. On the other hand, we find that well-configured task-specific models, particularly NHITS combined with QRA, achieve performance very close to TSFMs, and in some scenarios, such as when supplied with additional informative feature groups or adapted via few-shot learning from other European markets, they can even surpass TSFMs. Overall, our findings show that while TSFMs offer expressive modeling capabilities, conventional models remain highly competitive, emphasizing the need to weigh computational expense against marginal performance improvements in PEPF.

Consumer wearables enable continuous measurement of physiological data related to stress and recovery, but turning these streams into actionable, personalized stress-management recommendations remains a challenge. In practice, users often do not know how a given intervention, defined as an activity intended to reduce stress, will affect heart rate (HR), heart rate variability (HRV), or inter-beat intervals (BBI) over the next 15 to 120 minutes. We present a framework that predicts post-intervention trajectories and the direction of change for these physiological indicators across time windows. Our methodology combines a Transformer model for multi-horizon trajectories of percent change relative to a pre-intervention baseline, direction-of-change calls (positive, negative, or neutral) at each horizon, and an empirical study using wearable sensor data overlaid with user-tagged events and interventions. This proof of concept shows that personalized post-intervention prediction is feasible. We encourage future integration into stress-management tools for personalized intervention recommendations tailored to each person's day following further validation in larger studies and, where applicable, appropriate regulatory review.

Vision-Language-Action (VLA) models have emerged as a promising paradigm for building embodied agents that ground perception and language into action. However, most existing approaches rely on direct action prediction, lacking the ability to reason over long-horizon trajectories and evaluate their consequences, which limits performance in complex decision-making tasks. In this work, we introduce World-Value-Action (WAV) model, a unified framework that enables implicit planning in VLA systems. Rather than performing explicit trajectory optimization, WAV model learn a structured latent representation of future trajectories conditioned on visual observations and language instructions. A learned world model predicts future states, while a trajectory value function evaluates their long-horizon utility. Action generation is then formulated as inference in this latent space, where the model progressively concentrates probability mass on high-value and dynamically feasible trajectories. We provide a theoretical perspective showing that planning directly in action space suffers from an exponential decay in the probability of feasible trajectories as the horizon increases. In contrast, latent-space inference reshapes the search distribution toward feasible regions, enabling efficient long-horizon decision making. Extensive simulations and real-world experiments demonstrate that the WAV model consistently outperforms state-of-the-art methods, achieving significant improvements in task success rate, generalization ability, and robustness, especially in long-horizon and compositional scenarios.

To quantify the geometric expressivity of transformers, we introduce a tropical geometry framework to characterize their exact spatial partitioning capabilities. By modeling self-attention as a vector-valued tropical rational map, we prove it evaluates exactly to a Power Voronoi Diagram in the zero-temperature limit. Building on this equivalence, we establish a combinatorial rationale for Multi-Head Self-Attention (MHSA): via the Minkowski sum of Newton polytopes, multi-head aggregation expands the polyhedral complexity to $\mathcal{O}(N^H)$, overcoming the $\mathcal{O}(N)$ bottleneck of single heads. Extending this to deep architectures, we derive the first tight asymptotic bounds on the number of linear regions in transformers ($Θ(N^{d_{\text{model}}L})$), demonstrating a combinatorial explosion driven intrinsically by sequence length $N$, ambient embedding dimension $d_{\text{model}}$, and network depth $L$. Importantly, we guarantee that this idealized polyhedral skeleton is geometrically stable: finite-temperature soft attention preserves these topological partitions via exponentially tight differential approximation bounds.

Online anomaly detection (OAD) plays a pivotal role in real-time analytics and decision-making for evolving data streams. However, existing methods often rely on costly retraining and rigid decision boundaries, limiting their ability to adapt both effectively and efficiently to concept drift in dynamic environments. To address these challenges, we propose DyMETER, a dynamic concept adaptation framework for OAD that unifies on-the-fly parameter shifting and dynamic thresholding within a single online paradigm. DyMETER first learns a static detector on historical data to capture recurring central concepts, and then transitions to a dynamic mode to adapt to new concepts as drift occurs. Specifically, DyMETER employs a novel dynamic concept adaptation mechanism that leverages a hypernetwork to generate instance-aware parameter shifts for the static detector, thereby enabling efficient and effective adaptation without retraining or fine-tuning. To achieve robust and interpretable adaptation, DyMETER introduces a lightweight evolution controller to estimate instance-level concept uncertainty for adaptive updates. Further, DyMETER employs a dynamic threshold optimization module to adaptively recalibrates the decision boundary by maintaining a candidate window of uncertain samples, which ensures continuous alignment with evolving concepts. Extensive experiments demonstrate that DyMETER significantly outperforms existing OAD approaches across a wide spectrum of application scenarios.

Recent advances in query optimization have shifted from traditional rule-based and cost-based techniques towards machine learning-driven approaches. Among these, reinforcement learning (RL) has attracted significant attention due to its ability to optimize long-term performance by learning policies over query planning. However, existing RL-based query optimizers often exhibit unstable performance at the level of individual queries, including severe performance regressions, and require prolonged training to reach the plan quality of expert, cost-based optimizers. These shortcomings make learned query optimizers difficult to deploy in practice and remain a major barrier to their adoption in production database systems. To address these challenges, we present RELOAD, a robust and efficient learned query optimizer for database systems. RELOAD focuses on (i) robustness, by minimizing query-level performance regressions and ensuring consistent optimization behavior across executions, and (ii) efficiency, by accelerating convergence to expert-level plan quality. Through extensive experiments on standard benchmarks, including Join Order Benchmark, TPC-DS, and Star Schema Benchmark, RELOAD demonstrates up to 2.4x higher robustness and 3.1x greater efficiency compared to state-of-the-art RL-based query optimization techniques.

Vision State Space Models (SSMs) like Vim, VMamba, and SiMBA rely on complex scanning strategies to adapt sequential SSMs to process 2D images, introducing computational overhead and architectural complexity. We propose HAMSA, a scanning-free SSM operating directly in the spectral domain. HAMSA introduces three key innovations: (1) simplified kernel parameterization-a single Gaussian-initialized complex kernel replacing traditional (A, B, C) matrices, eliminating discretization instabilities; (2) SpectralPulseNet (SPN)-an input-dependent frequency gating mechanism enabling adaptive spectral modulation; and (3) Spectral Adaptive Gating Unit (SAGU)-magnitude-based gating for stable gradient flow in the frequency domain. By leveraging FFT-based convolution, HAMSA eliminates sequential scanning while achieving O(L log L) complexity with superior simplicity and efficiency. On ImageNet-1K, HAMSA reaches 85.7% top-1 accuracy (state-of-the-art among SSMs), with 2.2 X faster inference than transformers (4.2ms vs 9.2ms for DeiT-S) and 1.4-1.9X speedup over scanning-based SSMs, while using less memory (2.1GB vs 3.2-4.5GB) and energy (12.5J vs 18-25J). HAMSA demonstrates strong generalization across transfer learning and dense prediction tasks.

Large language models are increasingly used as natural-language interfaces to enterprise software, but their direct use as system operators remains unsafe. Model errors can propagate into unauthorized actions, malformed requests, cross-workspace execution, and other costly failures. We argue this is primarily an execution architecture problem. We present a bounded-autonomy architecture in which language models may interpret intent and propose actions, but all executable behavior is constrained by typed action contracts, permission-aware capability exposure, scoped context, validation before side effects, consumer-side execution boundaries, and optional human approval. The enterprise application remains the source of truth for business logic and authorization, while the orchestration engine operates over an explicit published actions manifest. We evaluate the architecture in a deployed multi-tenant enterprise application across three conditions: manual operation, unconstrained AI with safety layers disabled, and full bounded autonomy. Across 25 scenario trials spanning seven failure families, the bounded-autonomy system completed 23 of 25 tasks with zero unsafe executions, while the unconstrained configuration completed only 17 of 25. Two wrong-entity mutations escaped all consumer-contributed layers; only disambiguation and confirmation mechanisms intercept this class. Both AI conditions delivered 13-18x speedup over manual operation. Critically, removing safety layers made the system less useful: structured validation feedback guided the model to correct outcomes in fewer turns, while the unconstrained system hallucinated success. Several safety properties are structurally enforced by code and intercepted all targeted violations regardless of model output. The result is a practical, deployed architecture for making imperfect language models operationally useful in enterprise systems.

Transformers commonly exhibit an attention sink: disproportionately high attention to the first position. We study this behavior in GPT-2-style models with learned query biases and absolute positional embeddings. Combining structural analysis with causal interventions, validated across natural-language, mathematical, and code inputs, we find that the sink arises from the interaction among (i) a learned query bias, (ii) the first-layer MLP transformation of the positional encoding, and (iii) structure in the key projection. Crucially, each component we identify is individually dispensable: architectures omitting each of them robustly exhibit sinks. This indicates that attention sinks may arise through distinct circuits across architectures. These findings inform mitigation of sinks, and motivate broader investigation into why sinks emerge.

This article argues that the most important significance of the AI revolution, especially the rise of large language models, lies not simply in automation, but in a fundamental change in how complex information and human know-how are carried, replicated, and shared. From this perspective, AI for Science is especially important because it may transform not only the efficiency of research, but also the structure of scientific collaboration, discovery, publishing, and evaluation. The article outlines a gradual path from AI as a research tool to AI as a scientific collaborator, and discusses how AI is likely to fundamentally reshape scientific publication. It also argues that continuous learning and diversity of ideas are essential if AI is to play a meaningful role in original scientific discovery.

Persistent language-model agents increasingly combine tool use, tiered memory, reflective prompting, and runtime adaptation. In such systems, behavior is shaped not only by current prompts but by mutable internal conditions that influence future action. This paper introduces layered mutability, a framework for reasoning about that process across five layers: pretraining, post-training alignment, self-narrative, memory, and weight-level adaptation. The central claim is that governance difficulty rises when mutation is rapid, downstream coupling is strong, reversibility is weak, and observability is low, creating a systematic mismatch between the layers that most affect behavior and the layers humans can most easily inspect. I formalize this intuition with simple drift, governance-load, and hysteresis quantities, connect the framework to recent work on temporal identity in language-model agents, and report a preliminary ratchet experiment in which reverting an agent's visible self-description after memory accumulation fails to restore baseline behavior. In that experiment, the estimated identity hysteresis ratio is 0.68. The main implication is that the salient failure mode for persistent self-modifying agents is not abrupt misalignment but compositional drift: locally reasonable updates that accumulate into a behavioral trajectory that was never explicitly authorized.

LLM-powered systems require complex multi-step decision-making abilities to solve real-world tasks, yet current planning approaches face a trade-off between the high latency of inference-time search and the limited generalization of supervised fine-tuning. To address this limitation, we introduce \textbf{SGA-MCTS}, a framework that casts LLM planning as non-parametric retrieval. Offline, we leverage Monte Carlo Tree Search (MCTS) to explore the solution space and distill high-fidelity trajectories into State-Goal-Action (SGA) atoms. These atoms are de-lexicalized primitives that abstract concrete entities into symbolic slots, preserving reusable causal logic while discarding domain-specific noise. Online, a retrieval-augmented agent employs a hybrid symbolic-semantic mechanism to fetch relevant SGAs and re-ground them into the current context as soft reasoning hints. Empirical results on complex benchmarks demonstrate that this paradigm enables frozen, open-weights models to match the performance of SOTA systems (e.g., GPT-5) without task-specific fine-tuning. By effectively amortizing the heavy computational cost of search, SGA-MCTS achieves System 2 reasoning depth at System 1 inference speeds, rendering autonomous planning both scalable and real-time feasible.

Existing benchmarks for hardware design primarily evaluate Large Language Models (LLMs) on isolated, component-level tasks such as generating HDL modules from specifications, leaving repository-scale evaluation unaddressed. We introduce HWE-Bench, the first large-scale, repository-level benchmark for evaluating LLM agents on real-world hardware bug repair tasks. HWE-Bench comprises 417 task instances derived from real historical bug-fix pull requests across six major open-source projects spanning both Verilog/SystemVerilog and Chisel, covering RISC-V cores, SoCs, and security roots-of-trust. Each task is grounded in a fully containerized environment where the agent must resolve a real bug report, with correctness validated through the project's native simulation and regression flows. The benchmark is built through a largely automated pipeline that enables efficient expansion to new repositories. We evaluate seven LLMs with four agent frameworks and find that the best agent resolves 70.7% of tasks overall, with performance exceeding 90% on smaller cores but dropping below 65% on complex SoC-level projects. We observe larger performance gaps across models than commonly reported on software benchmarks, and difficulty is driven by project scope and bug-type distribution rather than code size alone. Our failure analysis traces agent failures to three stages of the debugging process: fault localization, hardware-semantic reasoning, and cross-artifact coordination across RTL, configuration, and verification components, providing concrete directions for developing more capable hardware-aware agents.

Human activity traces (HATs) are critical for many applications, including human mobility modeling and point-of-interest (POI) recommendation. However, growing privacy concerns have severely limited access to authentic large-scale HAT datasets. Recent advances in generative AI provide new opportunities to synthesize realistic and privacy-preserving HATs for such applications. Yet two major challenges remain: (i) HATs are highly irregular and dynamic, with long and varying time intervals, making it difficult to capture their complex spatio-temporal dependencies and underlying distributions; and (ii) generative models are often computationally expensive, making long-term, fine-grained HAT synthesis inefficient. To address these challenges, we propose SynHAT, a computationally efficient coarse-to-fine HAT synthesis framework built on a novel spatio-temporal denoising diffusion model. In Stage 1, we develop Coarse-HADiff, which models the overall spatio-temporal dependencies of coarse-grained latent spatio-temporal traces. It incorporates a novel Latent Spatio-Temporal U-Net with dual Drift-Jitter branches to jointly model smooth spatial transitions and temporal variations during denoising. In Stage 2, we introduce a three-step pipeline consisting of Behavior Pattern Extraction, Fine-HADiff, which shares the same architecture as Coarse-HADiff, and Semantic Alignment to generate fine-grained latent spatio-temporal traces from the Stage 1 outputs. We extensively evaluate SynHAT in terms of data fidelity, utility, privacy, robustness, and scalability. Experiments on real-world HAT datasets from four cities across three countries show that SynHAT substantially outperforms state-of-the-art baselines, achieving 52% and 33% improvements on spatial and temporal metrics, respectively.

Multiplicative gating is widely used in neural architectures and has recently been applied to attention layers to improve performance and training stability in large language models. Despite the success of gated attention, the mathematical implications of gated attention mechanisms remain poorly understood. We study attention through the geometry of its representations by modeling outputs as mean parameters of Gaussian distributions and analyzing the induced Fisher--Rao geometry. We show that ungated attention operator is restricted to intrinsically flat statistical manifolds due to its affine structure, while multiplicative gating enables non-flat geometries, including positively curved manifolds that are unattainable in the ungated setting. These results establish a geometric expressivity gap between ungated and gated attention. Empirically, we show that gated models exhibit higher representation curvature and improved performance on tasks requiring nonlinear decision boundaries whereas they provide no consistent advantage on tasks with linear decision boundaries. Furthermore, we identify a structured regime in which curvature accumulates under composition, yielding a systematic depth amplification effect.

Topology optimization is a computational method used to determine the optimal material distribution within a prescribed design domain, aiming to minimize structural weight while satisfying load and boundary conditions. For critical infrastructure applications, such as structural health monitoring of bridges and buildings, particularly in digital twin contexts, low-latency energy-efficient topology optimization is essential. Traditionally, topology optimization relies on finite element analysis (FEA), a computationally intensive process. Recent advances in deep neural networks (DNNs) have introduced data driven alternatives to FEA, substantially reducing computation time while maintaining solution quality. These DNNs have complex architectures and implementing them on inference-class GPUs results in high latency and poor energy efficiency. To address this challenge, we present a hardware accelerated implementation of a topology optimization neural network (CRONet) on the AMD Versal AI Engine-ML (AIE-ML) architecture. Our approach efficiently exploits the parallelism and memory hierarchy of AIE-ML engines to optimize the execution of various neural network operators. We are the first to implement an end-to-end neural network fully realized on the AIE-ML array, where all intermediate activations and network weights reside on-chip throughout inference, eliminating any reliance on DRAM for intermediate data movement. Experimental results demonstrate that our implementation achieves up to 2.49x improvement in latency and up to 4.18x improvement in energy efficiency compared to an inference-class ML-optimized GPU in the same power budget (Nvidia T4) after scaling for technology node. These results highlight the potential of Versal AIE-ML based acceleration for enabling low-latency energy-efficient topology optimization.

Diffusion models have recently emerged as expressive policy representations for online reinforcement learning (RL). However, their iterative generative processes introduce substantial training and inference overhead. To overcome this limitation, we propose to represent policies using MeanFlow models, a class of few-step flow-based generative models, to improve training and inference efficiency over diffusion-based RL approaches. To promote exploration, we optimize MeanFlow policies under the maximum entropy RL framework via soft policy iteration, and address two key challenges specific to MeanFlow policies: action likelihood evaluation and soft policy improvement. Experiments on MuJoCo and DeepMind Control Suite benchmarks demonstrate that our method, Mean Flow Policy Optimization (MFPO), achieves performance comparable to or exceeding current diffusion-based baselines while considerably reducing training and inference time. Our code is available at https://github.com/MFPolicy/MFPO.

LLM-empowered agent simulations are increasingly used to study social emergence, yet the micro-to-macro causal mechanisms behind macro outcomes often remain unclear. This is challenging because emergence arises from intertwined agent interactions and meso-level feedback and nonlinearity, making generative mechanisms hard to disentangle. To this end, we introduce \textbf{\textsc{CAMO}}, an automated \textbf{Ca}usal discovery framework from \textbf{M}icr\textbf{o} behaviors to \textbf{M}acr\textbf{o} Emergence in LLM agent simulations. \textsc{CAMO} converts mechanistic hypotheses into computable factors grounded in simulation records and learns a compact causal representation centered on an emergent target $Y$. \textsc{CAMO} outputs a computable Markov boundary and a minimal upstream explanatory subgraph, yielding interpretable causal chains and actionable intervention levers. It also uses simulator-internal counterfactual probing to orient ambiguous edges and revise hypotheses when evidence contradicts the current view. Experiments across four emergent settings demonstrate the promise of \textsc{CAMO}.

Monte-Carlo Tree Search (MCTS) is a fundamental sampling-based search algorithm widely used for online planning in sequential decision-making domains. Despite its success in driving recent advances in artificial intelligence, understanding the behavior of MCTS agents remains a challenge for both developers and users. This difficulty stems from the complex search trees produced through the simulation of numerous future states and their intricate relationships. A known weakness of standard MCTS is its reliance on highly selective tree construction, which may lead to the omission of crucial moves and a vulnerability to tactical traps. To resolve this, we incorporate shallow, full-width Minimax search into the rollout phase of multi-agent MCTS to enhance strategic depth. Furthermore, to demystify the resulting decision-making logic, we introduce \textsf{M2-PALE} (MCTS--Minimax Process-Aided Linguistic Explanations). This framework employs process mining techniques, specifically the Alpha Miner, iDHM, and Inductive Miner algorithms, to extract underlying behavioral workflows from agent execution traces. These process models are then synthesized by LLMs to generate human-readable causal and distal explanations. We demonstrate the efficacy of our approach in a small-scale checkers environment, establishing a scalable foundation for interpreting hybrid agents in increasingly complex strategic domains.

Deep Research Agents (DRAs) aim to solve complex, long-horizon research tasks involving planning, retrieval, multimodal understanding, and report generation, yet their evaluation remains challenging due to dynamic web environments and ambiguous task definitions. We propose DR$^{3}$-Eval, a realistic and reproducible benchmark for evaluating deep research agents on multimodal, multi-file report generation. DR$^{3}$-Eval is constructed from authentic user-provided materials and paired with a per-task static research sandbox corpus that simulates open-web complexity while remaining fully verifiable, containing supportive documents, distractors, and noise. Moreover, we introduce a multi-dimensional evaluation framework measuring Information Recall, Factual Accuracy, Citation Coverage, Instruction Following, and Depth Quality, and validate its alignment with human judgments. Experiments with our developed multi-agent system DR$^{3}$-Agent based on multiple state-of-the-art language models demonstrate that DR$^{3}$-Eval is highly challenging and reveals critical failure modes in retrieval robustness and hallucination control. Our code and data are publicly available.

Speculative decoding accelerates large language model (LLM) inference. It uses a small draft model to propose a tree of future tokens. A larger target model then verifies these tokens in a single batched forward pass. Despite the growing body of work on speculative methods, the degree to which the cognitive characteristics of a task affect acceptance probability remains largely unexplored. We present an empirical study of tree-based speculative decoding acceptance dynamics. Our study spans four well-established NLP benchmark domains: code generation, mathematical reasoning, logical reasoning, and open-ended chat. For this, we use TinyLlama-1.1B as the draft model against Llama-2-7B-Chat-GPTQ as the target. Over 99,768 speculative nodes collected from 200 prompts, we derive per-domain acceptance rates, expected accepted lengths, depth-acceptance profiles, and entropy-acceptance correlations. We find that task type is a stronger predictor of acceptance than tree depth. Furthermore, only the chat domain consistently yields an expected accepted length exceeding 1.0 token per step. We also show that the entropy-acceptance correlation is consistently negative but weak across all domains (rho in [-0.20, -0.15]). Counterintuitively, chat produces the highest entropy yet the highest acceptance rate. We attribute this divergence to the lexical predictability of RLHF-aligned register. These findings have direct implications for domain-aware speculation budgets and draft-model selection strategies. Index Terms--speculative decoding, large language model inference, tree attention, draft model, acceptance probability, LLM efficiency

Large language models (LLMs) are being increasingly used in urban planning, but since gendered space theory highlights how gender hierarchies are embedded in spatial organization, there is concern that LLMs may reproduce or amplify such biases. We introduce SPAGBias - the first systematic framework to evaluate spatial gender bias in LLMs. It combines a taxonomy of 62 urban micro-spaces, a prompt library, and three diagnostic layers: explicit (forced-choice resampling), probabilistic (token-level asymmetry), and constructional (semantic and narrative role analysis). Testing six representative models, we identify structured gender-space associations that go beyond the public-private divide, forming nuanced micro-level mappings. Story generation reveals how emotion, wording, and social roles jointly shape "spatial gender narratives". We also examine how prompt design, temperature, and model scale influence bias expression. Tracing experiments indicate that these patterns are embedded and reinforced across the model pipeline (pre-training, instruction tuning, and reward modeling), with model associations found to substantially exceed real-world distributions. Downstream experiments further reveal that such biases produce concrete failures in both normative and descriptive application settings. This work connects sociological theory with computational analysis, extending bias research into the spatial domain and uncovering how LLMs encode social gender cognition through language.

Zeroth-order (ZO) methods are widely used when gradients are unavailable or prohibitively expensive, including black-box learning and memory-efficient fine-tuning of large models, yet their optimization dynamics in deep learning remain underexplored. In this work, we provide an explicit step size condition that exactly captures the (mean-square) linear stability of a family of ZO methods based on the standard two-point estimator. Our characterization reveals a sharp contrast with first-order (FO) methods: whereas FO stability is governed solely by the largest Hessian eigenvalue, mean-square stability of ZO methods depends on the entire Hessian spectrum. Since computing the full Hessian spectrum is infeasible in practical neural network training, we further derive tractable stability bounds that depend only on the largest eigenvalue and the Hessian trace. Empirically, we find that full-batch ZO methods operate at the edge of stability: ZO-GD, ZO-GDM, and ZO-Adam consistently stabilize near the predicted stability boundary across a range of deep learning training problems. Our results highlight an implicit regularization effect specific to ZO methods, where large step sizes primarily regularize the Hessian trace, whereas in FO methods they regularize the top eigenvalue.

We analyze five photonic microring tensor core designs with a common optical power model. The results show that circuit ordering, unary encoding, and homodyne accumulation shape scalability, with the last two offering the strongest path to higher parallelism.

Edge AI model deployment is a multi-stage engineering process involving model conversion, operator compatibility handling, quantization calibration, runtime integration, and accuracy validation. In practice, this workflow is long, failure-prone, and heavily dependent on deployment expertise, particularly when targeting hardware-specific inference runtimes. This technical report presents AIPC (AI Porting Conversion), an AI agent-driven approach for constrained automation of AI model deployment. AIPC decomposes deployment into standardized, verifiable stages and injects deployment-domain knowledge into agent execution through Agent Skills, helper scripts, and a stage-wise validation loop. This design reduces both the expertise barrier and the engineering time required for hardware deployment. Using Qualcomm AI Runtime (QAIRT) as the primary scenario, this report examines automated deployment across representative vision, multimodal, and speech models. In the cases covered here, AIPC can complete deployment from PyTorch to runnable QNN/SNPE inference within 7-20 minutes for structurally regular vision models, with indicative API costs roughly in the range of USD 0.7-10. For more complex models involving less-supported operators, dynamic shapes, or autoregressive decoding structures, fully automated deployment may still require further advances, but AIPC already provides practical support for execution, failure localization, and bounded repair.

Most medical multimodal benchmarks focus on static tasks such as image question answering, report generation, and plain-language rewriting. Patient education is more demanding: systems must identify relevant evidence across images, show patients where to look, explain findings in accessible language, and handle confusion or distress. Yet most patient education work remains text-only, even though combined image-and-text explanations may better support understanding. We introduce MedImageEdu, a benchmark for multi-turn, evidence-grounded radiology patient education. Each case provides a radiology report with report text and case images. A DoctorAgent interacts with a PatientAgent, conditioned on a hidden profile that captures factors such as education level, health literacy, and personality. When a patient question would benefit from visual support, the DoctorAgent can issue drawing instructions grounded in the report, case images, and the current question to a benchmark-provided drawing tool. The tool returns image(s), after which the DoctorAgent produces a final multimodal response consisting of the image(s) and a grounded plain-language explanation. MedImageEdu contains 150 cases from three sources and evaluates both the consultation process and the final multimodal response along five dimensions: Consultation, Safety and Scope, Language Quality, Drawing Quality, and Image-Text Response Quality. Across representative open- and closed-source vision-language model agents, we find three consistent gaps: fluent language often outpaces faithful visual grounding, safety is the weakest dimension across disease categories, and emotionally tense interactions are harder than low education or low health literacy. MedImageEdu provides a controlled testbed for assessing whether multimodal agents can teach from evidence rather than merely answer from text.

We present AgentGA, a framework that evolves autonomous code-generation runs by optimizing the agent seed: the task prompt plus optional parent archives that initialize a fresh workspace. The outer loop searches over these reusable starting conditions rather than editing code directly. Each generation launches a fresh autonomous run from a reset workspace, while selected parent archives provide inherited artifacts that descendants can inspect and reuse. AgentGA couples a population-level genetic algorithm with long-horizon agents; selection uses deterministic 1:1 elite tournaments and operator allocation is adapted online with a modified Hedge controller. We instantiate the approach for tabular AutoML on the 16-competition Weco-Kaggle Lite benchmark. On the 10 benchmark runs reported here, AgentGA averages 74.52% Exceeds % of Human versus 54.15% for AIDE. Across 1135 parent-child comparisons, descendants given parent archives outperform runs started from scratch, indicating that inherited artifacts improve later autonomous runs. These findings support agent-seed optimization as a practical design point for autonomous code-search systems.

Clinical language models (LMs) are increasingly applied to support clinical risk prediction from free-text notes, yet their uncertainty estimates often remain poorly calibrated and clinically unreliable. In this work, we propose Clinical Uncertainty Risk Alignment (CURA), a framework that aligns clinical LM-based risk estimates and uncertainty with both individual error likelihoods and cohort-level ambiguities. CURA first fine-tunes domain-specific clinical LMs to obtain task-adapted patient embeddings, and then performs uncertainty fine-tuning of a multi-head classifier using a bi-level uncertainty objective. Specifically, an individual-level calibration term aligns predictive uncertainty with each patient's likelihood of error, while a cohort-aware regularizer pulls risk estimates toward event rates in their local neighborhoods in the embedding space and places extra weight on ambiguous cohorts near the decision boundary. We further show that this cohort-aware term can be interpreted as a cross-entropy loss with neighborhood-informed soft labels, providing a label-smoothing view of our method. Extensive experiments on MIMIC-IV clinical risk prediction tasks across various clinical LMs show that CURA consistently improves calibration metrics without substantially compromising discrimination. Further analysis illustrates that CURA reduces overconfident false reassurance and yields more trustworthy uncertainty estimates for downstream clinical decision support.

Video outpainting aims to expand the visible content of a video beyond the original frame boundaries while preserving spatial fidelity and temporal coherence across frames. Existing methods primarily rely on large-scale generative models, such as diffusion models. However, generationbased approaches suffer from implicit temporal modeling and limited spatial context. These limitations lead to intraframe and inter-frame inconsistencies, which become particularly pronounced in dynamic scenes and large outpainting scenarios. To overcome these challenges, we propose Seen-to-Scene, a novel framework that unifies propagationbased and generation-based paradigms for video outpainting. Specifically, Seen-to-Scene leverages flow-based propagation with a flow completion network pre-trained for video inpainting, which is fine-tuned in an end-to-end manner to bridge the domain gap and reconstruct coherent motion fields. To further improve the efficiency and reliability of propagation, we introduce a reference-guided latent propagation that effectively propagates source content across frames. Extensive experiments demonstrate that our method achieves superior temporal coherence and visual realism with efficient inference, surpassing even prior state-of-the-art methods that require input-specific adaptation.

Recent advances in reinforcement learning (RL) have improved the reasoning capabilities of large language models (LLMs) and vision-language models (VLMs). However, the widely used Group Relative Policy Optimization (GRPO) consistently suffers from entropy collapse, causing the policy to converge prematurely and lose diversity. Existing exploration methods introduce additional bias or variance during exploration, making it difficult to maintain optimization stability. We propose Unified Entropy Control for Reinforcement Learning (UEC-RL), a framework that provides targeted mechanisms for exploration and stabilization. UEC-RL activates more exploration on difficult prompts to search for potential and valuable reasoning trajectories. In parallel, a stabilizer prevents entropy from growing uncontrollably, thereby keeping training stable as the model consolidates reliable behaviors. Together, these components expand the search space when needed while maintaining robust optimization throughout training. Experiments on both LLM and VLM reasoning tasks show consistent gains over RL baselines on both Pass@1 and Pass@$k$. On Geometry3K, UEC-RL achieves a 37.9\% relative improvement over GRPO, indicating that it sustains effective exploration without compromising convergence and underscoring UEC-RL as a key for scaling RL-based reasoning in large models. Our code is available at https://github.com/597358816/UEC-RL.

Convolutional neural networks (CNNs) often exhibit poor generalisation in limited training data scenarios due to overfitting and insufficient feature diversity. In this work, a simple and effective chaos-based feature transformation is proposed to enhance CNN performance without increasing model complexity. The method applies nonlinear transformations using logistic, skew tent, and sine maps to normalised feature vectors before the classification layer, thereby reshaping the feature space and improving class separability. The approach is evaluated on greyscale datasets (MNIST and Fashion-MNIST) and an RGB dataset (CIFAR-10) using CNN architectures of varying depth under limited data conditions. The results show consistent improvement over the standalone (SA) CNN across all datasets. Notably, a maximum performance gain of 5.43% is achieved on MNIST using the skew tent map with a 3-layer CNN at 40 samples per class. A higher gain of 9.11% is observed on Fashion-MNIST using the sine map with a 3-layer CNN at 50 samples per class. Additionally, a strong gain of 7.47% is obtained on CIFAR-10 using the skew tent map at 200 samples per class. The consistent improvements across different chaotic maps indicate that the performance gain is driven by the shared nonlinear and dynamical properties of chaotic systems. The proposed method is computationally efficient, requires no additional trainable parameters, and can be easily integrated into existing CNN architectures, making it a practical solution for data-scarce image classification tasks.

The inability to filter out in advance all potentially problematic data from the pre-training of large language models has given rise to the need for methods for unlearning specific pieces of knowledge after training. Existing techniques overlook the need for continuous and immediate action, causing them to suffer from degraded utility as updates accumulate and protracted exposure of sensitive information. To address these issues, we propose Continual Unlearning in Real Time with Ensured Preservation of LLM Knowledge (CURaTE). Our method begins by training a sentence embedding model on a dataset designed to enable the formation of sharp decision boundaries for determining whether a given input prompt corresponds to any stored forget requests. The similarity of a given input to the forget requests is then used to determine whether to answer or return a refusal response. We show that even with such a simple approach, not only does CURaTE achieve more effective forgetting than existing methods, but by avoiding modification of the language model parameters, it also maintains near perfect knowledge preservation over any number of updates and is the only method capable of continual unlearning in real-time.

Adversarial attacks pose a severe threat to the reliability of deep learning models in remote sensing (RS) image classification. Most existing methods rely on direct pixel-wise perturbations, failing to exploit the inherent atmospheric characteristics of RS imagery or survive real-world image degradations. In this paper, we propose FogFool, a physically plausible adversarial framework that generates fog-based perturbations by iteratively optimizing atmospheric patterns based on Perlin noise. By modeling fog formations with natural, irregular structures, FogFool generates adversarial examples that are not only visually consistent with authentic RS scenes but also deceptive. By leveraging the spatial coherence and mid-to-low-frequency nature of atmospheric phenomena, FogFool embeds adversarial information into structural features shared across diverse architectures. Extensive experiments on two benchmark RS datasets demonstrate that FogFool achieves superior performance: not only does it exceed in white-box settings, but also exhibits exceptional black-box transferability (reaching 83.74% TASR) and robustness against common preprocessing-based defenses such as JPEG compression and filtering. Detailed analyses, including confusion matrices and Class Activation Map (CAM) visualizations, reveal that our atmospheric-driven perturbations induce a universal shift in model attention. These results indicate that FogFool represents a practical, stealthy, and highly persistent threat to RS classification systems, providing a robust benchmark for evaluating model reliability in complex environments.

When faced with complex spatial problems, humans naturally sketch layouts to organize their thinking, and the act of drawing further sharpens their understanding. In this work, we ask whether a similar principle holds for Large Language Models (LLMs): can learning to construct explicit visual layouts from spatial descriptions instill genuine spatial understanding? We introduce Text2Space, a dataset that pairs natural language descriptions with ground-truth ASCII grid layouts and spatial QA pairs, enabling us to separate failures in constructing spatial representations from failures in reasoning over them. We adopt ASCII because it is human-readable, operates entirely within the token space of language models, and encodes spatial relations in a structurally verifiable form. Our evaluation reveals a pronounced "Read-Write Asymmetry": LLMs interpret ASCII representations effectively but struggle to produce them from text, and these construction errors propagate to incorrect answers downstream. To address this limitation, we train models on layout construction (Text$\rightarrow$ASCII) and find that it significantly improves spatial reasoning from text alone, even without producing any ASCII at inference time. Combining construction with comprehension training further amplifies these gains. Crucially, these improvements transfer to three external spatial reasoning benchmarks, demonstrating that, much as sketching sharpens human spatial thinking, learning to construct explicit layouts instills spatial understanding that generalizes beyond the training format.

The proliferation of financial misinformation poses a severe threat to market stability and investor trust, misleading market behavior and creating critical information asymmetry. Detecting such misleading narratives is inherently challenging, particularly in real-world scenarios where external evidence or supplementary references for cross-verification are strictly unavailable. This paper presents our winning methodology for the "Reference-Free Financial Misinformation Detection" shared task. Built upon the recently proposed RFC-BENCH framework (Jiang et al. 2026), this task challenges models to determine the veracity of financial claims by relying solely on internal semantic understanding and contextual consistency, rather than external fact-checking. To address this formidable evaluation setup, we propose a comprehensive framework that capitalizes on the reasoning capabilities of state-of-the-art Large Language Models (LLMs). Our approach systematically integrates in-context learning, specifically zero-shot and few-shot prompting strategies, with Parameter-Efficient Fine-Tuning (PEFT) via Low-Rank Adaptation (LoRA) to optimally align the models with the subtle linguistic cues of financial manipulation. Our proposed system demonstrated superior efficacy, successfully securing the first-place ranking on both official leaderboards. Specifically, we achieved an accuracy of 95.4% on the public test set and 96.3% on the private test set, highlighting the robustness of our method and contributing to the acceleration of context-aware misinformation detection in financial Natural Language Processing. Our models (14B and 32B) are available at https://huggingface.co/KaiNKaiho.

Multiple choice evaluation is widely used for benchmarking large language models, yet near ceiling accuracy in low option settings can be sustained by shortcut strategies that obscure true competence. Therefore, we propose a massive option evaluation protocol that scales the candidate set to one hundred options and sharply reduces the impact of chance performance. We apply this framework to a Korean orthography error detection task where models must pick the single incorrect sentence from a large candidate set. With fixed targets and repeated resampling and shuffling, we obtain stable estimates while separating content driven failures from positional artifacts. Across experiments, results indicate that strong performance in low option settings can overstate model competence. This apparent advantage often weakens under dense interference at high $N$, revealing gaps that conventional benchmarks tend to obscure. We identify two failure modes, semantic confusion and position bias toward early options under uncertainty. To isolate the effect of context length, we run padding controlled and length matched tests, which suggest that the main bottleneck is candidate ranking rather than context length. Together, these findings support massive option evaluation as a general framework for stress testing model reliability under extreme distractor density, beyond what low option benchmarks can reveal.

Effective code generation requires both model capability and a problem representation that carefully structures how models reason and plan. Existing approaches augment reasoning steps or inject specific structure into how models think, but leave scattered problem conditions unchanged. Inspired by the way humans organize fragmented information into coherent explanations, we propose StoryCoder, a narrative reformulation framework that transforms code generation questions into coherent natural language narratives, providing richer contextual structure than simple rephrasings. Each narrative consists of three components: a task overview, constraints, and example test cases, guided by the selected algorithm and genre. Experiments across 11 models on HumanEval, LiveCodeBench, and CodeForces demonstrate consistent improvements, with an average gain of 18.7% in zero-shot pass@10. Beyond accuracy, our analyses reveal that narrative reformulation guides models toward correct algorithmic strategies, reduces implementation errors, and induces a more modular code structure. The analyses further show that these benefits depend on narrative coherence and genre alignment, suggesting that structured problem representation is important for code generation regardless of model scale or architecture. Our code is available at https://github.com/gu-ni/StoryCoder.

Recent advances in unsupervised video object segmentation have highlighted the potential of two-stream architectures that integrate appearance and motion cues. However, fully leveraging these complementary sources of information requires effectively modeling their interdependencies. In this paper, we introduce cross-modality token modulation, a novel approach designed to strengthen the interaction between appearance and motion cues. Our method establishes dense connections between tokens from each modality, enabling efficient intra-modal and inter-modal information propagation through relation transformer blocks. To improve learning efficiency, we incorporate a token masking strategy that addresses the limitations of relying solely on increased model complexity. Our approach achieves state-of-the-art performance across all public benchmarks, outperforming existing methods.

The exact cover problem is a classical NP-hard problem with broad applications in the area of AI. Algorithm DXZ is a method to count exact covers representing by zero-suppressed binary decision diagrams (ZBDDs). In this paper, we propose a zero-suppressed variant of decision decomposable negation normal form (in short, decision-ZDNNF), which is strictly more succinct than ZBDDs. We then design a novel parallel algorithm, namely DXD, which constructs a decision-ZDNNF representing the set of all exact covers. Furthermore, we improve DXD by dynamically updating connected components. The experimental results demonstrate that the improved DXD algorithm outperforms all of state-of-the-art methods.

Mixture-of-Experts (MoE) models have become the dominant architecture for large-scale language models, yet on-premises serving remains fundamentally memory-bound as batching turns sparse per-token compute into dense memory activation. Memory-centric architectures (PIM, NMP) improve bandwidth but leave compute underutilized under MoE's low arithmetic intensity at high batch sizes. Speculative decoding (SD) trades idle compute for fewer target invocations, yet verification must load experts even for rejected tokens, severely limiting its benefit in MoE especially at low batch sizes. We propose ELMoE-3D, a hybrid-bonding (HB)-based HW-SW co-designed framework that unifies cache-based acceleration and speculative decoding to offer overall speedup across batch sizes. We identify two intrinsic elasticity axes of MoE-expert and bit-and jointly scale them to construct Elastic Self-Speculative Decoding (Elastic-SD), which serves as both an expert cache and a strongly aligned self-draft model accelerated by high HB bandwidth. Our LSB-augmented bit-sliced architecture exploits inherent redundancy in bit-slice representations to natively support bit-nested execution. On our 3D-stacked hardware, ELMoE-3D achieves an average $6.6\times$ speedup and $4.4\times$ energy efficiency gain over naive MoE serving on xPU across batch sizes 1-16, and delivers $2.2\times$ speedup and $1.4\times$ energy efficiency gain over the best-performing prior accelerator baseline.

Humans often specify tasks incompletely, so assistants must know when and how to ask clarifying questions. However, effective clarification remains challenging in software engineering tasks as not all missing information is equally valuable, and questions must target information users can realistically provide. We study clarification in real software engineering tasks by quantifying which types of information most affect task success and which questions elicit useful responses from simulated users. Using Shapley attribution and distributional comparisons, we identify two key properties of effective clarification: task relevance (which information predicts success) and user answerability (what users can realistically provide). We operationalize these properties as multi-stage reinforcement learning rewards to train CLARITI, an 8B-parameter clarification module, that matches GPT-5's resolution rate on underspecified issues while generating 41% fewer questions. Our results suggest that grounding reward design in empirical analysis of information impact and user answerability improves clarification efficiency.

Conformal prediction (CP) has attracted broad attention as a simple and flexible framework for uncertainty quantification through prediction sets. In this work, we study how to deploy CP under differential privacy (DP) in a statistically efficient manner. We first introduce differential CP, a non-splitting conformal procedure that avoids the efficiency loss caused by data splitting and serves as a bridge between oracle CP and private conformal inference. By exploiting the stability properties of DP mechanisms, differential CP establishes a direct connection to oracle CP and inherits corresponding validity behavior. Building on this idea, we develop Differentially Private Conformal Prediction (DPCP), a fully private procedure that combines DP model training with a private quantile mechanism for calibration. We establish the end-to-end privacy guarantee of DPCP and investigate its coverage properties under additional regularity conditions. We further study the efficiency of both differential CP and DPCP under empirical risk minimization and general regression models, showing that DPCP can produce tighter prediction sets than existing private split conformal approaches under the same privacy budget. Numerical experiments on synthetic and real datasets demonstrate the practical effectiveness of the proposed methods.

In computational paralinguistics, detecting cognitive load and deception from speech signals is a heavily researched domain. Recent efforts have attempted to apply these acoustic frameworks to corporate earnings calls to predict catastrophic stock market volatility. In this study, we empirically investigate the limits of acoustic feature extraction (pitch, jitter, and hesitation) when applied to highly trained speakers in in-the-wild teleconference environments. Utilizing a two-stream late-fusion architecture, we contrast an acoustic-based stream with a baseline Natural Language Processing (NLP) stream. The isolated NLP model achieved a recall of 66.25% for tail-risk downside events. Surprisingly, integrating acoustic features via late fusion significantly degraded performance, reducing recall to 47.08%. We identify this degradation as Acoustic Camouflage, where media-trained vocal regulation introduces contradictory noise that disrupts multimodal meta-learners. We present these findings as a boundary condition for speech processing applications in high-stakes financial forecasting.

Clinical value set authoring -- the task of identifying all codes in a standardized vocabulary that define a clinical concept -- is a recurring bottleneck in clinical quality measurement and phenotyping. A natural approach is to prompt a large language model (LLM) to generate the required codes directly, but structured clinical vocabularies are large, version-controlled, and not reliably memorized during pretraining. We propose Retrieval-Augmented Set Completion (RASC): retrieve the $K$ most similar existing value sets from a curated corpus to form a candidate pool, then apply a classifier to each candidate code. Theoretically, retrieve-and-select can reduce statistical complexity by shrinking the effective output space from the full vocabulary to a much smaller retrieved candidate pool. We demonstrate the utility of RASC on 11,803 publicly available VSAC value sets, constructing the first large-scale benchmark for this task. A cross-encoder fine-tuned on SAPBert achieves AUROC~0.852 and value-set-level F1~0.298, outperforming a simpler three-layer Multilayer Perceptron (AUROC~0.799, F1~0.250) and both reduce the number of irrelevant candidates per true positive from 12.3 (retrieval-only) to approximately 3.2 and 4.4 respectively. Zero-shot GPT-4o achieves value-set-level F1~0.105, with 48.6\% of returned codes absent from VSAC entirely. This performance gap widens with increasing value set size, consistent with RASC's theoretical advantage. We observe similar performance gains across two other classifier model types, namely a cross-encoder initialized from pre-trained SAPBert and a LightGBM model, demonstrating that RASC's benefits extend beyond a single model class. The code to download and create the benchmark dataset, as well as the model training code is available at: \href{https://github.com/mukhes3/RASC}{https://github.com/mukhes3/RASC}.

Scientific discovery in digital health requires converting continuous physiological signals from wearable devices into clinically actionable biomarkers. We introduce CoDaS (AI Co-Data-Scientist), a multi-agent system that structures biomarker discovery as an iterative process combining hypothesis generation, statistical analysis, adversarial validation, and literature-grounded reasoning with human oversight using large-scale wearable datasets. Across three cohorts totaling 9,279 participant-observations, CoDaS identified 41 candidate digital biomarkers for mental health and 25 for metabolic outcomes, each subjected to an internal validation battery spanning replication, stability, robustness, and discriminative power. Across two independent depression cohorts, CoDaS surfaced circadian instability-related features in both datasets, reflected in sleep duration variability (DWB, ρ= 0.252, p < 0.001) and sleep onset variability (GLOBEM, ρ= 0.126, p < 0.001). In a metabolic cohort, CoDaS derived a cardiovascular fitness index (steps/resting heart rate; ρ= -0.374, p < 0.001), and recovered established clinical associations, including the hepatic function ratio (AST/ALT; ρ= -0.375, p < 0.001), a known correlate of insulin resistance. Incorporating CoDaS-derived features alongside demographic variables led to modest but consistent improvements in predictive performance, with cross-validated ΔR^2 increases of 0.040 for depression and 0.021 for insulin resistance. These findings suggest that CoDaS enables systematic and traceable hypothesis generation and prioritization for biomarker discovery from large-scale wearable data.

We give an algorithm for PAC learning intersections of $k$ halfspaces with a $ρ$ margin to within error $\varepsilon$ that runs in time $\textsf{poly}(k, \varepsilon^{-1}, ρ^{-1}) \cdot \exp \left(O(\sqrt{n \log(1/ρ) \log k})\right)$. Notably, this improves on prior work which had an exponential dependence on either $k$ or $ρ^{-1}$ and matches known cryptographic and Statistical Query lower bounds up to the logarithmic factors in $k$ and $ρ$ in the exponent. Our learning algorithm extends to the more general setting when we are only promised that most points have distance at least $ρ$ from the boundary of the polyhedron, making it applicable to continuous distributions as well.

Learning Path Recommendation (LPR) is critical for personalized education, yet current methods often fail to account for historical interaction uncertainty (e.g., lucky guesses or accidental slips) and lack adaptability to diverse learning goals. We propose U-GLAD (Uncertainty-aware Generative Learning Path Recommendation with Cognition-Adaptive Diffusion). To address representation bias, the framework models cognitive states as probability distributions, capturing the learner's underlying true state via a Gaussian LSTM. To ensure highly personalized recommendation, a goal-oriented concept encoder utilizes multi-head attention and objective-specific transformations to dynamically align concept semantics with individual learning goals, generating uniquely tailored embeddings. Unlike traditional discriminative ranking approaches, our model employs a generative diffusion model to predict the latent representation of the next optimal concept. Extensive evaluations on three public datasets demonstrate that U-GLAD significantly outperforms representative baselines. Further analyses confirm its superior capability in perceiving interaction uncertainty and providing stable, goal-driven recommendation paths.

Self-speculative decoding is an inference technique for large language models designed to speed up generation without sacrificing output quality. It combines fast, approximate decoding using a compact version of the model as a draft model with selective re-evaluation by the full target model. Some existing methods form the draft model by dynamically learning which layers to skip during inference, effectively creating a smaller subnetwork to speed up computation. However, using heuristic-based approaches to select layers to skip can often be simpler and more effective. In this paper, we propose ConfLayers, a dynamic plug-and-play approach to forming the draft model in self-speculative decoding via confidence-based intermediate layer skipping. The process iteratively computes confidence scores for all layers, selects layers to skip based on an adaptive threshold, evaluates the performance of the resulting set, and updates the best selection until no further improvement is achieved or a maximum number of iterations is reached. This framework avoids the overhead and complexity of training a layer skipping policy and can provide more consistent speed-quality trade-offs while preserving the adaptivity of the draft model to diverse tasks and datasets. The performance evaluation of ConfLayers across different models and datasets shows that our novel approach offers up to 1.4x speedup over vanilla LLM generation.

AI for science promises to accelerate the discovery process. The advent of large language models (LLMs) and agentic workflows enables the expediting of a growing range of scientific tasks. However, most of the current generation of agentic systems depend on static, hand-curated toolsets that hinder adaptation to new domains and evolving libraries. We present El Agente Forjador, a multi-agent framework in which universal coding agents autonomously forge, validate, and reuse computational tools through a four-stage workflow of tool analysis, tool generation, task execution, and iterative solution evaluation. Evaluated across 24 tasks spanning quantum chemistry and quantum dynamics on five coding agent setups, we compare three operating modes: zero-shot generation of tools per task, reuse of a curriculum-built toolset, and direct problem-solving with the coding agents as the baseline. We find that our tool generation and reuse framework consistently improves accuracy over the baseline. We also show that reusing a toolset built by a stronger coding agent can reduce API cost and substantially raises the solution quality for weaker coding agents. Case studies further demonstrate that tools forged for different domains can be combined to solve hybrid tasks. Taken together, these results show that LLM-based agents can use their scientific knowledge and coding capabilities to autonomously build reusable scientific tools, pointing toward a paradigm in which agent capabilities are defined by the tasks they are designed to solve rather than by explicitly engineered implementations.

We study the overall process of automatic formalization of GDPR provisions using large language models, within a human-in-the-loop verification framework. Rather than aiming for full autonomy, we adopt a role-specialized workflow in which LLM-based AI components, operating in a multi-agent setting with iterative feedback, generate legal scenarios, formal rules, and atomic facts. This is coupled with independent verification modules which include human reviewers' assessment of representational, logical, and legal correctness. Using this approach, we construct a high-quality dataset to be used for GDPR auto-formalization, and analyze both successful and problematic cases. Our results show that structured verification and targeted human oversight are essential for reliable legal formalization, especially in the presence of legal nuance and context-sensitive reasoning.

Modern Large audio-language models (LALMs) power intelligent voice interactions by tightly integrating audio and text. This integration, however, expands the attack surface beyond text and introduces vulnerabilities in the continuous, high-dimensional audio channel. While prior work studied audio jailbreaks, the security risks of malicious audio injection and downstream behavior manipulation remain underexamined. In this work, we reveal a previously overlooked threat, auditory prompt injection, under realistic constraints of audio data-only access and strong perceptual stealth. To systematically analyze this threat, we propose \textit{AudioHijack}, a general framework that generates context-agnostic and imperceptible adversarial audio to hijack LALMs. \textit{AudioHijack} employs sampling-based gradient estimation for end-to-end optimization across diverse models, bypassing non-differentiable audio tokenization. Through attention supervision and multi-context training, it steers model attention toward adversarial audio and generalizes to unseen user contexts. We also design a convolutional blending method that modulates perturbations into natural reverberation, making them highly imperceptible to users. Extensive experiments on 13 state-of-the-art LALMs show consistent hijacking across 6 misbehavior categories, achieving average success rates of 79\%-96\% on unseen user contexts with high acoustic fidelity. Real-world studies demonstrate that commercial voice agents from Mistral AI and Microsoft Azure can be induced to execute unauthorized actions on behalf of users. These findings expose critical vulnerabilities in LALMs and highlight the urgent need for dedicated defense.

The fundamental limit of natural signal compression has traditionally been characterized by classical rate-distortion (RD) theory through the tradeoff between coding rate and reconstruction distortion, while the rate-distortion-perception (RDP) framework introduces a divergence-based measure of perceptual quality as a modeling principle rather than a theoretically-derived principle, leaving its theoretical origin unclear. In this paper, motivated by a synonymity-based semantic information perspective, we reformulate perceptual reconstruction as recovering any admissible sample within an ideal synonymous set (synset) associated with the source, rather than the source sample itself, and correspondingly establish a synonymous source coding architecture. On this basis, we develop a synonymous variational inference (SVI) analysis framework with a synonymous variational lower bound (SVLBO) for tractable analysis of synset-oriented compression. Within this framework, we establish a synonymity-perception consistency principle, showing that optimal identification of semantic information is theoretically consistent with perceptual optimization. Based on its derivation result, we prove a synonymous RDP tradeoff for the proposed synonymous source coding. These analytical results show that the distributional divergence term arises naturally from the synset-based reconstruction objective, clarify its compatibility with existing RDP formulations and classical RD theory, and suggest the potential advantages of synonymous source coding.

Large language models (LLMs) frequently generate toxic content, posing significant risks for safe deployment. Current mitigation strategies often degrade generation quality or require costly human annotation. We propose CAUSALDETOX, a framework that identifies and intervenes on the specific attention heads causally responsible for toxic generation. Using the Probability of Necessity and Sufficiency (PNS), we isolate a minimal set of heads that are necessary and sufficient for toxicity. We utilize these components via two complementary strategies: (1) Local Inference-Time Intervention, which constructs dynamic, input-specific steering vectors for context-aware detoxification, and (2) PNS-Guided Fine-Tuning, which permanently unlearns toxic representations. We also introduce PARATOX, a novel benchmark of aligned toxic/non-toxic sentence pairs enabling controlled counterfactual evaluation. Experiments on ToxiGen, ImplicitHate, and ParaDetox show that CAUSALDETOX achieves up to 5.34% greater toxicity reduction compared to baselines while preserving linguistic fluency, and offers a 7x speedup in head selection.

This position paper argues that recent progress with diversity in NLP is disproportionately concentrated on a small number of areas surrounding fairness. We further argue that this is the result of a number of incentives, biases, and barriers which come together to disenfranchise marginalized researchers in non-fairness fields, or to move them into fairness-related fields. We substantiate our claims with an investigation into the demographics of NLP researchers by subfield, using our research to support a number of recommendations for ensuring that all areas within NLP can become more inclusive and equitable. In particular, we highlight the importance of breaking down feedback loops that reinforce disparities, and the need to address geographical and linguistic barriers that hinder participation in NLP research.

While Large Language Models (LLMs) demonstrate increasingly sophisticated affective capabilities, the internal mechanisms by which they process complex emotions remain unclear. Existing interpretability approaches often treat models as black boxes or focus on coarse-grained basic emotions, leaving the cognitive structure of more complex affective states underexplored. To bridge this gap, we propose a Cognitive Reverse-Engineering framework based on Representation Engineering (RepE) to analyze social-comparison jealousy. By combining appraisal theory with subspace orthogonalization, regression-based weighting, and bidirectional causal steering, we isolate and quantify two psychological antecedents of jealousy, Superiority of Comparison Person and Domain Self-Definitional Relevance, and examine their causal effects on model judgments. Experiments on eight LLMs from the Llama, Qwen, and Gemma families suggest that models natively encode jealousy as a structured linear combination of these constituent factors. Their internal representations are broadly consistent with the human psychological construct, treating Superiority as the foundational trigger and Relevance as the ultimate intensity multiplier. Our framework also demonstrates that toxic emotional states can be mechanically detected and surgically suppressed, suggesting a possible route toward representational monitoring and intervention for AI safety in multi-agent environments.

In modern data-streaming systems, alongside traditional programs, a new type of entity has emerged that can interact with streaming data: AI agents. Unlike traditional programs, AI agents use LLM reasoning to accomplish high-level tasks specified in natural language over streaming data. Unfortunately, current streaming systems cannot fully support agents: they lack the fundamental mechanisms to avoid the performance interference caused by agentic tasks and to safely handle agentic writes. We argue that the shared log, the core abstraction underlying streaming data, must support creating forks of itself, and that such a forkable shared log serves as a great substrate for agents acting on streaming data. We propose AgileLog, a new shared log abstraction that provides novel forking primitives for agentic use cases. We design Bolt, an implementation of the AgileLog abstraction, that uses novel techniques to make forks cheap, and provide logical and performance isolation.

Lion optimizer is a popular learning-based optimization algorithm in machine learning, which shows impressive performance in training many deep learning models. Although convergence property of the Lion optimizer has been studied, its generalization analysis is still missing. To fill this gap, we study generalization property of the Lion via algorithmic stability based on the mathematical induction. Specifically, we prove that the Lion has a generalization error of $O(\frac{1}{Nτ^T})$, where $N$ is training sample size, and $τ>0$ denotes the smallest absolute value of non-zero element in gradient estimator, and $T$ is the total iteration number. In addition, we obtain an interesting byproduct that the SignSGD algorithm has the same generalization error as the Lion. To enhance generalization of the Lion, we design a novel efficient Cautious Lion (i.e., CLion) optimizer by cautiously using sign function. Moreover, we prove that our CLion has a lower generalization error of $O(\frac{1}{N})$ than $O(\frac{1}{Nτ^T})$ of the Lion, since the parameter $τ$ generally is very small. Meanwhile, we study convergence property of our CLion optimizer, and prove that our CLion has a fast convergence rate of $O(\frac{\sqrt{d}}{T^{1/4}})$ under $\ell_1$-norm of gradient for nonconvex stochastic optimization, where $d$ denotes the model dimension. Extensive numerical experiments demonstrate effectiveness of our CLion optimizer.

The rapid expansion of gaming industry requires advanced recommender systems tailored to its dynamic landscape. Existing Graph Neural Network (GNN)-based methods primarily prioritize accuracy over diversity, overlooking their inherent trade-off. To address this, we previously proposed CPGRec, a balance-oriented gaming recommender system. However, CPGRec fails to account for critical disparities in player-game interactions, which carry varying significance in reflecting players' personal preferences and may exacerbate over-smoothness issues inherent in GNN-based models. Moreover, existing approaches underutilize the reasoning capabilities and extensive knowledge of large language models (LLMs) in addressing these limitations. To bridge this gap, we propose two new modules. First, Preference-informed Edge Reweighting (PER) module assigns signed edge weights to qualitatively distinguish significant player interests and disinterests while then quantitatively measuring preference strength to mitigate over-smoothing in graph convolutions. Second, Preference-informed Representation Generation (PRG) module leverages LLMs to generate contextualized descriptions of games and players by reasoning personal preferences from comparing global and personal interests, thereby refining representations of players and games. Experiments on \textcolor{black}{two Steam datasets} demonstrate CPGRec+'s superior accuracy and diversity over state-of-the-art models. The code is accessible at https://github.com/HsipingLi/CPGRec-Plus.

Prompt optimization in compound AI systems is statistically indistinguishable from a coin flip: across 72 optimization runs on Claude Haiku (6 methods $\times$ 4 tasks $\times$ 3 repeats), 49% score below zero-shot; on Amazon Nova Lite, the failure rate is even higher. Yet on one task, all six methods improve over zero-shot by up to $+6.8$ points. What distinguishes success from failure? We investigate with 18,000 grid evaluations and 144 optimization runs, testing two assumptions behind end-to-end optimization tools like TextGrad and DSPy: (A) individual prompts are worth optimizing, and (B) agent prompts interact, requiring joint optimization. Interaction effects are never significant ($p > 0.52$, all $F < 1.0$), and optimization helps only when the task has exploitable output structure -- a format the model can produce but does not default to. We provide a two-stage diagnostic: an \$80 ANOVA pre-test for agent coupling, and a 10-minute headroom test that predicts whether optimization is worthwhile -- turning a coin flip into an informed decision.

Decentralized Finance (DeFi) lending protocols like Aave v3 rely on over-collateralization to secure loans, yet users frequently face liquidation due to volatile market conditions. Existing risk management tools utilize static health-factor thresholds, which are reactive and fail to distinguish between administrative "dust" cleanup and genuine insolvency. In this work, we propose an autonomous agent that leverages time-to-event (survival) analysis and moves beyond prediction to execution. Unlike passive risk signals, this agent perceives risk, simulates counterfactual futures, and executes protocol-faithful interventions to proactively prevent liquidations. We introduce a return period metric derived from a numerically stable XGBoost Cox proportional hazards model to normalize risk across transaction types, coupled with a volatility-adjusted trend score to filter transient market noise. To select optimal interventions, we implement a counterfactual optimization loop that simulates potential user actions to find the minimum capital required to mitigate risk. We validate our approach using a high-fidelity, protocol-faithful Aave v3 simulator on a cohort of 4,882 high-risk user profiles. The results demonstrate the agent's ability to prevent liquidations in imminent-risk scenarios where static rules fail, effectively "saving the unsavable" while maintaining a zero worsening rate, providing a critical safety guarantee often missing in autonomous financial agents. Furthermore, the system successfully differentiates between actionable financial risks and negligible dust events, optimizing capital efficiency where static rules fail.

Large language models (LLMs) show promise in generating supportive responses for mental health and counseling applications. However, their responses often lack cultural sensitivity, contextual grounding, and clinically appropriate guidance. This work addresses the gap of how to systematically incorporate domain-specific, clinically validated knowledge into LLMs to improve counseling quality. We utilize and compare two approaches, retrieval-augmented generation (RAG) and a knowledge graph (KG)-based method, designed to support para-counselors. Our KG is constructed manually and clinically validated, capturing causal relationships between stressors, interventions, and outcomes, with contributions from multidisciplinary people. We evaluated multiple LLMs in both settings using BERTScore F1 and SBERT cosine similarity, as well as human evaluation across five metrics, which is designed to directly measure the effectiveness of counseling beyond similarity at the surface level. The results show that KG-based approaches consistently improve contextual relevance, clinical appropriateness, and practical usability compared to RAG alone, demonstrating that structured, expert-validated knowledge plays a critical role in addressing LLMs limitations in counseling tasks.

Data-driven operations management often relies on parameters estimated from costly human-generated labels. Recent advances in large language models (LLMs) and other AI systems offer inexpensive auxiliary data, but introduce a new challenge: AI outputs are not direct observations of the target outcomes, but could involve high-dimensional representations with complex and unknown relationships to human labels. Conventional methods leverage AI predictions as direct proxies for true labels, which can be inefficient or unreliable when this relationship is weak or misspecified. We propose Generative Augmented Inference (GAI), a general framework that incorporates AI-generated outputs as informative features for estimating models of human-labeled outcomes. GAI uses an orthogonal moment construction that enables consistent estimation and valid inference with flexible, nonparametric relationship between LLM-generated outputs and human labels. We establish asymptotic normality and show a "safe default" property: relative to human-data-only estimators, GAI weakly improves estimation efficiency under arbitrary auxiliary signals and yields strict gains whenever the auxiliary information is predictive. Empirically, GAI outperforms benchmarks across diverse settings. In conjoint analysis with weak auxiliary signals, GAI reduces estimation error by about 50% and lowers human labeling requirements by over 75%. In retail pricing, where all methods access the same auxiliary inputs, GAI consistently outperforms alternative estimators, highlighting the value of its construction rather than differences in information. In health insurance choice, it cuts labeling requirements by over 90% while maintaining decision accuracy. Across applications, GAI improves confidence interval coverage without inflating width. Overall, GAI provides a principled and scalable approach to integrating AI-generated information.

Retrieval-Augmented Generation (RAG) grounds LLM responses in external evidence but treats the model as a passive consumer of search results: it never sees how the corpus is organized or what it has not yet retrieved, limiting its ability to backtrack or combine scattered evidence. We present Corpus2Skill, which distills a document corpus into a hierarchical skill directory offline and lets an LLM agent navigate it at serve time. The compilation pipeline iteratively clusters documents, generates LLM-written summaries at each level, and materializes the result as a tree of navigable skill files. At serve time, the agent receives a bird's-eye view of the corpus, drills into topic branches via progressively finer summaries, and retrieves full documents by ID. Because the hierarchy is explicitly visible, the agent can reason about where to look, backtrack from unproductive paths, and combine evidence across branches. On WixQA, an enterprise customer-support benchmark for RAG, Corpus2Skill outperforms dense retrieval, RAPTOR, and agentic RAG baselines across all quality metrics.

Visual reasoning models (VRMs) have recently shown strong cross-modal reasoning capabilities by integrating visual perception with language reasoning. However, they often suffer from overthinking, producing unnecessarily long reasoning chains for any tasks. We attribute this issue to \textbf{Reasoning Path Redundancy} in visual reasoning: many visual questions do not require the full reasoning process. To address this, we propose \textbf{AVR}, an adaptive visual reasoning framework that decomposes visual reasoning into three cognitive functions: visual perception, logical reasoning, and answer application. It further enables models to dynamically choose among three response formats: Full Format, Perception-Only Format, and Direct Answer. AVR is trained with FS-GRPO, an adaptation of Group Relative Policy Optimization that encourages the model to select the most efficient reasoning format while preserving correctness. Experiments on multiple vision-language benchmarks show that AVR reduces token usage by 50--90\% while maintaining overall accuracy, especially in perception-intensive tasks. These results demonstrate that adaptive visual reasoning can effectively mitigate overthinking in VRMs. Code and data are available at: https://github.com/RunRiotComeOn/AVR.

Accurate temperature estimation of pouch cells with indirect liquid cooling is essential for optimizing battery thermal management systems for transportation electrification. However, it is challenging due to the computational expense of finite element simulations and the limitations of data-driven models. This paper presents a physics-informed machine learning (PIML) framework for the efficient and reliable estimation of steady-state temperature profiles. The PIML approach integrates the governing heat transfer equations directly into the neural network's loss function, enabling high-fidelity predictions with significantly faster convergence than purely data-driven methods. The framework is evaluated on a dataset of varying cooling channel geometries. Results demonstrate that the PIML model converges more rapidly and achieves markedly higher accuracy, with a 49.1% reduction in mean squared error over the data-driven model. Validation against independent test cases further confirms its superior performance, particularly in regions away from the cooling channels. These findings underscore the potential of PIML for surrogate modeling and design optimization in battery systems.

Reinforcement learning (RL) paradigms have demonstrated strong performance on reasoning-intensive tasks such as code generation. However, limited trajectory diversity often leads to diminishing returns, which constrains the achievable performance ceiling. Search-enhanced RL alleviates this issue by introducing structured exploration, which remains constrained by the single-agent policy priors. Meanwhile, leveraging multiple interacting policies can acquire more diverse exploratory signals, but existing approaches are typically decoupled from structured search. We propose \textbf{MARS$^2$} (Multi-Agent Reinforced Tree-Search Scaling), a unified RL framework in which multiple independently-optimized agents collaborate within a shared tree-structured search environment. MARS$^2$ models the search tree as a learnable multi-agent interaction environment, enabling heterogeneous agents to collaboratively generate and refine candidate solutions within a shared search topology. To support effective learning, we introduce a path-level group advantage formulation based on tree-consistent reward shaping, which facilitates effective credit assignment across complex search trajectories. Experiments on code generation benchmarks show that MARS$^2$ consistently improves performance across diverse model combinations and training settings, demonstrating the effectiveness of coupling multi-agent collaboration with tree search for enhancing reinforcement learning. Our code is publicly available at https://github.com/TsinghuaC3I/MARTI.

Accurate thermal modeling in metal additive manufacturing (AM) is essential for understanding the process-structure-performance relationship. While prior studies have explored generalization across unseen process conditions, they often require extensive datasets, costly retraining, or pre-training. Generalization across different materials also remains relatively unexplored due to the challenges posed by distinct material-dependent thermal behaviors. This paper introduces a parametric physics-informed neural network (PINN) framework for zero-shot generalization across arbitrary materials without labeled data, retraining, or pre-training. The framework adopts a decoupled parametric PINN architecture that separately encodes material properties and spatiotemporal coordinates, fusing them through conditional modulation to better align with the multiplicative role of material parameters in the governing equation and boundary conditions. Physics-guided output scaling derived from Rosenthal's analytical solution and a hybrid optimization strategy are further incorporated to enhance physical consistency, training stability, and convergence. Experiments on bare plate laser powder bed fusion (LPBF) across diverse metal alloys, including both in-distribution and out-of-distribution cases, demonstrate effective zero-shot generalizability along with superior training efficiency. Specifically, the proposed framework achieved up to a 64.2% reduction in relative L2 error compared to the non-parametric baseline while surpassing its performance within only 4.4% of the baseline training epochs. Ablation studies confirm that the proposed framework's components are broadly applicable to other PINN-based approaches. Overall, the proposed framework provides an efficient and scalable material-agnostic solution for zero-shot thermal modeling, contributing to more flexible and practical deployment in metal AM.

Diffusion Transformers (DiTs) are increasingly adopted in scientific computing, yet growing model sizes and resolutions make distributed multi-GPU inference essential. Ulysses sequence parallelism scales DiT inference but introduces frequent all-to-all collectives that dominate latency. Overlapping these with computation is difficult due to tight data dependencies, large message volumes, and asymmetric interconnect bandwidths. We introduce CoCoDiff, a distributed DiT inference engine exploiting two observations: (1) V requires only linear projection while Q/K need additional normalization and RoPE, creating opportunities to overlap V's communication with Q/K computation; (2) adjacent denoising steps produce similar tensors, yielding temporal redundancy. CoCoDiff introduces three mechanisms: Tile-Aware Parallel All-to-all (TAPA) decomposes collectives into topology-aligned phases; V-First scheduling hides V's communication behind Q/K computation; and V-Major selective communication transmits only active projections on slow interconnects. On the Aurora supercomputer with four DiT models across 1-8 nodes (up to 96 Intel GPU tiles), CoCoDiff achieves an average speedup of 3.6x, peaking at 8.4x.

Video object insertion is a critical task for dynamically inserting new objects into existing environments. Previous video generation methods focus primarily on synthesizing entire scenes while struggling with ensuring consistent object appearance, spatial alignment, and temporal coherence when inserting objects into existing videos. In this paper, we propose a novel solution for Video Object Insertion, which integrates multi-view object priors to address the common challenges of appearance inconsistency and occlusion handling in dynamic environments. By lifting 2D reference images into multi-view representations and leveraging a dual-path view-consistent conditioning mechanism, our framework ensures stable identity guidance and robust integration across diverse viewpoints. A quality-aware weighting mechanism is also employed to adaptively handle noisy or imperfect inputs. Additionally, we introduce an Integration-Aware Consistency Module that guarantees spatial realism, effectively resolving occlusion and boundary artifacts while maintaining temporal continuity across frames. Experimental results show that our solution significantly improves the quality of video object insertion, providing stable and realistic integration.

Modern datacenters increasingly rely on low-power, single-slot inference accelerators to balance performance, energy efficiency, and rack density constraints. The NVIDIA T4 GPU has become widely deployed due to strong performance per watt and mature software support. Its successor, the NVIDIA L4 GPU, introduces improvements in Tensor Core throughput, cache capacity, memory bandwidth, and parallel execution capability. However, limited empirical evidence quantifies the practical inference performance gap between these two generations under controlled and reproducible conditions. This work introduces DEEP-GAP, a systematic evaluation extending the GDEV-AI methodology to GPU inference. Using identical configurations and workloads, we evaluate ResNet18, ResNet50, and ResNet101 across FP32, FP16, and INT8 precision modes using PyTorch and TensorRT. Results show that reduced precision significantly improves performance, with INT8 achieving up to 58x throughput improvement over CPU baselines. L4 achieves up to 4.4x higher throughput than T4 while reaching peak efficiency at smaller batch sizes between 16 and 32, improving latency-throughput tradeoffs for latency-sensitive workloads. T4 remains competitive for large batch workloads where cost or power efficiency is important. DEEP-GAP provides practical guidance for selecting precision modes, batch sizes, and GPU architectures for modern inference deployments.

Generating synthesizable Verilog for large, hierarchical hardware designs remains a significant challenge for large language models (LLMs), which struggle to replicate the structured reasoning that human experts employ when translating complex specifications into RTL. When tasked with producing hierarchical Verilog, LLMs frequently lose context across modules, hallucinate interfaces, fabricate inter-module wiring, and fail to maintain structural coherence - failures that intensify as design complexity grows and specifications involve informal prose, figures, and tables that resist direct operationalization. To address these challenges, we present VeriGraphi, a framework that introduces a spec-anchored Knowledge Graph as the architectural substrate driving the RTL generation pipeline. VeriGraphi constructs a HDA, a structured knowledge graph that explicitly encodes module hierarchy, port-level interfaces, wiring semantics, and inter-module dependencies as first-class graph entities and relations. Built through iterative multi-agent analysis of the specification, this Knowledge Graph provides a deterministic, machine-checkable structural scaffold before code generation. Guided by the KG, a progressive coding module incrementally generates pseudo-code and synthesizable RTL while enforcing interface consistency and dependency correctness at each submodule stage. We evaluate VeriGraphi on a benchmark of three representative specification documents from the National Institute of Standards and Technology and their corresponding implementations, and we present a RV32I processor as a detailed case study to illustrate the full pipeline. The results demonstrate that VeriGraphi enables reliable hierarchical RTL generation with minimal human intervention for RISC-V, marking a significant milestone for LLM-generated hardware design while maintaining strong functional correctness.

As speech language models (SLMs) transition from personal devices into shared, multi-user environments, their responses must account for far more than the words alone. Who is speaking, how they sound, and where the conversation takes place can each turn an otherwise benign request into one that is unsafe, unfair, or privacy-violating. Existing benchmarks, however, largely focus on basic audio comprehension, study individual risks in isolation, or conflate content that is inherently harmful with content that only becomes problematic due to its acoustic context. We introduce VoxSafeBench, among the first benchmarks to jointly evaluate social alignment in SLMs across three dimensions: safety, fairness, and privacy. VoxSafeBench adopts a Two-Tier design: Tier1 evaluates content-centric risks using matched text and audio inputs, while Tier2 targets audio-conditioned risks in which the transcript is benign but the appropriate response hinges on the speaker, paralinguistic cues, or the surrounding environment. To validate Tier2, we include intermediate perception probes and confirm that frontier SLMs can successfully detect these acoustic cues yet still fail to act on them appropriately. Across 22 tasks with bilingual coverage, we find that safeguards appearing robust on text often degrade in speech: safety awareness drops for speaker- and scene-conditioned risks, fairness erodes when demographic differences are conveyed vocally, and privacy protections falter when contextual cues arrive acoustically. Together, these results expose a pervasive speech grounding gap: current SLMs frequently recognize the relevant social norm in text but fail to apply it when the decisive cue must be grounded in speech. Code and data are publicly available at: https://amphionteam.github.io/VoxSafeBench_demopage/

Objective: Post-traumatic epilepsy (PTE) is a debilitating neurological disorder that develops after traumatic brain injury (TBI). Early prediction of PTE remains challenging due to heterogeneous clinical data, limited positive cases, and reliance on resource-intensive neuroimaging data. We investigate whether routinely collected acute clinical records alone can support early PTE prediction using language model-based approaches. Methods: Using a curated subset of the TRACK-TBI cohort, we developed an automated PTE prediction framework that implements pretrained large language models (LLMs) as fixed feature extractors to encode clinical records. Tabular features, LLM-generated embeddings, and hybrid feature representations were evaluated using gradient-boosted tree classifiers under stratified cross-validation. Results: LLM embeddings achieved performance improvements by capturing contextual clinical information compared to using tabular features alone. The best performance was achieved by a modality-aware feature fusion strategy combining tabular features and LLM embeddings, achieving an AUC-ROC of 0.892 and AUPRC of 0.798. Acute post-traumatic seizures, injury severity, neurosurgical intervention, and ICU stay are key contributors to the predictive performance. Significance: These findings demonstrate that routine acute clinical records contain information suitable for early PTE risk prediction using LLM embeddings in conjunction with gradient-boosted tree classifiers. This approach represents a promising complement to imaging-based prediction.

Purpose. Athlete monitoring is constrained by small cohorts, heterogeneous biomarker scales, limited feasibility of repeated sampling, and the lack of reliable injury ground truth. These limitations reduce the interpretability and utility of traditional univariate and binary risk models. This study addresses these challenges by proposing an unsupervised multivariate framework to identify latent physiological states in athletes using real data. Methods. We propose a modular computational framework that operates in the joint biomarker space, integrating preprocessing, clinical safety screening, unsupervised clustering, and centroid-based physiological interpretation. Profiles are learned exclusively from amateur soccer players during a competitive microcycle. Synthetic data augmentation evaluates robustness and scalability. Ward hierarchical clustering supports monitoring and etiological differentiation, while Gaussian Mixture Models (GMM) enable structural stability analysis in high-dimensional settings. Results. The framework identifies coherent profiles that distinguish mechanical damage from metabolic stress while preserving homeostatic states. Synthetic data augmentation demonstrates feasibility and detection of latent silent risk phenotypes typically missed by univariate monitoring. Structural analyses indicate robustness under augmentation and higher-dimensional settings. Conclusion. The framework enables interpretable identification of latent physiological states from multivariate biomarker data without injury labels. By distinguishing mechanisms and revealing silent risk patterns not captured by conventional monitoring, it provides actionable insights for individualized athlete monitoring and decision making.

Accurate prediction of future risk and disease progression in sepsis is clinically important for early warning and timely intervention in intensive care. However, short-window sepsis prediction remains challenging, because shorter observation windows provide limited historical evidence, whereas longer prediction horizons reduce the number of patient trajectories with valid future supervision. To address this problem, we propose CSRA, a Controlled Spectral Residual Augmentation framework for short-window multi-system ICU time series. CSRA first groups variables by clinical systems and extracts system-level and global representations. It then performs input-adaptive residual perturbation in the spectral domain to generate structured and clinically plausible trajectory variations. To improve augmentation stability and controllability, CSRA is trained end-to-end with the downstream predictor under a unified objective, together with anchor consistency loss and controller regularization. Experiments on a MIMIC-IV sepsis cohort across multiple downstream models show that CSRA is consistently competitive and often superior, reducing regression error by 10.2\% in MSE and 3.7\% in MAE over the non-augmentation baseline, while also yielding consistent gains on classification. CSRA further maintains more favorable performance under shorter observation windows, longer prediction horizons, and smaller training data scales, while also remaining effective on an external clinical dataset~(ZiGongICUinfection), indicating stronger robustness and generalizability in clinically constrained settings.

Every call to an LLM classification endpoint produces a labeled input-output pair already retained in production logs. These pairs constitute a free, growing training set: a lightweight surrogate trained on them can absorb a significant portion of future traffic at near-zero marginal inference cost. The open questions are when the surrogate is reliable enough to deploy, what it handles versus defers, and how that boundary evolves as data accumulates. We introduce TRACER (Trace-based Adaptive Cost-Efficient Routing), an open-source system that trains ML surrogates on an LLM's own production traces and governs deployment through a parity gate: the surrogate is activated only when its agreement with the LLM exceeds a user-specified threshold α. To make the routing boundary transparent, TRACER generates interpretability artifacts describing which input regions the surrogate handles, where it plateaus, and why it defers. On a 77-class intent benchmark with a Sonnet 4.6 teacher, TRACER achieves 83-100% surrogate coverage depending on the quality target α; on a 150-class benchmark, the surrogate fully replaces the teacher. On a natural language inference task, the parity gate correctly refuses deployment because the embedding representation cannot support reliable separation. The system is available as open-source software.

We study the problem of constructing concurrent objects in a setting where $P$ processes run in parallel and interact through a shared memory that is subject to write contention. Our goal is to transform hardware primitives that are subject to write contention into ones that handle contention gracefully. We give contention-resolution algorithms for several basic primitives, and analyze them under a relaxed, roughly-synchronous stochastic scheduler, where processes run at roughly the same rate up to a constant factor with high probability. Specifically, we construct read/write registers and CAS registers that have latency $O(\log P)$ w.h.p. under our scheduler model, using $O(1)$ hardware read/write registers and, in the case of our CAS construction, one hardware CAS register. Our algorithms guarantee performance even when their operations are invoked by an adaptive adversary that is able to see the entire history of operations so far, including their timing and return values. This allows them to be used as building blocks inside larger programs; using this compositionality property, we obtain several other constructions (LL/SC, fetch-and-increment, bounded max registers, and counters). To complement our constructions, we give a trade-off showing that even under a perfectly synchronous schedule and even if each process only executes one operation, any algorithm that implements any of the primitives that we consider, uses space $M$, and has latency at most $L$ with high probability must have expected latency at least $Ω(\log_{ML} P)$.

Large Language Models (LLMs) achieve strong performance through extended inference-time deliberation, yet how their reasoning failures arise remains poorly understood. By analyzing model-generated reasoning trajectories, we find that errors are not uniformly distributed but often originate from a small number of early transition points, after which reasoning remains locally coherent but globally incorrect. These transitions coincide with localized spikes in token-level entropy, and alternative continuations from the same intermediate state can still lead to correct solutions. Based on these observations, we introduce GUARD, a targeted inference-time framework that probes and redirects critical transitions using uncertainty signals. Empirical evaluations across multiple benchmarks confirm that interventions guided by these failure dynamics lead to more reliable reasoning outcomes. Our findings highlight the importance of understanding when and how reasoning first deviates, complementing existing approaches that focus on scaling inference-time computation.

Large language models frequently produce mutually inconsistent answers when reasoning over multiple related queries. We study case-file logical consistency: maintaining a globally satisfiable belief state across interdependent queries. We introduce a benchmark of 390 multi-query reasoning instances with entailment/contradiction/unknown labels and propose set-level metrics including Case Satisfiability Rate, Contradiction Density and Revision Cost. Our solver-augmented approach extracts commitments, verifies global satisfiability and performs counterexample-guided repair. Across four reasoning domains, our method substantially reduces cross-query contradictions (SetCons: 0.56 to 0.94) while preserving per-query accuracy, demonstrating that global coherence is critical for robust multi-query reasoning.

Catastrophic forgetting remains a fundamental challenge in continual learning, in which models often forget previous knowledge when fine-tuned on a new task. This issue is especially pronounced in class incremental learning (CIL), which is the most challenging setting in continual learning. Existing methods to address catastrophic forgetting often sacrifice either model interpretability or accuracy. To address this challenge, we introduce ClassIncremental Concept Bottleneck Model (CI-CBM), which leverage effective techniques, including concept regularization and pseudo-concept generation to maintain interpretable decision processes throughout incremental learning phases. Through extensive evaluation on seven datasets, CI-CBM achieves comparable performance to black-box models and outperforms previous interpretable approaches in CIL, with an average 36% accuracy gain. CICBM provides interpretable decisions on individual inputs and understandable global decision rules, as shown in our experiments, thereby demonstrating that human understandable concepts can be maintained during incremental learning without compromising model performance. Our approach is effective in both pretrained and non-pretrained scenarios; in the latter, the backbone is trained from scratch during the first learning phase. Code is publicly available at github.com/importAmir/CI-CBM.

We present \textbf{Mind DeepResearch (MindDR)}, an efficient multi-agent deep research framework that achieves leading performance with only \textasciitilde30B-parameter models through a meticulously designed data synthesis and multi-stage training pipeline. The core innovation of MindDR lies in a collaborative three-agent architecture (Planning Agent, DeepSearch Agent, and Report Agent) and a four-stage agent-specialized training pipeline comprising SFT cold-start, Search-RL, Report-RL and preference alignment. With this regime, MindDR demonstrates competitive performance even with \textasciitilde30B-scale models. Specifically, MindDR achieves 45.7\% on BrowseComp-ZH, 42.8\% on BrowseComp, 46.5\% on WideSearch, 75.0\% on xbench-DS, and 52.5 on DeepResearch Bench, outperforming comparable-scale open-source agent systems and rivaling larger-scale models. MindDR has been deployed as an online product in Li Auto. Furthermore, we introduce \textbf{MindDR Bench}, a curated benchmark of 500 real-world Chinese queries from our internal product user interactions, evaluated through a comprehensive multi-dimensional rubric system rather than relying on a single RACE metric. On MindDR Bench, MindDR achieves a state-of-the-art score of 51.8.

Healthcare disparities persist across socioeconomic boundaries, often attributed to unequal access to screening, diagnostics, and therapeutics. However, this perspective highlights that critical biases can emerge much earlier, during data collection and research prioritization, long before clinical implementation in cases where the focus of the studies and the data that is collected is at the molecular level. A vast number of studies focus on collecting omics data but the demographic information associated with these datasets is often not reported in the studies, and when it is reported, it shows big biases. An automated analysis of 4719 PubMed-indexed omics publications from 2015 to 2024 reveals that only a small fraction report ancestry or ethnicity information, with ancestry reporting improving slightly. Analysis of large-scale datasets commonly used for model training, such as CellxGene and GEO, reveals substantial population bias where European-ancestry data dominates. As biomedical foundation models become central to biomedical discovery with a paradigm in which base models are pretrained on large datasets and reusing them time and again for many different downstream tasks, they risk perpetuating or amplifying these early-stage biases, leading to cascading inequities that regulatory interventions cannot fully reverse. We propose a community-wide focus on three foundational principles: Provenance, Openness, and Evaluation Transparency to improve equity and robustness in biomedical AI. This approach aims to foster biomedical innovation that more effectively serves underserved populations and improves health outcomes.

Large Language Models (LLMs) are increasingly used in scientific peer review, assisting with drafting, rewriting, expansion, and refinement. However, existing peer-review LLM detection methods largely treat authorship as a binary problem-human vs. AI-without accounting for the hybrid nature of modern review workflows. In practice, evaluative ideas and surface realization may originate from different sources, creating a spectrum of human-AI collaboration. In this work, we introduce PeerPrism, a large-scale benchmark of 20,690 peer reviews explicitly designed to disentangle idea provenance from text provenance. We construct controlled generation regimes spanning fully human, fully synthetic, and multiple hybrid transformations. This design enables systematic evaluation of whether detectors identify the origin of the surface text or the origin of the evaluative reasoning. We benchmark state-of-the-art LLM text detection methods on PeerPrism. While several methods achieve high accuracy on the standard binary task (human vs. fully synthetic), their predictions diverge sharply under hybrid regimes. In particular, when ideas originate from humans but the surface text is AI-generated, detectors frequently disagree and produce contradictory classifications. Accompanied by stylometric and semantic analyses, our results show that current detection methods conflate surface realization with intellectual contribution. Overall, we demonstrate that LLM detection in peer review cannot be reduced to a binary attribution problem. Instead, authorship must be modeled as a multidimensional construct spanning semantic reasoning and stylistic realization. PeerPrism is the first benchmark evaluating human-AI collaboration in these settings. We release all code, data, prompts, and evaluation scripts to facilitate reproducible research at https://github.com/Reviewerly-Inc/PeerPrism.

Agent communication languages (ACLs) enable heterogeneous agents to share knowledge and coordinate across diverse domains. This diversity demands extensibility, but expressive extension mechanisms can push the input language beyond the complexity classes where full validation is tractable. We present CBCL (Common Business Communication Language), an agent communication language that constrains all messages, including runtime language extensions, to the deterministic context-free language (DCFL) class. CBCL allows agents to define, transmit, and adopt domain-specific "dialect" extensions as first-class messages; three safety invariants (R1--R3), machine-checked in Lean 4 and enforced in a Rust reference implementation, prevent unbounded expansion, applying declared resource limits, and preserving core vocabulary. We formalize the language and its safety properties in Lean 4, implement a reference parser and dialect engine in Rust with property-based and differential tests, and extract a verified parser binary. Our results demonstrate that homoiconic protocol design, where extension definitions share the same representation as ordinary messages, can be made provably safe. As autonomous agents increasingly extend their own communication capabilities, formally bounding what they can express to each other is a precondition for oversight.

News recommender systems are devised to alleviate the information overload, attracting more and more researchers' attention in recent years. The lack of a dedicated learner-oriented news recommendation toolkit hinders the advancement of research in news recommendation. We propose a PyTorch-based news recommendation toolkit called NewsTorch, developed to support learners in acquiring both conceptual understanding and practical experience. This toolkit provides a modular, decoupled, and extensible framework with a learner-friendly GUI platform that supports dataset downloading and preprocessing. It also enables training, validation, and testing of state-of-the-art neural news recommendation models with standardized evaluation metrics, ensuring fair comparison and reproducible experiments. Our open-source toolkit is released on Github: https://github.com/whonor/NewsTorch.

As a classic vision task, anomaly detection has been widely applied in industrial inspection and medical imaging. In this task, data scarcity is often a frequently-faced issue. To solve it, the few-shot anomaly detection (FSAD) scheme is attracting increasing attention. In recent years, beyond traditional visual paradigm, Vision-Language Model (VLM) has been extensively explored to boost this field. However, in currently-existing VLM-based FSAD schemes, almost all perform anomaly inference only by pairwise feature matching, ignoring structural dependencies and global consistency. To further redound to FSAD via VLM, we propose a Heterogeneous Hypergraph Vision-Language Reasoning (H2VLR) framework. It reformulates the FSAD as a high-order inference problem of visual-semantic relations, by jointly modeling visual regions and semantic concepts in a unified hypergraph. Experimental comparisons verify the effectiveness and advantages of H2VLR. It could often achieve state-of-the-art (SOTA) performance on representative industrial and medical benchmarks. Our code will be released upon acceptance.

We study the expressive power and limitations of multi-layer state-space models (SSMs). First, we show that multi-layer SSMs face fundamental limitations in compositional tasks, revealing an inherent gap between SSMs and streaming models. Then, we examine the role of chain-of-thought (CoT), showing that offline CoT does not fundamentally increase the expressiveness, while online CoT can substantially increase its power. Indeed, with online CoT, multi-layer SSMs become equivalent in power to streaming algorithms. Finally, we investigate the tradeoff between width and precision, showing that these resources are not interchangeable in the base model, but admit a clean equivalence once online CoT is allowed. Overall, our results offer a unified perspective on how depth, finite precision, and CoT shape the power and limits of SSMs.

Expert specialization is fundamental to Mixture-of-Experts (MoE) model success, yet existing metrics (cosine similarity, routing entropy) lack theoretical grounding and yield inconsistent conclusions under reparameterization. We present an information-geometric framework providing the first rigorous characterization of MoE specialization dynamics. Our key insight is that expert routing distributions evolve on the probability simplex equipped with the Fisher information metric, enabling formal analysis via Riemannian geometry. We prove that standard heuristic metrics violate parameterization invariance (Theorem 1), establish that specialization corresponds to geodesic flow with quantified approximation bounds (Theorem 2), and derive a failure predictor with theoretical threshold justification (Theorem 3). The framework introduces two principled metrics: Fisher Specialization Index (FSI) achieving r=0.91+/-0.02 correlation with downstream performance, and Fisher Heterogeneity Score (FHS) predicting training failure at 10% completion with AUC=0.89+/-0.03 -- outperforming validation-loss-based early stopping by 23% while requiring 40x fewer compute cycles. We validate intervention protocols achieving 87% recovery rate when FHS>1 is detected. Comprehensive experiments across language modeling (WikiText-103, C4), vision MoE (ImageNet), and scaling studies (8-64 experts, 125M-2.7B parameters) validate our theoretical predictions.

Synthetic augmentation is increasingly used to mitigate data scarcity in financial machine learning, yet its statistical role remains poorly understood. We formalize synthetic augmentation as a modification of the effective training distribution and show that it induces a structural bias--variance trade-off: while additional samples may reduce estimation error, they may also shift the population objective whenever the synthetic distribution deviates from regions relevant under evaluation. To isolate informational gains from mechanical sample-size effects, we introduce a size-matched null augmentation and a finite-sample, non-parametric block permutation test that remains valid under weak temporal dependence. We evaluate this framework in both controlled Markov-switching environments and real financial datasets, including high-frequency option trade data and a daily equity panel. Across generators spanning bootstrap, copula-based models, variational autoencoders, diffusion models, and TimeGAN, we vary augmentation ratio, model capacity, task type, regime rarity, and signal-to-noise. We show that synthetic augmentation is beneficial only in variance-dominant regimes, such as persistent volatility forecasting-while it deteriorates performance in bias-dominant settings, including near-efficient directional prediction. Rare-regime targeting can improve domain-specific metrics but may conflict with unconditional permutation inference. Our results provide a structural perspective on when synthetic data improves financial learning performance and when it induces persistent distributional distortion.

Financial institutions face tension between maximizing data utility and mitigating the re-identification risks inherent in traditional anonymization methods. This paper explores Differentially Private (DP) synthetic data as a robust "Privacy by Design" framework to resolve this conflict, ensuring output privacy while satisfying stringent regulatory obligations. We examine two distinct generative paradigms: Direct Tabular Synthesis, which reconstructs high-fidelity joint distributions from raw data, and DP-Seeded Agent-Based Modeling (ABM), which uses DP-protected aggregates to parameterize complex, stateful simulations. While tabular synthesis excels at reflecting static historical correlations for QA testing and business analytics, the DP-Seeded ABM offers a forward-looking "counterfactual laboratory" capable of modeling dynamic market behaviors and black swan events. By decoupling individual identities from data utility, these methodologies eliminate traditional data-clearing bottlenecks, enabling seamless cross-institutional research and compliant decision-making in an evolving regulatory landscape.

Deploying high-quality automatic speech recognition (ASR) on edge devices requires models that jointly optimize accuracy, latency, and memory footprint while operating entirely on CPU without GPU acceleration. We conduct a systematic empirical study of state-of-the-art ASR architectures, encompassing encoder-decoder, transducer, and LLM-based paradigms, evaluated across batch, chunked, and streaming inference modes. Through a comprehensive benchmark of over 50 configurations spanning OpenAI Whisper, NVIDIA Nemotron, Parakeet TDT, Canary, Conformer Transducer, and Qwen3-ASR, we identify NVIDIA's Nemotron Speech Streaming as the strongest candidate for real-time English streaming on resource-constrained hardware. We then re-implement the complete streaming inference pipeline in ONNX Runtime and conduct a controlled evaluation of multiple post-training quantization strategies, including importance-weighted k-quant, mixed-precision schemes, and round-to-nearest quantization, combined with graph-level operator fusion. These optimizations reduce the model from 2.47 GB to as little as 0.67 GB while maintaining word error rate (WER) within 1% absolute of the full-precision PyTorch baseline. Our recommended configuration, the int4 k-quant variant, achieves 8.20% average streaming WER across eight standard benchmarks, running comfortably faster than real-time on CPU with 0.56 s algorithmic latency, establishing a new quality-efficiency Pareto point for on-device streaming ASR.

Topic modeling seeks to uncover latent semantic structure in text corpora with minimal supervision. Neural approaches achieve strong performance but require extensive tuning and struggle with lifelong learning due to catastrophic forgetting and fixed capacity, while classical probabilistic models lack flexibility and adaptability to streaming data. We introduce \textsc{CobwebTM}, a low-parameter lifelong hierarchical topic model based on incremental probabilistic concept formation. By adapting the Cobweb algorithm to continuous document embeddings, \textsc{CobwebTM} constructs semantic hierarchies online, enabling unsupervised topic discovery, dynamic topic creation, and hierarchical organization without predefining the number of topics. Across diverse datasets, \textsc{CobwebTM} achieves strong topic coherence, stable topics over time, and high-quality hierarchies, demonstrating that incremental symbolic concept formation combined with pretrained representations is an efficient approach to topic modeling.

In any domain where knowledge accumulates under formal authority -- law, drug regulation, software security -- a later document can formally void an earlier one while remaining semantically distant from it. We formalize this as Controlling Authority Retrieval (CAR): recovering the active frontier front(cl(A_k(q))) of the authority closure of the semantic anchor set -- a different mathematical problem from argmax_d s(q,d). The two central results are: Theorem 4 (CAR-Correctness Characterization) gives necessary-and-sufficient conditions on any retrieved set R for TCA(R,q)=1 -- frontier inclusion and no-ignored-superseder -- independent of how R was produced. Proposition 2 (Scope Identifiability Upper Bound) establishes phi(q) as a hard worst-case ceiling: for any scope-indexed algorithm, TCA@k <= phi(q) * R_anchor(q), proved by an adversarial permutation argument. Three independent real-world corpora validate the proved structure: security advisories (Dense TCA@5=0.270, two-stage 0.975), SCOTUS overruling pairs (Dense=0.172, two-stage 0.926), FDA drug records (Dense=0.064, two-stage 0.774). A GPT-4o-mini experiment shows the downstream cost: Dense RAG produces explicit "not patched" claims for 39% of queries where a patch exists; Two-Stage cuts this to 16%. Four benchmark datasets, domain adapters, and a single-command scorer are released at https://github.com/andremir/car-retrieval.

Quantization is a natural complement to the sparse, event-driven computation of Spiking Neural Networks, reducing memory bandwidth and arithmetic cost for deployment on resource-constrained hardware. However, existing SNN quantization evaluation focuses almost exclusively on accuracy, overlooking whether a quantized network preserves the firing behavior of its full-precision counterpart. We demonstrate that quantization method, clipping range, and bit-width can produce substantially different firing distributions at equivalent accuracy, differences invisible to standard metrics but relevant to deployment, where firing activity governs effective sparsity, state storage, and event-processing load. To capture this gap, we propose Earth Mover's Distance as a diagnostic metric for firing distribution divergence, and apply it systematically across weight and membrane quantization on SEW-ResNet architectures trained on CIFAR-10 and CIFAR-100. We find that uniform quantization induces distributional drift even when accuracy is preserved, while LQ-Net style learned quantization maintains firing behavior close to the full-precision baseline. Our results suggest that behavior preservation should be treated as an evaluation criterion alongside accuracy, and that EMD provides a principled tool for assessing it.

Behavior cloning (BC) policies on position-controlled robots inherit the closed-loop response of the underlying PD controller, yet the effect of controller gains on BC failure lacks a nonasymptotic theory. We show that independent sub-Gaussian action errors propagate through the gain-dependent closed-loop dynamics to yield sub-Gaussian position errors whose proxy matrix $X_\infty(K)$ governs the failure tail. The probability of horizon-$T$ task failure factorizes into a gain-dependent amplification index $Γ_T(K)$ and the validation loss plus a generalization slack, so training loss alone cannot predict closed-loop performance. Under shape-preserving upper-bound structural assumptions the proxy admits the scalar bound $X_\infty(K)\preceqΨ(K)\bar X$ with $Ψ(K)$ decomposed into label difficulty, injection strength, and contraction, ranking the four canonical regimes with compliant-overdamped (CO) tightest, stiff-underdamped (SU) loosest, and the stiff-overdamped versus compliant-underdamped ordering system-dependent. For the canonical scalar second-order PD system the closed-form continuous-time stationary variance $X_\infty^{\mathrm{c}}(α,β)=σ^2α/(2β)$ is strictly monotone in stiffness and damping over the entire stable orthant, covering both underdamped and overdamped regimes, and the exact zero-order-hold (ZOH) discretization inherits this monotonicity. The analysis provides the first nonasymptotic explanation of the empirical finding that compliant, overdamped controllers improve BC success rates.

Transparency of neural networks' internal reasoning is at the heart of interpretability research, adding to trust, safety, and understanding of these models. The field of mechanistic interpretability has recently focused on studying task-specific computational graphs, defined by connections (edges) between model components. Such edge-based circuits have been defined in the context of large language models, yet vision-based approaches so far only consider neuron-based circuits. These tell which information is encoded, but not how it is routed through the complex wiring of a neural network. In this work, we investigate whether useful mechanistic circuits can be identified through computational graphs in vision transformers. We propose an effective method for Automatic Visual Circuit Discovery (Vi-CD) that recovers class-specific circuits for classification, identifies circuits underlying typographic attacks in CLIP, and discovers circuits that lend themselves for steering to correct harmful model behavior. Overall, we find that insightful and actionable edge-based circuits can be recovered from vision transformers, adding transparency to the internal computations of these models.

Tool-augmented large language model (LLM) agents can orchestrate specialist classifiers, segmentation models, and visual question-answering modules to interpret chest X-rays. However, these agents still solve each case in isolation: they fail to accumulate experience across cases, correct recurrent reasoning mistakes, or adapt their tool-use behavior without expensive reinforcement learning. While a radiologist naturally improves with every case, current agents remain static. In this work, we propose Evo-MedAgent, a self-evolving memory module that equips a medical agent with the capacity for inter-case learning at test time. Our memory comprises three complementary stores: (1)~\emph{Retrospective Clinical Episodes} that retrieve problem-solving experiences from similar past cases, (2)~an \emph{Adaptive Procedural Heuristics} bank curating priority-tagged diagnostic rules that evolves via reflection, much like a physician refining their internal criteria, and (3)~a \emph{Tool Reliability Controller} that tracks per-tool trustworthiness. On ChestAgentBench, Evo-MedAgent raises multiple-choice question (MCQ) accuracy from 0.68 to 0.79 on GPT-5-mini, and from 0.76 to 0.87 on Gemini-3 Flash. With a strong base model, evolving memory improves performance more effectively than orchestrating external tools on qualitative diagnostic tasks. Because Evo-MedAgent requires no training, its per-case overhead is bounded by one additional retrieval pass and a single reflection call, making it deployable on top of any frozen model.

Traditional esports scouting workflows rely heavily on manual video review and aggregate performance metrics, which often fail to capture the nuanced decision-making patterns necessary to determine if a prospect fits a specific tactical archetype. To address this, we reframe style-based player evaluation in esports as an Inverse Reinforcement Learning (IRL) problem. In this paper, we introduce a novel player selection framework that learns professional-specific reward functions from logged gameplay demonstrations, allowing organizations to rank candidates by their stylistic alignment with a target star player. Our proposed architecture utilizes a multimodal, two-branch intake: one branch encodes structured state-action trajectories derived from high-resolution in-game telemetry, while the second encodes temporally aligned tactical pseudo-commentary generated by Vision-Language Models (VLMs) from broadcast footage. These representations are fused and evaluated via a Generative Adversarial Imitation Learning (GAIL) objective, where a discriminator learns to capture the unique mechanical and tactical signatures of elite professionals. By transitioning from generic skill estimation to scouting "by reward," this framework provides a scalable, workflow-aware digital twin system that enables data-driven roster construction and targeted talent discovery across massive candidate pools.

A common approach to personalization in large language models (LLMs) is to incorporate a subset of the user memory into the prompt at inference time to guide the model's generation. Existing methods select these subsets primarily using similarity between user memory items and input queries, ignoring how features actually affect the model's response distribution. We propose Response-Utility optimization for Memory Selection (RUMS), a novel method that selects user memory items by measuring the mutual information between a subset of memory and the model's outputs, identifying items that reduce response uncertainty and sharpen predictions beyond semantic similarity. We demonstrate that this information-theoretic foundation enables more principled user memory selection that aligns more closely with human selection compared to state-of-the-art methods, and models $400\times$ larger. Additionally, we show that memory items selected using RUMS result in better response quality compared to existing approaches, while having up to $95\%$ reduction in computational cost.

Physics-informed neural networks (PINNs) are often selected by a single scalar loss even when the quantity of interest is more specific. We study a hybrid design in which the governing PDE residual remains automatic-differentiation (AD) based, while finite differences (FD) appear only in a weak auxiliary term that penalizes gradients of the sampled residual field. The FD term regularizes the residual field without replacing the PDE residual itself. We examine this idea in two stages. Stage 1 is a controlled Poisson benchmark comparing a baseline PINN, the FD residual-gradient regularizer, and a matched AD residual-gradient baseline. Stage 2 transfers the same logic to a three-dimensional annular heat-conduction benchmark (PINN3D), where baseline errors concentrate near a wavy outer wall and the auxiliary grid is implemented as a body-fitted shell adjacent to the wall. In Stage 1, the FD regularizer reproduces the main effect of residual-gradient control while exposing a trade-off between field accuracy and residual cleanliness. In Stage 2, the shell regularizer improves the application-facing quantities, namely outer-wall flux and boundary-condition behavior. Across seeds 0-5 and 100k epochs, the most reliable tested configuration is a fixed shell weight of 5e-4 under the Kourkoutas-beta optimizer regime: relative to a matched run without the shell term, it reduces the mean outer-wall BC RMSE from 1.22e-2 to 9.29e-4 and the mean wall-flux RMSE from 9.21e-3 to 9.63e-4. Adam with beta2=0.999 becomes usable when the initial learning rate is reduced to 1e-3, although its shell benefit is less robust than under Kourkoutas-beta. Overall, the results support a targeted view of hybrid PINNs: an auxiliary-only FD regularizer is most valuable when it is aligned with the physical quantity of interest, here the outer-wall flux.

AI systems are increasingly used to assist humans in sequential decision-making tasks, yet determining when and how an AI assistant should intervene remains a fundamental challenge. A potential baseline is to recommend the optimal action according to a strong model. However, such actions assume optimal follow-up actions, which human decision makers may fail to execute, potentially reducing overall performance. In this work, we propose and study value-aware interventions, motivated by a basic principle in reinforcement learning: under the Bellman equation, the optimal policy selects actions that maximize the immediate reward plus the value function. When a decision maker follows a suboptimal policy, this policy-value consistency no longer holds, creating discrepancies between the actions taken by the policy and those that maximize the immediate reward plus the value of the next state. We show that these policy-value inconsistencies naturally identify opportunities for intervention. We formalize this problem in a Markov decision process where an AI assistant may override human actions under an intervention budget. In the single-intervention regime, we show that the optimal strategy is to recommend the action that maximizes the human value function. For settings with multiple interventions, we propose a tractable approximation that prioritizes interventions based on the magnitude of the policy-value discrepancy. We evaluate these ideas in the domain of chess by learning models of humans from large-scale gameplay data. In simulation, our approach consistently outperforms interventions based on the strongest chess engine (Stockfish) in a wide range of settings. A within-subject human study with 20 players and 600 games further shows that our interventions significantly improve performance for low- and mid-skill players while matching expert-engine interventions for high-skill players.

Large language models (LLMs) emulate a consistent human-like behavior that can be shaped through activation-level interventions. This paradigm is converging on additive residual-stream injections, which rely on injection-strength sweeps to approximate optimal intervention settings. However, existing methods restrict the search space and sweep in uncalibrated activation-space units, potentially missing optimal intervention conditions. Thus, we introduce a psychological steering framework that performs unbounded, fluency-constrained sweeps in semantically calibrated units. Our method derives and calibrates residual-stream injections using psychological artifacts, and we use the IPIP-NEO-120, which measures the OCEAN personality model, to compare six injection methods. We find that mean-difference (MD) injections outperform Personality Prompting (P$^2$), an established baseline for OCEAN steering, in open-ended generation in 11 of 14 LLMs, with gains of 3.6\% to 16.4\%, overturning prior reports favoring prompting and positioning representation engineering as a new frontier in open-ended psychological steering. Further, we find that a hybrid of P$^2$ and MD injections outperforms both methods in 13 of 14 LLMs, with gains over P$^2$ ranging from 5.6\% to 21.9\% and from 3.3\% to 26.7\% over MD injections. Finally, we show that MD injections align with the Linear Representation Hypothesis and provide reliable, approximately linear control knobs for psychological steering. Nevertheless, they also induce OCEAN trait covariance patterns that depart from the Big Two model, suggesting a gap between learned representations and human psychology.

Neuromotor decoding from upper-limb electromyography (sEMG) can enhance human-machine interfaces and offer a more natural means of controlling prosthetic limbs, virtual reality, and household electronics. Unfortunately, current sEMG technology does not always perform consistently across users because individual differences such as age and body mass index, among many others, can substantially alter signal quality. This variability makes sEMG characteristics highly idiosyncratic, often necessitating laborious personalization and iterative tuning to achieve reliable performance. This variability has particular import for sEMG-based assistive devices and neural interfaces, where demographic biases in sEMG features could undermine broad and fair deployment. In this study, we explore how demographic differences affect the sEMG signals produced and their implications for machine learning-based gesture decoding. We analyze the data set provided by, in which we derive 147 common sEMG features extracted from 81 demographically diverse individuals performing discrete hand gestures. Using mixed-effects linear models and partial least squares (PLS) analysis, which take into consideration demographic variables (including age, sex, height, weight, skin properties, subcutaneous fat, and hair density), we identify that 33\% (49 of 147) of commonly used sEMG features show significant associations with demographic characteristics. These results may help guide the development of fair and unbiased sEMG-based neural interfaces across a diverse population.

For humans, filler-gap dependencies require a shared representation across different syntactic constructions. Although causal analyses suggest this may also be true for LLMs (Boguraev et al., 2025), it is still unclear if such a representation also exists for language models trained on developmentally feasible quantities of data. We applied Distributed Alignment Search (DAS, Geiger et al. (2024)) to LMs trained on varying amounts of data from the BabyLM challenge (Warstadt et al., 2023), to evaluate whether representations of filler-gap dependencies transfer between wh-questions and topicalization, which greatly vary in terms of their input frequency. Our results suggest shared, yet item-sensitive mechanisms may develop with limited training data. More importantly, LMs still require far more data than humans to learn comparable generalizations, highlighting the need for language-specific biases in models of language acquisition.

Visualizing narratives is useful to writers to reflect on unfinished drafts and identify unintentional biases and inconsistencies. Literary scholars can use the visualizations to identify nuanced patterns and literary styles from written text. Current narrative visualization is limited to representing character and location co-occurrences in a timeline, omitting important and complex narrative components such as focalization, causality, and speech. This paper aims to capture and visualize underexplored, complex narrative components as a basis for narrative visualization. As a starting point, we propose a new narrative visualization, named FocalLens, that uses focalization, the component that establishes who sees or perceives the events in a narrative, for representing the narrative. We provide the theoretical foundation of focalization and describe various types and facets of focalization. The details are incorporated in the novel visualization that captures how different characters perceive an event, who directly participate in an event, who indirectly observe the event, and who narrate the event. We also developed a tool that provides fluid interaction between the text and the proposed visualization. The tool was evaluated with four writers and scholars in a qualitative study, where writers analyzed their draft stories and scholars analyzed well-known stories. The findings suggest the tool added a new dimension to the workflow for writers and scholars, an analytical lens that is not available otherwise. We conclude by identifying design implications and future directions.

AI models underpin modern intelligent systems, driving advances across science, medicine, finance, and technology. Yet developing high-performing AI models remains a labor-intensive process that requires expert practitioners to iteratively design architectures, engineer representations, implement training pipelines and refine approaches through empirical evaluation. Existing AutoML methods partially alleviate this burden but remain limited to narrow aspects such as hyperparameter optimization and model selection within predefined search spaces, leaving the full development lifecycle largely dependent on human expertise. To address this gap, we introduce AIBuildAI, an AI agent that automatically builds AI models from a task description and training data. AIBuildAI adopts a hierarchical agent architecture in which a manager agent coordinates three specialized sub-agents: a designer for modeling strategy, a coder for implementation and debugging, and a tuner for training and performance optimization. Each sub-agent is itself a large language model (LLM) based agent capable of multi-step reasoning and tool use, enabling end-to-end automation of the AI model development process that goes beyond the scope of existing AutoML approaches. We evaluate AIBuildAI on MLE-Bench, a benchmark of realistic Kaggle-style AI development tasks spanning visual, textual, time-series and tabular modalities. AIBuildAI ranks first on MLE-Bench with a medal rate of 63.1%, outperforming all existing baseline methods and matching the capability of highly experienced AI engineers. These results demonstrate that hierarchical agent systems can automate the full AI model development process from task specification to deployable model, suggesting a pathway toward broadly accessible AI development with minimal human intervention.

Weak gravitational lensing, the correlated distortion of background galaxy shapes by foreground structures, is a powerful probe of the matter distribution in our universe and allows accurate constraints on the cosmological model. In recent years, high-order statistics and machine learning (ML) techniques have been applied to weak lensing data to extract the nonlinear information beyond traditional two-point analysis. However, these methods typically rely on cosmological simulations, which poses several challenges: simulations are computationally expensive, limiting most realistic setups to a low training data regime; inaccurate modeling of systematics in the simulations create distribution shifts that can bias cosmological parameter constraints; and varying simulation setups across studies make method comparison difficult. To address these difficulties, we present the first weak lensing benchmark dataset with several realistic systematics and launch the FAIR Universe Weak Lensing Machine Learning Uncertainty Challenge. The challenge focuses on measuring the fundamental properties of the universe from weak lensing data with limited training set and potential distribution shifts, while providing a standardized benchmark for rigorous comparison across methods. Organized in two phases, the challenge will bring together the physics and ML communities to advance the methodologies for handling systematic uncertainties, data efficiency, and distribution shifts in weak lensing analysis with ML, ultimately facilitating the deployment of ML approaches into upcoming weak lensing survey analysis.

Quick and accurate emergency handling in Disaster Decision Support Systems (DDSS) is often hampered by network latency and suboptimal application accuracy. While Federated Learning (FL) addresses some of these issues, it is constrained by high communication costs and rigid synchronization requirements across heterogeneous convolutional neural network (CNN) architectures. To overcome these challenges, this paper proposes a decentralized ensembling framework based on asynchronous probability aggregation and feedback distillation. By shifting the exchange unit from model weights to class-probability vectors, our method maintains data privacy, reduces communication requirements by orders of magnitude, and improves overall accuracy. This approach enables diverse CNN designs to collaborate asynchronously, enhancing disaster image identification performance even in resource-constrained settings. Experimental tests demonstrate that the proposed method outperforms traditional individual backbones and standard federated approaches, establishing a scalable and resource-aware solution for real-time disaster response.

Recent advances in data-centric artificial intelligence highlight inherent limitations in object recognition datasets. One of the primary issues stems from the semantic gap problem, which results in complex many-to-many mappings between visual data and linguistic descriptions. This bias adversely affects performance in computer vision tasks. This paper proposes an image annotation methodology that integrates knowledge representation, natural language processing, and computer vision techniques, aiming to reduce annotator subjectivity by applying visual property constraints. We introduce an interactive crowdsourcing framework that dynamically asks questions based on a predefined object category hierarchy and annotator feedback, guiding image annotation by visual properties. Experiments demonstrate the effectiveness of this methodology, and annotator feedback is discussed to optimize the crowdsourcing setup.

Large language models (LLMs) are increasingly used as sources of information, yet their reliability depends on the ability to search the web, select relevant evidence, and synthesize complete answers. While recent benchmarks evaluate web-browsing and agentic tool use, multilingual settings, and Portuguese in particular, remain underexplored. We present \textsc{MARCA}, a bilingual (English and Portuguese) benchmark for evaluating LLMs on web-based information seeking. \textsc{MARCA} consists of 52 manually authored multi-entity questions, paired with manually validated checklist-style rubrics that explicitly measure answer completeness and correctness. We evaluate 14 models under two interaction settings: a Basic framework with direct web search and scraping, and an Orchestrator framework that enables task decomposition via delegated subagents. To capture stochasticity, each question is executed multiple times and performance is reported with run-level uncertainty. Across models, we observe large performance differences, find that orchestration often improves coverage, and identify substantial variability in how models transfer from English to Portuguese. The benchmark is available at https://github.com/maritaca-ai/MARCA

Probabilistic computers offer promising solutions for computationally hard problems in domains such as combinatorial optimization and machine learning. A key building block in these systems is the probabilistic bit (p-bit), which relies on superparamagnetic tunnel junctions (sMTJs) as its source of randomness. A challenging threshold to cross for scaling sMTJ-based p-bit systems is integration of sMTJs with CMOS technology. In this work, we present experimental results of a p-bit unit cell using sMTJs integrated with 130 nm CMOS technology and demonstrate that the sMTJ's resistance fluctuations can generate a corresponding fluctuating digital output voltage which is tunable via the input voltage. These findings establish the feasibility of CMOS-compatible, sMTJ-based probabilistic circuits and mark a key step toward scalable hardware for real-world probabilistic computing applications.

The growing volume of data in scientific domains has made spatial query processing increasingly challenging due to high data transfer costs across the memory hierarchy and limited memory bandwidth. To address these bottlenecks and reduce the energy consumed on data movement, this work explores Processing-in-Memory (PIM) systems by executing range queries directly inside memory chips. Unlike prior PIM studies centered on linear scans or hash-based queries, this work is the first to map R-tree range queries onto commercial PIM hardware. The proposed broadcast-based method constructs the R-tree bottom-up on the CPU, broadcasts top levels to UPMEM DPUs (DRAM Processing Units) for global filtering, and distributes lower levels for parallel batched queries in a CPU-DPU system. We evaluate our approach on two real spatial datasets, Sports (999K rectangles) and Lakes (8.4M rectangles), and assess scalability using a synthetic dataset with up to 16M rectangles and 3.9M queries on a commercial UPMEM PIM system with up to 2,540 DPUs. Across all datasets, broadcast-based execution consistently outperforms subtree partitioning by preventing communication from dominating execution. On the Lakes dataset, strong scaling from 512 to 2,540 DPUs reduces kernel time from 64.9 s to 17.6 s, yielding up to 3.66x kernel and 2.70x end-to-end speedup relative to the CPU R-tree search on the same system. The PIM kernel also consumes approximately 3.4x less energy than the corresponding CPU search (e.g., 59.6 kJ vs. 167.0 kJ on Lakes), demonstrating scalable and energy-efficient hierarchical spatial range queries.

Ensuring the reliability of machine learning-based intrusion detection systems remains a critical challenge in Internet of Things (IoT) environments, particularly as data poisoning attacks increasingly threaten the integrity of model training pipelines. This study evaluates the susceptibility of four widely used classifiers, Random Forest, Gradient Boosting Machine, Logistic Regression, and Deep Neural Network models, against multiple poisoning strategies using three real-world IoT datasets. Results show that while ensemble-based models exhibit comparatively stable performance, Logistic Regression and Deep Neural Networks suffer degradation of up to 40% under label manipulation and outlier-based attacks. Such disruptions significantly distort decision boundaries, reduce detection fidelity, and undermine deployment readiness. The findings highlight the need for adversarially robust training, continuous anomaly monitoring, and feature-level validation within operational Network Intrusion Detection Systems. The study also emphasizes the importance of integrating resilience testing into regulatory and compliance frameworks for AI-driven IoT security. Overall, this work provides an empirical foundation for developing more resilient intrusion detection pipelines and informs future research on adaptive, attack-aware models capable of maintaining reliability under adversarial IoT conditions.

We present an empirical study of whether hierarchically structured, shared-weight recurrence can match the representational quality of independent-layer stacking in a Transformer-based language model. HRM-LM replaces L independent Transformer layers with a two-speed recurrent pair: a Fast module operating at every step for local refinement, and a Slow module operating every T steps for global compression. This recurrent hierarchy is unrolled for M = N x T steps with shared parameters. The central and most robust finding, supported by a parameter-matched Universal Transformer ablation (UniTF, 1.2B) across five independent runs, is a sharp empirical gap between the two approaches.

Large Language Models (LLMs) have achieved impressive results on public benchmarks, often leading to claims of advanced reasoning and understanding. However, recent research in cognitive science reveals that these models sometimes rely on shallow heuristics and memorization, taking shortcuts rather than demonstrating genuine cognitive abilities. This paper investigates LLM behavior in automated test generation for software, contrasting performance on an open-source system (LevelDB) with SAP HANA, one of the most widely deployed commercial database systems worldwide, whose proprietary codebase is guaranteed to be absent from training data. We combine cognitive evaluation principles, drawing on Mitchell's mechanism-focused assessment methodology, with empirical software testing, employing mutation score and iterative compiler-feedback repair loops to assess both accuracy and underlying reasoning strategies. Results show that LLMs excel on familiar, open-source benchmarks but struggle with unseen, complex domains, often prioritizing compilability over semantic effectiveness. These findings provide independent software engineering evidence for the broader claim that current LLMs lack robust reasoning, and highlight the need for evaluation frameworks that penalize trivial shortcuts and reward true generalization.

The Variational Quantum Linear Solver (VQLS), a hybrid quantum-classical algorithm for solving linear systems, faces a practical scalability bottleneck: the Linear Combination of Unitaries (LCU) decomposition requires O(L^2) circuit evaluations per optimizer iteration, where $L$ can grow as 4^n for n-qubit systems for the worst case scenario. We address this computational bottleneck through two complementary strategies. First, we present a distributed VQLS (D-VQLS) framework, built on NVIDIA CUDA-Q, that enables asynchronous, scalable distribution of the O(L^2) cost-function evaluations. Second, a fast Walsh--Hadamard transform (FWHT)-based Pauli decomposition with 1% coefficient thresholding curbs the exponential growth of LCU terms, reducing L from O}(2^n) to O(1) for n > 6 qubits and compressing the per-iteration circuit complexity from O(n * 4^n) to O(n) for sparse, structured matrices. For a 10-qubit tridiagonal Toeplitz system, this yields a 256x reduction, from 23 million to 90,112 circuits per iteration, while preserving over $99.99\%$ solution fidelity. Additionally, to inform feasibility on early fault-tolerant QPUs, the paper provides resource estimates -- gate counts, qubit requirements, and circuit evaluations per iteration -- for VQLS applied to arbitrary matrices. The D-VQLS framework is validated on the NERSC Perlmutter supercomputer using multi-node, multi-GPU ideal state-vector simulations, achieving over 99.99% fidelity against classical solutions on tridiagonal Toeplitz and Hele--Shaw flow benchmarks, with near-ideal strong scaling up to 24 GPUs and 95.3% weak scaling efficiency at 96 GPUs processing 360,448 circuits per iteration for a 10-qubit system. Systematic profiling identifies the optimal resource allocation for distributed quantum circuit workloads, yielding a 2.52x speedup for the configurations studied.

Sparse Mixture-of-Experts (MoE) models scale parameters while fixing active computation per token, but the specialization of individual experts remains opaque. In a companion paper we showed that routing topology is quality-neutral: five structurally different configurations converge to statistically equivalent language modeling quality. Here we show that expert identity is nonetheless causally meaningful: individual rank-1 experts are monosemantic by construction, and cosine-similarity routing in a low-dimensional metric space makes their specialization directly inspectable. We present four lines of evidence. First, projecting expert output vectors through the unembedding matrix yields a Semantic Dictionary: 15% of experts are monosemantic specialists spanning 10 categories (temporal, geographic, cardinal, discourse, emotional, financial, military, scientific). Second, routing exhibits a frequency-to-syntax gradient: early layers separate tokens by word frequency, deeper layers by syntactic class (Zipf-confound controls, all $p < 0.001$). Third, causal interventions confirm these labels: steering toward a temporal expert's centroid increases P(temporal) by +321% (median across 44 prompts); suppressing a geographic expert drops P(geographic) by -23%; rewriting an expert's output vector halves target-category probability, and effects compose additively across layers. Fourth, the interventions are not unique to cosine routing: linear routers support comparable steering, but only cosine routing provides geometric transparency -- expert specialization is readable directly from the centroid matrix. MoE expert-level specialization is a first-class interpretability primitive: architecturally monosemantic, causally validated, and controllable at inference with zero overhead.

Zero-ablation -- replacing token activations with zero vectors -- is widely used to probe token function in vision transformers. Register zeroing in DINOv2+registers and DINOv3 produces large drops (up to $-36.6$\,pp classification, $-30.9$\,pp segmentation), suggesting registers are functionally indispensable. However, three replacement controls -- mean-substitution, noise-substitution, and cross-image register-shuffling -- preserve performance across classification, correspondence, and segmentation, remaining within ${\sim}1$\,pp of the unmodified baseline. Per-patch cosine similarity shows these replacements genuinely perturb internal representations, while zeroing causes disproportionately large perturbations, consistent with why it alone degrades tasks. We conclude that zero-ablation overstates dependence on exact register content. In the frozen-feature evaluations we test, performance depends on plausible register-like activations rather than on exact image-specific values. Registers nevertheless buffer dense features from \texttt{[CLS]} dependence and are associated with compressed patch geometry. These findings, including the replacement-control results, replicate at ViT-B scale.

Mobile applications rely on complex backends that introduce significant security risks, yet developers often lack the tools to assess these risks effectively. This paper presents AndroScanner, an automated pipeline for detecting vulnerabilities in Android application backends through combined static and dynamic analysis. AndroScanner extracts backend API calls from APK files using apktool, Androguard, and Frida-based dynamic instrumentation, then vets them against the OWASP API Security Top 10 using APIFuzzer. We evaluate AndroScanner on two Android applications: a purposely vulnerable bank application and a production recruitment application with over 50,000 downloads on Google Play Store. Across both applications, AndroScanner extracted 24 APIs and identified 5 vulnerabilities, including a previously unreported zero-day Excessive Data Exposure vulnerability (ranked 3rd in the OWASP API Security Top 10) in the production application. The vulnerability was responsibly disclosed to the development team prior to publication. AndroScanner is available upon request to assist developers in identifying and remediating backend security risks before deployment.

We present Three-Phase Transformer (3PT), a residual-stream structural prior for decoder-only Transformers on a standard SwiGLU + RMSNorm + RoPE + GQA backbone. The hidden vector is partitioned into N equally-sized cyclic channels, each maintained by phase-respecting ops: a per-channel RMSNorm, a 2D Givens rotation between attention and FFN that rotates each channel by theta + i*(2*pi/N), and a head-count constraint aligning GQA heads with the partition. The architecture is a self-stabilizing equilibrium between scrambling and re-imposition, not a bolted-on module. The partition carves out a one-dimensional DC subspace orthogonal to the channels, into which we inject a fixed Gabriel's horn profile r(p) = 1/(p+1) as an absolute-position side-channel composing orthogonally with RoPE's relative-position rotation. The canonical N=3 borrows its metaphor from balanced three-phase AC, where three sinusoids 120 degrees apart sum to zero with no anti-correlated pair. At 123M parameters on WikiText-103, 3PT achieves -7.20% perplexity (-2.62% bits-per-byte) over a matched RoPE-Only baseline at +1,536 parameters (0.00124% of total), with 1.93x step-count convergence speedup (1.64x wall-clock). N behaves as a parameter-sharing knob rather than a unique optimum: at 5.5M an N-sweep over {1,2,3,4,6,8,12} is near-monotone with N=1 winning; at 123M a three-seed sweep finds N=3 and N=1 statistically indistinguishable. The load-bearing mechanism is the channel-partitioned residual stream, per-block rotation, per-phase normalization, and horn DC injection. We characterize (a) self-stabilization of the geometry without explicit enforcement, a novel instance of the conservation-law framework for neural networks; (b) a U-shaped depth profile of rotation-angle drift at 12 layers; (c) orthogonal composition with RoPE, attention, and FFN.

Most practical engineering design problems involve nonlinear spatio-temporal dynamical systems. Multi-physics simulations are often performed to capture the fine spatio-temporal scales which govern the evolution of these systems. However, these simulations are often high-fidelity in nature, and can be computationally very expensive. Hence, generating data from these expensive simulations becomes a bottleneck in an end-to-end engineering design process. Spatio-temporal surrogate modeling of these dynamical systems has been a popular data-driven solution to tackle this computational bottleneck. This is because accurate machine learning models emulating the dynamical systems can be orders of magnitude faster than the actual simulations. However, one key limitation of purely data-driven approaches is their lack of generalizability to inputs outside the training distribution. In this paper, we propose a physics-informed spatio-temporal surrogate modeling (PISTM) framework constrained by the physics of the underlying dynamical system. The framework leverages state-of-the-art advancements in the field of Koopman autoencoders to learn the underlying spatio-temporal dynamics in a non-intrusive manner, coupled with a spatio-temporal surrogate model which predicts the behavior of the Koopman operator in a specified time window for unknown operating conditions. We evaluate our framework on a prototypical fluid flow problem of interest: two-dimensional incompressible flow around a cylinder.

Data analysts working with relational data often start with vague or underspecified questions and refine them iteratively as they explore the data. To support this iterative process, we demonstrate Pneuma-Seeker, a system that reifies a user's information need as explicit, inspectable relational specifications, enabling iterative refinement of the information need, targeted data discovery, and provenance-aware execution. Through two real-world procurement use cases, we show how Pneuma-Seeker leverages LLMs as transparent, interactive analytical collaborators rather than opaque answer engines.

Sparse Mixture-of-Experts (MoE) architectures employ increasingly sophisticated routing mechanisms -- learned routers, multi-hop trajectories, token-dependent gating. We ask: does routing topology actually determine language modeling quality? We build a geometric MoE (ST-MoE) using cosine-similarity routing against learned centroids in a low-dimensional space ($d_{space} = 64$), requiring 80% fewer routing parameters than standard linear routers. Through 62 controlled experiments on WikiText-103 at 76--84M parameters trained to convergence (50K steps, 1.64B tokens), we find that routing topology does not determine asymptotic perplexity (PPL): five cosine-routing variants are statistically equivalent within a 1-PPL margin (Two One-Sided Tests [TOST], $p < 0.05$ for all 10 pairwise comparisons; 15 runs across 3 seeds, observed range 33.93--34.72). The finding extends to hash, random-fixed, and top-1 routing (single-seed; graceful 1.1--2.2 PPL degradation) and replicates on OpenWebText (0.03 PPL gap, 6 runs, 3 seeds each). A standard linear router with 5.3$\times$ more routing parameters reaches PPL 32.76, but iso-parameter cosine routing closes 67% of this gap -- the true mechanism advantage is $\sim$1.2%. The mechanistic explanation is convergent redundancy: multi-hop updates are collinear ($\cos(Δh_0, Δh_1) = 0.805$), implementing magnitude amplification rather than compositional reasoning; a single learnable scalar replicates multi-hop performance. As a practical payoff, zero-shot relative-norm halting saves 25% of MoE FLOPs at +0.12% PPL. Expert-level specialization and causal controllability -- which coexist with topology-level equifinality -- are explored in a companion paper.

Turn-level metrics are widely used to evaluate properties of multi-turn human-LLM conversations, from safety and sycophancy to dialogue quality. However, consecutive turns within a conversation are not statistically independent -- a fact that virtually all current evaluation pipelines fail to correct for in their statistical inference. We systematically characterize the autocorrelation structure of 66 turn-level metrics across 202 multi-turn conversations (11,639 turn pairs, 5 German-speaking users, 4 LLM platforms) and demonstrate that naive pooled analysis produces severely inflated significance estimates: 42% of associations that appear significant under standard pooled testing fail to survive cluster-robust correction. The inflation varies substantially across categories rather than scaling linearly with autocorrelation: three memoryless families (embedding velocity, directional, differential) aggregate to 14%, while the seven non-memoryless families (thermo-cycle, frame distance, lexical/structural, rolling windows, cumulative, interaction, timestamp) aggregate to 33%, with individual category rates ranging from 0% to 100% depending on per-family effect size. We present a two-stage correction framework combining Chelton (1983) effective degrees of freedom with conversation-level block bootstrap, and validate it on a pre-registered hold-out split where cluster-robust metrics replicate at 57% versus 30% for pooled-only metrics. We provide concrete design principles, a publication checklist, and open-source code for the correction pipeline. A survey of ~30 recent papers at major NLP and AI venues that compute turn-level statistics in LLM evaluations finds that only 4 address temporal dependence at all, and 26 do not correct for it.

Hypergraph partitioning is a pervasive NP-hard problem, and accelerating its computation on GPU can both slice time-to-solution and raise quality of results. In this work, we implement a multi-level hypergraph partitioning algorithm on GPU targeting a specific set of problem constraints: bounded per-partition size and distinct inbound hyperedges. Manipulating hypergraphs requires long orders of nested iterations, and enforcing these constraints introduces further set operations amidst them. Hence, we design algorithms around our problem's specifics, materializing the hypergraph's incidence structure in memory and exploiting set sparsity. Our results show competitive speedups as high as 940x and 2-26% better results in connectivity over a sequential multi-level partitioner.

Toxic interactions during code reviews can undermine teamwork and hinder productivity in software engineering (SE) teams. While prior studies explore toxicity detection and empirical investigation, they lack real-time detoxification tools to support the SE community. To address this gap, we present ToxiShield, a browser extension for GitHub pull requests that is built using three modules: i) Toxicity Filter -- to identify whether a text is toxic, ii) Communication coach -- to facilitate just-in-time fine-grained toxicity categorization with explanations, and iii) The Reframer -- that generates a revised, constructive alternative of a toxic text. For each module, we trained and evaluated multiple deep learning and Large Language Models (LLMs) to identify the best choice. A BERT-based binary detection model, trained on 38,761 code review samples, achieves 98% accuracy and an F1-score of 97% and is the selected one for the Toxicity Filter module. For the Communication Coach, prompt-tuned Claude 3.5 Sonnet achieved the best performance with 39% MCC and 42% F1 in multiclass toxicity classification with detailed reasoning. For Reframer, we evaluated five LLMs using a fine-tuning strategy on a dataset of 10,120 code review comments. The fine-tuned Llama 3.2 model achieves 95.27% style transfer accuracy, 97.03% fluency, 67.07% content preservation, and an 84% J-score. We further validated ToxiShield through a human evaluation using the Technology Acceptance Model with 10 participants, confirming its perceived usefulness and ease of adoption. ToxiShield sets a benchmark for advancing constructive communication in software engineering, driving inclusivity and healthier collaboration in open-source communities.

Agentic AI systems are becoming commonplace in domains that require long-lived, stateful decision-making in continuously evolving conditions. As such, correctness depends not only on the output of individual model calls, but also on how to best adapt when incorporating new evidence or revising prior conclusions. However, existing frameworks rely on imperative control loops, ephemeral memory, and prompt-embedded logic, making agent behavior opaque, brittle, and difficult to verify. This paper introduces Credo, which represents semantic state as beliefs and regulates behavior using declarative policies defined over these beliefs. This design supports adaptive, auditable, and composable execution through a database-backed semantic control plane. We showcase these concepts in a decision-control scenario, where beliefs and policies declaratively guide critical execution choices (e.g., model selection, retrieval, corrective re-execution), enabling dynamic behavior without requiring any changes to the underlying pipeline code.

Autonomous on-orbit servicing demands embodied agents that perceive through visual sensors, reason about 3D spatial situations, and execute multi-phase tasks over extended horizons. We present SpaceMind, a modular and self-evolving vision-language model (VLM) agent framework that decomposes knowledge, tools, and reasoning into three independently extensible dimensions: skill modules with dynamic routing, Model Context Protocol (MCP) tools with configurable profiles, and injectable reasoning-mode skills. An MCP-Redis interface layer enables the same codebase to operate across simulation and physical hardware without modification, and a Skill Self-Evolution mechanism distills operational experience into persistent skill files without model fine-tuning. We validate SpaceMind through 192 closed-loop runs across five satellites, three task types, and two environments, a UE5 simulation and a physical laboratory, deliberately including degraded conditions to stress-test robustness. Under nominal conditions all modes achieve 90--100% navigation success; under degradation, the Prospective mode uniquely succeeds in search-and-approach tasks where other modes fail. A self-evolution study shows that the agent recovers from failure in four of six groups from a single failed episode, including complete failure to 100% success and inspection scores improving from 12 to 59 out of 100. Real-world validation confirms zero-code-modification transfer to a physical robot with 100% rendezvous success. Code: https://github.com/wuaodi/SpaceMind

Rotating detonation engines (RDEs) are a promising propulsion concept that may offer higher thermodynamic efficiency and specific impulse than conventional systems, but nonlinear phenomena, including transitions to oscillatory or chaotic propagation modes, can hinder practical operation. Deep Reinforcement Learning (DRL) has emerged as a promising method for controlling complex nonlinear dynamics such as those observed in RDEs. However, the multi-timescale nature of the RDE system makes direct application of DRL challenging. We address this challenge by reformulating the DRL problem in a moving reference frame that follows the detonation-wave pattern, making the wave structure appear quasi-steady to the agent. This reformulation enables scale separation between fast detonation propagation and slower operating-mode dynamics. We train DRL controllers to modulate spatially segmented injection pressure in a one-dimensional reduced-order RDE model and induce rapid transitions between different mode-locked states. Across a range of actuation periods, initial states, and target modes, controllers trained in the moving frame learn more reliably than those trained in a stationary frame and remain effective over a broader range of actuation periods. These results suggest that symmetry-aware moving reference frame formulations may be useful for related multiscale flow-control problems and that scale separation should be exploited whenever possible to enable DRL control of multi-timescale systems.

We study the task of automatically expanding WordNet-style lexical resources to new languages through sense generation. We generate senses by associating target-language lemmas with existing lexical concepts via semantic projection. Given a sense-tagged English corpus and its translation, our method projects English synsets onto aligned target-language tokens and assigns the corresponding lemmas to those synsets. To generate these alignments and ensure their quality, we augment a pre-trained base aligner with a bilingual dictionary, which is also used to filter out incorrect sense projections. We evaluate the method on multiple languages, comparing it to prior methods, as well as dictionary-based and large language model baselines. Results show that the proposed project-and-filter strategy improves precision while remaining interpretable and requiring few external resources. We plan to make our code, documentation, and generated sense inventories accessible.

Automated fact-checking in dialogue involves multi-turn conversations where colloquial language is frequent yet understudied. To address this gap, we propose a conservative rewrite candidate for each response claim via staged de-colloquialisation, combining lightweight surface normalisation with scoped in-claim coreference resolution. We then introduce BiCon-Gate, a semantics-aware consistency gate that selects the rewrite candidate only when it is semantically supported by the dialogue context, otherwise falling back to the original claim. This gated selection stabilises downstream fact-checking and yields gains in both evidence retrieval and fact verification. On the DialFact benchmark, our approach improves retrieval and verification, with particularly strong gains on SUPPORTS, and outperforms competitive baselines, including a decoder-based one-shot LLM rewrite that attempts to perform all de-colloquialisation steps in a single pass.

Large Language Model (LLM) agents are increasingly deployed in multi-agent systems requiring strategic coordination. While recent work has analyzed LLM behavior in two-player games, coalition formation, where $n$ agents dynamically form cooperative groups, remains theoretically uncharacterized. We present the first framework grounding coalition formation in LLM agent networks in hedonic game theory with formal stability guarantees. We introduce the LLM Coalition Formation Game (LCFG), establish sufficient conditions for Nash-stable partitions, and prove complexity results. Our analysis reveals that LLM agents exhibit bounded rationality characterized by $ε$-rational preferences; we provide both deterministic existence guarantees and consistency-driven stability bounds whose predictions are consistent with empirical outcomes. Experiments with GPT-4, Claude-3, and Llama-3 across 2,400 episodes validate our framework: LLM coalitions achieve Nash stability in 73.2% of cases under our Coalition-of-Thought (CoalT) protocol, compared to 58.4% under chain-of-thought and 41.8% under standard prompting ($p < 0.001$). Our framework provides theoretical foundations for designing stable multi-agent LLM systems.

Reinforcement learning (RL) has emerged as a powerful tool for aligning diffusion models with human preferences, typically by optimizing a single reward function under a KL regularization constraint. In practice, however, human preferences are inherently pluralistic, and aligned models must balance multiple downstream objectives, such as aesthetic quality and text-image consistency. Existing multi-objective approaches either rely on costly multi-objective RL fine-tuning or on fusing separately aligned models at denoising time, but they generally require access to reward values (or their gradients) and/or introduce approximation error in the resulting denoising objectives. In this paper, we revisit the problem of RL fine-tuning for diffusion models and address the intractability of identifying the optimal policy by introducing a step-level RL formulation. Building on this, we further propose Multi-objective Step-level Denoising-time Diffusion Alignment (MSDDA), a retraining-free framework for aligning diffusion models with multiple objectives, obtaining the optimal reverse denoising distribution in closed form, with mean and variance expressed directly in terms of single-objective base models. We prove that this denoising-time objective is exactly equivalent to the step-level RL fine-tuning, introducing no approximation error. Moreover, we provide numerical results, which indicate our method outperforms existing denoising-time approaches.

Catastrophic forgetting remains a primary hurdle in sequential task learning for artificial neural networks. We propose a silicon-native modular architecture that achieves structural parameter isolation using Task-Specific Experts and a distributed, outlier-based Gatekeeper. Moving beyond traditional sequential consolidation, our framework utilizes a Simultaneous Pipeline where Teacher learning, Student distillation, and Router manifold acquisition occur in parallel while raw data is present in a localized training session. This approach ensures computational efficiency and complies with privacy mandates like GDPR by deleting raw data as soon as a task is learned. We demonstrate that a Tight-Bottleneck Autoencoder (TB-AE) can effectively distinguish semantically crowded manifolds in high-dimensional latent spaces, overcoming the posterior collapse inherent to standard variational methods. By establishing strict topological boundaries, our TB-AE resolves latent space crowding in 4096-D LLM embeddings to provide a robust, unsupervised novelty signal. Furthermore, we validate an Autonomous Retrieval mechanism that confidently identifies returning manifolds, enabling stable lifelong learning without redundant module instantiation. Empirical results demonstrate that our ``Live Distillation'' approach acts as a natural regularizer, achieving strong retention across computer vision and natural language processing domains without suffering a student fidelity gap.

Rural environmental risks are shaped by place-based conditions (e.g., housing quality, road access, land-surface patterns), yet standard vulnerability indices are coarse and provide limited insight into risk contexts. We propose SatBLIP, a satellite-specific vision-language framework for rural context understanding and feature identification that predicts county-level Social Vulnerability Index (SVI). SatBLIP addresses limitations of prior remote sensing pipelines-handcrafted features, manual virtual audits, and natural-image-trained VLMs-by coupling contrastive image-text alignment with bootstrapped captioning tailored to satellite semantics. We use GPT-4o to generate structured descriptions of satellite tiles (roof type/condition, house size, yard attributes, greenery, and road context), then fine-tune a satellite-adapted BLIP model to generate captions for unseen images. Captions are encoded with CLIP and fused with LLM-derived embeddings via attention for SVI estimation under spatial aggregation. Using SHAP, we identify salient attributes (e.g., roof form/condition, street width, vegetation, cars/open space) that consistently drive robust predictions, enabling interpretable mapping of rural risk environments.

AI tools increasingly guide targeted interventions in healthcare, education, and recruiting. Algorithms score individuals, trigger outreach to those above a threshold (e.g., high-risk or high-value), and encourage them to request service; then providers deliver service to those who request. Standard practice sets the threshold and selects the algorithm to maximize predictive accuracy, assuming that better predictions yield better outcomes. We show that this approach is suboptimal when limited service capacity and probabilistic behavioral responses influence who receives service. In such settings, the optimal score threshold must balance two effects: ensuring all capacity is filled (utilization) and ensuring high-value individuals are served despite competition between requests (cannibalization). We characterize the optimal threshold and prove that policies based solely on predictive accuracy are generally suboptimal. Further, because optimal thresholds vary with service capacity, algorithm selection metrics like AUC, which weight all thresholds equally, are misaligned with operational performance. We introduce a new metric--Operational AUC (OpAUC)--and show it leads to optimal algorithm selection. Finally, we conduct a case study on sepsis early warning data and illustrate the magnitude of improvement that can be achieved from improved threshold and algorithm selection.

Multimodal language models systematically underperform on visual perception tasks, yet the structure underlying this failure remains poorly understood. We propose centroid replacement, collapsing each token to its nearest K-means centroid, as a controlled probe for modal dependence. Across seven models spanning three architecture families, erasing text centroid structure costs 4$\times$ more accuracy than erasing visual centroid structure, exposing a universal imbalance where language representations overshadow vision even on tasks that demand visual reasoning. We exploit this asymmetry through text centroid contrastive decoding, recovering up to +16.9% accuracy on individual tasks by contrastively decoding against a text-centroid-erased reference. This intervention varies meaningfully with training approaches: standard fine-tuned models show larger gains (+5.6% on average) than preference-optimized models (+1.5% on average). Our findings suggest that modal competition is structurally localized, correctable at inference time without retraining, and quantifiable as a diagnostic signal to guide future multimodal training.

Large language models still struggle with reliable long-term conversational memory: simply enlarging context windows or applying naive retrieval often introduces noise and destabilizes responses. We present APEX-MEM, a conversational memory system that combines three key innovations: (1) a property graph which uses domain-agnostic ontology to structure conversations as temporally grounded events in an entity-centric framework, (2) append-only storage that preserves the full temporal evolution of information, and (3) a multi-tool retrieval agent that understands and resolves conflicting or evolving information at query time, producing a compact and contextually relevant memory summary. This retrieval-time resolution preserves the full interaction history while suppressing irrelevant details. APEX-MEM achieves 88.88% accuracy on LOCOMO's Question Answering task and 86.2% on LongMemEval, outperforming state-of-the-art session-aware approaches and demonstrating that structured property graphs enable more temporally coherent long-term conversational reasoning.

Resilience in cyber-physical systems (CPS) is the fundamental ability to maintain safety and critical functionality despite adverse "perturbations," which includes security attacks, environmental disruptions, and hardware or software failures. This survey provides a comprehensive review of CPS resilience, framing the field through five interconnected themes that are required in an integrated whole to achieve real-world resilience. The article first posits that resilience is a system-wide property emerging from interactions between hardware, software, and human users. Second, it addresses the challenges of learning-enabled CPS, which often operate in data-scarce environments characterized by imbalanced or noisy data, requiring innovative solutions like synthetic data generation and foundation model adaptation. Third, the survey examines proactive measures for resilience, which include distinctive aspects of verification, testing, and redundancy. Fourth, it explores recovery mechanisms, moving beyond traditional fault models to design "just good enough" recovery strategies that prioritize safety-critical functions during perturbations. Finally, it highlights the central role of the human, focusing on the different levels of human intervention, the necessity of trust calibration, and the requirement for explainable AI to support human-CPS teaming. These themes are illustrated through representative application domains, primarily Connected and Autonomous Transportation Systems (CATS) and Medical CPS (MCPS). By integrating the five interconnected themes, this survey provides a systematic roadmap for achieving the resilient CPS in increasingly complex and adversarial environments.

Women with polycystic ovary syndrome (PCOS) face substantially elevated risks of body image distress, disordered eating, and metabolic challenges, yet existing natural language processing approaches for detecting these conditions lack transparency and cannot identify co-occurring presentations. We developed small, open-source language models to automatically detect this triple burden in social media posts with grounded explainability. We collected 1,000 PCOS-related posts from six subreddits, with two trained annotators labeling posts using guidelines operationalizing Lee et al. (2017) clinical framework. Three models (Gemma-2-2B, Qwen3-1.7B, DeepSeek-R1-Distill-Qwen-1.5B) were fine-tuned using Low-Rank Adaptation to generate structured explanations with textual evidence. The best model achieved 75.3 percent exact match accuracy on 150 held-out posts, with robust comorbidity detection and strong explainability. Performance declined with diagnostic complexity, indicating their best use is for screening rather than autonomous diagnosis.

Online A/B testing at scale relies on proxy metrics -- short-term, easily-measured signals used in place of slow-moving long-term outcomes. When the proxy-outcome relationship is heterogeneous across user segments, aggregate correlation can mask directional failures akin to Simpson's Paradox, leading to costly ship/no-ship errors. We introduce PROXIMA (Proxy Metric Validation Framework for Online Experiments), a lightweight diagnostic framework that scores proxy reliability through a composite of three complementary dimensions: normalised effect correlation, directional accuracy, and segment-level fragility rate. Unlike surrogate-index approaches that predict long-term treatment effects, PROXIMA directly audits whether a candidate proxy leads to correct launch decisions and flags the user segments where it fails. We validate PROXIMA on two public datasets -- the Criteo Uplift corpus (14M observations, advertising) and KuaiRec (7K users, video recommendation) -- using 80 simulated A/B tests. Early engagement metrics achieve a composite reliability of 0.80 on Criteo and 0.62 on KuaiRec, yielding 98.4% average decision agreement with an oracle policy. Fragility analysis reveals that recommendation domains exhibit substantially higher segment-level heterogeneity (68% fragility) than advertising (13%), yet directional accuracy remains above 96% in both cases. A sensitivity analysis over the weight space confirms that no single component suffices and that the composite provides substantially better discrimination between reliable and unreliable proxies than correlation alone. Code and reproduction scripts are available at: https://github.com/Avinash-Amudala/PROXIMA

As search depth increases in autonomous reasoning and embodied planning, the candidate action space expands exponentially, heavily taxing computational budgets. While heuristic pruning is a common countermeasure, it operates without formal safety guarantees when surrogate models (like LLMs) exhibit systematic evaluation biases. This paper frames the node expansion process as a localized Best-Arm Identification (BAI) problem over dynamic frontiers, subject to a bounded systematic bias $L$. By inverting the Lambert W function, we establish an additive sample complexity of $\mathcal{O}((Δ-4L)^{-2})$, which indicates that safe node elimination is only feasible when the empirical reward gap exceeds $4L$. We complement this with an information-theoretic lower bound of $Ω((Δ-2L)^{-2})$ to confirm the structural limits of biased search. Subsequent evaluations on both synthetic trees and complex reasoning tasks demonstrate that adhering to this local safety boundary successfully preserves optimal trajectories while maximizing sample allocation efficiency.

Large language models (LLMs) increasingly operate in settings that require reliable long-context understanding, such as retrieval-augmented generation and multi-document reasoning. A common strategy is to fine-tune pretrained short-context models at the target sequence length. However, we find that standard long-context adaptation can remain brittle: model accuracy depends strongly on the absolute placement of relevant evidence, exhibiting high positional variance even when controlling for task format and difficulty. We propose RoPE-Perturbed Self-Distillation, a training regularizer that improves positional robustness. The core idea is to form alternative "views" of the same training sequence by perturbing its RoPE indices -- effectively moving parts of the context to different positions -- and to train the model to produce consistent predictions across views via self-distillation. This encourages reliance on semantic signals instead of brittle position dependencies. Experiments on long-context adaptation of Llama-3-8B and Qwen-3-4B demonstrate consistent gains on long-context benchmarks, including up to 12.04% improvement on RULER-64K for Llama-3-8B and 2.71% on RULER-256K for Qwen-3-4B after SFT, alongside improved length extrapolation beyond the training context window.

We introduce path-sampled integrated gradients (PS-IG), a framework that generalizes feature attribution by computing the expected value over baselines sampled along the linear interpolation path. We prove that PS-IG is mathematically equivalent to path-weighted integrated gradients, provided the weighting function matches the cumulative distribution function of the sampling density. This equivalence allows the stochastic expectation to be evaluated via a deterministic Riemann sum, improving the error convergence rate from $O(m^{-1/2})$ to $O(m^{-1})$ for smooth models. Furthermore, we demonstrate analytically that PS-IG functions as a variance-reducing filter against gradient noise - strictly lowering attribution variance by a factor of 1/3 under uniform sampling - while preserving key axiomatic properties such as linearity and implementation invariance.

Synaptic plasticity is metabolically expensive, yet animals continuously update their internal models without exhausting energy reserves. However, when artificial neural networks are trained, the network parameters are typically updated on every sample that is presented, even if the sample was classified correctly. Inspired by the human negativity bias and error-related negativity, we propose 'memorized mistake-gated learning' -- a biologically plausible plasticity rule where synaptic updates are strictly gated by current and past classification errors. This reduces the number of updates the network needs to make by $50\%\sim80\%$. Mistake gating is particularly well suited in two cases: 1) For incremental learning where new knowledge is acquired on a background of pre-existing knowledge, 2) For online learning scenarios when data needs to be stored for later replay, as mistake-gating reduces storage buffer requirements. The algorithm can be implemented in a few lines of code, adds no hyper-parameters, and comes at negligible computational overhead. Learning on mistakes is an energy efficient and biologically relevant modification to commonly used learning rules that is well suited for continual learning.

Gradient saliency from deep sequence models surfaces candidate biomarkers efficiently, but the resulting gene lists are contaminated by tissue-composition confounders that degrade downstream classifiers. We study whether LLM chain-of-thought (CoT) reasoning can faithfully filter these confounders, and whether reasoning quality drives downstream performance. We train a Mamba SSM on TCGA-BRCA RNA-seq and extract the top-50 genes by gradient saliency; DeepSeek-R1 evaluates every candidate with structured CoT to produce a final 17-gene set. The raw 50-gene saliency set (no LLM) performs worse than a 5,000-gene variance baseline (AUC 0.832 vs. 0.903), while the LLM-filtered set surpasses it (AUC 0.927), using 294x fewer features. A faithfulness audit (COSMIC CGC, OncoKB, PAM50) reveals only 6 of 17 selected genes (35.3%) are validated BRCA biomarkers, yet 10 of 16 known BRCA genes in the input were missed - including FOXA1. This gap between downstream performance and reasoning faithfulness suggests selective faithfulness: targeted confounder removal is sufficient for performance gains even without comprehensive recall.

Key Opinion Leader (KOL) discourse on social media is widely consumed as investment guidance, yet turning it into executable trading strategies without injecting assumptions about unspecified execution decisions remains an open problem. We observe that the gaps in KOL statements are not random deficiencies but a structured separation: KOLs express directional intent (what to buy or sell and why) while leaving execution decisions (when, how much, how long) systematically unspecified. Building on this observation, we propose an intent-preserving policy completion framework that treats KOL discourse as a partial trading policy and uses offline reinforcement learning to complete the missing execution decisions around the KOL-expressed intent. Experiments on multimodal KOL discourse from YouTube and X (2022-2025) show that KICL achieves the best return and Sharpe ratio on both platforms while maintaining zero unsupported entries and zero directional reversals, and ablations confirm that the full framework yields an 18.9% return improvement over the KOL-aligned baseline.

Diffusion-model inference and overdamped Langevin dynamics are formally identical. A physical substrate that encodes the score function therefore equilibrates to the correct output by thermodynamics alone, requiring no digital arithmetic during inference and potentially achieving a $10{,}000\times$ reduction in energy relative to a GPU. Two fundamental barriers have until now prevented this equivalence from being realized at production scale: non-local skip connections, which locally coupled analog substrates cannot represent, and input conditioning, in which the coupling constants carry roughly $2{,}600\times$ too little signal to anchor the system to a specific input. We resolve both obstacles. \emph{Hierarchical bilinear coupling} encodes U-Net skip connections as rank-$k$ inter-module interactions derived directly from the singular structure of the encoder and decoder Gram matrices, requiring only $O(Dk)$ physical connections instead of $O(D^2)$. A \emph{minimal digital interface} -- a 4-dimensional bottleneck encoder together with a 16-unit transfer network, totalling \textbf{2,560 parameters} -- overcomes the conditioning barrier. When evaluated on activations drawn from a trained denoising U-Net, the complete system attains a decoder cosine similarity of \textbf{0.9906} against an oracle upper bound of 1.0000, while preserving theoretical net energy savings of approximately $10^7\times$ over GPU inference. These results constitute the first demonstration of trained-weight, production-scale thermodynamic diffusion inference.

Applying kernel methods to matchings is challenging due to their discrete, non-Euclidean nature. In this paper, we develop a principled framework for constructing geometric kernels that respect the natural geometry of the space of matchings. To this end, we first provide a complete characterization of stationary kernels, i.e. kernels that respect the inherent symmetries of this space. Because the class of stationary kernels is too broad, we specifically focus on the heat and Matérn kernel families, adding an appropriate inductive bias of smoothness to stationarity. While these families successfully extend widely popular Euclidean kernels to matchings, evaluating them naively incurs a prohibitive super-exponential computational cost. To overcome this difficulty, we introduce and analyze a novel, sub-exponential algorithm leveraging zonal polynomials for efficient kernel evaluation. Finally, motivated by the known bijective correspondence between matchings and phylogenetic trees-a crucial data modality in biology-we explore whether our framework can be seamlessly transferred to the space of trees, establishing novel negative results and identifying a significant open problem.

Large language models (LLMs) achieve strong performance and have revolutionized NLP, but their lack of explainability keeps them treated as black boxes, limiting their use in domains that demand transparency and trust. A promising direction to address this issue is post-hoc text-based explanations, which aim to explain model decisions in natural language. Prior work has focused on generating convincing rationales that appear to be subjectively faithful, but it remains unclear whether these explanations are epistemically faithful, whether they reflect the internal evidence the model actually relied on for its decision. In this paper, we first assess the epistemic faithfulness of LLM-generated explanations via counterfactuals and show that they are often unfaithful. We then introduce a training-free method that enhances faithfulness by guiding explanation generation through attention-level interventions, informed by token-level heatmaps extracted via a faithful attribution method. This method significantly improves epistemic faithfulness across multiple models, benchmarks, and prompts.

Large language models (LLMs) often exhibit hallucinations due to their inability to accurately perceive their own knowledge boundaries. Existing abstention fine-tuning methods typically partition datasets directly based on response accuracy, causing models to suffer from severe label noise near the decision boundaries and consequently exhibit high rates of abstentions or hallucinations. This paper adopts a latent space representation perspective, revealing a "gray zone" near the decision hyperplane where internal belief ambiguity constitutes the core performance bottleneck. Based on this insight, we propose the **GeoDe** (**Geo**metric **De**noising) framework for abstention fine-tuning. This method constructs a truth hyperplane using linear probes and performs "geometric denoising" by employing geometric distance as a confidence signal for abstention decisions. This approach filters out ambiguous boundary samples while retaining high-fidelity signals for fine-tuning. Experiments across multiple models (Llama3, Qwen3) and benchmark datasets (TriviaQA, NQ, SciQ, SimpleQA) demonstrate that GeoDe significantly enhances model truthfulness and demonstrates strong generalization in out-of-distribution (OOD) scenarios. Code is available at https://github.com/Notbesidemoon/GeoDe.

We derive a robust update rule for the online infinite hidden Markov model (iHMM) for when the streaming data contains outliers and the model is misspecified. Leveraging recent advances in generalised Bayesian inference, we define robustness via the posterior influence function (PIF), and provide conditions under which the online iHMM has bounded PIF. Imposing robustness inevitably induces an adaptation lag for regime switching. Our method, which is called Batched Robust iHMM (BR-iHMM), balances adaptivity and robustness with two additional tunable parameters. Across limit order book data, hourly electricity demand, and a synthetic high-dimensional linear system, BR-iHMM reduces one-step-ahead forecasting error by up to 67% relative to competing online Bayesian methods. Together with theoretical guarantees of bounded PIF, our results highlight the practicality of our approach for both forecasting and interpretable online learning.

We tasked GPT-4.1 to read what baseball fans wrote about their game-day experience and predict the overall experience rating each fan gave on a 0-10 survey scale. The model received only the text of a single open-ended response. These AI predictions were compared with the actual experience ratings captured by the survey instrument across approximately 10,000 fan responses from five Major League Baseball teams. In total two-thirds of predicted ratings fell within one point of self-reported fan ratings (67% within +/-1, 36% exact match), and the predicted measurement was near-deterministic across three independent scoring runs (87% exact agreement, 99.9% within +/-1). Predicted ratings aligned most strongly with the overall experience rating (r = 0.82) rather than with any specific aspect of the game-day experience such as parking, concessions, staff, etc. However, predictions were systematically lower than self-reported ratings by approximately one point, and this gap was not driven by any single aspect. Rather, our analysis shows that self-reported ratings capture the fan's verdict, an overall evaluative judgment that integrates the entire experience. While predicted ratings quantify the impact of salient moments characterized as memorable, emotionally intense, unusual, or actionable. Each measure contains information the other misses. These baseline results establish that a simple, unoptimized prompt can directionally predict how fans rate their experience from the text a fan wrote and that a gap between the two numbers can be interpreted as a construct difference worth preserving rather than an error to eliminate.

Agentic systems built on large language models (LLMs) are increasingly being used for complex security tasks, including binary reverse engineering (RE). Despite recent growth in popularity and capability, these systems continue to face limitations in realistic settings. Cutting-edge systems still fail in complex RE scenarios that involve obfuscation, timing, and unique architecture. In this work, we examine how agentic systems perform reverse engineering tasks with static, dynamic, and hybrid agents. Through an analysis of existing agentic tool usage, we identify several limitations, including token constraints, struggles with obfuscation, and a lack of program guardrails. From these findings, we outline current challenges and position future directions for system designers to overcome from a security perspective.

Large scale vision language models have shown promise in automating chest Xray interpretation, yet their clinical utility remains limited by a gap between model outputs and radiologist reasoning. Most systems optimize for semantic information without emulating how experts visually examine medical images, often overlooking critical findings or diverging from established diagnostic workflows. Radiologists follow structured protocols (e.g., the ABCDEF approach) that ensure all clinically relevant regions are systematically examined, reducing missed findings and supporting reliable diagnostic reasoning. We introduce GazeX, a vision language model that leverages radiologists' eye tracking data as a behavioral prior to model expert diagnostic reasoning. By incorporating gaze trajectories and fixation patterns into pretraining, GazeX learns to follow the spatial and temporal structure of radiologist attention and integrates observations in a clinically meaningful sequence. Using a curated dataset of over 30,000 gaze key frames from five radiologists, we demonstrate that GazeX produces more accurate, interpretable, and expert consistent outputs across radiology report generation, disease grounding, and visual question answering, utilizing 231,835 radiographic studies, 780,014 question answer pairs, and 1,162 image sentence pairs with bounding boxes. Unlike autonomous reporting systems, GazeX produces verifiable evidence artifacts, including inspection trajectories and finding linked localized regions, enabling efficient human verification and safe human AI collaboration. Learning through expert eyes provides a practical route toward more trustworthy, explainable, and diagnostically robust AI systems for radiology and beyond.

The modern news cycle has been fundamentally reshaped by the rapid exchange of information online. As a result, media framing shifts dynamically as new information, political responses, and social reactions emerge. Understanding how these narratives form, propagate, and evolve is essential for interpreting public discourse during moments of crisis. In this study, we examine the temporal and semantic dynamics of reporting for violent and catastrophic events using a large-scale corpus of 126,602 news articles collected from online publishers. We quantify narrative change through publication volume, semantic drift, semantic dispersion, and term relevance. Our results show that sudden events of impact exhibit structured and predictable news-cycle patterns characterized by rapid surges in coverage, early semantic drift, and gradual declines toward the baseline. In addition, our results indicate the terms that are driving the temporal patterns.

This manuscript introduces DharmaOCR Full and Lite, a pair of specialized small language models (SSLMs) for structured OCR that jointly optimize transcription quality, generation stability, and inference cost. It also presents DharmaOCR-Benchmark, a benchmark that covers printed, handwritten, and legal/administrative documents, and proposes a unified evaluation protocol that measures fidelity and structure while explicitly tracking text degeneration as a first-class benchmark metric (alongside unit cost). Beyond reporting degeneration rates, the manuscript empirically shows degeneration is not merely a quality failure, since it materially worsens production performance by increasing response time, reducing throughput, and inflating computational cost due to abnormally long generations. To the best of the author's knowledge, as a methodological contribution, this is the first application of Direct Preference Optimization (DPO) for OCR, explicitly using degenerate generations as rejected examples to penalize looping behavior. Combined with Supervised Fine-Tuning (SFT) for enforcing a strict JSON schema (header, margin, footer, and text), DPO consistently reduces degeneration rate across model families (up to 87.6% relative) while preserving or improving extraction quality. The resulting models, namely, DharmaOCR Full (7B) and DharmaOCR Lite (3B), set a new state-of-the-art on DharmaOCR-Benchmark, outperforming each open-source and commercial baseline model evaluated regarding extraction quality, reaching 0.925 and 0.911 scores with 0.40% and 0.20% degeneration rates. AWQ quantization reduced up to 22% per-page cost with negligible quality loss, enabling a strong quality-cost trade-off in comparison to proprietary OCR APIs and open-source alternatives.

To address high data traffic demands of sixth-generation (6G) networks, this paper proposes a novel architecture that integrates autonomous aerial vehicles (AAVs) and multi-functional reconfigurable intelligent surfaces (MF-RISs) as AM-RIS in fluid antenna (FA)-assisted full-duplex (FD) networks. The AM-RIS provides hybrid functionalities, including signal reflection, amplification, and energy harvesting (EH), potentially improving both signal coverage and sustainability. Meanwhile, FA facilitates fine-grained spatial adaptability at FD-enabled base station (BS), which complements residual self-interference (SI) suppression. We aim at maximizing the overall energy efficiency (EE) by jointly optimizing transmit DL beamforming at BS, UL user power, configuration of AM-RIS, and positions of the FA and AM-RIS. Owing to the hybrid continuous-discrete parameters and high dimensionality of the intractable problem, we have conceived a self-optimized multi-agent hybrid deep reinforcement learning (DRL) framework (SOHRL), which integrates multi-agent deep Q-networks (DQN) and multi-agent proximal policy optimization (PPO), respectively handling discrete and continuous actions. To enhance self-adaptability, an attention-driven state representation and meta-level hyperparameter optimization are incorporated, enabling multi-agents to autonomously adjust learning hyperparameters. Simulation results validate the effectiveness of the proposed AM-RIS-enabled FA-aided FD networks empowered by SOHRL algorithm. The results reveal that SOHRL outperforms benchmarks of the case without attention mechanism and conventional hybrid/multi-agent/standalone DRL. Moreover, AM-RIS in FD achieves the highest EE compared to half-duplex, conventional rigid antenna arrays, partial EH, and conventional RIS without amplification, highlighting its potential as a compelling solution for EE-aware wireless networks.

While Large Language Models (LLMs) have demonstrated high proficiency on English-centric medical examinations, their performance often declines when faced with non-English languages and multimodal diagnostic tasks. This study protocol describes the development of EuropeMedQA, the first comprehensive, multilingual, and multimodal medical examination dataset sourced from official regulatory exams in Italy, France, Spain, and Portugal. Following FAIR data principles and SPIRIT-AI guidelines, we describe a rigorous curation process and an automated translation pipeline for comparative analysis. We evaluate contemporary multimodal LLMs using a zero-shot, strictly constrained prompting strategy to assess cross-lingual transfer and visual reasoning. EuropeMedQA aims to provide a contamination-resistant benchmark that reflects the complexity of European clinical practices and fosters the development of more generalizable medical AI.

Targeted amplicon panels are widely used in oncology diagnostics, but providing per-gene performance guarantees for copy number variant (CNV) detection remains challenging due to amplification artifacts, process-mismatch heterogeneity, and limited validation sample sizes. While Bayesian CNV callers naturally quantify per-sample uncertainty, translating this into the frequentist population-level guarantees required for clinical validation, coverage rates, false-positive bounds, and minimum detectable copy-number changes, is a fundamentally different inferential problem. We show empirically that even robust Bayesian credible intervals, including coarsened posteriors and sandwich-adjusted intervals, are severely miscalibrated on panels with small amplicon counts per gene. To address this, we propose a hybrid framework that evaluates Bayesian posterior functionals on validation samples and models the resulting squared losses with a Gamma distribution, yielding tolerance intervals with valid frequentist coverage. Three components make the method practical under real-world constraints: (1) imputation that removes the influence of true CNV-positive samples without requiring known ground truth, (2) regularization to address small sample variability, and (3) evidence-based stratification on the log model evidence to accommodate non-exchangeable noise profiles arising from process mismatch. Evaluated on two targeted amplicon panels using leave-one-out cross-validation, the proposed method achieves single-digit mean absolute coverage error across all genes under both process-matched and unmatched conditions, whereas Bayesian comparators exhibit mean absolute errors exceeding 60\% on clinically relevant genes such as ERBB2.

Tensor networks were developed in the context of many-body physics as compressed representations of multiparticle quantum states. These representations mitigate the exponential complexity of many-body systems by capturing only the most relevant dependencies. Due to the formal similarity between quantum entanglement and statistical correlations, tensor networks have recently been integrated in machine learning, operating both as alternative learning architectures and as decompositions of components of neural networks. The expectation is that the theoretical understanding of tensor networks developed within quantum many-body physics leads to novel methods that offer advantages in terms of computational efficiency, explainability, or privacy. Here we review the use of tensor networks in the context of machine learning, providing a critical assessment of the state of the art, the potential advantages, and the challenges that must be overcome.

Spatial reasoning over three-dimensional scenes is a core capability for embodied intelligence, yet continuous model improvement remains bottlenecked by the cost of geometric annotation. The self-evolving paradigm offers a promising path, but its reliance on model consensus to construct pseudo-labels causes training to reinforce rather than correct the model's own geometric errors. We identify a property unique to 3D spatial reasoning that circumvents this limitation: ground truth is a deterministic consequence of the underlying geometry, computable exactly from point clouds and camera poses without any model involvement. Building on this insight, we present SpatialEvo, a self-evolving framework for 3D spatial reasoning, centered on the Deterministic Geometric Environment (DGE). The DGE formalizes 16 spatial reasoning task categories under explicit geometric validation rules and converts unannotated 3D scenes into zero-noise interactive oracles, replacing model consensus with objective physical feedback. A single shared-parameter policy co-evolves across questioner and solver roles under DGE constraints: the questioner generates physically valid spatial questions grounded in scene observations, while the solver derives precise answers against DGE-verified ground truth. A task-adaptive scheduler endogenously concentrates training on the model's weakest categories, producing a dynamic curriculum without manual design. Experiments across nine benchmarks demonstrate that SpatialEvo achieves the highest average score at both 3B and 7B scales, with consistent gains on spatial reasoning benchmarks and no degradation on general visual understanding.

While reinforcement learning with verifiable rewards (RLVR) significantly enhances LLM reasoning by optimizing the conditional distribution P(y|x), its potential is fundamentally bounded by the base model's existing output distribution. Optimizing the marginal distribution P(y) in the Pre-train Space addresses this bottleneck by encoding reasoning ability and preserving broad exploration capacity. Yet, conventional pre-training relies on static corpora for passive learning, leading to a distribution shift that hinders targeted reasoning enhancement. In this paper, we introduce PreRL (Pre-train Space RL), which applies reward-driven online updates directly to P(y). We theoretically and empirically validate the strong gradient alignment between log P(y) and log P(y|x), establishing PreRL as a viable surrogate for standard RL. Furthermore, we uncover a critical mechanism: Negative Sample Reinforcement (NSR) within PreRL serves as an exceptionally effective driver for reasoning. NSR-PreRL rapidly prunes incorrect reasoning spaces while stimulating endogenous reflective behaviors, increasing transition and reflection thoughts by 14.89x and 6.54x, respectively. Leveraging these insights, we propose Dual Space RL (DSRL), a Policy Reincarnation strategy that initializes models with NSR-PreRL to expand the reasoning horizon before transitioning to standard RL for fine-grained optimization. Extensive experiments demonstrate that DSRL consistently outperforms strong baselines, proving that pre-train space pruning effectively steers the policy toward a refined correct reasoning subspace.

As language models are increasingly deployed for complex autonomous tasks, their ability to reason accurately over longer horizons becomes critical. An essential component of this ability is planning and managing a long, complex chain-of-thought (CoT). We introduce LongCoT, a scalable benchmark of 2,500 expert-designed problems spanning chemistry, mathematics, computer science, chess, and logic to isolate and directly measure the long-horizon CoT reasoning capabilities of frontier models. Problems consist of a short input with a verifiable answer; solving them requires navigating a graph of interdependent steps that span tens to hundreds of thousands of reasoning tokens. Each local step is individually tractable for frontier models, so failures reflect long-horizon reasoning limitations. At release, the best models achieve <10% accuracy (GPT 5.2: 9.8%; Gemini 3 Pro: 6.1%) on LongCoT, revealing a substantial gap in current capabilities. Overall, LongCoT provides a rigorous measure of long-horizon reasoning, tracking the ability of frontier models to reason reliably over extended periods.

Evaluating LLMs is challenging, as benchmark scores often fail to capture models' real-world usefulness. Instead, users often rely on ``vibe-testing'': informal experience-based evaluation, such as comparing models on coding tasks related to their own workflow. While prevalent, vibe-testing is often too ad hoc and unstructured to analyze or reproduce at scale. In this work, we study how vibe-testing works in practice and then formalize it to support systematic analysis. We first analyze two empirical resources: (1) a survey of user evaluation practices, and (2) a collection of in-the-wild model comparison reports from blogs and social media. Based on these resources, we formalize vibe-testing as a two-part process: users personalize both what they test and how they judge responses. We then introduce a proof-of-concept evaluation pipeline that follows this formulation by generating personalized prompts and comparing model outputs using user-aware subjective criteria. In experiments on coding benchmarks, we find that combining personalized prompts and user-aware evaluation can change which model is preferred, reflecting the role of vibe-testing in practice. These findings suggest that formalized vibe-testing can serve as a useful approach for bridging benchmark scores and real-world experience.

We consider the block withholding attacks on pools, more specifically the state-of-the-art Power Adjusting Withholding (PAW) attack. We propose a generalization called Temporary PAW (T-PAW) where the adversary withholds a fPoW from pool mining at most $T$-time even when no other block is mined. We show that PAW attack corresponds to $T\to\infty$ and is not optimal. In fact, the extra reward of T-PAW compared to PAW improves by an unbounded factor as adversarial hash fraction $α$, pool size $β$ and adversarial network influence $γ$ decreases. For example, the extra reward of T-PAW is 22 times that of PAW when an adversary targets a pool with $(α,β,γ)=(0.05,0.05,0)$. We show that honest mining is sub-optimal to T-PAW even when there is no difficulty adjustment and the adversarial revenue increase is non-trivial, e.g., for most $(α,β)$ at least $1\%$ within $2$ weeks in Bitcoin even when $γ=0$ (for PAW it was at most $0.01\%$). Hence, T-PAW exposes a significant structural weakness in pooled mining-its primary participants, small miners, are not only contributors but can easily turn into potential adversaries with immediate non-trivial benefits.

Rhetorical questions are asked not to seek information but to persuade or signal stance. How large language models internally represent them remains unclear. We analyze rhetorical questions in LLM representations using linear probes on two social-media datasets with different discourse contexts, and find that rhetorical signals emerge early and are most stably captured by last-token representations. Rhetorical questions are linearly separable from information-seeking questions within datasets, and remain detectable under cross-dataset transfer, reaching AUROC around 0.7-0.8. However, we demonstrate that transferability does not simply imply a shared representation. Probes trained on different datasets produce different rankings when applied to the same target corpus, with overlap among the top-ranked instances often below 0.2. Qualitative analysis shows that these divergences correspond to distinct rhetorical phenomena: some probes capture discourse-level rhetorical stance embedded in extended argumentation, while others emphasize localized, syntax-driven interrogative acts. Together, these findings suggest that rhetorical questions in LLM representations are encoded by multiple linear directions emphasizing different cues, rather than a single shared direction.

While end-to-end Vision-Language-Action (VLA) models offer a promising paradigm for robotic manipulation, fine-tuning them on narrow control data often compromises the profound reasoning capabilities inherited from their base Vision-Language Models (VLMs). To resolve this fundamental trade-off, we propose HiVLA, a visual-grounded-centric hierarchical framework that explicitly decouples high-level semantic planning from low-level motor control. In high-level part, a VLM planner first performs task decomposition and visual grounding to generate structured plans, comprising a subtask instruction and a precise target bounding box. Then, to translate this plan into physical actions, we introduce a flow-matching Diffusion Transformer (DiT) action expert in low-level part equipped with a novel cascaded cross-attention mechanism. This design sequentially fuses global context, high-resolution object-centric crops and skill semantics, enabling the DiT to focus purely on robust execution. Our decoupled architecture preserves the VLM's zero-shot reasoning while allowing independent improvement of both components. Extensive experiments in simulation and the real world demonstrate that HiVLA significantly outperforms state-of-the-art end-to-end baselines, particularly excelling in long-horizon skill composition and the fine-grained manipulation of small objects in cluttered scenes.

LLM reasoning traces suffer from complex flaws -- *Step Internal Flaws* (logical errors, hallucinations, etc.) and *Step-wise Flaws* (overthinking, underthinking), which vary by sample. A natural approach would be to provide ground-truth labels to guide LLMs' reasoning. Contrary to intuition, we show that this yields no improvement in reasoning ability. We then propose CRAFT, a unified framework that mitigates both types of Step flaws, which builds a Reasoning Knowledge Graph (RKG) based on the consensus parts of multiple candidate traces, and synthesizes a high-quality trace through topological generation. Our approach improves label-prediction accuracy by 10+% on average, and consistently outperforms all baselines across both logical and mathematical reasoning benchmarks. Further, detailed benchmark evaluation proves that our method also improves the quality of LLMs' reasoning traces in multiple dimensions.

Given two symmetric positive-definite matrices $A, B \in \mathbb{R}^{n \times n}$, we study the spectral properties of the interpolation $A^{1-x} B^x$ for $0 \leq x \leq 1$. The presence of `common structures' in $A$ and $B$, eigenvectors pointing in a similar direction, can be investigated using this interpolation perspective. Generically, exact log-linearity of the operator norm $\|A^{1-x} B^x\|$ is equivalent to the existence of a shared eigenvector in the original matrices; stability bounds show that approximate log-linearity forces principal singular vectors to align with leading eigenvectors of both matrices. These results give rise to and provide theoretical justification for a multi-manifold learning framework that identifies common and distinct latent structures in multiview data.

While Large Language Models (LLMs) have empowered AI research agents to perform isolated scientific tasks, automating complex, real-world workflows, such as LLM training, remains a significant challenge. In this paper, we introduce TREX, a multi-agent system that automates the entire LLM training life-cycle. By orchestrating collaboration between two core modules-the Researcher and the Executor-the system seamlessly performs requirement analysis, open-domain literature and data research, formulation of training strategies, preparation of data recipes, and model training and evaluation. The multi-round experimental process is modeled as a search tree, enabling the system to efficiently plan exploration paths, reuse historical results, and distill high-level insights from iterative trials. To evaluate the capability of automated LLM training, we construct FT-Bench, a benchmark comprising 10 tasks derived from real-world scenarios, ranging from optimizing fundamental model capabilities to enhancing performance on domain-specific tasks. Experimental results demonstrate that the TREX agent consistently optimizes model performance on target tasks.

Search agents extend Large Language Models (LLMs) beyond static parametric knowledge by enabling access to up-to-date and long-tail information unavailable during pretraining. While reinforcement learning has been widely adopted for training such agents, existing approaches face key limitations: process supervision often suffers from unstable value estimation, whereas outcome supervision struggles with credit assignment due to sparse, trajectory-level rewards. To bridge this gap, we propose Contribution-Weighted GRPO (CW-GRPO), a framework that integrates process supervision into group relative policy optimization. Instead of directly optimizing process rewards, CW-GRPO employs an LLM judge to assess the retrieval utility and reasoning correctness at each search round, producing per-round contribution scores. These scores are used to rescale outcome-based advantages along the trajectory, enabling fine-grained credit assignment without sacrificing optimization stability. Experiments on multiple knowledge-intensive benchmarks show that CW-GRPO outperforms standard GRPO by 5.0\% on Qwen3-8B and 6.3\% on Qwen3-1.7B, leading to more effective search behaviors. Additional analysis reveals that successful trajectories exhibit concentrated contributions across rounds, providing empirical insight into search agent tasks.

Sequential recommendation has become increasingly prominent in both academia and industry, particularly in e-commerce. The primary goal is to extract user preferences from historical interaction sequences and predict items a user is likely to engage with next. Recent advances have leveraged contrastive learning and graph neural networks to learn more expressive representations from interaction histories -- graphs capture relational structure between nodes, while ID-based representations encode item-specific information. However, few studies have explored multi-view contrastive learning between ID and graph perspectives to jointly improve user and item representations, especially in settings where only interaction data is available without auxiliary information. To address this gap, we propose Multi-View Contrastive learning for sequential recommendation (MVCrec), a framework that integrates complementary signals from both sequential (ID-based) and graph-based views. MVCrec incorporates three contrastive objectives: within the sequential view, within the graph view, and across views. To effectively fuse the learned representations, we introduce a multi-view attention fusion module that combines global and local attention mechanisms to estimate the likelihood of a target user purchasing a target item. Comprehensive experiments on five real-world benchmark datasets demonstrate that MVCrec consistently outperforms 11 state-of-the-art baselines, achieving improvements of up to 14.44\% in NDCG@10 and 9.22\% in HitRatio@10 over the strongest baseline. Our code and datasets are available at https://github.com/sword-Lz/MMCrec.

GUI grounding, which localizes interface elements from screenshots given natural language queries, remains challenging for small icons and dense layouts. Test-time zoom-in methods improve localization by cropping and re-running inference at higher resolution, but apply cropping uniformly across all instances with fixed crop sizes, ignoring whether the model is actually uncertain on each case. We propose \textbf{UI-Zoomer}, a training-free adaptive zoom-in framework that treats both the trigger and scale of zoom-in as a prediction uncertainty quantification problem. A confidence-aware gate fuses spatial consensus among stochastic candidates with token-level generation confidence to selectively trigger zoom-in only when localization is uncertain. When triggered, an uncertainty-driven crop sizing module decomposes prediction variance into inter-sample positional spread and intra-sample box extent, deriving a per-instance crop radius via the law of total variance. Extensive experiments on ScreenSpot-Pro, UI-Vision, and ScreenSpot-v2 demonstrate consistent improvements over strong baselines across multiple model architectures, achieving gains of up to +13.4\%, +10.3\%, and +4.2\% respectively, with no additional training required.

Large Language Models (LLMs) are now capable of generating highly fluent, human-like text. They enable many applications, but also raise concerns such as large scale spam, phishing, or academic misuse. While much work has focused on detecting LLM-generated text, only limited work has gone into understanding the stylistic differences between human-written and machine-generated text. In this work, we perform a large scale analysis of stylistic variation across human-written text and outputs from 11 LLMs spanning 8 different genres and 4 decoding strategies using Douglas Biber's set of lexicogrammatical and functional features. Our findings reveal insights that can guide intentional LLM usage. First, key linguistic differentiators of LLM-generated text seem robust to generation conditions (e.g., prompt settings to nudge them to generate human-like text, or availability of human-written text to continue the style); second, genre exerts a stronger influence on stylistic features than the source itself; third, chat variants of the models generally appear to be clustered together in stylistic space, and finally, model has a larger effect on the style than decoding strategy, with some exceptions. These results highlight the relative importance of model and genre over prompting and decoding strategies in shaping the stylistic behavior of machine-generated text.

Recent work suggests that (stochastic) gradient descent self-organizes near an instability boundary, shaping both optimization and the solutions found. Momentum and mini-batch gradients are widely used in practical deep learning optimization, but it remains unclear whether they operate in a comparable regime of instability. We demonstrate that SGD with momentum exhibits an Edge of Stochastic Stability (EoSS)-like regime with batch-size-dependent behavior that cannot be explained by a single momentum-adjusted stability threshold. Batch Sharpness (the expected directional mini-batch curvature) stabilizes in two distinct regimes: at small batch sizes it converges to a lower plateau $2(1-β)/η$, reflecting amplification of stochastic fluctuations by momentum and favoring flatter regions than vanilla SGD; at large batch sizes it converges to a higher plateau $2(1+β)/η$, where momentum recovers its classical stabilizing effect and favors sharper regions consistent with full-batch dynamics. We further show that this aligns with linear stability thresholds and discuss the implications for hyperparameter tuning and coupling.

We study behavior-regularized reinforcement learning (RL), where regularization toward a reference distribution (the dataset in offline RL or the base model in LLM RL finetuning) is essential to prevent value over-optimization caused by erroneous out-of-distribution extrapolation. Existing methods either rely on reparameterized policy gradient, which are difficult to scale to large generative models, or on reject sampling, which can be overly conservative when attempting to move beyond the behavior support. In this paper, we propose Value Gradient Flow (VGF), a scalable new paradigm for behavior-regularized RL. VGF casts behavior-regularized RL as an optimal transport problem that maps the reference distribution to the value-induced optimal policy distribution. We solve this transport problem via discrete gradient flow, where value gradients guide particles initialized from the reference distribution. Our analysis shows that VGF imposes regularization implicitly by controlling the transport budget. VGF eliminates explicit policy parameterization while remaining expressive and flexible, this enables adaptive test-time scaling by adjusting the transport budget. Extensive experiments demonstrate that VGF significantly outperforms prior methods, achieving state-of-the-art results on offline RL benchmarks (D4RL, OGBench) and LLM RL tasks. Code and runs can be found at https://ryanxhr.github.io/vgf.

Post-training adaptation of language models is commonly achieved through parameter updates or input-based methods such as fine-tuning, parameter-efficient adaptation, and prompting. In parallel, a growing body of work modifies internal activations at inference time to influence model behavior, an approach known as steering. Despite increasing use, steering is rarely analyzed within the same conceptual framework as established adaptation methods. In this work, we argue that steering should be regarded as a form of model adaptation. We introduce a set of functional criteria for adaptation methods and use them to compare steering approaches with classical alternatives. This analysis positions steering as a distinct adaptation paradigm based on targeted interventions in activation space, enabling local and reversible behavioral change without parameter updates. The resulting framing clarifies how steering relates to existing methods, motivating a unified taxonomy for model adaptation.

We present UMI-3D, a multimodal extension of the Universal Manipulation Interface (UMI) for robust and scalable data collection in embodied manipulation. While UMI enables portable, wrist-mounted data acquisition, its reliance on monocular visual SLAM makes it vulnerable to occlusions, dynamic scenes, and tracking failures, limiting its applicability in real-world environments. UMI-3D addresses these limitations by introducing a lightweight and low-cost LiDAR sensor tightly integrated into the wrist-mounted interface, enabling LiDAR-centric SLAM with accurate metric-scale pose estimation under challenging conditions. We further develop a hardware-synchronized multimodal sensing pipeline and a unified spatiotemporal calibration framework that aligns visual observations with LiDAR point clouds, producing consistent 3D representations of demonstrations. Despite maintaining the original 2D visuomotor policy formulation, UMI-3D significantly improves the quality and reliability of collected data, which directly translates into enhanced policy performance. Extensive real-world experiments demonstrate that UMI-3D not only achieves high success rates on standard manipulation tasks, but also enables learning of tasks that are challenging or infeasible for the original vision-only UMI setup, including large deformable object manipulation and articulated object operation. The system supports an end-to-end pipeline for data acquisition, alignment, training, and deployment, while preserving the portability and accessibility of the original UMI. All hardware and software components are open-sourced to facilitate large-scale data collection and accelerate research in embodied intelligence: \href{https://umi-3d.github.io}{https://umi-3d.github.io}.

On-policy knowledge distillation (OPD) trains a student on its own rollouts under token-level supervision from a teacher. Not all token positions matter equally, but existing views of token importance are incomplete. We ask a direct question: which tokens carry the most useful learning signal in OPD? Our answer is that informative tokens come from two regions: positions with high student entropy, and positions with low student entropy plus high teacher--student divergence, where the student is overconfident and wrong. Empirically, student entropy is a strong first-order proxy: retaining $50\%$ of tokens with entropy-based sampling matches or exceeds all-token training while reducing peak memory by up to $47\%$. But entropy alone misses a second important region. When we isolate low-entropy, high-divergence tokens, training on fewer than $10\%$ of all tokens nearly matches full-token baselines, showing that overconfident tokens carry dense corrective signal despite being nearly invisible to entropy-only rules. We organize these findings with TIP (Token Importance in on-Policy distillation), a two-axis taxonomy over student entropy and teacher--student divergence, and give a theoretical explanation for why entropy is useful yet structurally incomplete. This view motivates type-aware token selection rules that combine uncertainty and disagreement. We validate this picture across three teacher--student pairs spanning Qwen3, Llama, and Qwen2.5 on MATH-500 and AIME 2024/2025, and on the DeepPlanning benchmark for long-horizon agentic planning, where Q3-only training on $<$$20\%$ of tokens surpasses full-token OPD. Our experiments are implemented by extending the OPD repository https://github.com/HJSang/OPSD_OnPolicyDistillation, which supports memory-efficient distillation of larger models under limited GPU budgets.

Accurate detection and segmentation of glomeruli in kidney tissue are essential for diagnostic applications. Traditional deep learning methods primarily rely on semantic segmentation, which often fails to precisely delineate adjacent glomeruli. To address this challenge, we propose a novel glomerulus detection and segmentation model that emphasises boundary separation. Leveraging pathology foundation models, the proposed U-Net-based architecture incorporates a specialised attention decoder designed to highlight critical regions and improve instancelevel segmentation. Experimental evaluations demonstrate that our approach surpasses state-of-the-art methods in both Dice score and Intersection over Union, indicating superior performance in glomerular delineation.

We introduce Multistage Conditional Compositional Optimization (MCCO) as a new paradigm for decision-making under uncertainty that combines aspects of multistage stochastic programming and conditional stochastic optimization. MCCO minimizes a nest of conditional expectations and nonlinear cost functions. It has numerous applications and arises, for example, in optimal stopping, linear-quadratic regulator problems, distributionally robust contextual bandits, as well as in problems involving dynamic risk measures. The naïve nested sampling approach for MCCO suffers from the curse of dimensionality familiar from scenario tree-based multistage stochastic programming, that is, its scenario complexity grows exponentially with the number of nests. We develop new multilevel Monte Carlo techniques for MCCO whose scenario complexity grows only polynomially with the desired accuracy.

Resolving and rewriting references is fundamental in programming languages. Motivated by a real-world decompilation task, we abstract reference rewriting into the problems of direct and indirect indexing by permutation. We create synthetic benchmarks for these tasks and show that well-known sequence-to-sequence machine learning architectures are struggling on these benchmarks. We introduce new sequence-to-sequence architectures for both problems. Our measurements show that our architectures outperform the baselines in both robustness and scalability: our models can handle examples that are ten times longer compared to the best baseline. We measure the impact of our architecture in the real-world task of decompiling switch statements, which has an indexing subtask. According to our measurements, the extended model decreases the error rate by 42%. Multiple ablation studies show that all components of our architectures are essential.

GUI grounding models report over 85% accuracy on standard benchmarks, yet drop 27-56 percentage points when instructions require spatial reasoning rather than direct element naming. Current benchmarks miss this because they evaluate each screenshot once with a single fixed instruction. We introduce GUI-Perturbed, a controlled perturbation framework that independently varies visual scenes and instructions to measure grounding robustness. Evaluating three 7B models from the same architecture lineage, we find that relational instructions cause systematic accuracy collapse across all models, a 70% browser zoom produces statistically significant degradation, and rank-8 LoRA fine-tuning with augmented data degrades performance rather than improving it. By perturbing along independent axes, GUI-Perturbed isolates which specific capability axes are affected-spatial reasoning, visual robustness, reasoning calibration-providing diagnostic signal that aggregate benchmarks cannot. We release the dataset, augmentation pipeline, and a fine-tuned model.

We propose novel particle swarm optimization (PSO) variants incorporated with deep neural networks (DNNs) for particles to pursue globally optimal positions in dynamic environments. PSO is a heuristic approach for solving complex optimization problems. However, canonical PSO and its variants struggle to adapt efficiently to dynamic environments, in which the global optimum moves over time, and to track them accurately. Many PSO algorithms improve convergence by increasing the swarm size beyond potential optima, which are global/local optima but are not identified until they are discovered. Additionally, in dynamic environments, several methods use multiple sub-population and re-diversification mechanisms to address outdated memory and local optima entrapment. To track the global optimum in dynamic environments with smaller swarm sizes, the DNNs in our methods determine particle movement by learning environmental characteristics and adapting dynamics to pursue moving optimal positions. This enables particles to adapt to environmental changes and predict the moving optima. We propose two variants: a swarm with a centralized network and distributed networks for all particles. Our experimental results show that both variants can track moving potential optima with lower cumulative tracking error than those of several recent PSO-based algorithms, with fewer particles than potential optima.

This paper provides a systematic comparison between Fitted Dynamic Programming (DP), where demand is estimated from data, and Reinforcement Learning (RL) methods in finite-horizon dynamic pricing problems. We analyze their performance across environments of increasing structural complexity, ranging from a single typology benchmark to multi-typology settings with heterogeneous demand and inter-temporal revenue constraints. Unlike simplified comparisons that restrict DP to low-dimensional settings, we apply dynamic programming in richer, multi-dimensional environments with multiple product types and constraints. We evaluate revenue performance, stability, constraint satisfaction behavior, and computational scaling, highlighting the trade-offs between explicit expectation-based optimization and trajectory-based learning.

Deep search agents have emerged as a promising paradigm for addressing complex information-seeking tasks, but their training remains challenging due to sparse rewards, weak credit assignment, and limited labeled data. Self-play offers a scalable route to reduce data dependence, but conventional self-play optimizes students only through sparse outcome rewards, leading to low learning efficiency. In this work, we observe that self-play naturally produces a question construction path (QCP) during task generation, an intermediate artifact that captures the reverse solution process. This reveals a new source of privileged information for self-distillation: self-play can itself provide high-quality privileged context for the teacher model in a low-cost and scalable manner, without relying on human feedback or curated privileged information. Leveraging this insight, we propose Privileged Information Self-Play ($π$-Play), a multi-agent self-evolution framework. In $π$-Play, an examiner generates tasks together with their QCPs, and a teacher model leverages QCP as privileged context to densely supervise a student via self-distillation. This design transforms conventional sparse-reward self-play into a dense-feedback self-evolution loop. Extensive experiments show that data-free $π$-Play surpasses fully supervised search agents and improves evolutionary efficiency by 2-3$\times$ over conventional self-play.

The rapid rise in AI conference submissions has driven increasing exploration of large language models (LLMs) for peer review support. However, LLM-based reviewers often generate superficial, formulaic comments lacking substantive, evidence-grounded feedback. We attribute this to the underutilization of two key components of human reviewing: explicit rubrics and contextual grounding in existing work. To address this, we introduce REVIEWBENCH, a benchmark evaluating review text according to paper-specific rubrics derived from official guidelines, the paper's content, and human-written reviews. We further propose REVIEWGROUNDER, a rubric-guided, tool-integrated multi-agent framework that decomposes reviewing into drafting and grounding stages, enriching shallow drafts via targeted evidence consolidation. Experiments on REVIEWBENCH show that REVIEWGROUNDER, using a Phi-4-14B-based drafter and a GPT-OSS-120B-based grounding stage, consistently outperforms baselines with substantially stronger/larger backbones (e.g., GPT-4.1 and DeepSeek-R1-670B) in both alignment with human judgments and rubric-based review quality across 8 dimensions. The code is available \href{https://github.com/EigenTom/ReviewGrounder}{here}.

Efficiency and safety of Large Language Models (LLMs), among other factors, rely on the quality of tokenization. A good tokenizer not only improves inference speed and language understanding but also provides extra defense against jailbreak attacks and lowers the risk of hallucinations. In this work, we investigate the efficiency of code tokenization, in particular from the perspective of data source diversity. We demonstrate that code tokenizers are prone to producing unused, and thus under-trained, tokens due to the imbalance in repository and language diversity in the training data, as well as the dominance of source-specific, repetitive tokens that are often unusable in future inference. By modifying the BPE objective and introducing merge skipping, we implement different techniques under the name Source-Attributed BPE (SA-BPE) to regularize BPE training and minimize overfitting, thereby substantially reducing the number of under-trained tokens while maintaining the same inference procedure as with regular BPE. This provides an effective tool suitable for production use.

Functional magnetic resonance imaging (fMRI) is widely used for studying and diagnosing brain disorders, with functional connectivity (FC) matrices providing powerful representations of large-scale neural interactions. However, existing diagnostic models are trained either on a single site or under full multi-site access, making them unsuitable for real-world scenarios where clinical data arrive sequentially from different institutions. This results in limited generalization and severe catastrophic forgetting. This paper presents the first continual learning framework specifically designed for fMRI-based diagnosis across heterogeneous clinical sites. Our framework introduces a structure-aware variational autoencoder that synthesizes realistic FC matrices for both patient and control groups. Built on this generative backbone, we develop a multi-level knowledge distillation strategy that aligns predictions and graph representations between new-site data and replayed samples. To further enhance efficiency, we incorporate a hierarchical contextual bandit scheme for adaptive replay sampling. Experiments on multi-site datasets for major depressive disorder (MDD), schizophrenia (SZ), and autism spectrum disorder (ASD) show that the proposed generative model enhances data augmentation quality, and the overall continual learning framework substantially outperforms existing methods in mitigating catastrophic forgetting. Our code is available at https://github.com/4me808/FORGE.

Large language models are typically post-trained using supervised fine-tuning (SFT) and reinforcement learning (RL), yet effectively unifying efficient knowledge injection with robust generalization remains challenging. In this work, we provide a training-dynamics analysis showing that SFT can be interpreted as a special case of policy gradient optimization with an extremely sparse implicit reward and unstable inverse-probability weighting, which together lead to single-path dependency, entropy collapse, and gradient explosion. Motivated by this diagnosis, we propose Group Fine-Tuning (GFT), a unified post-training framework that addresses these intrinsic limitations through two mechanisms: Group Advantage Learning, which constructs diverse response groups and derives normalized contrastive supervision to alleviate reward sparsity, and Dynamic Coefficient Rectification, which adaptively bounds inverse-probability weights to stabilize optimization while preserving efficient knowledge injection. Experiments demonstrate that GFT consistently surpasses SFT-based methods and yields policies that integrate more smoothly with subsequent RL training.

Modern information access ecosystems consist of mixtures of systems, such as retrieval systems and large language models, and increasingly rely on marketplaces to mediate access to models, tools, and data, making competition between systems inherent to deployment. In such settings, outcomes are shaped not only by benchmark quality but also by competitive pressure, including user switching, routing decisions, and operational constraints. Yet evaluation is still largely conducted on static benchmarks with accuracy-focused measures that assume systems operate in isolation. This mismatch makes it difficult to predict post-deployment success and obscures competitive effects such as early-adoption advantages and market dominance. We introduce Marketplace Evaluation, a simulation-based paradigm that evaluates information access systems as participants in a competitive marketplace. By simulating repeated interactions and evolving user and agent preferences, the framework enables longitudinal evaluation and marketplace-level metrics, such as retention and market share, that complement and can extend beyond traditional accuracy-based metrics. We formalize the framework and outline a research agenda, motivated by business and economics, around marketplace simulation, metrics, optimization, and adoption in evaluation campaigns like TREC.

The field of machine ethics aims to build Artificial Moral Agents (AMAs) to better understand morality and make AI agents safer. To do so, many approaches encode human moral intuition as a set of axioms on actions e.g., do not harm, you must help others. However, this introduces (at least) two limitations for future AMAs. First, it does not consider the agent's purposes in performing the action. Second, it assumes that we humans can enumerate our moral intuition. This paper explores formalizing a moral procedure that alleviates these two limitations. We specifically consider Kantian ethics and present a multi-sorted quantified modal logic we call the Formula of the Universal Law Logic (FULL). The FULL formalizes Kant's first formulation of the categorical imperative, the Formula of the Universal Law (FUL), and concepts such as causality and agency. We demonstrate on three cases from Kantian ethics that the FULL can reason to evaluate agents' actions for certain purposes without built-in moral intuition, given that it has sufficient (non-normative) background knowledge. Therefore, the FULL is a contribution towards more robust and autonomous AMAs, and a more formal understanding of Kantian ethics.

Driver drowsiness significantly impairs the ability to accurately judge safe braking distances and is estimated to contribute to 10%-20% of road accidents in Europe. Traditional driver-assistance systems lack adaptability to real-time physiological states such as drowsiness. This paper proposes a deep reinforcement learning-based autonomous braking system that integrates vehicle dynamics with driver physiological data. Drowsiness is detected from ECG signals using a Recurrent Neural Network (RNN), selected through an extensive benchmark analysis of 2-minute windows with varying segmentation and overlap configurations. The inferred drowsiness state is incorporated into the observable state space of a Double-Dueling Deep Q-Network (DQN) agent, where driver impairment is modeled as an action delay. The system is implemented and evaluated in a high-fidelity CARLA simulation environment. Experimental results show that the proposed agent achieves a 99.99% success rate in avoiding collisions under both drowsy and non-drowsy conditions. These findings demonstrate the effectiveness of physiology-aware control strategies for enhancing adaptive and intelligent driving safety systems.

This paper develops a unified Markov Decision Process (MDP) framework for optimising two recurring in-match decisions in T20 cricket, namely batting order selection and bowling plan assignment, directly in terms of win and defend probability rather than expected runs. A three-phase player profile engine (Powerplay, Middle, Death) with James-Stein shrinkage (a technique that blends a player's individual statistics toward the league average when their phase-specific data is sparse) is estimated from 1,161 IPL ball-by-ball records (2008-2025). Win/defend probabilities are evaluated using vectorised Monte Carlo simulation over N = 50,000 innings trajectories. Batting orders are evaluated by comparing all feasible arrangements of the remaining players and selecting the one that maximises win probability. Bowling plans are optimised through a guided search over possible over assignments, progressively improving the allocation while respecting constraints such as the prohibition on consecutive overs by the same bowler. Applied to two 2026 IPL matches, the optimal batting order improves Mumbai Indians' win probability by 4.1 percentage points (52.4% to 56.5%), and the optimal Gujarat Titans bowling plan improves defend probability by 5.2 percentage points (39.1% to 44.3%). In both cases, the observed sub-optimality is consistent with phase-agnostic deployment: decisions that appear reasonable under aggregate metrics are shown to be costly when phase-specific profiles are applied.

Personalized Large Language Models (LLMs) facilitate more natural, human-like interactions in human-centric applications. However, existing personalization methods are constrained by limited controllability and high resource demands. Furthermore, their reliance on static personality modeling restricts adaptability across varying situations. To address these limitations, we first demonstrate the existence of situation-dependency and consistent situation-behavior patterns within LLM personalities through a multi-perspective analysis of persona neurons. Building on these insights, we propose IRIS, a training-free, neuron-based Identify-Retrieve-Steer framework for advanced situational personality steering. Our approach comprises situational persona neuron identification, situation-aware neuron retrieval, and similarity-weighted steering. We empirically validate our framework on PersonalityBench and our newly introduced SPBench, a comprehensive situational personality benchmark. Experimental results show that our method surpasses best-performing baselines, demonstrating IRIS's generalization and robustness to complex, unseen situations and different models architecture.

Reward models are central to aligning large language models, yet they often overfit to spurious cues such as response length and overly agreeable tone. Most prior work weakens these cues directly by penalizing or controlling specific artifacts, but it does not explicitly encourage the model to ground preferences in the prompt's intent. We learn a decoder that maps a candidate answer to the latent intent embedding of the input. The reconstruction error is used as a signal to regularize the reward model training. We provide theoretical evidence that this signal emphasizes prompt-dependent information while suppressing prompt-independent shortcuts. Across math, helpfulness, and safety benchmarks, the decoder selects shorter and less sycophantic candidates with 0.877 accuracy. Incorporating this signal into RM training in Gemma-2-2B-it and Gemma-2-9B-it increases RewardBench accuracy from 0.832 to 0.868. For Best-of-N selection, our framework increases length-controlled win rates while producing shorter outputs, and remains robust to lengthening and mild off-topic drift in controlled rewrite tests.

Monitoring LLM safety at scale requires balancing cost and accuracy: a cheap latent-space probe can screen every input, but hard cases should be escalated to a more expensive expert. Existing cascades delegate based on probe uncertainty, but uncertainty is a poor proxy for delegation benefit, as it ignores whether the expert would actually correct the error. To address this problem, we introduce Calibrate-Then-Delegate (CTD), a model-cascade approach that provides probabilistic guarantees on the computation cost while enabling instance-level (streaming) decisions. CTD builds on a novel delegation value (DV) probe, a lightweight model operating on the same internal representations as the safety probe that directly predicts the benefit of escalation. To enforce budget constraints, CTD calibrates a threshold on the DV signal using held-out data via multiple hypothesis testing, yielding finite-sample guarantees on the delegation rate. Evaluated on four safety datasets, CTD consistently outperforms uncertainty-based delegation at every budget level, avoids harmful over-delegation, and adapts budget allocation to input difficulty without requiring group labels.

An important aspect of crowd monitoring is knowing how many people we are dealing with. Sometimes, knowing the size of a crowd in a single location and at a specific moment is enough. Matters become problematic when counting the same people across dif ferent locations or counting them over longer periods of time. In those cases, we need to identify and later reidentify a person, which immediately leads to privacy concerns. Until recently, solutions have been based on unique identification of carry-on devices, yet privacy improvements have caused transmitted information to be randomized, rendering this technique mostly useless. We propose to use biometric data instead. We introduce a pipeline that counts people based on face recognition, yet without ever being able to reveal the identity of individuals. To count, a camera initially detects a face, extracts its features, and derives an identifier using a fuzzy extractor. The original facial image is then deleted. Identifiers are inserted into homomorphically encrypted Bloom filters. This allows oblivious set membership testing directly on encrypted data, enabling the system to count across locations or across different moments, without revealing any identities. We provide an initial evaluation of our method that shows promising results.

We introduce Metric-Aware Principal Component Analysis (MAPCA), a unified framework for scale-invariant representation learning based on the generalised eigenproblem max Tr(W^T Sigma W) subject to W^T M W = I, where M is a symmetric positive definite metric matrix. The choice of M determines the representation geometry. The canonical beta-family M(beta) = Sigma^beta, beta in [0,1], provides continuous spectral bias control between standard PCA (beta=0) and output whitening (beta=1), with condition number kappa(beta) = (lambda_1/lambda_p)^(1-beta) decreasing monotonically to isotropy. The diagonal metric M = D = diag(Sigma) recovers Invariant PCA (IPCA), a method rooted in Frisch (1928) diagonal regression, as a distinct member of the broader framework. We prove that scale invariance holds if and only if the metric transforms as M_tilde = CMC under rescaling C, a condition satisfied exactly by IPCA but not by the general beta-family at intermediate values. Beyond its classical interpretation, MAPCA provides a geometric language that unifies several self-supervised learning objectives. Barlow Twins and ZCA whitening correspond to beta=1 (output whitening); VICReg's variance term corresponds to the diagonal metric. A key finding is that W-MSE, despite being described as a whitening-based method, corresponds to M = Sigma^{-1} (beta = -1), outside the spectral compression range entirely and in the opposite spectral direction to Barlow Twins. This distinction between input and output whitening is invisible at the level of loss functions and becomes precise only within the MAPCA framework.

SAKURAONE is a managed high performance computing (HPC) cluster developed and operated by the SAKURA Internet Research Center. It builds on the KOKARYOKU PHY bare metal GPU platform and is optimized for advanced workloads, including large language model (LLM) training. In ISC 2025 TOP500, SAKURAONE is ranked 49th by HPL and is the only top 100 system that uses a fully open networking stack - 800 GbE with SONiC - demonstrating the scalability of vendor-neutral technology. Measured performance is 33.95 PFLOP/s (HPL Rmax), 396.295 TFLOP/s (HPCG), and 339.86 PFLOP/s on HPL-MxP with FP8. The system consists of 100 nodes, each with eight NVIDIA H100 GPUs and a 2 PB all-flash Lustre file system, interconnected via a rail-optimized 800 GbE leaf-spine fabric with RoCEv2. Through exclusive use by a single research project, we observed the characteristics of development-related jobs. Consistent with previous HPC studies, small-scale jobs dominated in number, while a few large-scale jobs accounted for most GPU resource time. As the project progressed, resource use shifted from large-scale to mid-scale jobs, reflecting a transition from initial large-scale training to iterative refinement. These observations illustrate the real-world utilization dynamics of GPU clusters under unified project workloads.

Reasoning capabilities in large language models (LLMs) have generally advanced significantly. However, it is still challenging for existing reasoning-based LLMs to perform effective decision-making abilities in multi-agent environments, due to the absence of explicit foresight modeling. To this end, strategic reasoning, the most fundamental capability to anticipate the counterpart's behaviors and foresee its possible future actions, has been introduced to alleviate the above issues. Strategic reasoning is fundamental to effective decision-making in multi-agent environments, yet existing reasoning enhancement methods for LLMs do not explicitly capture its foresight nature. In this work, we introduce Foresight Policy Optimization (FoPO) to enhance strategic reasoning in LLMs, which integrates opponent modeling principles into policy optimization, thereby enabling explicit consideration of both self-interest and counterpart influence. Specifically, we construct two curated datasets, namely Cooperative RSA and Competitive Taboo, equipped with well-designed rules and moderate difficulty to facilitate a systematic investigation of FoPO in a self-play framework. Our experiments demonstrate that FoPO significantly enhances strategic reasoning across LLMs of varying sizes and origins. Moreover, models trained with FoPO exhibit strong generalization to out-of-domain strategic scenarios, substantially outperforming standard LLM reasoning optimization baselines.

High-level synthesis (HLS) tools offer limited support for Engineering Change Orders (ECOs), making late-stage design modifications challenging and costly. This paper introduces a graph-based ECO methodology tailored for Google XLS. A Graph Edit Distance (GED) algorithm is used to detect structural differences between original and revised intermediate representations (IRs), which are then transformed into patch operations. A patch application mechanism is developed to enforce XLS IR constraints while preserving semantic correctness, together with a schedule constraining scheme that maintains the original pipeline registers. Experiments across several XLS designs demonstrate high structural reuse ratios, effective schedule preservation, and full functional correctness, highlighting the practicality of the approach for production HLS flows.

Sparse Mixture-of-Experts (MoE) models have achieved remarkable scalability, yet they remain vulnerable to hallucinations, particularly when processing long-tail knowledge. We identify that this fragility stems from static Top-$k$ routing: routers tend to favor high-frequency patterns over rare factual associations. Consequently, ``specialist experts'' possessing critical long-tail knowledge are often assigned low gating scores and remain ``dormant'' -- under-prioritized for specific tokens despite their proven causal importance on other inputs. To address this, we propose Counterfactual Routing (CoR), a training-free inference framework designed to awaken these dormant experts. CoR integrates layer-wise perturbation analysis with the Counterfactual Expert Impact (CEI) metric to dynamically shift computational resources from syntax-dominant to knowledge-intensive layers while maintaining a constant total activation count, effectively retrieving causally decisive experts via virtual ablation. Extensive experiments on TruthfulQA, FACTOR, and TriviaQA demonstrate that CoR improves factual accuracy by 3.1\% on average without increasing the inference budget, establishing a superior Pareto frontier compared to static scaling strategies.

Real-world decision-making systems operate in environments where state transitions depend not only on the agent's actions, but also on \textbf{exogenous factors outside its control}--competing agents, environmental disturbances, or strategic adversaries--formally, $s_{h+1} = f(s_h, a_h, \bar{a}_h)+ω_h$ where $\bar{a}_h$ is the adversary/external action, $a_h$ is the agent's action, and $ω_h$ is an additive noise. Ignoring such factors can yield policies that are optimal in isolation but \textbf{fail catastrophically in deployment}, particularly when safety constraints must be satisfied. Standard Constrained MDP formulations assume the agent is the sole driver of state evolution, an assumption that breaks down in safety-critical settings. Existing robust RL approaches address this via distributional robustness over transition kernels, but do not explicitly model the \textbf{strategic interaction} between agent and exogenous factor, and rely on strong assumptions about divergence from a known nominal model. We model the exogenous factor as an \textbf{adversarial policy} $\barπ$ that co-determines state transitions, and ask how an agent can remain both optimal and safe against such an adversary. \emph{To the best of our knowledge, this is the first work to study safety-constrained RL under explicit adversarial dynamics}. We propose \textbf{Robust Hallucinated Constrained Upper-Confidence RL} (\texttt{RHC-UCRL}), a model-based algorithm that maintains optimism over both agent and adversary policies, explicitly separating epistemic from aleatoric uncertainty. \texttt{RHC-UCRL} achieves sub-linear regret and constraint violation guarantees.

The classic paradigm of structural biology is that the sequence of a biomolecule (protein, nucleic acid, lipid, etc) determines its conformation (shape) which determines its biological function. Protein folding programs like AlphaFold address this paradigm by predicting the single best conformation given a sequence that defines the molecule. However, biomolecules are not static structures, and their conformational ensemble determines their function. We present the Polyformer -- a generative framework for thermodynamic modeling of polymeric molecules. Given the sequence and temperature (or another thermodynamic variable), the Polyformer generates conformations faithful to the molecule's thermodynamic conformational ensemble. It is the first generative model that solves three problems simultaneously: how does a molecule fold, what is its conformational ensemble, and how does the conformational ensemble change as we change physical temperature. As a concrete test case, we apply Polyformer to protein domains with 50-111 residues and report good agreement of model predictions to Molecular Dynamics (MD) trajectories.

The simulation of complex systems increasingly relies on sophisticated but fundamentally opaque computational black-box simulators. Surrogate models play a central role in reducing the computational cost of complex systems simulations across a wide range of scientific and engineering domains. Notwithstanding, they inevitably inherit and often exacerbate this black-box nature, obscuring how input variables drive physical responses. Conversely, Explainable Artificial Intelligence (XAI) offers powerful tools to unpack these models. Yet, XAI methods struggle with engineering-specific constraints, such as highly correlated inputs, dynamical systems, and rigorous reliability requirements. Consequently, surrogate modeling and XAI have largely evolved as distinct fields of research, despite their strong complementarity. To reconnect these approaches, this state-of-the-art survey provides a structured perspective that maps existing XAI techniques onto the various stages of surrogate modeling workflows for design and exploration. To ground this synthesis, we draw upon illustrative applications across both equation-based simulations and agent-based modeling. We survey a broad spectrum of techniques, highlighting their strengths for revealing interactions and supporting human comprehension. Finally, we identify pressing open challenges, including the explainability of dynamical systems and the handling of mixed-variable systems, and propose a research agenda to make explainability a core, embedded element of simulation-driven workflows from model construction through decision-making. By transforming opaque emulators into explainable tools, this agenda empowers practitioners to move beyond accelerating simulations to extracting actionable insights from complex system behaviors.

Transistor topology optimization is a critical step in standard cell design, directly dictating diffusion sharing efficiency and downstream routability. However, identifying optimal topologies remains a persistent bottleneck, as conventional exhaustive search methods become computationally intractable with increasing circuit complexity in advanced nodes. This paper introduces TOPCELL, a novel and scalable framework that reformulates high-dimensional topology exploration as a generative task using Large Language Models (LLMs). We employ Group Relative Policy Optimization (GRPO) to fine-tune the model, aligning its topology optimization strategy with logical (circuit) and spatial (layout) constraints. Experimental results within an industrial flow targeting an advanced 2nm technology node demonstrate that TOPCELL significantly outperforms foundation models in discovering routable, physically-aware topologies. When integrated into a state-of-the-art (SOTA) automation flow for a 7nm library generation task, TOPCELL exhibits robust zero-shot generalization and matches the layout quality of exhaustive solvers while achieving an 85.91x speedup.

High-resolution data in spatial and temporal contexts is imperative for developing climate resilient cities. Current datasets for monitoring urban parameters are developed primarily using manual inspections, embedded-sensing, remote sensing, or standard street-view imagery (RGB). These methods and datasets are often constrained respectively by poor scalability, inconsistent spatio-temporal resolutions, overhead views or low spectral information. We present a novel method and its open implementation: a multi-spectral terrestrial-view dataset that circumvents these limitations. This dataset consists of 17,718 street level multi-spectral images captured with RGB, Near-infrared, and Thermal imaging sensors on bikes, across diverse urban morphologies (village, town, small city, and big urban area) in the Netherlands. Strict emphasis is put on data calibration and quality while also providing the details of our data collection methodology (including the hardware and software details). To the best of our knowledge, Spectrascapes is the first open-access dataset of its kind. Finally, we demonstrate two downstream use-cases enabled using this dataset and provide potential research directions in the machine learning, urban planning and remote sensing domains.

Fraud detection on graph data can be viewed as a demanding task that requires distinguishing between different types of nodes. Because graph neural networks (GNNs) are naturally suited for processing information encoded in graph form through their message-passing operations, methods based on GNN models have increasingly attracted attention in the fraud detection domain. However, fraud graphs inherently exhibit relation camouflage, high heterophily, and class imbalance, causing most GNNs to underperform in fraud detection tasks. To address these challenges, this paper proposes a Graph-Based Fraud Detection Model with Dual-Path Graph Filtering (DPF-GFD). DPF-GFD first applies a beta wavelet-based operator to the original graph to capture key structural patterns. It then constructs a similarity graph from distance-based node representations and applies an improved low-pass filter. The embeddings from the original and similarity graphs are fused through supervised representation learning to obtain node features, which are finally used by an ensemble tree model to assess the fraud risk of unlabeled nodes. Unlike existing single-graph smoothing approaches, DPF-GFD introduces a frequency-complementary dual-path filtering paradigm tailored for fraud detection, explicitly decoupling structural anomaly modeling and feature similarity modeling. This design enables more discriminative and stable node representations in highly heterophilous and imbalanced fraud graphs. Comprehensive experiments on four real-world financial fraud detection datasets demonstrate the effectiveness of our proposed method.

The IEC-61850 GOOSE protocol underpins time-critical communication in modern digital substations but lacks native security mechanisms, leaving it vulnerable to replay, masquerade, and data injection attacks. Intrusion detection in this setting is challenging due to strict latency constraints (sub-4ms) and limited availability of labeled attack data. This paper evaluates whether unsupervised temporal modeling can provide effective and deployable anomaly detection for GOOSE networks. Five models are compared on the ERENO IEC-61850 dataset: a supervised Random Forest baseline, a feedforward Autoencoder, and three recurrent sequence autoencoders (RNN, LSTM, and GRU). The supervised Random Forest achieves the highest detection performance (F1=0.9516) but fails to meet real-time constraints at 21.8ms per prediction. All four unsupervised models satisfy the 4ms requirement, with the GRU achieving the best accuracy to latency tradeoff among them (F1=0.8737 at 1.118ms). A cross-environment evaluation on an independent dataset shows that all models degrade under distribution shift. However, recurrent models retain substantially higher relative performance than the supervised baseline, suggesting that temporal sequence modeling generalizes better than fitting labeled attack distributions. Anomaly thresholds for the unsupervised models are selected on a held out validation partition to avoid test set leakage. These results support unsupervised temporal models as a practical choice for real-time GOOSE intrusion detection, particularly in environments where labeled training data may be unavailable or where large-scale deployment across diverse substations is required.

The Spatial-Temporal Graph Attention Network (ST-GAT) framework was created to serve as an explainable GNN-based solution for detecting bank distress early warning signs and for conducting macro-prudential surveillance of the interbank system in the United States. The ST-GAT framework models 8,103 FDIC insured institutions across 58 quarterly snapshots (2010Q1-2024Q2). Bilateral exposures were reconstructed from publicly available FDIC Call Reports using maximum entropy estimation to produce a dynamic directed weighted graph. The framework achieves the highest AUPRC among all GNN architectures (0.939 +/- 0.010), trailing only XGBoost (0.944). Ablation analysis confirms the BiLSTM temporal component contributes +0.020 AUPRC; temporal attention weights exhibit a monotonically decreasing pattern consistent with long-run structural vulnerability weighting. Permutation importance identifies ROA (0.309) and NPL Ratio (0.252) as dominant predictors, consistent with post-mortem analyses of the 2023 regional banking crisis. All data are publicly available FDIC Call Reports and FRED series; all code and results are released.

Financial crime costs U.S. institutions over $32 billion each year. Although AI tools for fraud detection have become more advanced, their use in real-world systems still faces a major obstacle: many of these models operate as black boxes that cannot provide the transparent, auditable explanations required by regulations such as OCC Bulletin 2011-12 and Federal Reserve SR 11-7. This study makes three main contributions. First, it offers a thorough evaluation of explanation quality across faithfulness (sufficiency and comprehensiveness at k=5, 10, and 15) and stability (Kendall's W across 30 bootstrap samples). XGBoost paired with TreeExplainer achieves near-perfect stability (W=0.9912), while LSTM with DeepExplainer shows weak results (W=0.4962). Second, the paper introduces the SHAP-Guided Adaptive Ensemble (SGAE), which dynamically adjusts per-transaction ensemble weights based on SHAP attribution agreement, achieving the highest AUC-ROC among all tested models (0.8837 held-out; 0.9245 cross-validation). Third, a complete three-architecture evaluation of LSTM, Transformer, and GNN-GraphSAGE on the full 590,540-transaction IEEE-CIS dataset is provided, with GNN-GraphSAGE achieving AUC-ROC 0.9248 and F1=0.6013. All results are mapped directly to OCC, SR 11-7, and BSA-AML regulatory compliance requirements.

Synthetic Aperture Radar (SAR) data is inherently complex-valued, while quantum machine learning (QML) models naturally operate in complex Hilbert spaces. This apparent alignment suggests that incorporating both magnitude and phase information into quantum encoding should improve performance in SAR Automatic Target Recognition (ATR). In this work, we systematically evaluate this assumption by comparing five quantum encoding strategies: magnitude-only, joint complex, I/Q-based, preprocessed phase, and pure quantum, under a unified experimental framework on the MSTAR benchmark dataset. Contrary to expectation, we observe a consistent pattern: in hybrid quantum-classical architectures, magnitude-only encoding outperforms all complex-valued strategies, achieving 99.57% accuracy on a 3-class task and 71.19% on an 8-class task, while phase-aware methods provide negligible (~0%) or negative improvements. In contrast, in purely quantum architectures with only 184-224 trainable parameters and no classical components, phase information becomes essential, contributing up to 21.65% improvement in accuracy. These results reveal that the utility of phase information is not inherent to the data, but depends critically on the model architecture. Hybrid models rely on classical components that compensate for missing phase information, whereas purely quantum models require phase to construct discriminative representations. Our findings provide practical design guidelines for encoding complex-valued data in QML and highlight the importance of encoding-architecture co-design in the NISQ era.

Claude Code is an agentic coding tool that can run shell commands, edit files, and call external services on behalf of the user. This study describes its comprehensive architecture by analyzing the publicly available TypeScript source code and further comparing it with OpenClaw, an independent open-source AI agent system that answers many of the same design questions from a different deployment context. Our analysis identifies five human values, philosophies, and needs that motivate the architecture (human decision authority, safety and security, reliable execution, capability amplification, and contextual adaptability) and traces them through thirteen design principles to specific implementation choices. The core of the system is a simple while-loop that calls the model, runs tools, and repeats. Most of the code, however, lives in the systems around this loop: a permission system with seven modes and an ML-based classifier, a five-layer compaction pipeline for context management, four extensibility mechanisms (MCP, plugins, skills, and hooks), a subagent delegation mechanism with worktree isolation, and append-oriented session storage. A comparison with OpenClaw, a multi-channel personal assistant gateway, shows that the same recurring design questions produce different architectural answers when the deployment context changes: from per-action safety classification to perimeter-level access control, from a single CLI loop to an embedded runtime within a gateway control plane, and from context-window extensions to gateway-wide capability registration. We finally identify six open design directions for future agent systems, grounded in recent empirical, architectural, and policy literature.

On-policy distillation (OPD) has become a core technique in the post-training of large language models, yet its training dynamics remain poorly understood. This paper provides a systematic investigation of OPD dynamics and mechanisms. We first identify that two conditions govern whether OPD succeeds or fails: (i) the student and teacher should share compatible thinking patterns; and (ii) even with consistent thinking patterns and higher scores, the teacher must offer genuinely new capabilities beyond what the student has seen during training. We validate these findings through weak-to-strong reverse distillation, showing that same-family 1.5B and 7B teachers are distributionally indistinguishable from the student's perspective. Probing into the token-level mechanism, we show that successful OPD is characterized by progressive alignment on high-probability tokens at student-visited states, a small shared token set that concentrates most of the probability mass (97%-99%). We further propose two practical strategies to recover failing OPD: off-policy cold start and teacher-aligned prompt selection. Finally, we show that OPD's apparent free lunch of dense token-level reward comes at a cost, raising the question of whether OPD can scale to long-horizon distillation.

Retrieval-Augmented Generation (RAG) is a key approach to mitigating the temporal staleness of large language models (LLMs) by grounding responses in up-to-date evidence. Within the RAG pipeline, re-rankers play a pivotal role in selecting the most useful documents from retrieved candidates. However, existing benchmarks predominantly evaluate re-rankers in static settings and do not adequately assess performance under evolving information -- a critical gap, as real-world systems often must choose among temporally different pieces of evidence. To address this limitation, we introduce FRESCO (Factual Recency and Evolving Semantic COnflict), a benchmark for evaluating re-rankers in temporally dynamic contexts. By pairing recency-seeking queries with historical Wikipedia revisions, FRESCO tests whether re-rankers can prioritize factually recent evidence while maintaining semantic relevance. Our evaluation reveals a consistent failure mode across existing re-rankers: a strong bias toward older, semantically rich documents, even when they are factually obsolete. We further investigate an instruction optimization framework to mitigate this issue. By identifying Pareto-optimal instructions that balance Evolving and Non-Evolving Knowledge tasks, we obtain gains of up to 27% on Evolving Knowledge tasks while maintaining competitive performance on Non-Evolving Knowledge tasks.

There is growing interest in leveraging large language models (LLMs) for text-to-model translation and optimization tasks. This paper aims to advance this line of research by introducing \textsc{Text2Model} and \textsc{Text2Zinc}. \textsc{Text2Model} is a suite of copilots based on several LLM strategies with varying complexity, along with an online leaderboard. \textsc{Text2Zinc} is a cross-domain dataset for capturing optimization and satisfaction problems specified in natural language, along with an interactive editor with built-in AI assistant. While there is an emerging literature on using LLMs for translating combinatorial problems into formal models, our work is the first attempt to integrate \textit{both} satisfaction and optimization problems within a \textit{unified architecture} and \textit{dataset}. Moreover, our approach is \textit{solver-agnostic} unlike existing work that focuses on translation to a solver-specific model. To achieve this, we leverage \textsc{MiniZinc}'s solver-and-paradigm-agnostic modeling capabilities to formulate combinatorial problems. We conduct comprehensive experiments to compare execution and solution accuracy across several single- and multi-call strategies, including; zero-shot prompting, chain-of-thought reasoning, intermediate representations via knowledge-graphs, grammar-based syntax encoding, and agentic approaches that decompose the model into sequential sub-tasks. Our copilot strategies are competitive, and in parts improve, recent research in this domain. Our findings indicate that while LLMs are promising they are not yet a push-button technology for combinatorial modeling. We contribute \textsc{Text2Model} copilots and leaderboard, and \textsc{Text2Zinc} and interactive editor to open-source to support closing this performance gap.

In advanced manufacturing systems, humans and robots collaborate to conduct the production process. Effective task planning and allocation (TPA) is crucial for achieving high production efficiency, yet it remains challenging in complex and dynamic manufacturing environments. The dynamic nature of humans and robots, particularly the need to consider spatial information (e.g., humans' real-time position and the distance they need to move to complete a task), substantially complicates TPA. To address the above challenges, we decompose production tasks into manageable subtasks. We then implement a real-time hierarchical human-robot TPA algorithm, including a high-level agent for task planning and a low-level agent for task allocation. For the high-level agent, we propose an efficient buffer-based deep Q-learning method (EBQ), which reduces training time and enhances performance in production problems with long-term and sparse reward challenges. For the low-level agent, a path planning-based spatially aware method (SAP) is designed to allocate tasks to the appropriate human-robot resources, thereby achieving the corresponding sequential subtasks. We conducted experiments on a complex real-time production process in a 3D simulator. The results demonstrate that our proposed EBQ&SAP method effectively addresses human-robot TPA problems in complex and dynamic production processes.

Human-robot collaborative manufacturing, a core aspect of Industry 5.0, emphasizes ergonomics to enhance worker well-being. This paper addresses the dynamic human-robot task planning and allocation (HRTPA) problem, which involves determining when to perform tasks and who should execute them to maximize efficiency while ensuring workers' physical fatigue remains within safe limits. The inclusion of fatigue constraints, combined with production dynamics, significantly increases the complexity of the HRTPA problem. Traditional fatigue-recovery models in HRTPA often rely on static, predefined hyperparameters. However, in practice, human fatigue sensitivity varies daily due to factors such as changed work conditions and insufficient sleep. To better capture this uncertainty, we treat fatigue-related parameters as inaccurate and estimate them online based on observed fatigue progression during production. To address these challenges, we propose PF-CD3Q, a safe reinforcement learning (safe RL) approach that integrates the particle filter with constrained dueling double deep Q-learning for real-time fatigue-predictive HRTPA. Specifically, we first develop PF-based estimators to track human fatigue and update fatigue model parameters in real-time. These estimators are then integrated into CD3Q by making task-level fatigue predictions during decision-making and excluding tasks that exceed fatigue limits, thereby constraining the action space and formulating the problem as a constrained Markov decision process (CMDP).

Traditional conversational travel recommender systems primarily optimize for user relevance and convenience, often reinforcing popular, overcrowded destinations and carbon-intensive travel choices. To address this, we present TRACE (Tourism Recommendation with Agentic Counterfactual Explanations), a multi-agent, LLM-based framework that promotes sustainable tourism through interactive nudging. TRACE uses a modular orchestrator-worker architecture where specialized agents elicit latent sustainability preferences, construct structured user personas, and generate recommendations that balance relevance with environmental impact. A key innovation lies in its use of agentic counterfactual explanations and LLM-driven clarifying questions, which together surface greener alternatives and refine understanding of intent, fostering user reflection without coercion. User studies and semantic alignment analyses demonstrate that TRACE effectively supports sustainable decision-making while preserving recommendation quality and interactive responsiveness. TRACE is implemented on Google's Agent Development Kit, with full code, Docker setup, prompts, and a publicly available demo video to ensure reproducibility. A project summary, including all resources, prompts, and demo access, is available at https://ashmibanerjee.github.io/trace-chatbot.

Retrieval-Augmented Generation (RAG) has become the standard paradigm for grounding Large Language Model outputs in external knowledge. Lumer et al. [1] presented the first systematic evaluation comparing vector-based agentic RAG against hierarchical node-based reasoning systems for financial document QA across 1,200 SEC filings, finding vector-based systems achieved a 68% win rate. Concurrently, the PageIndex framework [2] demonstrated 98.7% accuracy on FinanceBench through purely reasoning-based retrieval. This paper extends their work by: (i) implementing and evaluating three retrieval architectures: Vector RAG, Tree Reasoning, and the proposed Adaptive Hybrid Retrieval (AHR) across financial, legal, and medical domains; (ii) introducing a four-tier query complexity benchmark; and (iii) employing GPT-4-powered LLM-as-judge evaluation. Experiments reveal that Tree Reasoning achieves the highest overall score (0.900), but no single paradigm dominates across all tiers: Vector RAG wins on multi-document synthesis (Tier 4, score 0.900), while the Hybrid AHR achieves the best performance on cross-reference (0.850) and multi-section queries (0.929). Cross-reference recall reaches 100% for tree-based and hybrid approaches versus 91.7% for vector search, quantifying a critical capability gap. Validation on FinanceBench (150 expert-annotated questions on real SEC 10-K and 10-Q filings) confirms and strengthens these findings: Tree Reasoning scores 0.938, Hybrid AHR 0.901, and Vector RAG 0.821, with the Tree--Vector quality gap widening to 11.7 percentage points on real-world documents. These findings support the development of adaptive retrieval systems that dynamically select strategies based on query complexity and document structure. All code and data are publicly available.

Large language models (LLMs) face significant challenges in processing long contexts due to the linear growth of the key-value (KV) cache and quadratic complexity of self-attention. Existing approaches address these bottlenecks separately: Multi-head Latent Attention (MLA) reduces the KV cache by projecting tokens into a low-dimensional latent space, while sparse attention reduces computation. However, sparse methods cannot operate natively on MLA's compressed latent structure, missing opportunities for joint optimization. In this paper, we propose Latent-Condensed Attention (LCA), which directly condenses context within MLA's latent space, where the representation is disentangled into semantic latent vectors and positional keys. LCA separately aggregates semantic vectors via query-aware pooling and preserves positional keys via anchor selection. This approach jointly reduces both computational cost and KV cache without adding parameters. Beyond MLA, LCA's design is architecture-agnostic and readily extends to other attention mechanisms such as GQA. Theoretically, we prove a length-independent error bound. Experiments show LCA achieves up to 2.5$\times$ prefilling speedup and 90% KV cache reduction at 128K context while maintaining competitive performance.

Reliable evaluation of anomaly detection methods in multivariate time series remains an open challenge, largely due to the limitations of existing benchmark datasets. Current resources often lack fine-grained anomaly annotations, do not provide explicit intervariable and temporal dependencies, and offer little insight into the underlying generative mechanisms. These shortcomings hinder the development and rigorous comparison of detection models, especially those targeting interpretable and variable-specific outputs. To address this gap, we introduce Fun-TSG, a fully customizable time series generator designed to support high-quality evaluation of anomaly detection systems. Our tool enables both fully automated generation, based on randomly sampled dependency structures and anomaly types, and manual generation through user-defined equations and anomaly configurations. In both cases, it provides full transparency over the data generation process, including access to ground-truth anomaly labels at the variable and timestamp levels. Fun-TSG supports the creation of diverse, interpretable, and reproducible benchmarking scenarios, enabling fine-grained performance analysis for both classical and modern anomaly detection models.

This paper addresses two limitations of large language models (LLMs) in solving complex problems: (1) their reasoning processes exhibit Bayesian-like stochastic generation, where each token is sampled from a context-dependent probability distribution, leading to inherently random decision trajectories rather than deterministic planning; (2) the reasoning and decision-making mechanisms are statically decoupled, meaning dynamically retrieved domain knowledge fails to dynamically adjust the underlying reasoning strategy. These dual deficiencies result in initial decisions lacking strategic anchoring and reasoning chains often failing to converge on correct solutions, as stochastic generation lacks mechanisms for trajectory correction or knowledge-guided optimization during sequential reasoning. To resolve these issues, we propose a problem-solving method integrated into the LLM's generation process to guide reasoning. This method, compatible with numerous LLMs and featuring reusable solutions, is grounded in a novel Heuristic-Classification-of-Thoughts prompting schema (HCoT). HCoT synergizes the LLM's reasoning ability with a structured problem space via a heuristic classification model that controls the reasoning process and provides reusable abstract solutions. Evaluated on two complex inductive reasoning tasks with ill-defined search spaces, HCoT outperforms existing approaches (e.g., Tree-of-Thoughts and Chain-of-Thoughts prompting) in performance. On the well-structured 24 Game task, HCoT demonstrates significantly higher token efficiency compared to the state-of-the-art Tree-of-Thoughts-Breadth-First-Search. In terms of both accuracy and token usage, HCoT achieves a Pareto frontier balance, offering a strong trade-off between performance and computational cost.

This research paper addresses the limitations of semantic search in complex enterprise document ecosystems. Traditional RAG pipelines often fail to capture hierarchical and interconnected information, leading to retrieval inaccuracies. We propose Agentic Knowledge Graphs featuring Recursive Crawling as a robust solution for navigating superseding logic and multi-hop references. Our benchmark evaluation using the Code of Federal Regulations (CFR) demonstrates that this Knowledge Graph-enhanced approach achieves a 70% accuracy improvement over standard vector-based RAG systems, providing exhaustive and precise answers for complex regulatory queries.

Simulating Standardized Patients with cognitive impairment offers a scalable and ethical solution for clinical training. However, existing methods rely on discrete prompt engineering and fail to capture the heterogeneity of deficits across varying domains and severity levels. To address this limitation, we propose StsPatient for the fine-grained simulation of cognitively impaired patients. We innovatively capture domain-specific features by extracting steering vectors from contrastive pairs of instructions and responses. Furthermore, we introduce a Stochastic Token Modulation (STM) mechanism to regulate the intervention probability. STM enables precise control over impairment severity while mitigating the instability of conventional vector methods. Comprehensive experiments demonstrate that StsPatient significantly outperforms baselines in both clinical authenticity and severity controllability.

Although artificial intelligence (AI) agents are increasingly proposed to support potentially longitudinal health tasks, such as symptom management, behavior change, and patient support, most current implementations fall short of facilitating user intent and fostering accountability. This contrasts with prior work on supporting longitudinal needs, where follow-up, coherent reasoning, and sustained alignment with individuals' goals are critical for both effectiveness and safety. In this paper, we draw on established clinical and personal health informatics frameworks to define what it would mean to orchestrate longitudinal health interactions with AI agents. We propose a multi-layer framework and corresponding agent architecture that operationalizes adaptation, coherence, continuity, and agency across repeated interactions. Through representative use cases, we demonstrate how longitudinal agents can maintain meaningful engagement, adapt to evolving goals, and support safe, personalized decision-making over time. Our findings underscore both the promise and the complexity of designing systems capable of supporting health trajectories beyond isolated interactions, and we offer guidance for future research and development in multi-session, user-centered health AI.

This paper reports an unexpected finding: in a deterministic hyperdimensional computing (HDC) architecture based on Galois-field algebra, a path-dependent semantic selection mechanism emerges, equivalent to spike-timing-dependent plasticity (STDP), with magnitude predictable a priori by a closed-form expression matching large-scale measurements. This addresses limitations of modern AI including catastrophic forgetting, learning stagnation, and the Binding Problem at an algebraic level. We propose VaCoAl (Vague Coincident Algorithm) and its Python implementation PyVaCoAl, combining ultra-high-dimensional memory with deterministic logic. Rooted in Sparse Distributed Memory, it resolves orthogonalisation and retrieval in high-dimensional binary spaces via Galois-field diffusion, enabling low-load deployment. VaCoAl is a memory-centric architecture prioritising retrieval and association, enabling reversible composition while preserving element independence and supporting compositional generalisation with a transparent reliability metric (CR score). We evaluated multi-hop reasoning on about 470k mentor-student relations from Wikidata, tracing up to 57 generations (over 25.5M paths). Using HDC bundling and unbinding with CR-based denoising, we quantify concept propagation over DAGs. Results show a reinterpretation of the Newton-Leibniz dispute and a phase transition from sparse convergence to a post-Leibniz "superhighway", from which structural indicators emerge supporting a Kuhnian paradigm shift. Collision-tolerance mechanisms further induce path-based pruning that favors direct paths, yielding emergent semantic selection equivalent to STDP. VaCoAl thus defines a third paradigm, HDC-AI, complementing LLMs with reversible multi-hop reasoning.

We introduce Context Kubernetes, an architecture for orchestrating enterprise knowledge in agentic AI systems, with a prototype implementation and eight experiments. The core observation is that delivering the right knowledge, to the right agent, with the right permissions, at the right freshness -- across an entire organization -- is structurally analogous to the container orchestration problem Kubernetes solved a decade ago. We formalize six core abstractions, a YAML-based declarative manifest for knowledge-architecture-as-code, a reconciliation loop, and a three-tier agent permission model where agent authority is always a strict subset of human authority. On synthetic seed data, we compare four governance baselines of increasing strength: ungoverned RAG, ACL-filtered retrieval, RBAC-aware routing, and the full architecture. Each layer contributes a different capability: ACL filtering eliminates cross-domain leaks, intent routing reduces noise by 19 percentage points, and only the three-tier model blocks all five tested attack scenarios -- the one attack RBAC misses is an agent sending confidential pricing via email, which RBAC cannot distinguish from ordinary email. TLA+ model-checking verifies safety properties across 4.6 million reachable states with zero violations. A survey of four major platforms (Microsoft, Salesforce, AWS, Google) documents that none architecturally isolates agent approval channels. We identify four properties that make context orchestration harder than container orchestration, and argue these make the solution more valuable.

LLM evaluations drive which models get deployed, which safety standards get adopted, and which research conclusions get published. Yet these scores carry hidden uncertainty: rephrasing the prompt, switching the judge model, or changing the temperature can shift results enough to flip rankings and reverse conclusions. Standard confidence intervals ignore this variance, producing under-coverage that worsens with more data. The same unmeasured variance creates an exploitable surface for benchmarks: model developers can optimize against measurement noise rather than genuine performance (some have infamously done so, see \citep{boyeau2025leaderboard}). This paper decomposes LLM pipeline uncertainty into its sources, distinguishes variance that shrinks with more data from sensitivity to researcher design choices, and uses design-study projections to reduce total error. Across ideology annotation, safety classification, MMLU benchmarking, and a human-validated propaganda audit, the decomposition reveals that the dominant variance source differs by domain and scoring method. On MMLU, optimized budget allocation halves estimation error at equivalent cost. On the propaganda task, the recommended pipeline outperforms 73\% of single-configuration alternatives against a human baseline. A small-sample pilot is sufficient to derive confidence intervals that approach nominal coverage and to identify which design changes yield the largest precision gains.

Ecological Momentary Assessment (EMA) is widely used to study adolescents' experiences; yet, how the design of EMA platforms shapes engagement, research practices, and power dynamics in youth studies remains under-examined. We developed a youth-centered EMA platform prioritizing youth engagement and researcher support, and evaluated it through a case study on a longitudinal investigation with adolescent twins focused on mental health and sleep behavior. Interviews with the research team examined how the platform design choices shaped participant onboarding, sustained engagement, risk monitoring, and data interpretation. The app's teen-centered design and gamified features sustained teen engagement, while the web portal streamlined administrative oversight through a centralized dashboard. However, technical instability and rigid data structures created significant hurdles, leading to privacy concerns among parents and complicating the researchers' ability to analyze raw usage metadata. We provide actionable interaction design guidelines for developing EMA platforms that prioritize youth agency, ethical practice, and research goals.

Ensuring the timeliness and reliability of master data remains a persistent challenge for many organizations. To mitigate these quality deficits, organizations frequently rely on commercial data brokers. However, this practice creates strategic dependencies and poses significant business risks, particularly as providers typically disclaim liability for the accuracy of the supplied data. In contrast, modern data ecosystems enable the trusted sharing of data assets with strong data sovereignty. In this paper, we address this paradigm shift by deriving a nascent design theory for trustworthy master data management based on self-sovereign identity. The theory is grounded through a hermeneutic literature review combined with industry expert interviews and instantiated through integration into a reference architecture for data spaces. Following an evaluation through additional industry expert interviews, our work provides a framework for a trustworthy master data management in data ecosystems that is reliable, sovereign, and accountable.

Fine-tuning pretrained image classifiers is standard practice, yet which individual samples are forgotten during this process, and whether forgetting patterns are stable or architecture dependent, remains unclear. Understanding these dynamics has direct implications for curriculum design, data pruning, and ensemble construction. We track per-sample correctness at every epoch during fine-tuning of ResNet-18 and DeiT-Small on a retinal OCT dataset (7 classes, 56:1 imbalance) and CUB-200-2011 (200 bird species), fitting Ebbinghaus-style exponential decay curves to each sample's retention trace. Five findings emerge. First, the two architectures forget fundamentally different samples: Jaccard overlap of the top 10 percent most-forgotten is 0.34 on OCTDL and 0.15 on CUB-200. Second, ViT forgetting is more structured (mean $R^2 = 0.74$) than CNN forgetting ($R^2 = 0.52$). Third, per-sample forgetting is stochastic across random seeds (Spearman $ρ\approx 0.01$), challenging the assumption that sample difficulty is an intrinsic property. Fourth, class-level forgetting is consistent and semantically interpretable: visually similar species are forgotten most, distinctive ones least. Fifth, a sample's loss after head warmup predicts its long-term decay constant ($ρ= 0.30$ to $0.50$, $p < 10^{-45}$). These findings suggest that architectural diversity in ensembles provides complementary retention coverage, and that curriculum or pruning methods based on per-sample difficulty may not generalize across runs. A spaced repetition sampler built on these decay constants does not outperform random sampling, indicating that static scheduling cannot exploit unstable per-sample signals.

Contextual causal reasoning is a critical yet challenging capability for Large Language Models (LLMs). Existing benchmarks, however, often evaluate this skill in fragmented settings, failing to ensure context consistency or cover the full causal hierarchy. To address this, we pioneer METER to systematically benchmark LLMs across all three levels of the causal ladder under a unified context setting. Our extensive evaluation of various LLMs reveals a significant decline in proficiency as tasks ascend the causal hierarchy. To diagnose this degradation, we conduct a deep mechanistic analysis via both error pattern identification and internal information flow tracing. Our analysis reveals two primary failure modes: (1) LLMs are susceptible to distraction by causally irrelevant but factually correct information at lower level of causality; and (2) as tasks ascend the causal hierarchy, faithfulness to the provided context degrades, leading to a reduced performance. We belive our work advances our understanding of the mechanisms behind LLM contextual causal reasoning and establishes a critical foundation for future research. Our code and dataset are available at https://github.com/SCUNLP/METER .

Dual-encoder Vision-Language Models (VLMs) such as CLIP are often characterized as bag-of-words systems due to their poor performance on compositional benchmarks. We argue that this limitation may stem less from deficient representations than from the standard inference protocol based on global cosine similarity. First, through controlled diagnostic experiments, we show that explicitly enforcing fine-grained region-segment alignment at inference dramatically improves compositional performance without updating pretrained encoders. We then introduce a lightweight transformer that learns such alignments directly from frozen patch and token embeddings. Comparing against full fine-tuning and prior end-to-end compositional training methods, we find that although these approaches improve in-domain retrieval, their gains do not consistently transfer under distribution shift. In contrast, learning localized alignment over frozen representations matches full fine-tuning on in-domain retrieval while yielding substantial improvements on controlled out-of-domain compositional benchmarks. These results identify global embedding matching as a key bottleneck in dual-encoder VLMs and highlight the importance of alignment mechanisms for robust compositional generalization.

Developing non-collaborative dialogue agents traditionally requires the manual, unscalable codification of expert strategies. We propose \ours, a method that leverages large language models to autonomously induce both strategy actions and planning logic directly from raw transcripts. METRO formalizes expert knowledge into a Strategy Forest, a hierarchical structure that captures both short-term responses (nodes) and long-term strategic foresight (branches). Experimental results across two benchmarks show that METRO demonstrates promising performance, outperforming existing methods by an average of 9%-10%. Our further analysis not only reveals the success behind METRO (strategic behavioral diversity and foresight), but also demonstrates its robust cross-task transferability. This offers new insights into building non-collaborative agents in a cost-effective and scalable way. Our code is available at https://github.com/Humphrey-0125/METRO.

Accurate air traffic prediction in the terminal airspace (TA) is pivotal for proactive air traffic management (ATM). However, existing data-driven approaches predominantly rely on time series-based forecasting paradigms, which inherently overlook critical aircraft state information, such as real-time kinematics and proximity to airspace boundaries. To address this limitation, we propose \textit{AeroSense}, a direct state-to-flow modeling framework for air traffic prediction. Unlike classical time series-based methods that first aggregate aircraft trajectories into macroscopic flow sequences before modeling, AeroSense explicitly represents the real-time airspace situation as \textit{a dynamic set of aircraft states}, enabling the direct processing of a variable number of aircraft instead of time series as inputs. Specifically, we introduce a situation-aware state representation that enables AeroSense to sense the instantaneous terminal airspace situation directly from microscopic aircraft states. Furthermore, we design a model architecture that incorporates masked self-attention to capture inter-aircraft interactions, together with two decoupled prediction heads to model heterogeneous flow dynamics across two key functional areas of the TA. Extensive experiments on a large-scale real-world airport dataset demonstrate that AeroSense consistently achieves state-of-the-art performance, validating that direct modeling of microscopic aircraft states yields substantially higher predictive fidelity than time series-based baselines. Moreover, the proposed framework exhibits superior robustness during peak traffic periods, achieves Pareto-optimal performance under dayparting multi-object evaluation, and provides meaningful interpretability through attention-based visualizations.

Hate speech detection in Devanagari-scripted social media memes presents compounded challenges: multimodal content structure, script-specific linguistic complexity, and extreme data scarcity in low-resource settings. This paper presents our system for the CHiPSAL 2026 shared task, addressing both Subtask A (binary hate speech detection) and Subtask B (three-class sentiment classification: positive, neutral, negative). We propose a hybrid cross-modal attention fusion architecture that combines CLIP (ViT-B/32) for visual encoding with BGE-M3 for multilingual text representation, connected through 4-head self-attention and a learnable gating network that dynamically weights modality contributions on a per-sample basis. Systematic evaluation across eight model configurations demonstrates that explicit cross-modal reasoning achieves a 5.9% F1-macro improvement over text-only baselines on Subtask A, while uncovering two unexpected but critical findings: English-centric vision models exhibit near-random performance on Devanagari script, and standard ensemble methods catastrophically degrade under data scarcity (N nearly equal to 850 per fold) due to correlated overfitting. The code can be accessed at https://github.com/Tri-Yantra-Technologies/MEME-Fusion/

Customer service chatbots are increasingly expected to serve not merely as reactive support tools for users, but as strategic interfaces for harvesting high-value information and business intelligence. In response, we make three main contributions. 1) We introduce and define a novel task of Proactive Information Probing, which optimizes when to probe users for pre-specified target information while minimizing conversation turns and user friction. 2) We propose PROCHATIP, a proactive chatbot framework featuring a specialized conversation strategy module trained to master the delicate timing of probes. 3) Experiments demonstrate that PROCHATIP significantly outperforms baselines, exhibiting superior capability in both information probing and service quality. We believe that our work effectively redefines the commercial utility of chatbots, positioning them as scalable, cost-effective engines for proactive business intelligence. Our code is available at https://github.com/SCUNLP/PROCHATIP.

We study the problem of characterizing the stability of Kullback-Leibler (KL) divergence under Gaussian perturbations beyond Gaussian families. Existing relaxed triangle inequalities for KL divergence critically rely on the assumption that all involved distributions are Gaussian, which limits their applicability in modern applications such as out-of-distribution (OOD) detection with flow-based generative models. In this paper, we remove this restriction by establishing a sharp stability bound between an arbitrary distribution and Gaussian families under mild moment conditions. Specifically, let $P$ be a distribution with finite second moment, and let $\mathcal{N}_1$ and $\mathcal{N}_2$ be multivariate Gaussian distributions. We show that if $KL(P||\mathcal{N}_1)$ is large and $KL(\mathcal{N}_1||\mathcal{N}_2)$ is at most $ε$, then $KL(P||\mathcal{N}_2) \ge KL(P||\mathcal{N}_1) - O(\sqrtε)$. Moreover, we prove that this $\sqrtε$ rate is optimal in general, even within the Gaussian family. This result reveals an intrinsic stability property of KL divergence under Gaussian perturbations, extending classical Gaussian-only relaxed triangle inequalities to general distributions. The result is non-trivial due to the asymmetry of KL divergence and the absence of a triangle inequality in general probability spaces. As an application, we provide a rigorous foundation for KL-based OOD analysis in flow-based models, removing strong Gaussian assumptions used in prior work. More broadly, our result enables KL-based reasoning in non-Gaussian settings arising in deep learning and reinforcement learning.

We present a discriminative multimodal reward model that scores all candidate responses in a single forward pass. Conventional discriminative reward models evaluate each response independently, requiring multiple forward passes, one for each potential response. Our approach concatenates multiple responses with separator tokens and applies cross-entropy over their scalar scores, enabling direct comparative reasoning and efficient $N$-way preference learning. The multi-response design also yields up to $N\times$ wall-clock speedup and FLOPs reduction over conventional single-response scoring. To enable $N$-way reward evaluation beyond existing pairwise benchmarks, we construct two new benchmarks: (1) MR$^2$Bench-Image contains human-annotated rankings over responses from 8 diverse models; (2) MR$^2$Bench-Video is a large-scale video-based reward benchmark derived from 94K crowdsourced pairwise human judgments over video question-answering spanning 19 models, denoised via preference graph ensemble. Both benchmarks provide 4-response evaluation variants sampled from the full rankings. Built on a 4B vision-language backbone with LoRA fine-tuning and a lightweight MLP value head, our model achieves state-of-the-art results on six multimodal reward benchmarks, including MR$^2$Bench-Image, MR$^2$Bench-Video, and four other existing benchmarks. Our model outperforms existing larger generative and discriminative reward models. We further demonstrate that our reward model, when used in reinforcement learning with GRPO, produces improved policy models that maintain performance across standard multimodal benchmarks while substantially improving open-ended generation quality, outperforming a single-response discriminative reward model (RM) baseline by a large margin in both training stability and open-ended generation quality.

We develop a queueing-theoretic framework to model the temporal evolution of cyber-attack surfaces, where the number of active vulnerabilities is represented as the backlog of a queue. Vulnerabilities arrive as they are discovered or created, and leave the system when they are patched or successfully exploited. Building on this model, we study how automation affects attack and defense dynamics by introducing an AI amplification factor that scales arrival, exploit, and patching rates. Our analysis shows that even symmetric automation can increase the rate of successful exploits. We validate the model using vulnerability data collected from an open source software supply chain and show that it closely matches real-world attack surface dynamics. Empirical results reveal heavy-tailed patching times, which we prove induce long-range dependence in vulnerability backlog and help explain persistent cyber risk. Utilizing our queueing abstraction for the attack surface, we develop a systematic approach for cyber risk mitigation. We formulate the dynamic defense problem as a constrained Markov decision process with resource-budget and switching-cost constraints, and develop a reinforcement learning (RL) algorithm that achieves provably near-optimal regret. Numerical experiments validate the approach and demonstrate that our adaptive RL-based defense policies significantly reduce successful exploits and mitigate heavy-tail queue events. Using trace-driven experiments on the ARVO dataset, we show that the proposed RL-based defense policy reduces the average number of active vulnerabilities in a software supply chain by over 90% compared to existing defense practices, without increasing the overall maintenance budget. Our results allow defenders to quantify cumulative exposure risk under long-range dependent attack dynamics and to design adaptive defense strategies with provable efficiency.

Interpreting chest X-rays is inherently challenging due to the overlap between anatomical structures and the subtle presentation of many clinically significant pathologies, making accurate diagnosis time-consuming even for experienced radiologists. Recent radiology-focused foundation models, such as LLaVA-Rad and Maira-2, have positioned multi-modal large language models (MLLMs) at the forefront of automated radiology report generation (RRG). However, despite these advances, current foundation models generate reports in a single forward pass. This decoding strategy diminishes attention to visual tokens and increases reliance on language priors as generation proceeds, which in turn introduces spurious pathology co-occurrences in the generated reports. To mitigate these limitations, we propose Category-Wise Contrastive Decoding (CWCD), a novel and modular framework designed to enhance structured radiology report generation (SRRG). Our approach introduces category-specific parameterization and generates category-wise reports by contrasting normal X-rays with masked X-rays using category-specific visual prompts. Experimental results demonstrate that CWCD consistently outperforms baseline methods across both clinical efficacy and natural language generation metrics. An ablation study further elucidates the contribution of each architectural component to overall performance.

We propose a Hanbury Brown-Twiss interferometer for a $ν=2/5$ fractional quantum Hall edge system, in which quasiparticles tunnel between two co-propagating edge modes. In contrast to the previously studied anyonic Fabry-Pérot and Mach-Zehnder interferometers, the proposed setup relies purely on two-particle interference rather than single-particle interference. In the weak-tunneling regime, we employ a bosonized edge theory together with Keldysh perturbation theory to evaluate the cross-correlation of the tunneling currents. In the large-device limit, we obtain an analytic expression for the flux-dependent noise, whose structure closely resembles that of an electronic HBT interferometer, but with the electron charge replaced by the fractional charge $e^{\star}=e/3$ and with scaling dimensions characteristic of the fractional edge modes. In this limit, the explicit anyonic exchange phases cancel, whereas when the device size becomes comparable to the thermal length, the cross-correlation may recover a more explicit dependence on the anyonic statistical angle.

We study dense mixtures of passive and active self-aligning disks with isotropic or anisotropic mobility. We find that the passive fraction controls an order-disorder transition that is continuous in the isotropic case and discontinuous in the anisotropic one. A mean-field equation derived from the microscopic heading dynamics captures this dichotomy. Near the transition, both ordered regimes can exhibit multiple metastable oscillating or rotating states, depending on the spatial arrangement of passive particles and lattice defects, but with different transient dynamics: Systems with isotropic mobility visit multiple long-lived attractors during each simulation while systems with anisotropic mobility are trapped by a single attractor. Our results reveal the passive fraction as a physically relevant control parameter in active systems, leading to rich self-organizing dynamics.

Kardar-Parisi-Zhang (KPZ) scaling has been observed in discrete polariton lattices, enabled by engineered band structures that stabilize the condensate. Whether this universality extends to intrinsically continuous systems with natural noise regularization remains an open question. We propose and numerically demonstrate KPZ scaling in a continuous quasi-one-dimensional polariton condensate stabilized by optical confinement in the transversal direction. Large-scale simulations of the stochastic Gross-Pitaevskii equation, with experimentally relevant parameters, reveal temporal and spatial scaling exponents of the two-point phase correlation function betaC = 0.30(5) and alfaC =0.46(8), and Tracy-Widom one-point phase fluctuation statistics, yielding robust KPZ dynamics intrinsic to the continuous polariton fluid.

Collective spin systems -- spin ensembles coupled to a common reservoir and effectively described by a single macrospin -- play an important role in both atomic and solid-state physics. Their intrinsic nonlinearity gives rise to multiple long-lived metastable states that ultimately relax to a unique most probable state. This dominant state can change with a control parameter, leading to first-order phase transitions. We develop a real-time instanton approach based on quantum quasiprobability dynamics that captures the stationary state in the large-spin limit and the asymptotic scaling of relaxation rates. We further show that these features are not accurately described by the previously applied semiclassical Wigner approach due to its neglect of non-Gaussian fluctuations.

Hydrodynamic electron transport, namely, the electric behaviors in solid materials at the macroscopic level are similar to the fluid hydrodynamics when the momentum-conserving electron-electron scattering plays the leading role, has got much attention in the past ten years. However, most of previous studies mainly focus on the electric properties. In this work, the thermal behaviors of hydrodynamic electron transport in a homogeneous 2D Corbino disk geometry is studied by the electron Boltzmann transport equation (eBTE) coupled with the Poisson equation under the magnetic field perpendicular to disk plane. Results show that in the electron hydrodynamic regime, the heat flux deflection phenomenon appears under the radial electric field or temperature gradient, namely, the heat flux no longer flows only along the radial direction and there is heat flux in the tangential direction of the radius. Heat flux deflection phenomenon is suppressed by momentum-relaxing scattering process and promoted by momentum-conserving scattering process. When an electric potential gradient or temperature gradient in the same direction is applied separately, the direction of heat flux is reversed in the electron hydrodynamic regime.

Quantum dot based platforms offer a promising route towards realizing the Kitaev chain Hamiltonian hosting Majorana bound states (MBSs). Poor man's MBSs arise in a two-site Kitaev chain when the parameters of the system are fine-tuned to the sweet spot. Based on our previous work [Phys. Rev. B 111, 155410 (2025)], we consider a microscopic model for the Kitaev chain based on quantum dots with proximity effect embedded in a photonic cavity. We find that the photon coupling in the microscopic model yields an effective Hamiltonian where the cavity affects the pairing term. However, we demonstrate that even in this case, it is possible to screen particle interactions and reach the sweet spot condition for the emergence of the poor man's MBSs. In particular, we find that attractive particle interactions can be canceled for the cavity prepared in the zero-photon state, while repulsive ones can be screened with a cavity prepared in the one-photon state. Furthermore, in case of a large number of photons in the cavity, we find that the hopping amplitudes are suppressed resulting in a degenerate spectrum. This motivates the use of quantum light for engineering poor man's MBSs with cavity embedding.

DNA-coated particles are promising as building blocks for functional and finite-sized assemblies because they can be programmed with orthogonal interactions owing to the sequence-specific hybridization of DNA strands. To fully exploit this programmability, it is important to develop particles with coatings that incorporate multiple distinct DNA sequences in tunable ratios and to understand how the coating composition influences self-assembly. Here, we compared two strategies to graft multiple DNA sequences in tunable and well-defined ratios on micron-sized colloidal particles. We found that a method based on click chemistry yielded mixed coatings with large batch-to-batch variation in the composition, while a method based on isothermal DNA polymerization produced coatings of predictable composition with a precision of a few percent, but requires reaction rate measurements for each new sequence in the coating. Our self-assembly experiments showed that, even with precise control over coating composition, equilibrium co-assembly of multiple types of DNA-coated particles is limited by the number of interactions that are reversible within the same narrow temperature window. This finding highlights the need to explicitly incorporate sequential assembly pathways into structure design, with coating composition dictating the order of binding events, Together, our results show how systematic tuning of interaction strength and sequential assembly through multispecific DNA coatings is a prerequisite for the experimental realization of finite-sized and dynamic structures that have so far remained largely theoretical.

Magnetic skyrmions are emerging as promising candidates for next-generation information technologies, while the realization of scalable skyrmion lattices with tailored configurations is essential for advancing fundamental skyrmion physics and developing future applications. Here we achieved the controllable generation and regulation of a large-area, highly oriented skyrmion track array (STA) in ferromagnetic Fe3GaTe2 using a vector magnetic field manipulation technique. The orientation and ordering of STA, along with the types and density of skyrmions, are precisely controlled by modulating parameters during the manipulation. The critical roles of in-plane magnetic fields and Dzyaloshinskii-Moriya interaction in STA generation is further confirmed by micromagnetic simulation. Our findings develop a strategy for engineering large-area and highly-oriented skyrmion configurations, offering a new pathway for the future application of next-generation spintronic and information technologies.

While the geometry of equilibrium states and driven non-equilibrium processes is clearly understood, a geometric description for relaxation towards equilibrium is still lacking. Here, we propose a thermo-geometric measure based on the Rayleigh quotient, reformulating relaxation as a fundamental competition between entropic stiffness and frictional dissipation. Taking a van der Waals gas with two dissipation channels as an example, we explicitly demonstrate its relaxation landscape. Particularly, we find that upon approaching the critical temperature $T_c$, the slow-mode relaxation rate vanishes linearly as $λ_s \propto (T-T_c)/T_c$, indicating critical slowing down. This study completes the thermodynamic geometry framework, providing a general tool for characterizing the relaxation dynamics of complex systems.

Fluctuation theorems show how coarse graining transforms microscopic symmetry into observable irreversibility. Here we ask whether an analogous symmetrybased diagnostic can be constructed for financial markets. At the microscopic level, each transaction pairs a buyer and a seller, whereas trading decisions are typically made from coarse-grained price histories. Using symmetric takeprofit and stop-loss rules, we compare the holding-time distributions of long and short trading ensembles generated from the same price series. Across equityindices, individual stocks and cryptocurrencies, the log-ratio of the two distributions shows a robust crossover. It remains nearly constant at short durations but becomes linear in the tail, implying an exponential directional asymmetry. The tail slope defines an effective market temperature, an operational measure of fluctuation intensity on the chosen observation scale. A Bachelier first-passage benchmark captures the exponential tails but not the asymmetry, because long and short positions share the same leading decay rate. By contrast, short-time correlations between overlapping positions provide a minimal mechanism for the asymmetry by generating direction-dependent subleading relaxation spectra in a coarse-grained Markov description. Together, these results establish a fluctuation-theorem-like diagnostic of irreversibility in financial markets and, more broadly, in complex systems accessible only through coarse-grained observables.

Interactions between fluids and granular materials are prevalent on the Earth's surface. In the case of fluid flow over a sediment bed, the fluid imparts a shear stress to the granular materials. When the applied shear stress is above a critical value, the grains become entrained in the fluid flow. Prior experimental studies have shown that granular beds subjected to a sub-critical fluid flow can strengthen in the same direction as the sub-critical flow. In contrast, granular beds can become weaker in the direction opposite to the sub-critical fluid flow. To investigate the grain-scale mechanisms that control directional strengthening and weakening, we perform discrete element method (DEM) simulations of granular beds subjected to model fluid flows in two (2D) and three (3D) dimensions with varied inter-particle static friction coefficients and conditioning flow speeds. In these studies, the sub-critical grain motion does not cause significant bed compaction. Instead, we find that the strength of a granular bed in a particular direction is highly correlated with the fraction of {\it surface} grains that can be dislodged by a fluid force applied in that direction. Further, the anisotropic bed strength only persists over a finite time scale that is set by the Shields number. We also show that inter-particle static friction is not required for bed strength anisotropy, but varying the friction affects the magnitude of the anisotropy. This research enhances the grain-scale understanding of erosion of granular beds caused by fluid flows and underscores the importance of tracking the history of the fabric of the bed surface since it couples strongly to bed strength.

We present a meshless numerical method for simulating the dynamics of axisymmetric vesicles in a viscous medium. Key innovations include: (1) adaptive reparameterization based on local length scales, reducing the number of required harmonics; (2) gauge dynamics for maintaining optimal parameterization; (3) error control near the symmetry axis; and (4) spectrally accurate quadrature schemes for singular integrals. The method achieves high accuracy and computational efficiency for simulating lipid bilayer dynamics and related problems in soft matter physics.

Active colloidal microparticles require reliable actuation to sustain directed motion. Light-based propulsion is particularly attractive as it provides persistent energy supply and enables direct spatiotemporal control. Here, we introduce 3D-printable particles with symmetry-broken refractive index profiles (SBRIP particles) that achieve propulsion through direct momentum transfer from asymmetric light refraction. Internal refractive-index gradients provide optical symmetry breaking independent of external shape, fundamentally decoupling propulsion from particle geometry. Geometrically symmetry-broken particles with a homogeneous refractive index are another special case, where propulsion originates from refractive contrast at the boundary instead of within the particle. Unlike conventional systems relying on absorption or reflection, this transparency-based mechanism minimizes heating and mitigates shadowing in bulk suspensions. We present a theoretical framework for refractive propulsion as well as numerical simulations of the SBRIP particles using raytracing and the finite volume method. This is complemented by experiments, validating the momentum transfer mechanism using particles with geometric symmetry breaking. The high transparency of our particles ensures deep light penetration, enabling the realization of volumetric active matter. This opens pathways toward adaptive nonlinear optical materials where light-driven particle reorganization modulates the local refractive index, establishing a dynamic feedback loop between the optical field and the material structure.

Modelling micro- and meso-scopic scale thermodynamic and transport properties of soft condensed matter hinges upon its representation. This is especially relevant for polar solvents such as water, since these require effective representation of their dielectric nature as driven by molecular charge distributions and molecular network structuring. The dielectric nature of a medium leads to complex phenomena such as local polarisability response and restructuring near interfaces in reaction to changes in local charge distributions. Inclusion of such phenomena when using larger-than-atomistic techniques such as coarse-grained molecular dynamics (CG-MD) and dissipative particle dynamics (DPD) is still an open question, to which we provide a novel way to consider and justify the necessary and suitable coarse-graining level, enabling us to compare new polar CG models' performance against that of an underlying atomistic model. We polarise our previous non-polar nDPD water model to prepare it for use in simulations of liquid electrolytes as well as solvated organic membranes and measure its fitness to serve as a dielectric medium by comparing its properties to those of the TIP3P water model, while simultaneously observing changes to properties already represented well by the non-polar model.

We address a fundamental question: under which conditions do the dynamics and thermodynamics of open chemical reaction networks (CRNs), grounded on the notion of idealized chemostats that exchange selected species, emerge from underlying closed CRNs? While open CRNs provide the standard framework to describe out-of-equilibrium chemical systems, real systems are finite and ultimately relax to equilibrium, leaving the status of this description conceptually unresolved. Here we show that open-CRN behavior arises as an asymptotic regime of closed CRNs when two minimal and physically transparent conditions are met: a time-scale separation, whereby fast reactions effectively act as exchange mechanisms, and an abundance separation, whereby a subset of species behaves as chemostats with diverging chemical capacity. In this regime, both the stochastic dynamics and the thermodynamic structure \ -- including local detailed balance, entropy production, and free-energy balance \ -- emerge to leading order from the underlying closed CRN. Our results apply to arbitrary stoichiometries. They show that chemostats need not be introduced as external idealizations, but instead arise as emergent thermodynamic structures within closed systems, providing a unified and physically grounded foundation for the nonequilibrium thermodynamics of CRNs.

The response of a complex system to a slow varying external force often displays a jump discontinuity in the order parameter near the critical point. However, this discontinuity is not usually a single jump but rather breaks into smaller jumps which makes it difficult to locate the critical point on approaching its vicinity based only on simulations, in the absence of exact results. Our work is a small effort in understanding these breaks in jump through the hysteretic response of a classical Ising spin system to an external field, $h$, in the context of a nonequilibrium zero-temperature random field Ising model on dilute systems. We consider a Bethe lattice with coordination number, $z = 4$, and dilute a fraction $(1-c)$ of the sites. Therefore the lattice now consists of sites with varying $z = 4, 3, 2, 1$ and possibly few isolated sites $(z=0)$, depending on the concentration $c$. We obtain the exact solution of the magnetization curve, $m(h)$ vs $h$, for the entire lattice as well as for each sublattice of different $z$ coordinated sites, $m_4(h), m_3(h), m_2(h), m_1(h), m_0(h)$. The discontinuity in total magnetization is the result of the superposition of the jumps of different $z$ coordinated sites and observed at the same value of external field, $h_{crit}$. The dominant contribution to the jump comes from those sites with higher concentration and larger $z$. However, the triggering sites responsible for large jumps are mostly $z\ge3$. We test this on cubic lattices as well, where exact results are not available. We hope our analysis will help in understanding fluctuations around a jump in numerical simulations as well as experiments.

Magnetic tunnel junction (MTJ) is the key component to enable information access and increasing number of MTJs is integrated to develop high-density spintronic devices. However, continuous miniaturization of the conventional MTJs is hindered by stray magnetic fields. Altermagnets, combining the advantages of both ferromagnets and antiferromagnets, provide a promising alternative to fabricate versatile MTJs with exotic properties, such as giant spin splitting, high intrinsic frequency, and absence of stray fields. Inspired by the altermagnetic metal candidate KV2Se2O reported recently, we design an altermagnetic tunnel junction (AMTJ) based on KV2Se2O/SrTiO3/KV2Se2O. Using density functional theory combined with non-equilibrium Green's function, we investigate the layer-dependent quantum transport properties and the tunneling magnetoresistance (TMR) of such AMTJ device. Our calculated results reveal that the transmission of the AMTJ device exhibits a pronounced oscillation behavior dependent on the number of layers of the SrTiO3 semiconductor, which is attributed to the interface configuration determined by parity of the layer number. In odd-layer devices, the electron-rich O-Se interface exhibits a smooth effective potential and enables transverse momentum (k||) transport channels, leading to enhanced transmission. In contrast, in even-layer devices, the Ti-Se interface presents a steeper effective potential, impeding quantum transport through transverse momentum (k||) channels. A giant TMR of 4.6*10^7% is predicted to be realized by using a 4-layer SrTiO3. Our findings not only provide physical understanding relevant to the quantum transport in AMTJs, but also unveil that the barrier interface engineering is a strategy to tune the magnetoelectric performance.

While significant advancements have been made in weak magnetic field detection, conventional high-sensitivity techniques are often limited by requirements for cryogenic operation or bulky setups. In this work, we develop a sensitive alternating magnetic field sensor based on a coupled system of an active microwave cavity and yttrium iron garnet (YIG), with the components implemented on printed circuit boards (PCBs). By introducing electrically tunable gain to compensate for cavity losses, we substantially improve both the quality factor and the signal intensity. Under the coupled system, Floquet modulation is induced by the alternating magnetic field, allowing for weak field detection by driving a specific hybrid mode and measuring the resulting Floquet sidebands. This miniaturized device operates at room temperature, achieving a detection limit of 121 pT/\sqrt{Hz}.

Nanofluidic devices prompts unconventional ion transports appealing to energy and information technologies, thanks to the susceptibility of confined electric double layers (EDL) to various external physical fields. Although experimental studies advance rapidly, the rationalization of field-tunable nanofluidic transports has not reached a formalized and unified level. Here we formally reformulate the Poisson-Boltzmann theory and reveal distinct EDL regimes on the parameter space. Based on the regime classification, we establish a formal framework for the tunable nanofluidic transport, which reproduces the observed conductivity-concentration scaling behaviors, rationalizes the ionic transistors with reconfigurable polarities, and predicts two fundamental thermodynamic limits for electrostatic modulation (60 mV/dec and 120 mV/dec). Being accurate, generalizable and extensible, this framework can account for a wide range of ion transports in confined spaces.

We study the finite-size Kuramoto model of all-to-all coupled phase oscillators with heterogeneous natural frequencies and characterize the minimal coupling strength required for the existence of a fully phase-locked equilibrium (in a co-rotating frame). To remove the degeneracy due to uniform phase shifts, we move to a reduced co-rotating frame and assess stability through the Jacobian of the reduced system: a fully phase-locked state is stable when this Jacobian is negative definite. This defines a stability region in the phase space. The Kuramoto vector field maps this region to a convex set in frequency space, so a fully-locked state at coupling $K$ exists exactly when the rescaled frequency vector $\hat{\mathbfω}/K$ lies inside that convex image. The critical coupling $K_{\ell}$ is defined as the smallest coupling strength for which a fully phase-locked equilibrium exists; geometrically, it corresponds to the first intersection of the ray $t\hat{\mathbfω}$ with the boundary of this convex set. Building on this convex-geometric structure, we construct an explicit polytope from analytically computable boundary points of the stability region, providing a closed-form upper bound $K_b \ge K_{\ell}$. The bound is exact for frequencies aligned with polytope vertices and offers a fully explicit outer approximation for general frequency vectors. While not uniformly sharp in a quantitative sense, this construction exposes the underlying geometry of stable fully phase-locking solutions. These results provide a practical use the convex-geometric structure underlying stable fully-locked states in the Kuramoto model.

Phononic silicon structures have emerged as an integrable and scalable nanosystem for tailoring thermal transport. However, their widespread adoption has been limited by their complex fabrication pathways. Alongside, the reliable characterization of thermal properties in suspended nanostructured films remains challenging, as thermal contact resistances often hinder the accuracy of measurements. In this work, we demonstrate a clear and controllable reduction of thermal conductivity in nanopatterned silicon membranes. A block copolymer self-assembly approach is employed to fabricate nanoholed silicon films with a pitch of 63 nm and hole diameters of 35 nm. Additionally, we introduce an extension of the three-probe technique that enables robust, quantitative, and spatially resolved thermal conductivity measurements in complex thin-film systems, accounting for thermal contact artifacts. The method is validated through measurements on unpatterned 40 nm-thick silicon thin films between 30 and 350 K, yielding a room-temperature thermal conductivity of 46.5 W/m.K. Finally, we further show that controlled etching of the nanoholes provides a powerful means to tune thermal transport in the overall studied temperature range, establishing hole etch depth control as an effective parameter in phononic silicon. Specifically, a fivefold reduction in thermal conductivity is achieved, reaching 7.3 W/m.K for fully etched-through membranes at room temperature.

Advances in nanotechnology now allow the creation of artificial atoms - engineered structures whose electronic states closely mimic those of real atoms. Understanding how these artificial atoms interact and bond is key to designing new materials with tailored electronic properties. Here, we use scanning tunnelling microscopy to visualise the bound states of nanostructures patterned in a two-dimensional molecular film featuring a parabolic band with multiple partial energy gaps. The lowest-energy states split off from the bottom of the band and resemble the familiar $s$ and $p$ orbitals of natural atoms, even bonding in the same way. Yet, artificial atoms go beyond this analogy: the gapped two-dimensional vacuum in which they reside gives rise to entirely new orbitals with no counterparts in real atoms. These quasi-one-dimensional localised states enrich the orbital vocabulary of chemistry, adding a new class of orbitals that are predominantly shaped by the surrounding electronic vacuum.

Understanding how the interaction range and various types of disorder affect the level statistics of many-body quantum systems and lead to the emergence of many-body localization (MBL) is a challenging open frontier. We study the level statistics of a variant of the spin-$1/2$ Haldane-Shastry model with $1/r^α$ interactions, where $α{\geq}0$ parametrizes the range of the interactions, in the presence of position disorder and/or random magnetic fields. We find that neither position disorder nor random magnetic fields alone yields pristine Poisson statistics in this long-range interacting system; however, Poisson statistics emerge in their combined presence, suggesting the emergence of MBL when both types of disorder coexist. Interestingly, once random magnetic fields break the $SU(2)$ symmetry, the strength of the position disorder, $δ$, appears to play an important role, as evidenced by an approximate scaling collapse of the disorder-averaged gap ratios that is parametrized in terms of a single parameter, $αδ$.

Utilization of Faraday rotation (FR) properties of topological materials offers a promising route toward novel magneto-optical devices. We systematically investigated the effect of Rashba spin-orbit coupling (SOC) on FR spectra in an extended Haldane model, which incorporates Rashba SOC and exchange splitting into the original spinless Haldane framework. Using the Kubo formalism, we calculated the FR spectra across the model's rich topological phase diagram. We found that in the Chern number C=2 region, in the absence of exchange splitting, the FR angle can exceed 4$^\circ$ and its peak position is tunable by the Rashba SOC. In contrast, with the inclusion of exchange splitting, a nearly flat FR profile emerges over a broad frequency range, and the FR peak values increase monotonically with the Rashba SOC strength. The Rashba SOC opens additional transition channels, whose net contribution constructively enhances the FR peak. Furthermore, we derived a low-energy effective Hamiltonian expanded up to quadratic terms, the results of which are in good agreement with tight-binding model calculations, thereby validating our numerical results. Our findings suggest that magneto-optical device characteristics can be designed and optimized through Rashba SOC engineering.

Solid phase of dense granular matter is inevitable because of jamming transition when the packing fraction or the pressure suffered is high enough. The experiment suggests that active Brownian granular matter will keep fluid phase even under the highest packing fraction (higher than the packing fraction of crystallization) if crystallization is prevented by mixing granular particles of different sizes. The findings encourage us to reconsider the role of activity in affecting the global dynamical properties of matter.

Chitosan is a highly versatile and sustainable polymer with a broad range of potential biological and materials engineering applications. Despite its versatility, the native brittleness of chitosan limits its broader utilization. This limitation can be addressed by blending chitosan with small-molecule additives to modulate its thermomechanical properties. We employ molecular dynamics (MD) simulations to investigate the mechanism underlying antiplasticization followed by plasticization at increasing water content. Decomposition of the elastic moduli reveals a competition between weakened polymer-polymer interactions and enhanced polymer-water interactions, with their relative strengths governing the resulting properties. We introduce a simple model incorporating dynamically accessible free volume regions as a key driver of polymer mobility, effectively capturing the (anti-)plasticization of elastic properties. We show that accessibility of free volume regions is enabled by connectivity of additive-accessible volume regions. This study provides new insights into the molecular interactions that dictate the properties of chitosan-water mixtures and may inform the rational design of chitosan-based materials and other hydrated biopolymers.

The spontaneous breaking of symmetries is a cornerstone of physics, defining the phases of matter from the cosmological scale to the quantum realm. In condensed matter, electronic orders are classified by their behavior under fundamental symmetries like spatial inversion (parity). While even-parity orders, such as conventional superconductivity and charge density waves, are ubiquitous, their odd-parity counterparts--predicted to host exotic phenomena such as gapless quasiparticle excitations and novel collective modes--are comparatively elusive states of quantum matter. Here, using high-resolution scanning tunneling microscopy and angle-resolved photoemission spectroscopy on the kagome metal CsV$_3$Sb$_5$, we report the discovery of an inversion symmetry-breaking $f$-wave charge bond order. We show that this phase, which preserves translation symmetry, is stabilized by the spontaneous opening of a spectral gap at a previously overlooked Dirac point, providing a textbook condensed-matter realization of the Gross-Neveu model for dynamical mass generation and parity breaking. Intriguingly, this $f$-wave order is itself a intervening phase, vanishing abruptly below a temperature of 10\,K and pointing to a subsequent transition into a `hidden' electronic state that is invisible to local STM probes. Our findings establish odd-parity charge order as a novel phase of matter, here, embedded within the intricate hierarchy of correlated electronic orders on the kagome lattice.

The single orbital, one-dimensional, Hatano-Nelson Hamiltonian provides deep insight into the physics of non-Hermiticity, resulting from asymmetric left/right hopping, and its connections to localization. In the absence of disorder, its single particle eigenvalues $E_α$ lie on an ellipse in the complex plane whose extent in the imaginary direction is controlled by the degree of asymmetry. When randomness is introduced, two sets of real eigenvalues emerge at the extremes of the largest and smallest real part of $E_α$. These real eigenvalues are associated with localized eigenvectors. For spinless fermions, increasing near-neighbor interactions first cause a transition to a charge density wave phase, and ultimately, on finite lattices, a collapse of all eigenvalues to the real axis. In this paper, we explore the presence of real eigenvalues in the interacting, two-particle sector for the spinful case (Hubbard model) in a two-chain (two-band) geometry with a Hermitian interchain hopping. Our key results are to obtain the ``phase" diagrams for the existence of a purely real spectrum, as a function of the interaction strength, degree of non-Hermiticity, and interchain hopping. We study the sensitivity to boundary conditions of the spectral properties of our two-chain model with winding number analysis and explore the relationship between PBC doublon states and OBC skin modes. To address the question of stability in such non-equilibrium systems, we solve the dynamics at low filling according to Lindbladian evolution and find that the non-Hermitian description is able to qualitatively describe such systems.

We theoretically study the direct and inverse spin Hall effects in a superconductor-normal metal-superconductor junction induced by a spin-orbit interaction that is invariant under spatial inversion. We show that a supercurrent induces a spin Hall effect, leading to a static spin accumulation with opposite polarizations at the two edges, analogous to that in normal conductors. For the inverse effect, we consider a spatially inhomogeneous static magnetic field and show that it induces an anomalous phase shift, which, in the presence of higher harmonics, results in a diode effect. Unlike Rashba systems, the present mechanism does not require broken structural inversion symmetry.

We present a theory of ballistic N/F/S and S/F/S junctions with a uniformly precessing magnetization, which generates long-range equal-spin superconducting correlations [Takahashi et al., Phys. Rev. Lett. 99, 057003 (2007), Houzet, Phys. Rev. Lett. 101, 057009 (2008)]. The non-equilibrium distribution of Andreev bound states leads to a strongly non-sinusoidal current-phase relationship for large precession angles. We derive detailed results for ballistic junctions involving partially and fully polarized ferromagnets. In the fully polarized half-metal limit, the magnetization precession switches the junction from an "off" state with vanishing subgap current to an "on" state with finite Andreev conductance and finite Josephson current.

Recent work has emphasized a subtlety of large- $N$ limits in AdS/CFT: a sequence of pure states in the microscopic theory need not remain pure with respect to the emergent algebra of observables. We study this phenomenon for Kourkoulou-Maldacena (KM) states in the double-scaling limit of the SYK model, and show that their ensemble-averaged algebraic description depends crucially on which observables survive the limit. For fermionic operators of size $N^{1/2}$, generic operators converge to the usual chord operators of double-scaled SYK. The resulting von Neumann algebra is the standard Type II$_1$ factor, and the KM pure states at infinite temperature converge to the tracial state, so generic probes lose access to microscopic purity. We then identify a class of operators adapted to the KM state that also survives the double-scaling limit. Since the KM state may be viewed as a projection inside the tracial state, these become dressed chord creation and annihilation operators. Once included, the limiting algebra becomes Type I$_\infty$ and the limiting state becomes pure. This gives a concrete example in which adding a sufficiently state-adapted operator to the emergent algebra restores access to the purity of the underlying state. We further show that correlators of the dressed operators admit exact modified chord-diagram rules, derive analytic expressions for uncrossed $2n$-point and crossed four-point functions, analyze their finite-temperature semiclassical and Schwarzian limits, study a deformation of the chord Hamiltonian that produces bound states and extends the correspondence with JT gravity plus an EOW brane to general brane tension, and identify an emergent $U(1)$ symmetry together with its finite-$N$ violation. Finally, we discuss analogies with boundary algebras proposed for black hole interiors and closed universes, and suggest lessons from our construction for both.

The Lindblad (GKLS) master equation, which represents the mathematical form for the general evolution of a density matrix, is a versatile and widely-used tool in open quantum systems. In contrast with the typical approach of imposing additional conditions on the system, such as weak coupling or energy conservation, we explore the structure of the equation with no assumptions. We demonstrate that general quantum dynamics can be expressed through a combination of free evolution, exchanges of some physical quantities (generalised charges), not necessarily commuting with the Hamiltonian, between the system and the bath, and pure dephasing. This result comprises a novel perspective on quantum master equations, employing physical processes as elemental parts. We use it to explore the dynamics and stationary states of a two-level system and show that strong coupling, particle exchange, and non-Abelian effects all share the same physical origin. Moreover, we demonstrate that the generalised Gibbs state for all three cases contains a non-commutation term, which has not been previously considered.

We investigate the thermo solutal transport phenomena and deposition patterns during the evaporation of surfactant laden droplets experimentally and through theoretical scaling based analysis. Experiments were conducted using the sessile droplet configuration in the acrylic chamber for both hydrophilic and hydrophobic substrates. Infrared thermography and particle image velocimetry measurements were conducted during evaporation to illustrate the temperature and velocity distributions, respectively. Sodium dodecyl sulphate SDS surfactant molecules enhanced the evaporation rate with an increase in concentration for the hydrophobic surface. In contrast, the evaporation rate increased up to 0.5 CMC and then decreased for droplets on a hydrophilic substrate. The evaporation rates computed from the shadowgraphy imaging were explained using the average velocities obtained from the PIV analysis. It was found that advection within the droplet is strongly dependent on surfactant concentration and wettability. Further, the theoretically obtained Marangoni velocities were in close agreement with the experimental values. It was found that Marangoni solutal advection dominates other advection mechanisms, such as Marangoni thermal advection and buoyancy driven flow. However, surfactant crowding and viscous resistance with increasing surfactant concentration can dampen the increase in solutal advection. The surface tension and viscosity measurements were also conducted with variation in surfactant concentration to understand the suppression of advection by viscous forces. The computation of contact line velocities showed sudden fluctuations, illustrating stick slip behaviour during droplet drying, complementing microscopic visual observations.

The Klein Paradox -- the anomalous scattering of relativistic fermions off a high potential step -- signals the limit of the single-particle interpretation of the Dirac equation. While Quantum Field Theory (QFT) resolves this via pair production, the microscopic mechanism is often obscured by abstract formalism. In this work, we investigate this phenomenon through the framework of Analog Gravity and Condensed Matter Physics. We utilize a hydrodynamic model wherein a relativistic particle is treated as a localized elastic excitation (defect) within a continuous linear medium. We demonstrate that when the external stress (potential) exceeds the medium's binding energy threshold ($V > 2mc^2$), the system undergoes a mechanical instability analogous to dielectric breakdown. This instability naturally generates modes with inverted topological winding, which we identify as antiparticles. By solving the boundary conditions for this elastic system, we reproduce the transmission coefficients of Hansen and Ravndal and recover the Schwinger limit for pair production rates. This approach provides a clear pedagogical model based on continuum mechanics to visualize vacuum decay processes, suggesting that the "paradox" is simply the elastic response of a medium under supercritical stress. This mechanical analogy serves as a pedagogical bridge for graduate students in condensed matter physics and advanced materials science, offering a concrete visualization of vacuum instability that complements standard abstract QFT derivations.

We derive the reduced Dirac Hamiltonian on the non-rotating BTZ background and use its redshift structure to construct a gauge-covariant single-band lattice model on the constant-time BTZ cylinder. In equal-area coordinates the AdS radius $L$ fixes the local Gaussian curvature, while the horizon radius $r_h$ fixes the throat size and the strength of the near-horizon redshift. The lattice model therefore has a direct geometric interpretation and is not presented as an unshown reduction of the two-component Dirac lattice. Its angular Fourier transform yields an exact curved Harper equation with BTZ-dependent hopping amplitudes and a consistent dimensionless angular quasi-momentum. We then supplement global parameter scans with state-resolved diagnostics: spectra color-coded by mean radius, local density of states, direct flux-response versus radius correlations, and Aharonov--Bohm spectral flow and persistent current on the BTZ cycle. These results show that weaker curvature sharpens the butterfly-like fragmentation, whereas larger horizons suppress both magnetic and Aharonov--Bohm response by creating weakly dispersing near-horizon states.

We develop a method for simulating colloidal suspensions using multiparticle collision dynamics (MPCD) with a discrete particle model represented as a rigid body. The key steps for incorporating the rigid-body constraints are to thermalize the velocities of the discrete sites before they participate in the MPCD collision step, then transfer momentum from the sites to the rigid body. We demonstrate that the rigid-body model produces the expected statistics for a single spherical particle and the same transport properties for a hard-sphere colloidal suspension as an equivalent model using harmonic bonds to maintain the site geometry. Importantly, the rigid-body model has less computational overhead and permits a larger simulation timestep than the harmonic-bond model, leading to a nearly order of magnitude speedup in benchmark simulations of hard-sphere colloidal suspensions. Our method is compatible with arbitrary discretization, so it enables more efficient MPCD simulations of suspensions of colloidal particles with complex shapes.

We explore the mechanics of a terminally loaded buckled elastica under frictionless self-contact. With the aid of two integrals associated with the elastica, we propose a scale-invariant condition necessary for the onset of contact. The condition is independent of the boundary conditions, does not involve the position vectors of the material points, and delivers the value of the compressive load at which self-contact initiates. Furthermore, we show that one of the two integrals, namely the \emph{Hamiltonian}, persists after contact. We compute post-contact configurations of modes three through ten for a pinned-pinned buckled elastica. At a given value of the compressive load, we report multiple post-contact configurations for modes eight and nine. Finally, we show that an infinite force is required to transition from a point contact to a line contact in symmetric post-contact configurations of odd modes.

Boundary time crystals (BTCs) are prominent examples of continuous time crystals in collective spin systems governed by Lindbladian evolution. To date, their analysis has mostly relied on semiclassical and numerical approaches. Here, we develop a fully quantum-compatible framework to classify collective spin dynamics and show that BTC behavior emerges from the absence of non-trivial weak symmetries of the Liouvillian. To this end, we introduce an irreducible tensor representation of operator space, in which the Lindbladian dynamics maps onto a non-Hermitian hopping problem. Within this picture, the dynamics corresponds to the transport of operator weight across tensor sectors. This mapping allows an identification of distinct dynamical regimes, including collective precession, pure relaxation, and the BTC phase, within a single unified framework. We show that BTCs arise from non-reciprocal transport in operator space, which delocalizes Liouvillian eigenmodes across multiple tensor sectors. This non-reciprocal transport provides a microscopic mechanism for the insensitivity to initial conditions of BTC oscillations. More broadly, our results establish operator space transport as a perspective for understanding dissipative many-body dynamics and highlights connections to non-Hermitian phenomena.

Unlike conventional charge-2e superconductors, a charge-4e superconductor exhibits long-range coherence of electron quartets rather than Cooper pairs. Clear zero-temperature realizations of charge-4e superconductivity remain rare. Here, we investigate the zero-temperature phase diagram of the attractive SU(4) Hubbard model with numerically exact, large-scale quantum Monte Carlo (QMC) simulations overcoming major technical hurdles. We identify both charge-2e and charge-4e superconducting phases. Upon increasing interaction, charge-2e correlations are suppressed and eventually vanish, while the charge-4e correlations remain robust and converge with system size, signaling the onset of a quartet-condensed phase. Interestingly, across the charge-2e--charge-4e transition, single electrons remain gapped, while charge-2e correlations exhibit a scaling behavior inconsistent with a conventional Landau description. These features are naturally captured by a fractionalized framework in which the physical charge-2e order parameter is a composite field coupled to an emergent non-Abelian gauge structure. We formulate an Sp(4) gauge-Higgs theory that realizes deconfined quantum pseudocriticality between the Higgs (charge-2e) phase and the confined (charge-4e) phase. The Sp(4) gauge-Higgs theory yields pseudocriticality through a fixed-point collision, and its one-loop collision-point exponents quantitatively track the QMC results. Our results establish charge-4e superconductivity as a bona fide zero-temperature phase, provide a simple model for future studies in a numerically exact framework, and reveal an unconventional route to superconducting criticality.

Investigating the interplay among topology and electron-electron interactions is an intriguing research quest which has recently gathered steam across the community of condensed-matter physics. In the present work, we study the fate of a three-dimensional Berry-dipole semimetal, lying at the topological quantum critical point separating a Hopf insulator from a trivial insulator, in the presence of long-range Coulomb interactions. Utilizing large-$N_f$ analysis at three spatial dimensions and an $ε$-expansion within the renormalization-group scheme, we uncover the emergence of a spatially \textit{anisotropic} non-Fermi liquid with enhanced Berry-dipole moment. We further derive the corresponding scaling relations of certain physical observables as functions of the probed energy and temperature scale, and we provide a simple observational criterion for distinguishing the onset of the topological anisotropic non-Fermi liquid from a Berry-dipole semimetal.

We investigate the thermal responses of a harmonic oscillator chain coupled at its boundaries to heat baths held at different temperatures. This setup sustains a steady energy flux, continuously dissipating heat into both reservoirs. By introducing slow variations in the bath temperatures, we quantify the resulting excess heat currents and thereby obtain the nonequilibrium heat capacity matrix at fixed but arbitrary temperature differences. We demonstrate the existence of a well-defined thermodynamic limit for long chains. The specific heat associated with energy exchanges with a single bath depends on the difference in friction coefficients governing the system-bath couplings. That thermokinetic effect is typical for nonequilibrium response. When the couplings with the thermal baths acquire temperature dependence, the specific heat correspondingly inherits a nontrivial temperature dependence, in sharp contrast with equilibrium. Our results provide the first explicit determination of specific heat(s) in a locally interacting, spatially extended driven system. Beyond its exact solvability, the model may offer a natural nonequilibrium extension of the Dulong-Petit law, capturing the high-temperature behavior of driven molecules.

The supercurrent diode effect (SDE), characterized by unequal critical currents in opposite directions, has been observed with or without magnetic fields, yet mechanisms enabling zero-field SDE without explicit symmetry breaking remain underexplored. Here we investigate a Josephson junction with strong electron-electron interaction modeled by a Hubbard $U$ term and an odd number of electrons. We find that strong correlations induce spontaneous breaking of time-reversal and mirror symmetries, forming a $\varphi$-junction with degenerate energy minima at $\pm\varphi$, resulting in zero-field Josephson diode effect (JDE) without magnetic order. Spin-orbit coupling breaks SU(2) symmetry but does not determine diode polarity, contrasting with magneto-chiral mechanisms. We further show that applying a tiny Zeeman field enables controllable JDE with sizable efficiency due to the enhancement by the strong magnetic correlation, and the JDE strength peaks when the field induces a level-crossing transition. These findings establish strong electron correlation as a distinct mechanism for nonreciprocal superconducting transport, broadening the understanding of SDE origins.

We study the role of heterogeneous volatility in a networked wealth dynamics model and its impact on the wealth condensation transition. Extending the Bouchaud--M{é}zard framework, we introduce binary volatility in networks and investigate how its configuration affects the effective power-law tail exponent of the wealth distribution. Using a stochastic block model, we control the mixing between volatility groups and show that the effective exponent is governed not only by the global parameter $Λ=2J/β^2$ but also by the volatility configuration in the network. We find that local interactions between nodes with different volatility induce a neutralization of group-wise exponents, which lowers the aggregate tail exponent and can drive a condensation transition across $γ_{\rm c}=2$. Our results identify volatility mixing as another control mechanism for wealth condensation and highlight the importance of noise heterogeneity in nonequilibrium systems on networks.

Granular flows are ubiquitous in nature and industrial applications, yet a complete continuum theory remains a long-standing challenge. The leading empirical approach, μ(I) rheology, lacks microscopic foundations and becomes multivalued in dense, slowly sheared flows where nonlocal corrections are required. Exploiting state-of-the-art high-speed X-ray tomography to investigate microscopic dynamics of dense granular flows in a Couette geometry, we establish a new, universal constitutive law spanning quasi-static to inertial regimes based on structural relaxation, resolving the fundamental difficulty in the original μ(I) framework. By further establishing a non-equilibrium statistical framework for granular flows, we demonstrate an intrinsic analogy between driven granular matter and hard-sphere liquids owing to their identical Carnahan-Starling equation of state, naturally explaining our rheological approach and the emergence of glassy behaviors. Our framework unifies granular rheology with the broader physics of disordered systems and provides a complete, microscopically-based theoretical framework for dense granular flow.

A time-integrated measure of complexity is proposed for diagnosing the degree of quantum ergodicity. The scrambling dynamics of maximally entangled states within the ensemble of unitary evolutions are quantified by applying numerical bootstrapped realizations of an ensemble of Hamiltonians, probing different unitary paths. Using Rosenzweig-Porter ensembles, we show that the integrated spread complexity provides a fine-grained resolution across different ergodic regimes. This approach offers a robust method for diagnosing quantum chaos from early to late times.

Being motivated by recent synthesis of a monolayer of gold, named goldene, from the nano-laminated ternary ceramic phase of Ti3AuC2, we are proposing two phases of goldene viz. goldene-I and goldene-II as anode material for Lithium-Ion batteries using first principles study. This innovative goldene-I monolayer, composed of triangular motifs of gold atoms, exhibits remarkable properties owing to its unique geometric configuration and intrinsic stability. In contrast, a theoretical structure known as goldene-II, featuring a combination of triangular and hexagonal motifs, has been proposed. This structure possesses intrinsic, periodically distributed pores among Au atoms and demonstrates structural integrity and mechanical robustness, even under lithium adsorption. The electronic band spectra and projected density of states reveal the metallic nature of both phases of goldene. Electrochemical evaluations reveal that goldene-II offers favorable lithium-ion adsorption energies, efficient charge transfer, and volumetric capacities. Goldene-I achieves a volumetric capacity of 0.713 Ah/cm3, while goldene-II reaches 0.783 Ah/cm3, confirming its high suitability for lithium storage volumetric capability. Moreover, goldene-I has an ultra-low barrier height of 15 meV, which supports rapid lithium-ion transport.

Recent investigations of turbulent circulation fluctuations have uncovered substantial insights into the statistical organization of flow structures and revealed unexpected geometric features of turbulent intermittency. Of particular interest here is the observation that circulation probability distribution functions admit a superstatistical representation, namely a description based on "ensembles of Boltzmann-Gibbs ensembles". A fundamental phenomenological ingredient of this approach, which serves as a natural starting point for modeling, relies on the strong correlation between the dissipation field and the spatial distribution of elementary circulation-carrying structures, i.e., small-scale vortices. Within the language of superstatistics, this corresponds to characterizing circulation statistics through an appropriate choice of conditioned (Boltzmann-like) distributions and mixing distributions. We show that the superstatistical class of q-exponentials, known to have broad applicability in a wide range of multiscale and non-equilibrium systems, provides an accurate description of the observed circulation statistics in homogeneous and isotropic turbulence. This finding opens avenues for exploring the statistical structure of the turbulent cascade in the context of non-extensive statistical mechanics, rooted in the concept of non-additive entropies.

Determining whether two graphs are isomorphic is a fundamental problem with practical applications in areas such as molecular chemistry or social network analysis, yet it remains a challenging task, with exact solutions often being computationally expensive. We address this task using persistent homology built on motif-based filtrations of graphs, a method from topological data analysis that summarizes the shape of data by tracking the persistence of structural features along filtrations. Specifically, we use edge-weighting schemes based on the densities of triangles, chordless squares, and chordless pentagons, which have been shown to be effective for detecting network dimensionality. Our cycle-density filtrations distinguish non-isomorphic graphs perfectly or nearly perfectly across four demanding graph families, many of which exhibit symmetries. We outperform curvature-based, degree-based, and Vietoris--Rips filtrations, and match or exceed the accuracy of egonet-distance methods while incurring a lower computational cost. The expressive power of our filtrations goes beyond isomorphism testing: because they capture rich structural information from graphs, they consistently achieve top performance on property prediction tasks using real-world data, and exhibit high sensitivity to edge rewiring and removal. Together, these findings establish cycle-density filtrations as an effective and computationally tractable framework for graph comparison and characterization, bridging topological data analysis and network science.

We investigate numerically the blocking of two-dimensional bistable reaction diffusion fronts by geometric obstacles. Our goal is to derive quantitative criteria for front propagation in the presence of spatial heterogeneities. Using a conservation law approach, we show that the integral of the reaction term acts as an effective driving force for the front. Combining this insight with the exact one-dimensional traveling wave solution, we construct a reduced analytical model that predicts blocking thresholds. In particular, we obtain explicit conditions for front propagation in a waveguide connected to a conical region of angle theta, valid for widths w less than 4. The model captures the influence of both geometry and nonlinearity, and shows good agreement with numerical simulations. Finally, we extend the analysis to more complex geometries, including checkerboard-like obstacles, and derive simple heuristic rules governing front propagation. ~

Shape adaptive shape memory alloy hybrid composites (SMAHCs) are composites that incorporate shape memory alloys (SMAs) to realize shape transformation. Despite the availability of numerous analytical and finite element models for predicting the transient response of SMAHCs, many approaches exhibit limitations with respect to the thermomechanical coupling and comprehensive experimental validation. Therefore, this paper presents a coupled, multiphysics, 3D finite element approach for the simulation of a SMAHC actuator, integrating mechanical, thermal and electromagnetic solvers in the Finite Element Code ANSYS LS-DYNA. The proposed approach employs a micromechanical constitutive model implemented in ANSYS LS-DYNA, to accurately capture the complex thermomechanical phase transformation of SMAs. A key feature of the model is the ability to prescribe a defined martensitic pre-strain through a preceding simulation step, in which an initially scaled SMA wire is mechanically loaded and stretched to its nominal length. This procedure enables partial detwinning of the martensitic microstructure and provides a physically motivated initialization of the material state. Joule heating of the SMA wires, as well as varying mechanical loads and ambient temperature conditions, are explicitly considered. The simulation results are validated against experimental data and a fully coupled transient staggered scheme model to assess the predictive capability of the 3D approach. The results show good qualitative agreement, reproducing the characteristic hysteresis of actuator deflection as a function of temperature. Quantitatively, the predicted deflections are of the correct order of magnitude, although marginally outside the 95 % experimental confidence interval. Overall, a consistent trend between simulation and experiment is observed, giving rise to possibility of simulating more complex SMAHC systems.

Active transmissive metasurfaces are central building blocks for future compact, cascadable optical systems, enabling the stacking of multiple functional layers for advanced dynamic beam shaping, photonic neural networks, depth sensing, and holography. We present a transmissive electro-optic metasurface based on silicon-on-lithium-niobate, where an array of silicon waveguides with periodic perturbations, individually controlled at the 100 pm scale, supports well-defined high-Q (>2000) guided-mode resonances (GMRs). We incorporate interdigitated push-pull electrodes between subwavelength-spaced GMR elements to locally tune the refractive index in the lithium niobate substrate, thereby shifting the GMR resonance and enabling opposite phase and amplitude modulation between neighboring radiative elements. In a geometrically symmetric metasurface, this effect introduces electro-optic beam splitting via diffraction, with diffraction efficiencies as high as 3%. By introducing controlled passive resonance detuning via 100 pm scale perturbation shifts, we increase the efficiency of amplitude modulation six-fold through geometrical symmetry breaking, achieving amplitude modulation depths of 40% at $\pm$30 V. This work demonstrates the potential of active and passive resonance control enabled by high-Q GMR structures for efficient electro-optic modulation or multifunctional sensing.

In the context of waves in space time modulated materials, we ask two questions how much energy does it cost to break time reversal symmetry and transition to a PT symmetric state. and can a PT symmetric system have a ground state in the sense that no photons are present. Our model system is a periodic metamaterial set in virtual motion with velocity cg to become a space-time crystal. We find that the expectation of energy content is always increased on breaking symmetry. At the same time breaking T symmetry introduces photon-pairs even when we start from a T symmetric ground state empty of photons, except in certain pathological examples which we describe. For a range of velocities, PT symmetry is broken so that energy must be continuously invested to preserve motion, creating a trail of photon pairs. Here energy must be continuously invested to preserve motion. We make an analogy with acoustic radiation generated from breaking the sound barrier.

This paper is devoted to studies of the mechanical deformation of the S. aureus cell wall. The bacterium is modelled as a thin elastic membrane containing cytoplasm, which is treated as an incompressible fluid. Deformation occurs via Van der Waals interactions between the bacterium and a solid metallic surface, both with and without the influence of surface plasmon resonance (SPR). Our modelling results indicate that the excitation of surface plasmons significantly increases the effective interaction area between the bacterial membrane and the nanostructured surface. The elastic and dielectric properties of the bacterium's components are uninvestigated. Therefore, theoretical calculations are performed in wide, physically meaningful ranges. Thus, the results of studies give only a qualitative estimation. However, they are novel and, with further experiments, can solve the inverse problem of obtaining physical properties. The paper highlights the potential of SPR to enhance antibacterial strategies, inspiring further research and innovation.

We introduce a Hybrid High-Order (HHO) method for the Schrödinger equation in the presence of a magnetic vector potential. In quantum mechanics, physical observables are invariant under continuous gauge transformations, which must be kept at the discrete level to avoid unphysical artifacts. To address this, we construct a discrete covariant gradient operator on arbitrary polyhedral meshes. We prove that the resulting discrete bilinear form guarantees gauge covariance asymptotically at the discrete level. The resulting scheme achieves optimal convergence rates and preserves a discrete Garding inequality, guaranteeing a stable ground state. The theoretical properties of the scheme are corroborated by numerical experiments, including the computation of the Fock-Darwin fundamental energy and replicating the Aharonov-Bohm effect.

Twenty-five years have passed since the first experimental demonstration of attosecond pulses, marking the advent of our ability to resolve and control electron motion in real time. What began as a technological breakthrough - generating the shortest flashes ever produced - has evolved into a powerful approach for probing and steering electronic dynamics in atoms, molecules, and solids. This roadmap, authored by leading experts in the field, surveys the recent rapid progress in the generation and characterization of attosecond pulses, emerging attosecond measurement and control techniques, and their expanding range of applications. It reviews current and future developments in attosecond light sources, including novel laser technologies, waveform synthesizers, new schemes for high-order harmonic generation, attosecond pulse generation at free-electron lasers, and structured light. Advances in attosecond measurement methodologies are also discussed, encompassing all-attosecond pump-probe spectroscopy, attosecond four-wave mixing, attosecond microscopy, spectroscopy with light transients, and attosecond interferometry. Furthermore, the roadmap addresses applications of attosecond spectroscopy to reveal electron dynamics in molecules and condensed matter systems from both theoretical and experimental perspectives, and highlights emerging directions at the interface with quantum optics and quantum entanglement. Overall, this work aims to serve as a comprehensive resource for navigating the evolving landscape of attosecond science.

Actively tunable photonic devices are vital for next-generation optoelectronics requiring rapid switching and high bandwidth. Although organic optoelectronic devices have found wide application, their use as optical modulators has been limited by low absorption in the critical near-infrared (NIR) region, slow response time, and weak nonlinearities. To address these limitations, we developed a scheme based on intermediate exciton-photon coupling in a NIR absorbing squaraine-dye based photonic structure. Using energy-momentum resolved pump-probe spectroscopy, we show that the sign and magnitude of the optical response of our system depends strongly on the energy detuning between the excitonic and photonic modes. These data are analyzed using temporal coupled-mode theory to show that near resonance, a distinct energy exchange process emerges in the cross-over regime between strong and weak light-matter coupling. This effect enables dynamical control over the photoinduced response, providing a pathway for broadband optical signal modulation extending into the NIR spectral region.

Near-infrared (NIR) emitters operating in the 650-900 nm range are highly attractive for imaging and sensing in turbid media; however, cadmium-free InP-based quantum dots (QDs) often suffer from limited brightness due to nonradiative pathways and inefficient photon outcoupling. In particular, heterostructured InP QDs can exhibit band alignments that induce partial spatial separation of charge carriers, leading to reduced electron-hole wavefunction overlap. This modifies intrinsic recombination dynamics and enhances the sensitivity of their emission to the surrounding photonic environment. Here, we investigate silver toroidal plasmonic nanoantenna dimers (Ag TPNDs) through finite-difference-time-domain (FDTD) simulations as a geometry-tunable platform for enhancing NIR emission of heterostructured InP-based QDs. The coupled toroidal geometry supports strongly confined bonding modes that generate intense nanogap hotspots, while its resonance can be systematically tuned through the toroid aspect ratio. By spectrally aligning the antenna response with QD emission bands (675-845 nm), we achieve large Purcell enhancements together with high quantum efficiencies, demonstrating efficient conversion of enhanced decay rates into radiative emission. We further show that nanometer-scale variations in emitter-antenna separation strongly modulate the radiative rates and spectral response. These results establish toroidal plasmonic nanodimers as a topology-driven platform for controlling emission in NIR quantum emitters and for advancing NIR nanophotonic applications.

Understanding the thermodynamics of radiation and the quantum-mechanical interactions between light and matter is important both for theoretical purposes and for technological advances, such as determining the limits of key processes like light-to-usable-energy conversion efficiencies. In this report, we discuss the physics of these two aspects, considering spontaneous emission as a pathway, and highlight the limitations of such descriptions in assessing energy-harvesting efficiency. In view of these limitations, we adopt a simplified approach to evaluate the free energy of radiation, providing a framework to assess various aspects of light-to-usable-energy conversion efficiencies. Our approach allows a theoretical estimate of the thermodynamic maximum limit for light-to-usable-energy conversion, which is approximately 74%. We validate this free energy estimate by modeling and accurately reproducing the Shockley-Queisser limit (~ 33%), which imposes a practical constraint on solar-to-usable-energy conversion efficiency. Beyond free-energy considerations, our model incorporates various processes, such as spontaneous emission, nonradiative thermal losses, and photon upconversion, allowing us to evaluate their roles. The model further suggests that, under certain conditions, the maximum conversion efficiency can reach approximately 48%, for example with multijunction solar cells or via photon upconversion. These findings further suggest that the true thermodynamic limit for light-to-usable-energy conversion may be much higher (approximately 74%). However, accurately estimating this limit requires a more complete understanding of the thermodynamics of light, light-matter interactions, and the connection between them.

Non-van der Waals two-dimensional (2D) materials derived from strongly bonded non-layered crystals have recently emerged as a novel and rising platform for nanoscale research. While uncovering and tuning their (opto-)electronic, catalytic, and magnetic properties has been the focus of intense research, the impact of spin-orbit coupling (SOC) onto their electronic structure has not yet been explored in detail. Studying these effects is, however, particularly relevant due to their surface cation termination and the presence of heavy elements in several representative compounds. Here, we investigate the effect of SOC onto the electronic structure of 2D AgBiO3, NaBiO3, and SbTlO3. While the first two systems show negligible band renormalization upon inclusion of relativistic effects around the band gap, SbTlO3 showcases a large SOC induced splitting (229meV) for the lowest conduction bands associated with a band inversion. Substitution of Tl with Pb forming SbPbO3 brings the band-inverted feature to the Fermi level. Analysis of topological invariants and investigation of edge states of zig-zag and armchair ribbons within the 200meV gap confirms the topological nature of the band splitting. Our work thus establishes a foundation for the systematic study of robust non-van der Waals 2D topological insulators.

We demonstrate unconventional photon blockade in a symmetric Kerr dimer driven with equal-amplitude fields at a $90^\circ$ phase difference. The minimum inter-cavity coupling is $J_{\min} = γ/4$ at a Kerr nonlinearity $U \ll γ$ achievable in standard photonic molecules. The quadrature-driven site emits strongly antibunched light with a smooth, oscillation-free second-order correlator directly resolvable with standard detectors. The scheme operates under continuous-wave and pulsed excitation, and fabrication disorder can be fully compensated by re-tuning the drive phase, removing the need for post-fabrication cavity trimming.

This work presents a compact, low-crosstalk CWDM4 MUX/DEMUX utilizing cascaded multimode waveguide grating filters. The individual filters are designed with finite Gaussian apodization and positive dispersion, enabling strong unilateral sidelobe suppression while maintaining a minimum feature size compatible with UV lithography. By cascading these filters, we demonstrate a DEMUX that achieves channel crosstalk below -25 dB, insertion loss under 1 dB, and a flat-top bandwidth of approximately 18 nm. The entire device occupies a compact footprint of only 1.6 mm x 40 um, with a channel spacing compatible with commercial TIA and driver chips. Furthermore, a series-parallel hybrid cascade configuration can further suppress the crosstalk to -40 dB.

Altermagnets are a new class of magnets accompanying global time-reversal symmetry breaking (TRSB) without net magnetization. The TRSB results in formation of novel altermagnetic domains. Features of altermagnetic domains, in particular their responses to external stimuli, are essentially important but yet unexplored. Here, we report visualization of bulk altermagnetic domains in MnTe based on scanning magneto-optical Kerr-effect microscopy using telecom infrared wavelength. We found two distinct TRSB domains with large Kerr rotations that do not scale with its tiny bulk magnetization. We also revealed controllability and stability of domains against magnetic or thermal perturbations. Our first observation of altermagnetic domains using a laboratory-scale simple optical technique showing their movable nature provide firm bases for future fundamental and application studies of altermagnets.

Unstructured meshes are among the most versatile approaches for capturing non-canonical geometries in fluid dynamics simulations. Despite this, most high-fidelity first-principles phase-change models are developed and applied on structured meshes. We present a phase-change simulation method for unstructured meshes that combines the algebraic Volume-of-Fluid (VOF) technique with geometric interface reconstruction, implemented in an in-house open-source CFD code. Phase-change rates are computed from local temperature gradients evaluated at the reconstructed interface, without empirical closure models, using a reconstruction procedure that operates on arbitrary polyhedral cells. Because the method relies on the standard finite-volume framework, it can be integrated into other cell-centred codes supporting unstructured meshes. The approach is validated against the one-dimensional Stefan and Sucking problems and the three-dimensional Scriven bubble growth on both hexahedral and polyhedral meshes, showing good agreement with analytical solutions in all three cases. A detailed analysis of the Scriven problem reveals that the interface-modified least-squares gradient stencil on Cartesian meshes overestimates the interfacial temperature gradient, producing a persistent overshoot of the analytical bubble radius and a coherent four-fold anisotropy that elongates the bubble along grid diagonals. On polyhedral meshes, the irregular face orientations eliminate both effects, yielding isotropic growth and monotonic convergence. Finally, we demonstrate the framework on turbulent upward co-current annular boiling flow, where early transient results are qualitatively consistent with a previous LES study and experimental observations of wave-modulated evaporation.

We present a hardware demonstration and resource analysis of split-evolution quantum phase estimation (SE-QPE) on a Quantinuum System Model H2 quantum computer. SE-QPE is a modification to canonical QPE for particle-conserving Hamiltonians in which controlled time evolution is replaced by CSWAP-based interference between a target register and a reference register. For factorizations of time evolution with a shared eigenbasis, SE-QPE preserves the phase-register outcome distribution of canonical QPE and, unlike with compute--uncompute substitutions, it remains compatible with non-exact eigenstates. The substitution removes controlled-simulation overhead and enables parallel evolution on two registers, reducing the depth of each phase-kickback block. Resource analysis for Trotterized double-factorized chemistry Hamiltonians shows that the substitution becomes increasingly favorable at higher phase powers, as such combining QPE and SE-QPE implementations can be a useful option. Over a range of FeMoco active spaces, SE-QPE reduces time evolution resources, with asymptotic reductions of about 33% in CX count, 25% in $T$ count, and an asymptotic depth ratio of $3/N$ for CX layers. On Quantinuum H2-2, a four-qubit model ethylene demonstration with explicit inverse QFT and repeated phase-kickback steps up to 6 phase bits yields distinct energies and shows the auxiliary registers provide useful error detection filters.

Recent advances in the field of active soft matter promise a lot. Both, experimental advances and theoretical understanding point towards new material classes in reach, for example self-healing materials that might switch their properties from elastic to solid easily or switch their macroscopic shapes. All these materials require an active force to propel parts of themselves on the micrometer scale. While chemical fuels are often used to generate these active forces, applying energy in a simple and continuous way remains unsolved. Here we explore using light as such an energy source. Overall, generating active driven, self-propelled particles is hence not only of great interest but also a general challenge. Moreover, controlling such particles even within living tissue would open new worlds, for example to enable specific drug delivery or the design of micro-robots. One recently proposed method to establish light driven self propelled particles is to create specific shaped and transparent objects, that move when illuminated with homogeneous light. In these particles, the refraction of the light leads to a momentum transfer, which then drives the active movement. Here, we show both in simulation and experiments that the production of such particles is possible and demonstrate the feasibility of this propulsion effect, while investigating different shapes. Our experiments show that breaking the shape-symmetry of the particles creates a refraction-based propulsion under homogeneous illumination. Subsequent simulations reveal that total reflection leads to the largest momentum transfer among all different geometries considered. Overall, our study introduces the proof-of-principle for refraction-propelled particles, which has the potential to benefit many fields of study including cellular behaviour, collective dynamics and the understanding of disease mechanisms.

Snapshot full-Stokes polarimetry across multiple wavelengths remains challenging because conventional architectures rely on multiplexed measurements and bulky optics. We present an end-to-end framework that reconstructs RGB full-Stokes images from a single monochrome sensor measurement. The system combines a differentiable 4f optical frontend with a U-Net backend for joint optimization. Notably, a metasurface modeled by the multilayer perceptron (MLP) is employed to encode the full-Stokes polarization information. We implement the design in two stages: first in a hybrid metasurface-refractive 4f architecture, and then in a pure meta-optic configuration. On a real-world dataset, the hybrid metasurface-refractive system achieves 30.00 dB peak signal-to-noise ratio (PSNR) and 0.8291 structural similarity index measure (SSIM) for monochromatic imaging in the visible range, and 26.71 dB and 0.7044 for RGB-achromatic imaging. The pure meta-optic system yields 26.94 dB/0.7184 for the monochromatic case and 24.10 dB/0.6015 for the RGB-achromatic case. These results show that end-to-end optical-digital co-design enables compact snapshot full-Stokes polarimetric imaging at a high compression ratio of 12.

Modern communication networks demand ever-increasing transmission bandwidth, placing stringent requirements on low-cost, high-performance electro-optic modulators. Substantial advances have been made in integrated photonics employing lithium niobate on insulator. In contrast, photonic integrated circuits based on lithium tantalate -- a material already commercially adopted for wireless filters -- have been developed, offering reduced DC drift, higher optical power handling, and lower birefringence. These advantages enable more complex and dense photonic integrated circuits, and make lithium tantalate a promising material platform for next-generation integrated electro-optic modulators. However, in contrast to the extensively studied thin-film lithium niobate platform, thin-film lithium tantalate modulators have only been explored on silicon substrates. Here, we report the first fabrication and characterization of thin-film lithium tantalate electro-optic modulators manufactured on a 4-inch (100 mm) fused-silica substrate for adapting a low-loss slow-wave microwave electrode to improve the electro-optic bandwidth. By employing a slow-wave electrode design to achieve velocity matching between microwave and optical signals, the demonstrated modulator achieves a 3-dB electro-optic bandwidth of 64 GHz with a low half-wave voltage of 1.53 V, with potential to operate at the measured 100 GHz electrical bandwidth, if the employed spectral biasing is removed. The modulator moreover exhibits low bias drift, with a constant switching voltage down to 10 mHz. This performance enables high-speed data transmission comparable to state-of-the-art lithium niobate modulators fabricated on quartz substrates. Using the fabricated devices, a net single lane data rate of 440.6 Gbps is achieved using PAM8 signaling.

Similar to microspheres, thin-walled microbubble resonators support whispering gallery modes (WGMs) that combine ultrahigh Q-factors and small effective mode volumes. In contrast, their hollow nature enables enhanced interactions with encapsulated materials and lower spectral mode density due to the tight radial confinement of the optical modes. However, the existence of a high axial-mode density still leads to significant mode mixing and modal interference that can complicate spectral shift measurements, thereby limiting sensing applications. To address this limitation, we have fabricated geometric filters directly on the surface of microbubbles using focused ion beam (FIB) milling. Based on numerical calculations, we first designed and then fabricated large tapered patterns, such as circular surface dips or holes, that could effectively filter modes while minimizing optical scattering losses. Local lateral mode confinement and partial recovery of high Q-factors were experimentally achieved by adding shallow slit patterns. Using few-mode engineered microbubble resonators, we subsequently demonstrated pressure sensing and wide spectral tuning of WGMs free from mode-mixing artifacts. This precision engineering approach promises improved mode isolation, tunable directional emission, and ultrasensitive measurements in microbubble resonator devices.

This paper extends and analyzes the high-order kernel regularization framework of Beale & Tlupova (arXiv:2510.13639) to all four boundary integral operators of the Helmholtz Calderon calculus in three dimensions: the single-layer, double-layer, adjoint double-layer, and hypersingular operators. To the best of our knowledge, this work provides the first high-order kernel regularization of the hypersingular operator for both the Helmholtz and Laplace equations in three dimensions. The regularization replaces each singular kernel with a smooth modification constructed from error functions together with a polynomial correction whose coefficients are determined through moment conditions. Alongside the derivation of the regularizing functions, the paper provides a unified error analysis of the combined regularization and quadrature discretization procedure. By coupling the regularization parameter to the mesh size, the two error contributions can be balanced, leading to explicit overall convergence rates that depend jointly on the order of the regularization and the degree of exactness of the surface quadrature rule. A key practical feature of the method is its implementation simplicity. Once the regularizing functions are determined, the numerical task reduces entirely to the evaluation of smooth surface integrals using standard quadrature, without the need for element-local solves, singularity-specific precomputations, or specialized quadrature rules. Although the modified kernel is generally incompatible with kernel-specific fast methods, this limitation is addressed through H-matrix acceleration, applicable in a black-box manner. Numerical examples -- including verification of the predicted convergence rates and solution of sound-soft and sound-hard scattering problems by smooth obstacles -- demonstrate the accuracy and practicality of the proposed methodology.

Source localization in complex networks is a rapidly advancing field with numerous real-world applications, including determining the source of misinformation. In this work, we model information spread across several real-world and synthetic complex networks using our Generalized Fractional Susceptible-Infected-Recovered (GFSIR) model, which incorporates the information overload (IOL) phenomenon. Then, we use Pearson's correlation algorithm to identify information sources in these networks and investigate how information overload affects localization quality. Numerical simulations have shown that localization effectiveness decreases with the parameter $α$, which controls the strength of the IOL, and increases with the spreading rate $β$. Our comparison across various topologies reveals that localization is generally more effective in synthetic structures, with Erdős-Rényi networks exhibiting greater resilience to IOL than Barabási-Albert models. Furthermore, we identified a critical reversal in the impact of network density: while a higher average degree enhances localization when IOL is negligible, less dense networks perform better under strong overload. This phenomenon represents a significant departure from the behavior observed in standard epidemic models.

Rutile RuO$_2$ has attracted great interest recently, as its magnetic ground state remains controversial. Experimental studies have reported either nonmagnetic or altermagnetic (AM) ground states in different crystalline samples of RuO$_2$, highlighting the need for a reasonable explanation to resolve this contradiction. In this study, density functional theory calculations are performed to reveal the correlation-sensitive and strain-dependent magnetism of bulk RuO$_2$. On one hand, multiple AM phases with different magnitudes of the spin magnetic moment are identified in the Hubbard parameter space for RuO$_2$. On the other hand, when appropriate strains which significantly change the crystal cell volume are applied, the ground state of RuO$_2$ can undergo transitions between the nonmagnetic state with no spin splitting and the magnetic states with spin splitting in the band structure. These findings not only demonstrate intriguing physics in 4d-electron-correlated RuO$_2$, but also retain its potential for spintronic applications.

Superluminous supernovae (SLSNe) are one of the most luminous stellar explosions known, yet they remain poorly understood. Because they are intrinsically rare, efficiently identifying them in the large alert streams produced by modern time-domain surveys is essential for enabling spectroscopic follow-up. We present NOMAI, a machine learning classifier designed to identify SLSN candidates directly from photometric alerts in the ZTF stream, using light curves accumulated over at least 30 days. It does not require any spectroscopic redshift and is running in real time within the Fink broker. ZTF light curves are transformed into a set of physically motivated features derived primarily from model-fitting procedures using SALT2 and Rainbow, a blackbody-based multi-band fitting framework. These features are used to train an XGBoost classifier on a curated dataset of labeled ZTF sources constructed using literature samples of SLSNe, along with TNS and internal ZTF labeled sources. The final training dataset contains 5280 unique sources, including 225 spectroscopically classified SLSNe. On the training sample, the classifier reaches 66% completeness and 58% purity. Deployed within the Fink broker, NOMAI has been running continuously since 18/12/2025 on the ZTF alert stream and publicly reports SLSN candidates every night by automatically posting them to dedicated communication channels. Based on this, we also report the first two-month as an evaluation period, where the classifier successfully recovered 22 of the 24 active SLSNe reported on the Transient Name Server. The achieved performances demonstrate that the classifier provides a valuable tool for experts to efficiently scan the alert stream and identify promising candidates. In the near future, NOMAI is intended to be adapted to operate on the Legacy Survey of Space and Time conducted by the Vera C. Rubin Observatory.

Exceptional points (EPs) in non-Hermitian photonic systems enable unconventional control of wave amplitude and phase. However, identifying the EPs in multidimensional parameter space of a system can be nontrivial and, in some cases, even infeasible. Here we propose an inverse-design method to efficiently locate the scattering EPs for a two-port resonant system supporting a single mode. The proposed method provides a direct guide for tuning of geometric parameters to realize scattering EPs, as confirmed by both full-wave simulation and equivalent circuit model. In principle, our method is compatible with multi-mode system, making it broadly applicable to a broad class of resonant systems.

A novel approach for measuring the electrical resistivity of non-magnetic materials using magnetic force microscopy (MFM) is discussed. In this method, MFM detects magnetic fields generated by eddy currents induced by the oscillation of a magnetized probe tip. To enhance measurement sensitivity, it is essential to increase the magnitude of these eddy currents. It is discussed that introducing controlled sample vibration amplifies eddy current generation by increasing the relative velocity between the probe tip and the sample surface. Theoretical analysis predicts increase of the phase shift by sample vibration, and experimental validation using a modified MFM system confirms the improvement in sensitivity. The calculated and experimental results exhibit relatively good agreement, establishing that sample vibration excitation is an effective strategy for high-sensitivity resistivity measurements.

Arbitrary Detuning ASynchronous OPtical Sampling (ADA-SOPS) is an emerging technique for extending standard pump--probe experiments performed with two femtosecond lasers to multitimescale experiments, which are of great interest for the study of complex systems. Although no specific requirements are needed for laser repetition rates, their ratio determines the achievable delay distribution and therefore is strongly related to the temporal resolution of the technique. We report a detailed theoretical analysis of measurement performances with respect to laser repetition rates, and we validate our model with experimental data. In the case of amplified laser systems, we demonstrate that achieved delays are inherently correlated to the time interval between amplified pulses, which affects the pulse energy and can generate artifacts. Nevertheless, a deep understanding of the origin of such artifacts allows to suggest several compensation strategies, either during data analysis or at the conception of the experimental setup. Finally we present a new algorithm integrated into the ADASOPS device: by selecting pairs of probe pulses having the same elapsed time with respect to the previous pulse, it automatically compensates any effect of energy fluctuation.

Orbital angular momentum (OAM) multiplexing offers a promising approach to high-capacity optical communication by harnessing the orthogonality of vortex beams. However, its practical deployment is severely limited in real-world settings where dynamic scattering media, such as turbulent atmosphere, distort multiplexed fields into random speckles and disrupt OAM demultiplexing. Although existing wavefront shaping and deep learning methods can mitigate static distortions, they fail under time-varying scattering conditions, leading to significant crosstalk and unreliable recovery. Here, we introduce a new concept, correlation invariance, which enables scattering-immune, robust OAM multiplexed transmission through dynamic media. By capturing orthogonally polarized speckle holograms in a compact common-path geometry and computing their intensity cross-correlation, dynamically imposed scattering phases are cancelled out while deterministic object information is preserved. This allows single-shot reconstruction of both amplitude and phase of the input OAM-multiplexed fields, without any pre-calibration or training. As a proof of principle, we demonstrate high-fidelity transmission of 24-bit RGB data with 99.61% accuracy under static scattering and 98.97% accuracy under dynamic scattering. This approach addresses a long-standing barrier in OAM-based systems and opens avenues for robust high-capacity optical communications, encryption, and imaging in dynamic scattering environments.

Ensuring the reproducibility of physics results is one of the crucial challenges in high-energy physics (HEP). In this study, we develop a proof-of-concept system that uses large language models (LLMs) to extract analysis procedures from HEP publications and generate executable analysis code for reproducing published results. Our method consists of two stages. In the first stage, open-weight LLMs extract event selection criteria, object definitions, and other relevant analysis information from a target paper and, when necessary, from its referenced publications, and then produce a structured selection list. In the second stage, the structured selection list is used to generate analysis code, which is then executed and validated iteratively. As a benchmark, we use the ATLAS $H \to ZZ^{*} \to 4\ell$ analysis based on proton-proton collision data recorded in 2015 and 2016 and released as ATLAS Open Data. This benchmark allows direct comparison between the generated results and the published analysis, as well as comparison with a manually developed baseline implementation. We separately evaluate selection extraction and code generation in order to clarify the current capabilities and limitations of open-weight LLMs for HEP analysis reproduction. Our initial results show that recent open-weight models can recover many documented selection criteria from papers and references, and that in some runs they can generate event selections fully matching a baseline implementation at the event level. At the same time, stochasticity, hallucination, and execution failure remain significant challenges. These results suggest that LLMs are already promising as human-in-the-loop tools for reproducibility support, although they are not yet reliable as fully autonomous HEP analysis agents. In this paper, we report the design of the prototype system and its initial performance evaluation.

Exceptional points (EPs) have attracted extensive research interest due to their intriguing properties. One of the hallmarks of EP physics is that dynamically encircling the EPs induces chiral mode switching, arising from the breakdown of adiabaticity due to the presence of a complex spectrum in the system's Hamiltonian. While such chiral mode behavior has been widely observed experimentally, achieving truly adiabatic, and thus symmetric, state transfer, regardless of the winding direction, in time-modulated non-Hermitian systems has remained elusive. In this work, we demonstrate that this long-sought adiabatic state dynamics can indeed be restored. By steering a two-mode photonic setup along specifically designed trajectories in parameter space, we realize conditions where the associated non-Hermitian evolution operator acquires a purely real spectrum. Moreover, our experimental platform enables controlled switching between symmetric (adiabatic) and chiral (non-adiabatic) state-transfer regimes for the same set of initial modes, thus effectively implementing a universal symmetric-asymmetric two-mode switch. Our results therefore open new avenues for harnessing unique topological spectral properties of non-Hermitian systems, paving the way for the practical design of versatile optical wave-manipulation devices and for advancing both classical and quantum information technologies.

Chiral photonics provides powerful routes for controlling the light handedness, yet nonlinear chiral responses are typically associated with intricate three-dimensional systems. Here, we demonstrate that strong nonlinear chirality can emerge and be precisely tuned in planar metasurfaces. We study free-standing membrane metasurfaces composed of periodic lattices of tilted elliptic holes, which preserve out-of-plane mirror symmetry while breaking all in-plane mirror symmetries through the in-plane rotation of the meta-atoms. We demonstrate that optical resonances play a decisive role in governing the nonlinear chiral response, enabling pronounced circular dichroism in third-harmonic generation even when symmetry is broken only in plane. We experimentally reveal strong nonlinear chiral response from the metasurfaces and a striking swapping of nonlinear chiral channels for complementary meta-atom rotation angles. This behaviour arises from the interplay between lattice symmetry and meta-atom orientation, which controls the symmetry of the resonant modes and the resulting nonlinear selection rules. Our results establish meta-atom rotation as a powerful mechanism for engineering nonlinear chiral responses in planar metasurfaces, opening new opportunities for tunable chiral nonlinear metaphotonics devices.

Optical vortices are singularity lines where the light field intensity vanishes and its phase is undefined. These threads of darkness are adorned by Gauss's law as lines of pure longitudinal polarization where the polarization plane tilts and winds around. We unveil the resulting spin field as a unique structure which unifies both topological textures and defects, as it includes a point defect of undefined spin surrounded by a meronic texture which spans half the spin unit sphere. Moreover, this intricate topological structure of transverse spin does not spread in propagation, is localized arbitrarily below the wavelength of light and presents highly anisotropic features. Here we describe these hidden topologies of transverse spin embedded in simple scalar vortex beams, highlighting the diversity of topological structures that arise in two different spaces -- the spin unit sphere and the transverse-axial Poincaré sphere -- and discuss the underlying aspects behind their subwavelength localization.

We study the navigation of a self-propelled inertial particle in two-dimensional Rayleigh--Bénard convection at Prandtl number $Pr = 0.71$ and cell aspect ratio $Γ= 4$ for Rayleigh numbers $Ra$ ranging from $10^{7}$ to $10^{11}$. A reinforcement-learning (RL) controller selects the propulsive acceleration, subject to an upper bound $\mathcal{A}_{\max}$, to achieve a prescribed horizontal displacement. We find that the success rate increases abruptly with $\mathcal{A}_{\max}$ at moderate $Ra$, whereas at higher $Ra$ the transition becomes more gradual and shifts to larger $\mathcal{A}_{\max}$. Moreover, although the completion time increases with $Ra$, the propulsion energy required for successful traversal decreases. Proper orthogonal decomposition (POD) reveals that these performance differences arise from reorganisation of the carrier flow. At moderate $Ra$, the dominant large-scale circulation partitions the domain through robust transport barriers, requiring a finite thrust surplus to cross them; at higher $Ra$, energy is distributed across many modes, the barriers fragment, and transient plume-assisted pathways emerge. Compared with a constant-heading baseline, the learned policy aligns with local currents and consumes significantly less energy. Lagrangian coherent structure (LCS) analysis further shows that the RL agent inherently learns to cross repelling barriers and surf along attracting pathways. Finally, by mapping these behaviours onto the local Eulerian flow topology using Voronoi tessellation and the $Q$-criterion, we distil an interpretable, physics-based heuristic strategy that achieves robust navigability. These results connect turbulent-flow organisation with autonomous navigation under bounded actuation.

We demonstrate a rapid, maskless fabrication method for superconducting terahertz Josephson plasma emitters (JPEs) based on direct ultraviolet laser micromachining of Bi$_2$Sr$_2$CaCu$_2$O$_{8+δ}$ (Bi-2212) single crystals. Although machining debris is formed near the processed regions, uniform stacks of intrinsic Josephson junctions are preserved inside the crystal, enabling stable terahertz emission. Devices fabricated with Ag, Cu, and Cr electrodes all exhibited terahertz radiation, with Cu electrodes showing performance comparable to Ag while offering a low-cost alternative. Spectroscopic and polarization analyses indicate that the emitted radiation is elliptically polarized and dominated by the geometrical cavity resonance mode. Structural and electrical characterizations reveal that the machining width and depth are not limited by the optical spot size but are governed by the anisotropic thermal conductivity of Bi-2212, consistent with a thermally dominated laser ablation process. This direct laser micromachining approach provides a fast and versatile fabrication technique for JPEs and is broadly applicable to superconducting electronics and terahertz devices.

This paper presents the experimental verification of a planar guided-wave terahertz (THz) spoof surface plasmon polariton (SSPP) bandpass filter (BPF) using a coplanar stripline (CPS) with internal grooves and periodic gaps. The proposed BPF operates by combining the low-pass behavior from the SSPP's band edge and the high-pass behavior from the gaps that act as series capacitors. The higher and lower cut-off frequencies can be tailored by the appropriate selection of the unit cell geometry. For demonstration, a BPF with a center frequency of approximately 1 THz and a bandwidth of 0.25 THz was designed, fabricated, and experimentally validated. The passband around 1 THz is observed in the measurements, along with the lower and higher cut-off frequencies at approximately 0.91 THz and 1.16 THz, respectively, in agreement with simulation results.

We establish a decomposition of the intensity-intensity correlation of a scalar optical beam carrying orbital angular momentum (OAM) across multiple modes into intermodal contributions, thereby linking it, within the framework of the Hanbury Brown-Twiss effect, to the underlying modal coherence structure. Upon filtering the spiral phase dependence, the intensity correlations are governed by OAM coherence and orbital anisotropy reflecting classical entanglement between spatial and OAM degrees of freedom. These results extend intensity interferometry to structured light fields and provide direct access to modal coherence properties without phase-sensitive measurements.

This paper presents a hybrid switched-capacitor converter (HSCC) with a novel multi-mode modulation (3M) scheme for wide-range voltage regulation in 48-V automotive power systems. By introducing a three-state operating sequence beyond the conventional 2:1 resonant operation, the proposed converter achieves variable step-down conversion ratios while preserving soft-switching operation in most transitions. The proposed modulation supports both zero-current switching (ZCS) and zero-voltage switching (ZVS) modes, enabling efficient operation over a broad range of load and conversion conditions. To enable voltage regulation, a closed-loop control configuration is proposed with a linear proportional-integral (PI) controller, with gain tuning assisted by reinforcement learning (RL) to address the converter's nonlinear and variable-frequency nature while maintaining good transient performance. A hardware prototype was built to validate the proposed modulation scheme. The measured results verify ZCS operation over voltage conversion ratios of 0.2--0.4, with a peak efficiency exceeding 92\% at 100~W, and efficiency above 88\% over a wide operating range for 3:1 conversion. The feasibility of both ZCS and ZVS operation is also experimentally demonstrated. These results show that the proposed HSCC significantly extends the practical regulation range of resonant switched-capacitor converters while maintaining high efficiency.

The large-area synthesis of high-crystalline-quality two-dimensional (2D) materials is at the core of novel material integration for semiconductor technology. This effort relies on developing fabrication and characterization techniques that can uncover the material's intrinsic properties by preserving its pristine conditions. In this article, we present an all ultra-high-vacuum cluster for the growth using molecular beam epitaxy of 2D semiconductors that are unstable under ambient conditions and optical spectroscopy using low temperature (20 K) photoluminescence and Raman scattering. The optical chamber of the setup provides micrometer scale spatial resolution and the ability to scan the entire wafer. The performance of its setup regarding spatial resolution, temperature control over a temperature range of 20-300 K using a closed-cycle cryostat and long-term preservation are demonstrated using as-grown post-transition metal monochalcogenides. Furthermore, we introduce a deconvolution-based algorithm to recover spatial information under vibration using a system-specific point-spread function. This enables in situ analysis of the structural and optoelectronic properties of as-grown materials in their pristine form, providing rich and reproducible feedback for both fundamental studies and the optimization of scalable 2D material growth toward integration in advanced devices.

Stereolithography (SLA) and Tailored Fiber Placement (TFP) were combined to fabricate shape memory alloy hybrid composites (SMAHC) featuring a three-layer structure and exhibiting out of plane bending deformation when activated, in a fully integrated, additive manufacturing process. SMA wires as active elements were attached to a textile reinforcement layer, which then was embedded within a UV-curable polymer matrix and combined with a geometrically tailored toplayer, featuring the re-entrant honeycomb architecture. Exploiting the design freedom of SLA, the overall mechanical response of the SMAHC can be systematically adjusted, enabling controlled out-of-plane bending during thermal activation. Two different SMA integration strategies - manual embedding and automated TFP were investigated to assess their influence on actuation behavior, reproducibility, and deformation behaviour. A total of eight geometric configurations were manufactured and experimentally characterized using synchronized optical measurements. The results demonstrate that the combination of SLA-based fabrication and textile-mediated SMA integration enables precise control over the actuation response, while the use of re-entrant honeycomb structures provides an effective approach to tailor stiffness and deformation characteristics. In particular, the automated TFP integration yields improved reproducibility and more symmetric deformation behavior compared to manual fabrication. The presented approach establishes a fully additive manufacturing route for SMAHCs, enabling the realization of structurally integrated, morphing composite systems with programmable mechanical properties.

Nonreciprocal conversion is essential for protecting sources and enabling unidirectional signal routing in photonic, phononic, electronics, and quantum systems, yet conventional implementations rely on magnetic bias that could be challenging to integrate on chip. We propose a magnet-free scheme for frequency-domain nonreciprocity based on sequential, time-gated couplings in a three-mode system. By activating interactions in a fixed temporal order, the forward and reverse frequency conversion pathways acquire unequal dwell times in a lossy intermediate mode, producing strong nonreciprocity without requiring nonlinearities or magnetic materials. Using a harmonic-balance formulation and a Dyson-Born expansion, we derive a compact analytical expression for the isolation ratio that reveals the roles of Floquet sidebands, duty-cycle control, modulation frequency, and dissipation. The results are confirmed by direct time-domain simulations over a wide parameter range. From these results, we extract practical design rules for optimizing isolation through temporal sequencing, loss engineering, and modulation timing. The framework is general and directly applicable to integrated platforms in photonics, phononics, microwave electronics, and superconducting circuits.

Surface-enhanced Raman scattering (SERS) combines analyte-specificity and single-molecule sensitivity, but its potential is limited by slow readout where sophisticated nanosensors are analysed in a serial fashion, one particle at a time. We introduce SERS lock-in sampling to resolve the decades-old trade-off between spectral resolution and widefield imaging. By leveraging the inherent sparsity of Raman spectra, we demonstrate that a simple digital lock-in scheme allows high-quality chemical imaging far beyond the Nyquist-Shannon limit. Our approach integrates an in-situ temporal reference to transform mechanical jitter into an exploitable feature, enabling near-random sampling. We validate SERS lock-in sampling through the multiplexed and simultaneous imaging of thousands of individual SERS-encoded sensors, achieving an orders-of-magnitude throughput-increase over the state-of-the-art. Furthermore, we demonstrate volumetric 3D chemical imaging in biomedically relevant matrices. This robust, computationally simple strategy transforms SERS from a point-observation tool to an imaging modality for clinical diagnostics and real-time chemical observations.

Urban venues serve as arenas for social mixing, where individuals from different socioeconomic backgrounds share space during daily activities. While residential and activity-space segregation have been extensively studied, less is known about how public transit infrastructure shapes the spatial structure of social mixing at specific locations. This study examines how transit accessibility influences visitor diversity at points of interest (POIs) in nine cities in Sweden and the United States (New York, Washington DC, Atlanta), using mobile phone GPS traces and aggregated foot traffic data in 2024. We employ spatial regression models and geographically weighted regression (GWR) to test whether transit catchment diversity, the socioeconomic heterogeneity of populations reachable by public transit, predicts visitor diversity, and the spatial heterogeneity of this relationship. Transit catchment diversity is a consistent positive predictor of visitor diversity across nearly all cities, with the strongest effects in large metropolitan areas and where transit links demographically distinct neighborhoods. However, the relationship is much more spatially clustered in US cities, where transit-mixing hotspots emerge in peripheral, lower-diversity areas, whereas in Swedish cities diversity is more spatially diffuse and hotspots occur in already-diverse neighborhoods, suggesting that transit compensates for local homogeneity in the former but amplifies existing diversity in the latter. These results suggest that transit infrastructure functions as a spatial bridge, channeling diverse populations to common destinations and reshaping the geography of social mixing in urban space, particularly for venues located farther from city centers and in less-dense areas.

This work presents an updated version of the previously released freeware MAN (Modal Analysis of Nanoresonators) [Comput Phys Commun 284, 108627 (2023)], a software package designed to efficiently compute and normalize quasinormal modes (QNMs) and to use them for the analysis of the optical response of electromagnetic resonators. The current release introduces three major enhancements. First, Version 9 extends the capabilities of Version 8 by incorporating new models and functions dedicated to systems involving two-dimensional materials, such as graphene. Second, based on a newly developed coupled-QNM theory, a new toolbox has been implemented that enables the computation of the complex coupling coefficients between the modes of coupled resonators using the individual QNMs of the uncoupled cavities. Finally, we introduce new functions that allow for the direct evaluation of Fano parameters for the extinction cross-section directly from the QNM field distribution.

Dynamic soaring allows seabirds to harvest mechanical energy from vertical wind shear, but field trajectories lack a benchmark for comparing flight performances across species. We derive a reduced lower bound on transport effort from a simplified Hamilton-Jacobi-Bellman optimal-control model in which slow flight incurs an induced-drag penalty, fast flight incurs a dissipative penalty, and wind shear supplies an effective energetic subsidy. After species-specific normalization of transport speed and an accelerometer-based effort proxy, we map wandering albatrosses, Cory's shearwaters, and Eurasian oystercatchers into a common reduced speed-effort plane and estimate their empirical lower frontiers. The albatross frontier lies closest to the reduced bound, consistent with near-optimal wind-energy harvesting. The shearwater frontier is systematically displaced above it, and oystercatchers occupy a distinct non-soaring regime. The resulting framework places specialist dynamic soaring, mixed flap-gliding, and non-soaring flight in a common mechanical representation and provides a reduced benchmark for comparing wind-assisted flight across species using field trajectories.

We introduce a cyclotomic representation for finite $q$-hypergeometric series and $q$-deformed amplitudes that separates algebraic structure from evaluation. By expressing each summand in a sparse exponent basis over irreducible cyclotomic polynomials, all products and ratios of quantum factorials reduce to integer vector arithmetic. This ensures that cancellations between numerator and denominator are resolved exactly prior to any evaluation. This formulation yields the deferred cyclotomic representation (DCR), a parameter-independent combinatorial object of the series, from which evaluation in any target field is realized as a ring homomorphism. For quantum recoupling coefficients, we demonstrate that this framework achieves linear memory scaling in the compilation phase, eliminates intermediate expression swell in exact arithmetic, and substantially extends the range of reliable double-precision computation by reducing cancellation-induced error amplification. Beyond its computational advantages, the DCR provides a unified perspective on $q$-deformed amplitudes. Structural properties like admissibility at roots of unity, and the classical limit all emerge as intrinsic properties of a single underlying combinatorial object.

Pinching antenna (PA) systems have recently gained significant attention. While their physical-layer security (PLS) is being explored, most studies rely on idealized lossless models, ignoring practical waveguide attenuation. In this paper, we investigate the PLS performance of PA systems under a more realistic attenuation-incorporated waveguide model. Specifically, we investigate a PA system-based secure communication scenario consisting of a base station (BS), a legitimate user, and a passive eavesdropper. We derive expressions for closed-form upper and lower bounds on both the secrecy outage probability (SOP) and ergodic secrecy capacity (ESC). The results indicate that the PA system outperforms conventional fixed-antenna systems.

Objective. Decoding visual attention from brain signals during naturalistic video viewing has emerged as a new direction in brain-computer interface research. Current methods assume that stronger coupling between object motion and neural activity indicates higher attention, but this can be confounded by eye movement artifacts and stimulus properties. This study investigates how visual eccentricity (the distance between a visual object and the fixation point) affects neural responses when eye movement artifacts are controlled. Approach. EEG signals were recorded across three tasks that manipulated object eccentricity and attention conditions while participants maintained gaze fixation. Correlation analysis and match-mismatch decoding were performed to quantify the neural tracking of object motion. Main results. The analysis supports three conclusions: (1) neural tracking of object motion in natural videos works under gaze fixation; (2) the strength of neural tracking under gaze fixation is predictive of attention; and (3) there exists a significant eccentricity confound in the EEG responses, with poorer neural tracking of motion at larger eccentricities. Significance. These results provide critical evidence that findings from previous free-viewing studies reflect genuine neural processing rather than mere oculomotor artifacts. However, the identified eccentricity effect highlights a major limitation for current decoding approaches that assume coupling strength reflects attention levels alone.

By exploiting noise as an information-bearing resource, noise-driven communication offers a promising framework for low-complexity and secure wireless system design. In this letter, the scheme of ternary noise modulation (T-NoiseMod) is proposed for noise-based wireless communication scenarios, where information is encoded into the statistical characteristics of artificial noise. Unlike conventional binary NoiseMod, which employs two variance levels, the proposed scheme introduces a third transmission state: intentional silence. By pairing two consecutive noise blocks, the signaling scheme is expanded to eight valid state combinations, enabling the transmission of three information bits per signaling interval. In our proposed scheme, the two-stage receiver is developed, consisting of mean-based silent-state detection followed by variance-based low/high classification. An analytical expression for the bit error probability (BEP) is derived for Rayleigh fading. Our computer simulation results match closely with our theoretical results and show the effects of key system parameters. Furthermore, comparisons with binary NoiseMod demonstrate the inherent trade-off between reliability and rate.

In the sixth generation (6G) wireless communication networks, the device density, antenna number, and the complexity of communication scenarios will significantly increase, which brings great challenges for system design and network optimization. By obtaining channel information in advance, channel map has become a promising solution to these challenges in 6G era. However, conventional channel maps cannot be updated in time as physical environment changes. To solve the problem, a novel dynamic channel map (DCM) is proposed in this work. For DCM construction, we further present a ray tracing (RT) and geometric stochastic hybrid channel model (RT-GSHCM), which pre-constructs the DCM offline by RT and updates it online by geometry-based stochastic channel model (GBSM). By this way, the DCM can provide time-varying channel information and channel properties while matintaining accuracy. Next, a channel measurement campaign is conducted, and the measurement results are compared with the RT-GSHCM, RT, and GBSM. The comparison results validate the accuracy of DCM. Meanwhile, the time cost on DCM update is compared with that of conventional channel maps, illustrating the time-efficiency of DCM. Finally, important statistical channel properties of RT-GSHCM are further derived, analyzed, and compared under different configurations of interaction objects in physical environment.

Time-frequency representations, such as the short-time Fourier transform (STFT), are fundamental tools for analyzing non-stationary signals. However, their ability to achieve sharp localization in both time and frequency is inherently limited by the Gabor-Heisenberg uncertainty principle. In this paper, we address this limitation by introducing a method to generate super-resolution spectrograms through the fusion of two or more spectrograms with varying resolutions. Specifically, we compute the super-resolution spectrogram as the barycenter of input spectrograms using optimal transport (OT) divergences. Unlike existing fusion approaches, our method does not require the input spectrograms to share the same time-frequency grid. Instead, the input spectrograms can be computed using any STFT parameters, and the resulting super-resolution spectrogram can be defined on an arbitrary user-specified grid. We explore various OT divergences based on different transportation costs. Notably, we introduce a novel transportation cost that preserves time-frequency geometry while significantly reducing computational complexity compared to standard Wasserstein barycenters. We adopt the unbalanced OT framework and derive a new block majorization-minimization algorithm for efficient barycenter computation. We validate the proposed method on controlled synthetic signals and recorded speech using both quantitative and qualitative evaluations. The results show that our approach combines the best localization properties of the input spectrograms and outperforms an unsupervised state-of-the-art fusion method.

Reliable communication in confined environments, such as blood vessels or industrial pipelines, remain challenging due to signal attenuation and limited sensor accessibility. Therefore, this work investigates microbubbles as robust information carriers within the Internet of Bio-Nano Things (IoBNT) paradigm, leveraging their established use as ultrasound contrast agents. It presents a combined experimental and numerical analysis characterizing microbubble transport under varying flow conditions relevant to biomedical and industrial applications. Experiments with SonoVue microbubbles in a recirculating water channel validate an OpenFOAM-based Computational Fluid Dynamics (CFD) simulation using the incompressibleDenseParticleFluid solver. Key cases examine water vs. blood-like media and high vs. physiological flow velocities, analyzing the relative influence of fluid properties and advection on microbubble dynamics. Recirculation effects are considered in relation to in vivo circulation timescales.

Sub-THz bands are promising high bandwidth and data rates, and in the recent years the device technologies made large progress and provided a multitude of transceiver, power amplifier (PA) and phased array devices supporting the frequency bands above 100 GHz. The more painful aspect of sub-THz transmission is the increased power consumption, caused by the large data rates and the related data conversion and processing effort, and on the analog side the low achievable PA efficiency and the reduced achievable output power. When planning a deployment of sub-THz communication systems, the target coverage and throughput can be achieved with a variety of scenarios, which will be different with respect to locations and number of base stations and system architectures. Although leading to similar performance, they will differ significantly in the overall power consumption. With an accurate power consumption model, including also baseband (BB) processing functionality, and system level simulations for different hybrid beamforming and MIMO schemes the related variations in power consumption in relation to a given performance are evaluated. This paper shows the critical design aspects for energy efficient sub-THz deployments by highlighting the sub- THz specific trade-offs between different number of BS with different transmit powers but also changing number of BB units and RF chains.

Sub-Terahertz radio-stripe and distributed MIMO architectures promise extreme spatial reuse and multi-GHz bandwidths, but the cascaded fiber front-haul and RF hardware impairments strongly shape end-to-end performance. This paper presents an open-source, configuration-driven simulator that models the full waveform-level signal chain from CP-OFDM baseband generation in the central unit, through measurement-parameterized polymer microwave fiber and coupler links, to booster/active Radio Units (RUs) with configurable nonlinearity, noise, in-phase and quadrature imbalance, and oscillator phase noise and carrier frequency offset. Wireless propagation is supported via lightweight deterministic and stochastic per-subcarrier channel models as well as site-specific ray-tracing datasets generated with a companion Sionna ray-tracer module. The simulator exports intermediate waveforms and system metrics (e.g., normalised mean square error, signal-to-noise-and-distortion ratio, bit error rate) to enable reproducible studies of impairment accumulation, calibration, and algorithmic choices such as RU selection and beam management.

We study matched and Euclidean-mismatched decoding on finite Fourier-curve constellations with tangent-space artificial noise. Each hypothesis induces a Gaussian law with symbol-dependent rank-one covariance. We derive exact Euclidean pairwise errors for arbitrary pairs and an exact Gaussian-expectation representation for matched decoding on bilaterally tangent-orthogonal pairs. For uniform even constellations, the Euclidean side yields explicit distance spectra and symbol-error bounds across all offset classes; the matched side is exact on antipodal pairs and benchmarked numerically at the full-codebook level via Monte Carlo. By isolating the detection-theoretic consequence of tangent-space artificial noise, these results clarify analytically how noise fraction and constellation density enter the mismatch behavior; secrecy-rate implications require additional channel and adversary modeling.

To mitigate the residual interference from imperfect successive interference cancellation (SIC) in Rate-Splitting Multiple Access (RSMA), this paper incorporates improper Gaussian signaling (IGS) into the downlink RSMA framework. Unlike existing RSMA--IGS works that embed impropriety within IQ-imbalanced frameworks, we show that IGS alone effectively counters SIC-induced residual interference. For a basic SISO setup with IGS on the common stream and PGS on private streams, we establish three key results: the optimal impropriety degree for private rate maximization attains its maximum; closed-form optimal solutions with rigorous monotonicity conditions are derived for common rate maximization; and a soft actor-critic (SAC) algorithm is developed for the non-convex sum rate problem. Numerical results show that IGS consistently outperforms PGS, with the gain widening as SIC imperfection increases.

To meet the urgent demands for spectral efficiency and multi-user access in high-frequency application scenario for the sixth-generation wireless communication, this paper investigates a rate splitting multiple access (RSMA) system assisted by pinching antennas (PAs) with multiple waveguides and multiple carriers, aiming to maximize the overall system sum rate. To address the high sensitivity of high-frequency signals to PA movement in the overloaded scenarios, a two-stage PA position optimization method based on both path loss and phase shift error minimization is proposed under RSMA framework. Specifically, the first step is to perform coarse adjustment by minimizing large-scale path loss. Then, based on the derivation of a closed-form solution for the ideal phase shift in a single-user single-carrier case, the fine-grained positions of PAs are optimized via a one-dimensional line search to minimize the composite phase shift error across all users and carriers. In order to meet the quality of service requirements, the Lagrange dual method is employed to obtain the closed form of beamforming vectors after the PA positions are determined. Simulation results demonstrate that the proposed scheme achieves significant improvement in sum rate and confirm that RSMA exhibits stronger robustness to inaccurate PA positions caused by both discrete position channel estimation and physical hardware compared to other multiple-access techniques in PA-assisted systems. Furthermore, the results validate that fine-grained PA position adjustment is particularly crucial in high-frequency bands.

The globally optimal robust adaptive beamforming (RAB) solution is studied for worst-case signal-to-interference-plus-noise ratio (SINR) maximization (the maximin SINR problem) under convex and closed uncertainty sets for the desired signal covariance and interference-plus-noise covariance (INC) matrices, considering a general-rank signal model. First, the corresponding minimax SINR problem is reformulated as a convex optimization problem. In particular, this problem becomes a semidefinite programming (SDP) problem when the uncertainty sets can be represented by finitely many linear matrix inequality constraints. It is then shown that, for a general-rank signal model, the maximin and minimax SINR problems are equivalent when the uncertainty sets are convex and closed, in the sense that they share the same optimal value and the same set of optimal solutions. The requirement of closedness is weaker than the compactness assumption previously used to establish the equivalence between minimax and maximin SINR problems for the rank-one signal model, a state-of-the-art result reported approximately two decades ago. Consequently, an optimal solution to the minimax SINR problem is also globally optimal for the maximin SINR problem, and this solution can be obtained by solving the equivalent SDP of the minimax problem in a single step. In contrast, existing iterative approximation algorithms for the maximin SINR problem yield only locally optimal solutions. Simulation results demonstrate that these approximation algorithms return suboptimal values that can be strictly smaller than the optimal value of the minimax problem, and that the beamformer output SINR obtained via the minimax formulation is higher than that achieved by beamformers derived from the maximin problem using approximation algorithms.

This paper explores an integrated sensing and communication (ISAC) system with backscattering RFID tags. In this setup, an access point employs communication beams to serve communication users while leveraging a sensing beam to interrogate RFID tags. Under the total transmit power constraint of the system, our objective is to design a joint sensing and communication beamforming codebook by considering the tag interrogation and communication requirements. To lay a foundation for the codebook design problem, we first study the beamforming design problem in a single-tag scenario and investigate two approaches: (i) a zero-forcing approach with optimized sensing/communication power allocation, for which a closed-form solution is derived under a dominant sensitivity condition, and (ii) a joint sensing and communication beamforming design obtained by transmit power minimization. Then, we investigate the codebook design problem in a multi-tag scenario. To resolve this, we propose a sector-based joint sensing and communication beamforming codebook that scans the region of interest. For each sector, semidefinite relaxation and generalized Benders decomposition are employed to handle the resulting optimization. The simulation results show that the proposed joint beamforming designs can effectively mitigate the mutual interference between sensing and communication functionalities, thus enhancing the interrogation range of the tags with minimized transmit power. Also, the efficacy of the proposed sector-based codebook design has been demonstrated in terms of interrogation success rate, offering a promising approach for the ISAC-backscattering systems.

Affine frequency division multiplexing (AFDM), an emerging multi-carrier modulation scheme, has garnered significant attention due to its resilience to Doppler shifts and capability to achieve full diversity in doubly dispersive channels. However, existing data detection algorithms for AFDM systems face a significant trade-off between computational complexity and accuracy. In this paper, a novel low-complexity data detection scheme, termed the soft-feedback detector (SFD), is proposed. Particularly, building upon a maximum ratio combining (MRC) estimator framework, the SFD leverages the a priori symbol distribution to mitigate error propagation during iterative detection. Specifically, soft-decision feedback is incorporated as extrinsic information derived from the log-likelihood ratios of the transmitted symbols. As a result, the proposed detector significantly enhances detection accuracy while maintaining low computational complexity. Simulation results demonstrate that the SFD consistently outperforms benchmark decision-feedback detectors. In particular, compared with the conventional MRC detector, the proposed scheme achieves approximately a 3 dB signal-to-noise ratio (SNR) gain at the bit error rate (BER) of $10^{-3}$.

Beam squint, the frequency-dependent shift of the main beam, poses a major challenge for wideband antenna arrays. This paper focuses on the beam squint effects in super wideband (SW) systems, where high mutual coupling (MC) effects are present. These high MC effects complicate beamforming (BF) by creating frequency-dependent phase relationships that invalidate conventional approaches. To accurately model MC effects, this paper uses a circuit-theoretic framework for tightly coupled SW uniform linear arrays (ULAs). We derive closed-form expressions for the average received signal-to-noise ratio (SNR) with BF in conventional half-wavelength spaced, weakly coupled arrays and validate them. Extending our analysis to tightly coupled SW arrays, we demonstrate that, in contrast to conventional weakly coupled arrays, the effective true time delays exhibit a nonlinear dependence on frequency due to coupling-induced phase shifts. A comparative analysis reveals that strong MC in SW arrays significantly reduces squint in phase-controlled BF, extending the usable bandwidth considerably.

A site-specific Type-II codebook design is proposed for downlink massive multiple-input multiple-output (MIMO) limited-feedback beamforming. The key idea is to embed a learned site-specific propagation prior into the Type-II channel state information (CSI) feedback pipeline. Specifically, the base station (BS) uses a low-overhead reference signal received power (RSRP) fingerprint collected during synchronization signal block (SSB) probing to infer a user equipment (UE)-dependent dominant beam subspace before explicit CSI acquisition. The UE then estimates and feeds back only the low-dimensional effective channel coefficients within this inferred subspace, thereby avoiding full-dimensional online subspace discovery while retaining a rich multi-beam representation capability. To analyze the proposed design and compare it with standardized feedback mechanisms, a unified subspace-projection framework is developed by jointly characterizing CSI acquisition, UE-side compression, BS-side reconstruction, and effective spectral efficiency. Under this framework, Type-I, Type-II, port-selection feedback, and the proposed scheme are interpreted as different ways of inducing a feedback representation subspace. The probing codebook and the BS-side subspace inference network are then formulated as a coupled task-oriented design problem and are optimized end-to-end by maximizing the normalized CSI-capture efficiency. Extensive simulation results demonstrate that the proposed feedback scheme achieves Type-II-comparable CSI-capture capability with substantially lower online overhead and UE-side complexity, thereby improving the effective spectral efficiency.

Angle-of-arrival (AoA) estimation is a crucial function in wireless communications used for localization, beam-forming, interference management, and other applications. Deep learning (DL) solutions have been proposed for AoA to mitigate limitations of traditional AoA estimation techniques such as sensitivity to noise and the inability to generalize across different array characteristics. A challenge, however, of DL-based approaches is their reliance on large data collection campaigns and model training. This paper proposes the application of Prototypical Networks (PN) to address this challenge and utilizes a real-world dataset collected on a software defined radio (SDR) testbed to validate the effectiveness of the proposed solution. Prototypical Networks excel in extracting representative embeddings from unstructured input data, establishing class prototypes during training that can be few-shot trained on unseen classes. We demonstrate the efficacy of PNs for AoA classification using complex IQ samples, focusing on its ability to correctly classify new, unseen angles that the model was not trained on previously. Our results show that training our proposed ProtoAoA on only 23% of the AoA dataset classes can attain a mean absolute error (MAE) of 3 degrees with only 4-shots of training on the unseen angles - and an MAE of 2 degrees with 32-shots of training data. These results demonstrate that the developed prototypical network architecture requires remarkably few data samples to achieve reliable AoA estimation - and highlights its potential for other wireless applications where data availability is limited.

Foundation models show strong potential for large-scale, high-dimensional biomedical applications, yet their ability to capture relevant neurobiological characteristics remains underexplored. We systematically evaluate embeddings from two neuroimaging foundation models, BrainLM and SwiFT, across multi-site fMRI datasets using a comprehensive evaluation framework. Our results show that foundation model embeddings encode substantial batch-related variability, often dominating diagnosis-related information across heterogeneous datasets. We further investigate how harmonization, applied to reduce batch effects, influences these embeddings. In addition, we find that BrainLM prefers to capture fine-grained regional activity, whereas SwiFT tends to represent interactions between regions, consistent with their respective model architectures. Our study highlights the importance of accounting for batch effects in foundation models and motivates future work on disentangling biologically meaningful signals from acquisition-related variability.

Scalable estimation of quantum states with readout errors is a central challenge in large multiqubit systems. Existing overlapping-tomography methods improve scalability by working with local subsystems, but they usually assume known or separately calibrated measurements. At the same time, readout-estimation methods model measurement errors without enforcing consistency among overlapping regional states. In this context, the present paper introduces a unified framework for joint regional quantum state tomography with readout errors. A multiqubit system is partitioned in overlapping regions, each region is assigned to a local density operator and a local confusion matrix, and neighboring regions are coupled through reduced-state consistency on shared subsystems. This leads to a structured bilinear optimization problem. To solve it, a distributed alternating method is developed in which the state-update step is handled by the alternating direction method of multipliers (ADMM), while the confusion-matrix updates are carried out locally in parallel. Analytical guarantees are also established, including a sufficient condition for local identifiability, local quadratic growth of the population misfit, and convergence of the inner state-update procedure. Simulations on Ring, Ladder, Torus, and Hub graph geometries show that joint estimation improves state recovery over fixed-readout reconstruction, recovers a substantial portion of oracle performance, and reveals a clear tradeoff between state estimation performance, communication, and computation.

Reliable geolocation of non-cooperative emitters in environments where Global Navigation Satellite Systems (GNSS) are unavailable or degraded is a key enabler for spectrum regulation, emergency response, autonomous mobility, and Integrated Sensing and Communication (ISAC) services in 5G/6G systems. Doppler-based techniques - from single-receiver Signal Doppler Frequency (SDF) fixes through multi-node Frequency Difference of Arrival (FDOA) and Direct Position Determination (DPD) to derivative-enhanced and learning-assisted hybrids - exploit radial-velocity-induced frequency shifts as a passive, high-resolution localization cue accessible with commodity software-defined radios, millimeter-wave access points, or acoustic sensors. This review consolidates over a decade of research across radio, acoustic, and satellite domains. It introduces a unifying taxonomy that divides the field into five technique families, outlining their evolution, measurement models, and estimator archetypes. It then compares algebraic, Bayesian, convex, and neural inference frameworks under realistic impairments such as oscillator drift, multipath, and asynchronous clocks, highlighting conditions where derivative Doppler metrics tighten the Cramer-Rao bound with minimal hardware cost. Environment-specific deployments are examined, from urban canyons and GNSS-denied tunnels to underwater, radar, UAV-swarm, and multi-orbit satellite scenarios, with prototype accuracies reaching meter scale using low-size, weight, and power payloads. Finally, the survey distils design recommendations for mobile and tactical operations and identifies open research challenges in frequency-reference integrity, multipath-aware modelling, edge-constrained computation, and trajectory-aware sensing.

We develop a framework to describe gravitational wave propagation through a stochastic distribution of weak gravitational lenses beyond the geometric optics limit. We model the lens distribution as a static random background field and formulate the problem in the language of quenched disorder, treating the disorder averaged density matrix as the fundamental object from which observables are computed. Using the Schwinger Keldysh formalism, we construct a path-integral representation of the averaged density matrix and derive its explicit form perturbatively for a suitable class of couplings. The result naturally separates into a quadratic exponential term, which governs the suppression of phase sensitive contributions in the averaged description, and a purely oscillatory contribution, which modifies coherent propagation through a disorder-induced correction to the propagation kernel. This provides a unified description of interference, diffraction, and statistical fluctuations of the lens distribution within a single framework. We also identify the physical scales controlling the onset of coherence loss and illustrate the formalism in the case of Gaussian wave packets. More generally, the derivation applies to any system described by the same class of actions, making the framework relevant beyond gravitational wave lensing to wave propagation in disordered media.

This paper investigates certified upper bounds on the minimum distance of an explicit family of Calderbank-Shor-Steane quantum LDPC codes constructed from affine permutation matrices. All codes considered here have active Tanner graphs of girth eight. Rather than attempting to prove a general lower bound for the full code distance, we focus on constructing low-weight non-stabilizer logical representatives, which yield valid upper bounds once they are verified to lie in the opposite parity-check kernel and outside the stabilizer row space. We develop a unified framework for such witnesses arising from latent row relations, restricted-lift subspaces including block-compressed, selected-fiber, and CRT-stripe constructions, cycle- 8 elementary trapping-set structures, and decoder-failure residuals. In every case, search is used only to generate candidates; the reported bounds begin only after explicit kernel and row-space exclusion tests have been passed. For the latent part, we also identify a block-compression criterion under which the certification becomes exact. Applying these methods to representative APM-LDPC codes sharpens previously reported upper bounds and provides concrete certified values across the explored parameter range.

A standard approach to generate random pure quantum states relies on sampling from the Haar measure. However, the entanglement properties of such states present a fundamental challenge for their general applicability. Here, we introduce the $σ$-ensembles $\unicode{x2013}$ a family of random quantum states with only a single control parameter. Crucially, these states are designed such that they can be tuned between volume-law and area-law behavior, which has been a major obstacle thus far. We construct representatives of this ensemble by imposing a probability distribution on the eigenvalues of the successive subsystems, and subsequently reconstructing a compatible global state using the matrix product state (MPS) formalism. Due to their area-law entanglement, our approach circumvents the intractability of Haar-random pure states in classical simulations of quantum systems and is more representative of typical Hamiltonian ground states.

An $n$-qubit Dicke state of weight $k$, is the uniform superposition over all $n$-bit strings of Hamming weight $k$. Dicke states are an entanglement resource with important practical applications in the NISQ era and, for instance, play a central role in Decoded Quantum Interferometry (DQI). Furthermore, any symmetric state can be expressed as a superposition of Dicke states. First, we give explicit constant-depth circuits that prepare $n$-qubit Dicke states for all $k \leq \text{polylog}(n)$, using only multi-qubit Toffoli gates and single-qubit unitaries. This gives the first $\text{QAC}^0$ construction of super-constant weight Dicke states. Previous constant-depth constructions for any super-constant $k$ required the FANOUT$_n$ gate, while $\text{QAC}^0$ is only known to implement FANOUT$_k$ for $k$ up to $\text{polylog}(n)$. Moreover, we show that any weight-$k$ Dicke state can be constructed with access to FANOUT$_{\min(k,n-k)}$, rather than FANOUT$_n$. Combined with recent hardness results, this yields a tight characterization: for $k \leq n/2$, weight-$k$ Dicke states can be prepared in $\text{QAC}^0$ if and only if FANOUT$_k \in \text{QAC}^0$. We further extend our techniques to show that, in fact, \emph{any} superposition of $n$-qubit Dicke states of weight at most $k$ can be prepared in $\text{QAC}^0$ with access to FANOUT$_k$. Taking $k = n$, we obtain the first $O(1)$-depth unitary construction for arbitrary symmetric states. In particular, any symmetric state can be prepared in constant depth on quantum hardware architectures that support FANOUT$_n$, such as trapped ions with native global entangling operations.

We theoretically investigate the generation of non-Gaussian quantum states, specifically Schrödinger cat-like states (SCLSs), via degenerate dual-pump spontaneous four-wave mixing in a $χ^{(3)}$-based microring resonator. By introducing a unitary transformation that exactly decouples the self-phase modulation (SPM) and cross-phase modulation (XPM) terms, we reduce the full nonlinear Hamiltonian to an effective three-mode interaction. The resulting dynamics (decoupled and full Hamiltonians) are studied using the Lindblad master equation, accounting for cavity losses. Unlike semiclassical or parametric approximations, our full quantum mechanical approach explicitly includes quantum pump depletion, which enables the emergence and observation of non-Gaussian features. We compute the Wigner function, photon number distributions, quadrature variances, Fano factor, Schmidt number, and fidelity to characterize the generated states. For the non-dissipative case, we find that the signal mode $\hat{b}_3$ or $\hat{a}_3$ exhibits clear non-Gaussian features with a structured Wigner function and even-dominated photon number distribution, characteristic of an even coherent state. In the presence of dissipation ($γ_j = 0.2$), the interference fringes become faint, odd photon numbers appear, and the fidelity with the ideal state remains high ($>0.9$), indicating robustness. The pump mode $\hat{b}_1$ or $\hat{a}_1$ remains Gaussian, while both modes display super-Poissonian statistics and entanglement ($>2$). Our results demonstrate that degenerate dual-pump spontaneous four-wave mixing in microring resonators is a promising platform for generating and controlling cat-like states under dissipative conditions.

Node embeddings act as the information interface for graph neural networks, yet their empirical impact is often reported under mismatched backbones, splits, and training budgets. This paper provides a controlled benchmark of embedding choices for graph classification, comparing classical baselines with quantum-oriented node representations under a unified pipeline. We evaluate two classical baselines alongside quantum-oriented alternatives, including a circuit-defined variational embedding and quantum-inspired embeddings computed via graph operators and linear-algebraic constructions. All variants are trained and tested with the same backbone, stratified splits, identical optimization and early stopping, and consistent metrics. Experiments on five different TU datasets and on QM9 converted to classification via target binning show clear dataset dependence: quantum-oriented embeddings yield the most consistent gains on structure-driven benchmarks, while social graphs with limited node attributes remain well served by classical baselines. The study highlights practical trade-offs between inductive bias, trainability, and stability under a fixed training budget, and offers a reproducible reference point for selecting quantum-oriented embeddings in graph learning.

The impossibility of simultaneously cloning non-orthogonal states lies at the foundations of quantum theory. Even when allowing for approximation errors, cloning an arbitrary unknown pure state requires as many initial copies as needed to fully learn the state. Rather than arbitrary unknown states, modern quantum learning theory often considers structured classes of states and exploits such structure to develop learning algorithms that outperform general-state tomography. This raises the question: How do the sample complexities of learning and cloning relate for such structured classes? We answer this question for an important class of states. Namely, for $n$-qubit stabilizer states, we show that the optimal sample complexity of cloning is $Θ(n)$. Thus, also for this structured class of states, cloning is as hard as learning. To prove these results, we use representation-theoretic tools in the recently proposed Abelian State Hidden Subgroup framework and a new structured version of the recently introduced random purification channel to relate stabilizer state cloning to a variant of the sample amplification problem for probability distributions that was recently introduced in classical learning theory. This allows us to obtain our cloning lower bounds by proving new sample amplification lower bounds for classes of distributions with an underlying linear structure. Our results provide a more fine-grained perspective on No-Cloning theorems, opening up connections from foundations to quantum learning theory and quantum cryptography.

We introduce a systematic framework to construct nonlocal observables with extensive quantum Fisher information (QFI) density in stabilizer codes. The construction maps stabilizer generators to dual Ising spins whose correlators equal string order parameters, converting hidden nonlocal order into a metrologically accessible observable. Applying this to monitored cluster codes and the toric code, we identify transitions in the QFI scaling from an extensive regime, where long-range string order prevails, to an intensive one driven by competing single-site measurements.

We develop a quantum algorithm for estimating the free energy as well as the total Gibbs state of interacting quantum Coulomb gases and molecular systems in dimensions $d \in \{2,3\}$ at finite temperature. These systems lie beyond the reach of existing methods due to their singular interactions and infinite-dimensional Hilbert space structure. First, we show that the free energy of the full many-body Hamiltonian can be approximated by that of the same Hamiltonian with a finite-rank low-energy truncation of the interaction, with an explicit error bound polynomial in the particle number. This reduces the problem to a controlled finite-rank perturbation problem. Second, we introduce a quantum Gibbs sampling scheme tailored to this truncated system, based on a class of quantum Markov semigroups. Our main analytical result establishes that the associated generator has a strictly positive spectral gap for every truncation, implying exponential convergence to the target Gibbs state. This provides, to our knowledge, the first rigorous mixing-time guarantee for Gibbs sampling in a Coulomb interacting continuous-variable quantum system. Finally, we give an explicit quantum circuit implementation of the dynamics and derive an end-to-end complexity bound for approximating the free energy and the Gibbs state itself. Our results provide a mathematically rigorous route to quantum algorithms for free energy estimation in interacting quantum systems, without relying on classical approximations such as the Born-Oppenheimer reduction.

Stationary black hole geometries with non-degenerate Cauchy horizons are classically unstable due to mass inflation. At extremality, mass inflation is absent, but a different dynamical instability arises: the Aretakis instability. In this work, we investigate the properties of degenerate horizons and their associated Aretakis instabilities. By studying examples with increasingly higher-order horizon degeneracy, we show that the Aretakis instability weakens as the degree of degeneracy grows. Motivated by these results, we propose a new black hole geometry characterized by an infinitely degenerate horizon, which we argue is stable under Aretakis-type perturbations and may therefore provide a concrete realization of a "graveyard" end state for these objects.

Photonic architectures are one of the leading platforms for demonstrating quantum computational advantage, with Boson Sampling and Gaussian Boson Sampling as the primary schemes. Yet, we lack for these photonic primitives a systematic theoretical understanding of linear cross-entropy benchmarking (LXEB), which is a central tool for testing quantum advantage proposals. In this work, we develop a representation-theoretic framework for the classical computation of average LXEB scores and second moments of output probability distributions, covering a range of quantum advantage experiments based on scattering $n$-photon states through $m$-mode Haar-random interferometers. Our methods apply in any regime, including the saturated regime, where the (expected) number of photons is comparable to the number of optical modes. The same second-moment techniques also allow us to prove anticoncentration for traditional Fock-state Boson Sampling in the saturated regime. Interestingly, for Gaussian Boson Sampling second moments are not sufficient to establish a meaningful anticoncentration statement. The technical core of our approach rests on decomposing two copies of the $n$-particle bosonic space $\mathrm{Sym}^n(\mathbb{C}^m)$ into irreducible representations of $\mathrm{U}(m)$. This reduces two-copy Haar averages to computing purities of initial states after partial traces over particles, highlighting the role that particle entanglement plays for LXEB and anticoncentration.

In this work, we build a novel frequency-momentum space for $(d+1)$-dimensional de Sitter (dS) correlators from first principles. This construction follows directly from the decomposition into unitary irreducible representations (UIRs) of the spacetime isometry group $\mathrm{SO}(1,d+1)$. While the spatial momentum space is given by the standard $d$-dimensional Fourier transform, the frequency space arises from diagonalising the quadratic Casimir operator, leading to the $(d+1)$-dimensional Kontorovich-Lebedev-Fourier (KLF) transform. We show that square-integrable functions decompose only along the principal series, whereas more general functions can receive discrete contributions from other UIRs. Applying this framework to the bulk CFT two-point function reproduces its Källén-Lehmann representation. Using the path integral formulation, we derive the Feynman rules for in-in perturbation theory in KLF space, leading to the introduction of KLF-space correlators, which are simply related to late-time correlation functions through a reduction formula. Furthermore, the KLF-space formulation sheds light on the simple mathematical structure of perturbative computations. In particular, the propagators take the form of simple rational functions, and tree-level diagrams can be written as spectral integrals over known meromorphic functions, as demonstrated in the example of the single-exchange four-point function. At the loop level, we show, through the example of the self-energy correction to the scalar propagator, that the group-theoretical nature of the construction allows the momentum integral to be recast as an orthogonality relation among $\mathrm{SO}(1,d+1)$ Clebsch-Gordan coefficients.

The 2-Forrelation problem provides an optimal separation between classical and quantum query complexity and is also the problem used for separating $\mathsf{BQP}$ and $\mathsf{PH}$ relative to an oracle. A natural question is therefore to ask what are the minimal quantum resources needed to solve this problem. We show that 2-Forrelation can be solved using Instantaneous Quantum Polynomial-time ($\mathsf{IQP}$) circuits, a restricted model of quantum computation in which all gates commute. Concretely, two $\mathsf{IQP}$ circuits with two quantum queries and efficient classical processing suffice. For the signed variant of 2-Forrelation, even a single $\mathsf{IQP}$ circuit and query suffices. This answers a recent open question of Girish (arXiv:2510.06385) on the power of commuting quantum computations. We use this to show that $(\mathsf{BPP}^{\mathsf{IQP}})^O \not\subseteq \mathsf{PH}^O$ relative to an oracle $O$, strengthening the result of Raz and Tal (STOC 2019). Our results show that $\mathsf{IQP}$ circuits can be used for classically hard decision problems, thus providing a new route for showing quantum advantage with $\mathsf{IQP}$ circuits, avoiding the verification difficulties associated with sampling tasks. We also prove Fourier growth bounds for $\mathsf{IQP}$ circuits in terms of the size of their accepting set. The key ingredient is an algebraic identity of the quadratic function $Q(x) = \sum_{i < j} x_ix_j$ that allows extracting inner-product phases within an $\mathsf{IQP}$ circuit.

We present a numerical-relativity study of comparable-mass boson star-black hole (BS-BH) head-on collisions, focusing on both initial-data construction and gravitational-wave (GW) phenomenology. We show that plain superposition can strongly perturb the BS core, leading to large constraint violations and unphysical radial oscillations. To remedy this problem, we introduce a one-body conformal-factor correction and find that it robustly suppresses these artifacts. Using the improved initial data, we analyze GW emission from equal- and unequal-mass BS-BH binaries and compare with matched BS-BS and BH-BH baselines. For equal masses, the BS-BH radiated energy increases with BS compactness and approaches the BH-BH limit for highly compact stars. For unequal masses, the dominant $(2,0)$ mode often remains close to the BH-BH morphology, whereas the subdominant $(3,0)$ mode provides clear discriminatory power when the BH is the heavier companion. Our results identify higher multipoles as a key observable for distinguishing mixed BS-BH mergers from pure BH binaries.

Quantum state purification, which operates not by identifying and correcting specific errors but by repeatedly projecting multiple noisy copies onto special subspaces, provides a syndrome-free alternative to quantum error correction. Existing purification protocols, however, generally assume unconstrained operations and thus overlook the energetic restrictions inherent in realistic quantum devices. Here, we establish a general framework for universal state purification under energy-conservation constraints for depolarizing noise. We derive a necessary and sufficient condition for the nonexistence of universal energy-preserving purification and, whenever such purification is feasible, analytically determine the optimal performance and the corresponding protocols. We further show how the optimal protocols can be systematically implemented using only energy-preserving operations. Numerical results confirm the effectiveness of the proposed scheme. Our framework recovers the standard purification setting as a special case and naturally extends to scenarios assisted by external energy resources. These results identify fundamental physical limits on state distillation and provide an energy-efficient route to quantum error mitigation.

Inflationary models that contain a transient ultra-slow-roll phase can exhibit strong non-perturbative dynamics, making the usual perturbative treatment of cosmological fluctuations incomplete. In such regimes, quantum diffusion and the nonlinear gravitational response of the background can both play important roles, motivating a framework that treats them systematically within quantum field theory in curved spacetime. In this work, we derive the first-order stochastic equations in quasi-de Sitter spacetime from the Schwinger-Keldysh formalism and develop a practical procedure to obtain compact stochastic equations that consistently incorporate metric perturbations via the classical Arnowitt-Deser-Misner equations. Our approach systematically captures classical non-perturbative effects while retaining the leading first-order quantum diffusion. We apply the formalism to two inflationary scenarios with an ultra-slow-roll phase, namely the Starobinsky piecewise-linear model and critical Higgs inflation. For the Starobinsky model, numerical lattice simulations validate the stochastic description and agree well with analytical results. For critical Higgs inflation, we find that the dynamics lead to a minor suppression of the power spectrum with an additional oscillation feature. Throughout, our analysis is restricted to the regime of small metric perturbations, ensuring the self-consistency of the perturbative stochastic treatment. These results establish a concrete bridge between first-principles quantum field theory in curved spacetime and the stochastic-$δN$ formalism for investigating non-perturbative inflationary dynamics.

Quantum kernel methods are among the leading candidates for achieving quantum advantage in supervised learning. A key bottleneck is the cost of inference: evaluating a trained model on new data requires estimating a weighted sum $\sum_{i=1}^N α_i k(x,x_i)$ of $N$ kernel values to additive precision $\varepsilon$, where $α$ is the vector of trained coefficients. The standard approach estimates each term independently via sampling, yielding a query complexity of $O(N\lVertα\rVert_2^2/\varepsilon^2)$. In this work, we identify two independent axes for improvement: (1) How individual kernel values are estimated (sampling versus quantum amplitude estimation), and (2) how the sum is approximated (term-by-term versus via a single observable), and systematically analyze all combinations thereof. The query-optimal combination, encoding the full inference sum as the expectation value of a single observable and applying quantum amplitude estimation, achieves a query complexity of $O(\lVertα\rVert_1/\varepsilon)$, removing the dependence on $N$ from the query count and yielding a quadratic improvement in both $\lVertα\rVert_1$ and $\varepsilon$. We prove a matching lower bound of $Ω(\lVertα\rVert_1/\varepsilon)$, establishing query-optimality of our approach up to logarithmic factors. Beyond query complexity, we also analyze how these improvements translate into gate costs and show that the query-optimal strategy is not always optimal in practice from the perspective of gate complexity. Our results provide both a query-optimal algorithm and a practically optimal choice of strategy depending on hardware capabilities, along with a complete landscape of intermediate methods to guide practitioners. All algorithms require only amplitude estimation as a subroutine and are thus natural candidates for early-fault-tolerant implementations.

We explore the expected performance of a semiconducting spin cQED quantum processor for Multiple Hypothesis Tracking (MHT) algorithm via a quantum annealing procedure. From two different benchmarking scenarios we evaluate this type of quantum annealer on a quantum emulator in which we incorporated both dynamical coherent errors and incoherent errors. From estimate of the reset, measurement and annealing time of the processor, we find that cQED-spin processors could reach a total run time of around 50 ms. This makes this technology promising for potential real time application such as radar tracking.

We investigate the generation of quantum states for precision metrology in noisy two-level systems. These states are obtained by optimizing a variational quantum circuit to maximize the quantum Fisher information (QFI) of the output state for a given decoherence rate and interaction Hamiltonian. The circuit architecture, inspired by twist-and-turn schemes, features a sequence of $n$ entangling layers, each consisting of entangling gates followed by a global rotation. We observe notable improvements in the QFI as the circuit layer depth increases, even for appreciable noise rates, demonstrating that our entangle-rotate architecture expands the accessible state space under realistic noise conditions. Our approach thus provides a general and efficient framework for generating quantum-enhanced sensing states. Our analysis extends to systems of power-law interactions spanning from all-to-all to nearest-neighbor interactions. We also analyze the capabilities of our circuit to prepare states for system sizes greater than $8$ qubits.

We revisit static tidal perturbations of relativistic stars with emphasis on two technical issues in the standard quadrupolar formulation. First, we derive the regular-center Frobenius expansion of the interior even-parity master function and obtain a corrected subleading coefficient, which differs from the expression commonly used in the literature. Second, we derive the static even-parity master equation on a Schwarzschild-de Sitter background, extending the usual asymptotically flat problem to a two-horizon geometry. To place these results on a common footing, we also show how the general interior even-parity system in Regge-Wheeler gauge reduces to the standard quadrupolar equation used in Love-number calculations. Numerical integrations for polytropic equations of state show that the corrected center coefficient affects only subleading initial data and leaves the extracted Love number $k_2$ unchanged within numerical accuracy. Taken together, these results fix the regular-center input to the standard quadrupolar problem and extend the static even-parity formalism to Schwarzschild-de Sitter backgrounds.

We present a solvable same-sector effective theory for anomaly-inspired axion inflation, in which a heavy trace-anomaly mode dynamically backreacts on the axion potential. The tree-level elimination of the radial field resums the backreaction into a closed-form Lambert-$W$ potential, naturally flattening the hilltop potential without external plateau operators. By deriving the exact trough metric, we evaluate all the observables on the fully reduced one-field action, bypassing uncontrolled kinetic approximations. Calibrated at $N_\star=56$, reheating-compatible branches yield $r\simeq0.033$--$0.036$ and $α_s\simeq-(4.6$--$4.7)\times10^{-4}$, comfortably satisfying the current ACT/SPT/BICEP constraints. The evolution remains strictly adiabatic ($m_\perp^2/H^2\gtrsim6.1$, $Ω/H\lesssim7.6\times10^{-4}$) with negligible sound-speed and metric corrections. We provide analytic control over the constant-$w_{\rm eff}$ reheating map, the $N_{\rm re}=0$ boundary, and robustness against vacuum-offset deformations. This Lambert-$W$ backbone establishes a precise, deformable benchmark for confining axion inflation, with microscopic matching and reheating microphysics accessible as systematic EFT refinements.

The Metropolis-Hastings algorithm is a cornerstone of Markov Chain Monte Carlo methods, underpinning a wide range of applications in computational physics, Bayesian inference, and machine learning. Quantum variants of Metropolis-Hastings promise accelerated mixing through quantum walks, but their practical realisation remains challenging. In this work, we construct and simulate an explicit circuit level implementation of a quantum Metropolis-Hastings algorithm based on the framework introduced by Claudon \emph{et al.} (arXiv:2506.11576). We present the full quantum workflow required to prepare a stationary distribution, including a number of modifications required to make the algorithm implementable in a realistic quantum circuit model. Our results demonstrate that these modifications are essential to recover the correct stationary behaviour and highlight both the potential and current limitations of quantum Metropolis-Hastings algorithms, which are expected to become practically relevant in the fault tolerant quantum computing regime.

We propose a coherent-control scheme for engineering quantum correlations in a cavity optomechanical (COM) system consisting of a driven optical cavity with an embedded nonlinear medium and a membrane, assisted by a coherent feedback loop. The nonlinear medium and the membrane are pumped to implement optical and mechanical parametric amplifications with controllable modulation frequencies and pump amplitudes. Through the combined modulation of the two parametric amplifications and the coherent feedback loop, we engineer the effective cavity decay rate and the distribution of quantum fluctuations, thereby strengthening quantum correlations and improving their robustness against thermal noise. Our scheme provides an efficient route to realizing highly tunable, strong, thermally robust quantum correlations in COM systems, which is promising for the protection of fragile quantum resources.

We establish existence and uniqueness results for asymptotically locally Euclidean (ALE) and asymptotically locally flat (ALF) gravitational instantons. In particular, we prove the existence of a unique, Ricci-flat, toric ALE and ALF gravitational instanton, for every admissible rod structure, that is smooth up to possible conical singularites. We also give an elementary proof that any toric ALE or ALF self-dual instanton is a multi-Eguchi-Hanson or multi-Taub-NUT solution.

We present a minimal implementation of SU($N$) pure Yang-Mills theory in $3+1$ dimensions for digital quantum simulation, designed to enable quantum advantage. Building on the orbifold lattice simulation protocol with logarithmic scaling in the local Hilbert-space truncation, we introduce further simplified Hamiltonians. Furthermore, we test simple methods that improve the convergence to the infinite mass limit, thereby removing the requirement of a large scalar mass to obtain the Kogut-Susskind Hamiltonian. For the SU(2) theory, we can cut the resource requirement further by utilizing the embedding of $\mathrm{SU}(2)\cong\mathrm{S}^3$ into $\mathbb{R}^4$. Monte Carlo simulations of the Euclidean path integral were used to benchmark the accuracy of these new analytical improvements to the theory. These results provide further support for the noncompact-variable-based approach as a practical framework for quantum simulation of non-Abelian gauge theories.

We present a theoretical analysis of the WDW approach to quantum cosmology extended to gravity theories with torsion. The dynamics of the FLRW universe is formulated as a classical Hamiltonian problem of point particle mechanics. Unlike in the WDW formalism, the Hamiltonian is not zero, though, and the 3rd quantization does not enforce the cosmic time to vanish. The wave function of the Universe appears as a superposition of eigenfunctions of the quantum Hamiltonian with the cosmic time being the conjugate to its eigenvalues, spatial curvatures. The notion of weak measurement is then introduced to avoid the collapse of the total universal wave function upon measurements of the parameter set describing matter and spacetime. The collapse postulate of the standard Copenhagen quantum theory is discussed and the de Broglie-Bohm interpretation of the effective wave function introduced. The question of the boundary conditions for both, the wave function and the Bohmian guidance equation, is addressed. The corresponding numerical calculations will be published in a separate paper.

Executing a logical quantum circuit fault-tolerantly incurs a large spacetime overhead. Recent work has proposed and investigated phantom codes, defined by the property that every in-block logical $\mathrm{CNOT}$ circuit can be implemented with a physical permutation, a property that has the potential to greatly reduce the depth of compiled circuits. Here we show that phantomness comes at the cost of low encoding rate. Specifically, we prove that any binary phantom code encoding $k$ logical qubits into $n$ physical qubits with distance $d\geq 2$ obeys the bound $k\leq \log_2(n+1)$ for all $k\neq 4$. For $k=4$ we explicitly construct a nonstabiliser $(\!(8, 2^4, 2)\!)$ phantom code that violates the bound and has a transversal non-Clifford gate. We further show that, within the class of nontrivial CSS phantom codes with $k\neq 4$, there is a unique family of codes saturating this bound. In addition, we prove that this logarithmic ceiling cannot be circumvented by permitting additional local unitary gates, or by making use of subsystem codes: any subspace or subsystem code admitting a $\mathrm{SWAP}$-transversal implementation of every logical $\mathrm{CNOT}$ circuit is constrained to satisfy the same bound. These bounds follow from a general theorem relating the length of a quantum code to the structure of its automorphism group, a result which may find applications beyond phantom codes.

We present a systematic study of static solutions to the source-free SU(2) Yang-Mills equations, in which the gauge potential explicitly depends on spin operators. By employing the \emph{vector potential extraction approach} (VPEA) -- which requires the total angular momentum operator (orbital plus spin) to satisfy the standard angular momentum algebra -- we derive the most general form of the spin vector potential. This leads to the static ansatz $\{ \vec{A} = [k_1(\hat{r}\times\vecΓ) + k_2\vecΓ + k_3(\vecΓ\cdot\hat{r})\hat{r}]/r, \varphi = f_1(r)\,(\vecΓ\cdot\hat{r}) + f_2(r)\}$, parametrized by three constants $\{k_1, k_2, k_3\}$ and two radial functions $\{f_1(r), f_2(r)\}$. Substituting this ansatz into the Yang-Mills equations and imposing the angular momentum constraints from the VPEA yields a set of consistency equations. Solving these equations provides a complete classification of static solutions, including both real and complex families. Known simple SU(2) static solutions are recovered as special cases. Our classification reveals new static configurations that could be valuable for non-perturbative studies and for models where spin degrees of freedom couple to non-Abelian gauge fields.

We present a framework based on the determinantal geometry of two-qubit gates. Combining the Weyl chamber representation with operator Schmidt theory, we interpret gate synthesis as a distance problem to determinantal varieties. This gives an operational geometry to the Weyl chamber, quantifying nonlocal complexity. We show that the square root iSWAP gate is the closest perfect entangler to the variety of local operations, and that no perfect entangler can be approximated by a local gate with average gate fidelity above 79.8%. The three different determinantal costs form a synthesis-adapted coordinate system that encodes nonlocal complexity and generally reconstructs the Weyl chamber.

Toward the large-scale, practical realization of quantum computing, quantum error correction is essential. Among various quantum error-correcting codes, the surface code stands out as a leading candidate, and lattice surgery based on surface codes has emerged as a promising technique for fault-tolerant quantum computation (FTQC). However, implementing quantum algorithms using lattice surgery introduces both resource and time overhead. Existing approaches typically focus on large layout designs, with compiler passes aimed primarily at optimizing time overhead. This often overlooks the trade-off between rotation bottlenecks and movement distance, which leads to inefficient resource utilization and prevents further reduction of the quantum computation failure rate. To address these challenges, we introduce O3LS, a framework for optimizing lattice surgery through automatic layout search and loose scheduling. O3LS achieves an optimal balance by automatically generating squeezed data layouts to reduce space requirements and employing loose scheduling algorithms combined with circuit synthesis techniques to reduce time overhead, thereby effectively minimizing overall logical error rates. Numerical results indicate that O3LS can reduce space overhead by 28.0% over standard layouts and 46.7% over sparse layouts without increasing the number of time steps, leading to suppression of logical error rates by up to 16% relative to larger data layout designs. O3LS can also achieve time overhead reductions of 36.07% and 24.76% in compact and standard data layout designs, respectively. It suppresses logical error rates by up to an order of magnitude compared to prior compilers that focus primarily on maximizing parallelism.

To address the urgent need in the NISQ era for high-performance, scalable quantum compilers and to advance the integration of classical and quantum computing, we present QLLVM, an advanced Quantum-Classical co-compilation framework built on LLVM. To our knowledge, QLLVM delivers an end-to-end, LLVM-based compilation workflow that unifies the build of classical high-performance programs, including CUDA, MPI, and C++, together with quantum programs into a single executable. For quantum program compilation, QLLVM adopts a three-stage design: high-level optimizations are implemented in the MLIR Quantum dialect and then lowered to QIR, an LLVM IR-based representation, for low-level optimization and hardware mapping. Its extensible architecture and seamless interoperability with classical high-performance computing provide an efficient, flexible, industrial-grade compilation infrastructure for future quantum software development. Experimental results show that, on the MQTBench benchmark suite, QLLVM reduces circuit depth and gate counts compared with state-of-the-art compilers and demonstrates clear advantages in compiling hybrid classical-quantum programs.

We report a hardware validation of the DAGI (Directed Acyclic Graph Information) framework on IBM Quantum hardware using a small, controlled experiment whose ideal output distribution is constrained to a low-dimensional modular manifold (a "ridge"). For two $n$-bit registers $(u,v)$ with $n=4$ (modulus 16), each key instance $k$ induces an ideal relation $v \equiv k \cdot u \pmod{16}$, producing a visually distinct ridge in the joint $(u, v)$ distribution. Executed on ibm\_torino in a single Sampler V2 job (8 keys, 1024 shots/key, $N=8192$ total shots), the ridge persists under hardware noise with ridge-hit probability $p_{hit} = 0.1830$ (uniform baseline $1/16$), corresponding to a ridge contrast of $2.93\times$ (95\% bootstrap CI [2.80, 3.06]). Key recovery exceeds chance: per-shot accuracy 0.1689 (chance 0.125, 95\% Wilson CI [0.1610, 0.1772]), and per-group dictionary recovery 0.375 (chance 0.125). To test the central DAGI hypothesis -- that recoverable key information is predominantly high-order/synergistic rather than visible in low-order marginals -- we compute a Möbius-based information decomposition of $I(K;D_S)$ over detector-bit subsets $S$ via a Möbius inversion pipeline and evaluate targeted positive synergy $CPS_K$ at order $k_{max}=3$. We observe $CPS_K(k=3) = 0.08788$ with significance under label-shuffle permutation tests (accuracy $p=0.001996$, $CPS_K$ $p=0.004975$). Uniformity diagnostics show near-uniform single-bit marginals while correlation concentrates in specific low-order pairs, and a bootstrap reliability sweep confirms order-3 targeted synergy remains statistically reliable at the full 1024-shot target budget. These results support the claim that DAGI detects and quantifies nontrivial, hardware-resilient, higher-order information structure associated with a known global algebraic constraint.

Hybrid Quantum Neural Networks (HQNNs) combine classical learning with parameterized quantum circuits, but their practical performance is often limited by (i) the noise of Noisy Intermediate-Scale Quantum (NISQ) devices and (ii) the large, discrete design space of quantum circuit architectures. Moreover, HQNNs are commonly trained using a fixed circuit and a single backend, even though deployment frequently targets heterogeneous backends where compilation and execution characteristics may differ. To address these challenges, we propose GAT-QNN, a genetic algorithm (GA)-based framework that trains a macroCircuit (search space) by iteratively sampling microCircuits (subcircuits), training them, and reintegrating their learned parameters into the macroCircuit. After training, we run an independent GA-driven inference stage that evaluates candidate microCircuits using the trained macroCircuit weights and selects top-performing architectures for deployment. This two-stage approach enables backend-aware microCircuit selection without retraining each candidate architecture and can also reduce computational resources (gate count) by deploying smaller microCircuits derived from the macroCircuit. We validate the approach on MNIST classification (four classes) and report consistent 22-23% test accuracy gains for GA-driven inference across multiple backends.

The classical simulation of universal quantum circuits is crucial both fundamentally and practically for quantum computation. We propose SyQMA, a simulator with several convenient features, particularly suited for quantum error correction (QEC). SyQMA simulates universal quantum circuits with incoherent Pauli noise and computes exact expectation values and measurement probabilities as symbolic functions of circuit parameters: rotation angles, measurement outcomes, and noise rates. This simulator can sample measurement outcomes, enabling the simulation of dynamic quantum programs where circuit composition depends on prior measurement outputs. For QEC, it performs circuit-level maximum-likelihood decoding, provides exact symbolic expressions for logical error rates, and verifies the fault distance of fault-tolerant (FT) stabiliser and magic state preparation protocols. These features are enabled by an intuitive extension of stabiliser simulators, where each non-Clifford Pauli rotation and incoherent Pauli channel is compactly represented via auxiliary qubits and a modified trace. Representing the state requires only polynomial memory and time, while computing expectation values and measurement probabilities takes exponential time in the number of non-Clifford rotations and deterministic measurements, but only polynomial memory. The FT preparation of stabiliser and magic states, including the first stage of magic state cultivation, is analysed without approximations. We also exactly convert the disjoint error probabilities of a general multi-qubit Pauli channel to independent ones, a key step for creating and sampling from detector error models. The code is publicly available and open-source.

We present a novel lackadaisical alternating quantum walk (LAQW) algorithm whose circuit depth scales as $\mathcal{O}(n^2+nt)$ for a $n\times n$ lattice over $t$ time steps. We show that this is a significant depth reduction compared to the existing controlled alternating quantum walk (CAQW) model, which has a circuit depth that scales as $\mathcal{O}(n^2t)$ (Li et al., 2017, arXiv:1707.07389). This makes the implementation of the LAQW viable for Noisy Intermediate-scale Quantum (NISQ) devices. We then showcase the applicability of the LAQW algorithm by proposing a chaos-based symmetric-key generation scheme. Our approach uses the LAQW as a quantum entropy source from which reproducible random bitstring sequences are generated using the underlying probability distribution and subsequent post-processing methods. We provide a comprehensive evaluation of the LAQW algorithm and demonstrate the reproducibility of 128-bit keys under simulated quantum noise provided by IBM's FakeTorino backend. A direct comparison with the CAQW model, which has been used in image encryption and hash function schemes (Li et al., 2017, arXiv:1707.07389; Abd EL-Latif et al., 2020, ScienceDirect; Abd El-Latif, Abd El-Atty, and Venegas-Andraca, 2020, ScienceDirect), highlights the potential and usefulness of the LAQW model in cryptographic applications.

The certification of quantum systems is essential for emerging quantum technologies, particularly in quantum communication, networks, and distributed computing, where maintaining a common reference frame across distant nodes poses significant challenges. Reference frame independent approaches, such as randomized measurement schemes, offer a promising route by reducing experimental demands while granting access to basis-independent quantities, including entanglement. However, the efficiency of such schemes in measuring such local invariants has remained unclear. In this work, we determine the minimal number of measurement settings required to access all two-qubit invariants. We further demonstrate that entanglement certification necessarily involves the most demanding invariants, establishing it as a maximally difficult task. Our results reveal a fundamental hierarchy among invariants, with direct implications for experimental feasibility and theoretical understanding of quantum certification. Finally, we extend our analysis beyond bipartite systems by applying it to the Kempe invariant in three-qubit systems, improving known measurement protocols and providing a first step toward uncovering similar hierarchies in higher dimensions.

Where is the true boundary of the quantum advantage region of decoded quantum interferometry (DQI)? The best existing answer is provided by Theorem 10.1 of Jordan et al., yet we show that this answer systematically underestimates the extent of quantum advantage. On the standard partial-win LDPC benchmark instance, there exist 26 consecutive parameter points (l in [642, 667]) at which Jordan's analysis declares no quantum advantage (<s>/m < 0.5), while quantum advantage is in fact present with an approximation ratio reaching 0.66. The root cause is that Jordan's bound penalizes the entire system with the worst-case Hamming-layer decoding failure rate epsilon = max_k epsilon_k, discarding the spectral structure of the DQI tridiagonal matrix. Exploiting the concentration of the Perron eigenvector, we replace the uniform penalty with the weighted average epsilon_bar = sum_k epsilon_k w_k^2 and establish a unified lower bound (Master Theorem) valid over arbitrary finite fields F_q, proving that it strictly improves upon the original bound from three independent sources.

Partial quantum error correction and quantum error mitigation are expected to coexist in the pre-fault-tolerant regime, yet the resource advantage of combining them remains insufficiently quantified. We study zero-noise extrapolation constructed from mixed datasets that contain a small number of error-corrected data points together with data obtained without error correction. The low-noise logical points anchor the extrapolation, while the higher-noise physical points enlarge the noise baseline at a much smaller runtime cost. Under a simple model in which error correction suppresses the effective gate error rate from p to $γ$p, we derive the variance of the zero-noise estimator and compare the physical runtime required to reach a target precision. For Richardson extrapolation, the mixed-data strategy reduces variance amplification and can lower the required physical runtime by several orders of magnitude when $γ\leq 0.1$. As a proof of principle, we apply the method to digital quantum simulation of a six-spin transverse-field Ising model and find that mixed physical/logical datasets yield lower-variance zero-noise estimates and outperform extrapolation based only on error-corrected data in the parameter regime studied here. These results identify hybrid error correction and error mitigation as a practical route to resource-efficient quantum computation before full fault tolerance.

We construct an analytic geodesic-optics description of quasinormal ringing, black-hole shadows, strong lensing, and grey-body factors for the static spherical metric introduced in Eq.~(9) of Ref.~\cite{BakopoulosEtAl2024}. Working in a weak-hair regime for the coupling combination $β\equivηq^4$, we derive closed first-order formulas for the photon-sphere radius, orbital frequency $Ω_{\text{ph}}$, and Lyapunov exponent $λ_{\text{ph}}$. These invariants are then employed within the Schutz--Will WKB approach to obtain eikonal quasinormal frequencies, mapped to shadow and strong-deflection observables through exact identities for static spherical geometries, and used to build a closed analytic form for the transmission probability $Γ_\ell(ω)$. At leading eikonal order, these relations are controlled by null geodesics and are therefore spin-universal for test scalar/electromagnetic/gravitational sectors, up to subleading corrections. Besides the standard ringdown--shadow correspondence, we present three additional results: (i) an explicit quality-factor correction, (ii) limiting core-size expansions that show when damping ratios are nearly insensitive to the scalarized core, and (iii) a comparative study of grey-body factors for moderate multipoles ($\ell=3,4$) and several core-size ratios. The resulting construction provides a concise one-parameter connection from the metric function to ringdown, lensing, and scattering observables.

Mechanically isolated defect centers in hexagonal boron nitride are promising coherent quantum emitters, yet spectral instabilities persist, and their spin-related nature remains unclear. Here we investigate a single mechanically isolated quantum emitter in hBN integrated onto a coplanar waveguide. The emitter exhibits exceptionally bright resonant fluorescence with saturation count rates exceeding $10\,\mathrm{Mc/s}$. High-resolution spectroscopy reveals two closely spaced zero-phonon-line transitions originating from the same defect complex. Time-resolved spectroscopy shows that these transitions exhibit markedly different spectral diffusion dynamics, consistent with distinct donor-acceptor-pair-like recombination pathways with different sensitivities to local electrostatic fluctuations. Off-resonant blue illumination redistributes emission between the two transitions and increases the emission duty cycle without significantly modifying the dominant spectral diffusion rates at low temperature, indicating repumping from long-lived shelving states. Magnetic-field-dependent photoluminescence, optically detected magnetic resonance, and pump-probe measurements reveal millisecond-scale relaxation dynamics and magnetic-field-dependent fluorescence contrast, demonstrating spin-dependent population dynamics in the metastable shelving state. These results clarify how charge-driven spectral fluctuations and spin-dependent shelving jointly shape the optical cycling dynamics.

We study the time it takes for all states of a finite quantum system to return simultaneously to their original configuration. In particular, we define the recurrence time for a quantum system to be the time at which all time-evolved states are close to their initial configuration, and at least one state has deviated significantly during this interval. Considering finite-dimensional quantum systems evolving unitarily, we find bounds on this notion of recurrence time, for continuous time and discrete time, by using Dirichlet's approximation theorem. We show how the problem of finding a bound on recurrence time can be related to approximating the difference of real numbers by rationals. We present a mathematical result on the latter, which we then use to obtain tighter bounds on recurrence time.

Light's intrinsic carrier frequency of hundreds of terahertz theoretically enables information processing at terahertz clock rates. In optical quantum computing, continuous-variable quantum teleportation is the fundamental building block for deterministic logic operations. This protocol transfers unknown quantum states between nodes using quantum entanglement and real-time feedforward of measurement outcomes. However, electrical feedforward bottlenecks currently restrict operational bandwidths to approximately 100 megahertz, preventing the exploitation of light's ultimate speed. Here we show 1-terahertz-bandwidth all-optical quantum teleportation, completely bypassing this electronic limitation. By transferring Bell measurement outcomes optically, we successfully teleported vacuum states across the terahertz band and real-time random coherent wavepackets with a 42-picosecond temporal width. Evaluating the intrinsic state transfer quality, we achieved teleportation fidelities of $\mathcal{F}=0.784$ for the broadband vacuum states and $\mathcal{F}=0.770$ for the dynamic coherent wavepackets. Both results strictly surpass the classical limit of $\mathcal{F}=0.5$, demonstrating genuine quantum teleportation at ultrafast speeds. Our results establish that optical quantum processing speeds are constrained solely by the nonlinear medium's 1-picosecond-scale response, rather than classical electrical interfaces. This methodology provides a cornerstone for terahertz-clock quantum computers capable of overcoming Moore's law, and paves the way for a high-capacity, telecom-compatible quantum internet.

As quantum computing (QC) technologies mature, their integration into established high-performance computing (HPC) infrastructures is becoming a central objective for next-generation computing systems. However, unlocking the potential of hybrid platforms for computationally demanding workloads remains challenging. The mismatch between quantum and classical programming models, the limited maturity of quantum software stacks, and the scarcity of quantum processing units (QPUs) above all, necessitate scheduling strategies that go beyond standard HPC mechanisms to manage such heterogeneous and constrained resources. To address this issue, we investigate three distinct methodologies for HPC-QC resource scheduling: time-based multiplexing, dynamic resource management, and workflow decomposition. Experimental validation on production HPC clusters and real quantum hardware demonstrates the effectiveness of these approaches under different workload scenarios. Malleability and workflow strategies significantly optimize classical resource utilization, reducing consumption by up to 45.7% and 64% respectively, proving to be best fitted for hybrid jobs where quantum and classical workloads are evenly balanced. Conversely, time-multiplexing enhances QPU utilization and reduces execution time at the cluster level, making it the optimal strategy for the opposite context, which is characterized by high classical-quantum workload imbalances. These findings underscore the practical viability of tailored scheduling strategies for hybrid HPC-QC environments and highlight their complementarity in building efficient, scalable software stacks for next-generation quantum-accelerated facilities.

Levitated mesoscopic particles hold the promise of revolutionizing gravity sensing by using quantum effects. However, conventional quantum gravimeters based on such systems fail to harness the intrinsic large-mass advantage of the particles, because their commonly utilized auxiliary quantum systems counteract the role of mass as a resource. To overcome this limitation, we propose a quantum gravimetry by directly using the mechanical qubit (QM) formed by a levitated particle as the gravity sensor. Without resorting to the auxiliary quantum system, our scheme enables a straightforward readout of the particle's motion under gravitational influence. The obtained sensitivity behaves as a $m^{-1/2}$-scaling with the mass $m$. We also generalize our scheme to the \textit{mechanical cat qubit} as the gravity sensor. The sensitivity further scales as $N^{-1/2}$ with the mean phonon number $N$. In the experimentally realizable parameter regime, a sensitivity on the order of $0.1~ \text{\textmu}\text{Gal}/\sqrt{\text{Hz}}$ can be achieved, which outperforms the traditional schemes by two orders of magnitude. Reaching the \textit{double standard quantum limits} with $m$ and $N$ simultaneously, our scheme provides a feasible route toward compact high-sensitivity quantum gravimetry.

Floquet dynamical quantum phase transitions (FDQPTs) reveal many nonequilibrium critical phenomena in periodically driven quantum systems, and their underlying mechanisms have attracted deep attention in recent years. In this paper, we consider an extended XY spin chain under a periodic flux-quench protocol, and demonstrate the effect of the flux difference within each micromotion period on the emergence of FDQPTs, by analyzing physical quantities such as the Loschmidt echo, rate function, and dynamical topological order parameter (DTOP), etc. We also generalize the concept of quench fidelity to periodically driven systems, i.e., Floquet quench fidelity, and discuss the necessary and sufficient conditions for FDQPTs. In contrast to conventional single-quench scenarios, the occurrence of FDQPTs is determined by the requirement of Floquet fidelity condition and segment duration. Our framework may be applied generally to arbitrary periodically driven parameters, providing fundamental insights into how periodic protocols control nonequilibrium phase transitions in quantum many-body systems.

In a quantum LiDAR system, to achieve a better resolution and sensitivity, detection scheme plays an important role. We propose a novel detection scheme in which the photo detector considers only the $4n$ number of photons, where $n \in \mathbb{N}$, as a click and the rest of them as a no-click. Similar to the $Z$-detection scheme, where we get a click for any number of photons, we termed this measurement as $Z_{4n}$-detection scheme. By employing superposition of four coherent states (SFCS) and vacuum as input we investigate the performance of Mach-Zehnder interferometer (MZI) based quantum LiDAR systems. We found a significant enhancement in resolution and broader working point for the phase sensitivity in comparison to the $Z$-detection scheme. Our findings highlight the advantages of our approach and suggest promising advancements in the field of quantum LiDAR sensing technology, providing a pathway for more accurate and sensitive measurement capabilities.

We analyse the effect of expansion in Newtonian cosmology on the asymptotic behaviour of charged self-interacting plasmas close to Poisson equilibria. To this end, we study the Vlasov-Poisson system on the phase space of a $3$-torus which is expanding with respect to the scale factor $a(t)$. We show that, for $a(t)=t^q$ with $q\in(0,\frac12)$, solutions to this system exhibit nonlinear Landau damping for initial data that is small with respect to a suitably strong Gevrey class, i.e., the charge density contrast of the plasma decays superpolynomially. For larger choices of $q$ within this range, the initial data requirements become stricter while the decay weakens. To our knowledge, this is the first result showing Landau damping in a cosmological setting.

We investigate the non-monotonic temperature sensitivity of a coherently driven two-level quantum system coupled to an Ohmic phonon environment. By employing a unitary polaron transformation, we account for phonon-induced renormalization effects that go beyond the standard weak-coupling approximations. Our analysis reveals that the Quantum Fisher Information (QFI) exhibits a prominent peak at an intermediate system-environment coupling strength, identifying an optimal regime for thermal sensing. This behavior emerges from a fundamental competition between environment-induced dissipation enhancement and the exponential suppression of system parameters due to phonon dressing. We demonstrate that while thermometric precision vanishes in both the ultra-weak and strong coupling limits, a properly tuned nonequilibrium steady state can significantly enhance sensitivity. These results suggest that environmental interactions, often viewed as detrimental decoherence sources, can be engineered as a resource to optimize the performance of solid-state quantum thermometers.

Weyl conformal gravity was originally proposed in the early twentieth century as an attempt to unify gravitation and electromagnetism. Since 1989, renewed interest in this fourth-order theory of gravity has emerged following the discovery of several exact black hole solutions. In this work, we investigate the timelike circular geodesics of a spherically symmetric Weyl black hole. The effective potential, the circular geodesics and their Jacobi and Lyapunov stability are discussed. Our analysis provides new insights into the stability properties of Weyl black holes and the role of the free parameters appearing in their solutions.

We analytically study the effect of backreaction from analog Hawking radiation on its entanglement entropy in the Bose-Einstein condensate (BEC). The backreaction is expected to play an essential role in the decrease of the entanglement entropy and in realizing the Page curve. Since the BEC theory has microscopic Hamiltonian and thus exhibits unitarity, it is desirable to reproduce the Page curve explicitly by using the Hamiltonian. In order to analyze this in a concrete example, we study the BEC with a step-like configuration that has been extensively studied in the literature. By using the microscopic theory, we derive an explicit form of backreaction from analog Hawking radiation. Combining it with the known results of the Bogoliubov coefficients, we analytically compute the entanglement entropy of the Hawking radiation, and show that it decreases as expected due to the backreaction for sufficiently low energy modes over a wide range of the parameter characterizing the step-like configuration.

We present a time-dependent extension of logarithmic perturbation theory for nonrelativistic quantum dynamics governed by the Schrödinger equation, in which the logarithm of the wave function is expanded in powers of a coupling constant. The resulting hierarchy of equations defining the perturbative corrections is governed by a gauge-rotated Hamiltonian of the unperturbed system and leads to closed-integral expressions for the time-dependent corrections based on Duhamel's formula. This closed-integral structure of corrections is a hallmark of time-independent logarithmic perturbation theory and is preserved in the present extension. This structure, in particular, provides a computable expression for the instantaneous energy shift. Furthermore, dynamic energy shifts arise naturally within this framework in the form of time-averaged expectation values of pseudopotentials and can be related, for example, to AC Stark shifts and electric polarizabilities. As an illustration, we apply the method to the harmonic oscillator and the hydrogen atom, both driven by a time-dependent laser field. The harmonic oscillator provides a proof of principle for which the exact solution is recovered, while the hydrogen atom illustrates the method applied to atomic systems. Supported by numerical simulations, we demonstrate the applicability to obtain relevant physical observables with high accuracy. The present approach offers a promising alternative for analytical studies of time-dependent multi-photon processes in the perturbative regime.

Quantum coherence is a fundamental issue in quantum mechanics and quantum information processing. We explore the coherence dynamics of the evolved states in HHL quantum algorithm for solving the linear system of equation $A\overrightarrow{x}=\overrightarrow{b}$. By using the Tsallis relative $α$ entropy of coherence and the $l_{1,p}$ norm of coherence, we show that the operator coherence of the phase estimation $P$ relies on the coefficients $β_{i}$ obtained by decomposing $|b\rangle$ in the eigenbasis of $A$. We prove that the operator coherence of the inverse phase estimation $\widetilde{P}$ relies on the coefficients $β_{i}$, eigenvalues of $A$ and the success probability $P_{s}$, and it decreases with the increase of the probability when $α\in(1,2]$. Moreover, the variations of coherence deplete with the increase of the success probability and rely on the eigenvalues of $A$ as well as the success probability.

We demonstrate that the Ising all-to-all (ATA) model exhibits a range of dynamics, from integrable to chaotic, including mixed behaviour across symmetry blocks within a single system. While other works have explored the dynamics of all-to-all systems by varying parameters, we analyse a fixed set of parameters and examine the dynamics within different blocks. In addition to investigating the dynamical properties, we show that the system remains resilient to noise when the norm of the Hamiltonian representing the noise is close to 1. Our results are presented by mapping each symmetry sector of the system to a kicked top (KT) and observing that KT parameters for each sector depend on its dimension. This system, similar to the Bunimovich billiard for classical chaos, provides a new platform for studying dynamics determined by the symmetry sector, advancing quantum chaos research.

The plethora of possible ground states of spinor bosons placed in an external lattice and a cavity is revisited. We discuss the simplest case when the external lattice nodes coincide with the antinodes of the cavity field. We analyze the problem within the grand-canonical mean-field approach, considering both the homogeneous system and the nonhomogeneous case with a harmonic trapping potential. Due to the spin degree of freedom, in the homogeneous case we treat the system in a twofold manner: we impose the physically relevant total-magnetization constraint, while also discussing the minimization landscape for the full unconstrained problem. In the latter, by combining analytical arguments with numerical calculations based on the Gutzwiller ansatz, we show that the system exhibits two types of magnetic phases: an antiferromagnetic Mott insulator (AFM) and a ferromagnetic density wave (FDW). In addition, three distinct supersolid phases emerge, characterized by different patterns of spin and density imbalances. In case of the zero total magnetization, only two of the three supersolid regimes survive, and the FDW phases are replaced by entangled density waves (EDW). These new ground states present density-modulated superpositions of the underlying spin components of the bosons. Finally, we present the phase diagram of the trapped system, which is directly relevant for future experiments.

We introduce a general witness-based framework for quantifying and operationally activating continuous-variable (CV) resources into discrete-variable (DV) bipartite entanglement or Einstein- Podolsky-Rosen (EPR) steering. For the three standard CV resource theories associated with Wigner negativity (WN), genuine non-Gaussianity (GNG), and standard non-Gaussianity (SNG), we define infinite families of bounded-witness monotones indexed by box constraints on the witness operators. For closed convex free sets, these monotones are faithful, strongly monotonic under free instruments, Lipschitz continuous, and convex. For closed nonconvex free sets, we show that faithfulness requires a two-copy lift and formulate the corresponding strong-monotonicity statement in the lifted theory. We further construct witness-dependent completely positive trace-preserving (CPTP) measure-and-prepare channels whose outputs are two-qubit Werner states. For the representative case n = m = 1, the optimal entanglement and EPR steering attainable within this witness-dependent activation family are exactly proportional to the underlying monotones. We illustrate the framework with odd-parity states, pure-loss single-photon states, and Gottesman- Kitaev-Preskill (GKP) states, and derive explicit lower bounds for pure-state GNG and SNG. More broadly, our results show that closed CV free sets admit witness-based quantifiers with a direct operational interpretation in terms of experimentally accessible DV correlations.

The quantum Mpemba effect (QMpE) describes an anomalous thermalization phenomenon in which quantum states initially far from equilibrium can approach thermal equilibrium faster than states that begin closer to it. While this effect has been extensively studied in various frameworks, its practical implications for quantum information processing remain largely unexplored. We investigate the relationship between QMpE and quantum thermometry, focusing on non-equilibrium scenarios where measurements are performed during early-stage thermalization. In a Markovian model, we rigorously prove that the initial states that are optimal for thermometry exhibit QMpE with high probability and thermalize faster than most initial states. Our results reveal a fundamental connection between quantum thermodynamics and thermometry, suggesting that QMpE can be harnessed to enhance temperature estimation with quantum probes.

The Casimir-Polder (CP) effect -- the force between a neutral atom and an uncharged conducting plate in empty space -- is an intriguing consequence of quantum vacuum fluctuations. The typically attractive CP potential crosses over from a scaling of $z^{-3}$ at short separations to $z^{-4}$ at long distances, where retardation effects due to the finite speed of light become important. At intermediate distances, where the atom--surface separation is of the order of the wavelength of the dominant atomic transition, experiments have so far relied on indirect methods, such as diffraction or quantum reflection, to observe the CP effect. Here, we directly reveal the CP force between strontium atoms and a dielectric surface via the induced shifts in the atomic energy levels in the intermediate regime. We spectroscopically probe the CP-induced kHz-frequency shift of ultracold atoms confined by a magic-wavelength optical lattice at 189(2)~nm from the surface -- on the scale of the dominant 461-nm transition. Our measurements agree well with QED calculations and differ from the short-range approximation, while excluding the long-distance one. This paves the way for studying the CP effect across various surface properties and geometries, as well as exploring the tensor nature of the atom-surface potential -- all important for the development of hybrid atomic optical-magnetic quantum devices.

Photonic state engineering in waveguide QED is typically based on local light-matter interactions. This limits its control over the spatial structure of bound photonic states. Here, we demonstrate a distinct mechanism arising from the interplay between nonlocal giant-atom coupling and topological band structure. Specifically, we consider giant atoms coupled to a Su-Schrieffer-Heeger waveguide and show that this configuration enables a controllable energy-level crossing protected by the topological gap. Adiabatically sweeping the atomic detuning across the crossing leads to a controlled exchange between distinct photonic bound states. In a two-giant-atom configuration, this mechanism achieves high-fidelity conversion of a spatially splitting state into a combining state. Extending this scheme to three-giant atoms, we further realize robust, shape-preserving photon transfer mediated by sequential in-gap crossings. Our results demonstrate how topology and nonlocal light-matter coupling can be combined to achieve programmable control of bound photonic states in waveguide QED platforms.

Recent results from the Atacama Cosmology Telescope (ACT), indicating a higher and more tightly constrained scalar spectral index, $n_s = 0.9743 \pm 0.0034$, place several inflationary models under tension, with quintessential inflation pushed close to or beyond the $2σ$ boundary in the $r$--$n_s$ plane. In this work, we revisit quintessential inflation within the framework of Einstein--Gauss--Bonnet (EGB) gravity, where a scalar field non-minimally coupled to the Gauss--Bonnet invariant modifies the inflationary dynamics. We consider three representative coupling functions -- exponential, hyperbolic secant, and hyperbolic tangent -- and show that the exponential and sech-type couplings can shift the predicted values of $r$ and $n_s$ into the $1σ$ region allowed by ACT, thereby restoring consistency with observations. In contrast, the tanh-type coupling remains disfavored, underscoring the sensitivity of inflationary observables to the coupling structure. We further investigate the reheating phase using a model-independent parametrization and demonstrate that viable thermal histories can be realized even in the absence of a potential minimum, with reheating temperatures consistent with Big Bang nucleosynthesis bounds. Overall, our analysis shows that EGB corrections provide a viable and robust extension that reconciles quintessential inflation with current precision cosmological data, and we identify the corresponding allowed parameter space.

We investigate an analytical framework for reconstructing bulk geometries from pole-skipping data. Previously, this method enabled the recursive recovery of near-horizon metric derivatives in static, planar-symmetric black holes. Building on this framework, we systematically extend it to more intricate geometries, specifically static topological black holes and rotating black holes. For three-dimensional rotating black holes, we demonstrate that the metric can be fully reconstructed from boundary pole-skipping data. For four-dimensional rotating spacetimes admitting a separable coordinate system (such as the Kerr family), standard near-horizon pole-skipping successfully reconstructs the purely radial metric functions. To recover the remaining angular metric functions, we introduce a mathematical counterpart termed "angular pole-skipping," defined via a near-axis analysis. Although its precise holographic dictionary remains an open question, this bulk-side formalism completes the geometric reconstruction algorithm. Furthermore, we demonstrate that the vacuum Einstein equations can be recast as a set of algebraic equations governing the pole-skipping data and that the null energy condition imposes algebraic inequalities on this boundary data. Finally, we establish general polynomial constraints dictated by the overdetermined nature of the metric reconstruction, highlighting the highly redundant encoding of bulk geometry in boundary data.

This study maps the quantum landscape of superconducting diodes (SDs) \cite{nadeem23} onto the quantum technology architecture, which is currently constrained by fundamental challenges in control and scalability. In the existing non-integrated quantum technology hardware, control and scalability related issues emerge at two fronts: First, nonlinear and nonreciprocal circuit elements, which are essential building blocks for quantum processors, are often complex, bulky, and dissipative. Second, the temperature gradient between classical control electronics ($T_C\gtrsim$ K), which is also dissipative, and the quantum processor at cryogenic temperatures ($T_Q\sim$ mK) makes scalability even more challenging. The main focus is to reveal how the built-in nonlinearity, nonreciprocity, and quantum functionalities of SDs are significant for on-chip integrated circuit quantum electrodynamics (c-QED), enabling scalable integration of noise-resilient qubit and qubit-interfaces for efficient power delivery, coherent control and memory, high-fidelity readout, and quantum-limited amplification. To this end, this study will also shed light on how thermodynamic constraints and field effects can be harnessed within a quantum-enhanced SD platform, thereby enabling thermal compatibility between classical and quantum workflows, isothermal all-electrical control, and on-chip scalability. This perspective is expected to play a pivotal role in the advancement of superconducting circuit-based quantum hardware with temperature-matched classical, quantum, and hybrid workflows.

We establish a sharp logarithmic trace inequality that strengthens recent bounds of Cheng et al.(arXiv:2507.07961) by replacing their prefactor $c_s/s$ with the strictly smaller constant $G_s$. The constant $G_s$ is defined via the scalar inequality $\log(1+r)\le G_s r^s$ and admits a closed-form expression in terms of the Lambert $W$ function. Our approach introduces an iterative integration-by-parts procedure that lifts optimal scalar bounds to the operator level without loss. We prove that $G_s$ is the optimal universal constant, in the sense that no smaller constant satisfies the inequality for all positive operators. For density matrices, this optimality persists up to $s\le \frac{1}{2\log(2)}$, while beyond this threshold the commuting case exhibits a strictly smaller optimal constant and the noncommuting case remains open. In the regime $s\to0$, our result improves the prefactor $c_s/s$ of Cheng et al. by a factor of $1/e$. These sharper inequalities enhance key primitives in quantum information theory, including decoupling, convex-splitting, and covering lemmas, leading to tighter finite-resource bounds.

Cavity quantum electrodynamics (QED) with quantum emitters coupled to resonators provides a powerful platform for engineering light-matter interactions and exploring collective phenomena. In particular, superradiance, arising from collective quantum interference among emitters, has been explored as a route to ultrastable continuous radiation. However, engineering superradiance in the time domain to realize periodic pulsed sources or frequency combs remains largely unexplored. Here, we investigate the non-equilibrium many-body dynamics of a driven spin ensemble coupled to an on-chip superconducting resonator and uncover a dynamical phase transition from continuous-wave to periodic pulsed superradiant emission. To quantitatively capture the observed dynamical phases, we introduce a driven-dissipative cavity-QED model that elucidates how the periodic pulsed superradiant phase emerges from collective, periodically repeating spin dynamics stabilized by the interplay of coherence growth, disorder, and dissipation. We also find that rare-earth ion spin systems exhibiting both optical and microwave transitions enable phase-synchronized, dual-rail superradiant frequency combs in both the microwave and optical domains. Our results not only open new avenues for dual-rail frequency-comb applications in quantum metrology and information processing, but also establish a fundamental connection between periodic pulsed superradiance and the emergence of a continuous time crystal as a novel nonequilibrium phase in driven open systems.

We develop integration-by-parts (IBP) reduction and differential equations for massive loop integrals of cosmological correlators in de Sitter (dS) spacetime, demonstrating the feasibility of this approach. We identify a structural property of the dS IBP system: for an \(n\)-propagator family, it splits into \(2^n\) closed subsystems classified by the parity of the propagator indices. We further formulate a Baikov representation for loop integrals in dS space and derive the corresponding dimensional recurrence relations. In flat spacetime, intersection theory shows that \(\di\log\)-form master integrands lead to \(\di\log\)-form differential equations. Motivated by fibration intersection theory, we conjecture that this construction extends to dS integrands involving Hankel functions. We verify this conjecture in the one-loop bubble family and determine the associated alphabet.

We derive the density of states along with the $2$- and $4$-point functions of embedded ensembles for both fermions and bosons in the double-scaled limit. It is shown the models are equivalent to the double-scaled Sachdev-Ye-Kitaev model, expanding the double-scaled universality class to include both fermionic and bosonic systems. The models can be solved by introducing the Wick product of non-commuting Gaussian random variables. We show that deriving the Wick product is sufficient for computing the density of states, and properties of the Wick product can be used to compute $n$-point functions directly in the energy basis. In this context, the Wick product is equivalent to normal ordering of $q$-oscillators, which leads to the duality between moments of double-scaled models and expectation values in the chord Hilbert space. By considering operator probes as a second set of oscillators, we extend the duality to compute $n$-point functions. Embedded ensembles are equivalent to complex SYK at fixed charge, and we show working directly with embedded ensembles streamlines the derivations.

Multipartite entanglement is an essential quantum resource for various distributed quantum applications. One promising method for preparing multipartite entanglement is to interfere independent photons at linear optical interference setups. While heralding the successful interference and thereby the state generation is often costly, postselecting entangled states provides an achievable alternative in this framework. We introduce a family of interference schemes for postselecting symmetric qudit Dicke states, useful resources in quantum communication and variational quantum computing. We present schemes with and without ancillary photons and show that using ancillary photons can exceed the upper bound on the success probability of schemes without ancillary photons. Our results accommodate a wide range of linear optical schemes, providing multiple viable approaches for postselecting Dicke states.

We theoretically investigate a quadratic optomechanical system comprising a single-mode optical cavity linearly coupled to one mechanical resonator and quadratically coupled to a second resonator. By tuning the cavity detuning and optomechanical coupling strengths, we demonstrate the transition from optical bistability to multistability with up to seven steady-state solutions. Notably, simultaneous ground-state cooling of both mechanical resonators occurs on the dynamically stable branch of the nonlinear steady-state solutions, offering new opportunities for combined nonlinear optical and quantum cooling functionalities. Beyond the multistable regime, we systematically study dual-mode ground-state cooling and find that robust simultaneous cooling can be achieved over a broad parameter range, except when the linear and quadratic couplings become comparable, where a dark-mode effect arises. In this case, tuning the second-order optomechanical-induced frequency shifts effectively suppresses dark-mode interference, enabling controllable and simultaneous ground-state cooling. Our results provide a versatile framework for engineering multimode quantum states in optomechanical systems and open new avenues for the development of multifunctional quantum devices, including ultra-sensitive sensors, scalable quantum memories, and integrated quantum networks.

The quantum random number generation based on laser phase noise, which is featured with high generation rate and ease for photonic integration, has been extensively investigated and demonstrated. Despite these advancements, a theoretical model to achieve optimal performance in terms of maximizing the generation rate or entropy is still incomplete. In this work, a comprehensive physical model for this scheme is introduced to accurately predict the power spectrum and probability distribution of raw data, based on which the entropy source bandwidth and quantum min-entropy can be accordingly calculated and thus the system performance can be quantitatively evaluated. The model is sufficiently validated through both simulation and experiment with significant agreement under various setups. Furthermore, our proposal enables the priori configuration of experimental setups to achieve designed power spectrum and probability distribution of the raw data, thereby maximizing the generation rate or the min-entropy for system performance optimization.

We examine the gravitational lensing signatures of a Hayward-like regular black hole and its potential observational distinction from a Schwarzschild black hole. In the weak-field limit, the deflection angle includes a small positive correction proportional to $m \ell^2/b^3$, indicating slightly stronger light bending than in Schwarzschild, though the effect remains observationally negligible at large impact parameters. Current galaxy-scale Einstein-ring data, such as from ESO325-G004, cannot yet constrain the regular-core scale $\ell$. In the strong-deflection regime, for Sgr A* and M87*, the asymptotic position $θ_{\infty}$ is identical to Schwarzschild's. Nevertheless, $\ell$ modifies strong-lensing coefficients $\bar a, \bar b$, influencing angular separations s, relative flux ratio $r_\mathrm{mag}$, and time delays $ΔT_{2,1}$. Our predicted values for these observables remain consistent with current data, suggesting that future high-precision measurements of strong-field lensing may distinguish Hayward-like from Schwarzschild black holes.

We revisit the proposal to cure the negative density of states in the three-dimensional gravitational path integral by adding spinning states whose spin scales with the central charge. We show that sub-extremal and extremal spinning states below the black hole threshold can cancel the known negativities, and interpret these states as bulk spinning defects. Additionally, certain overspinning states above the black hole threshold can cure these negativities while preserving the spectral gap. Previously interpreted as classical spinning strings, we instead identify these overspinning states with overspinning BTZ geometries, which are smooth pure gravity quotients of AdS$_3$ with no fixed points. All of these spinning geometries exhibit causal pathologies in their Lorentzian continuations. Moreover, the overspinning geometries arise from mixed elliptic-hyperbolic identifications and contain a right-moving temperature and quasinormal modes. We also generalize the computation of scalar correlators to the extremal and overspinning backgrounds.

There has been considerable interest in constructing modified gravity theories that propagate only two degrees of freedom (DOFs), corresponding to the tensorial gravitational waves of general relativity. Within the framework of spatially covariant gravity (SCG), the conditions for obtaining 2-DOF theories can be derived from Hamiltonian constraint analysis, but it is generally difficult to translate those conditions into explicit SCG Lagrangians, especially when the Lagrangian depends nonlinearly on the extrinsic curvature. In this work, we adopt an alternative perturbative approach. We consider polynomial-type SCG Lagrangians up to $d=3$, where $d$ denotes the total number of derivatives in each monomial, and expand them around a cosmological background. By requiring the scalar mode to be eliminated up to cubic order in perturbations, we derive the corresponding conditions on the coefficient functions in the Lagrangian. We find five explicit Lagrangians that propagate only 2 DOFs up to cubic order in perturbations around a cosmological background. These theories therefore provide concrete candidate 2-DOF SCG models, at least at the perturbative level up to cubic order.

Many body gravity (MBG) is a novel modified theory of gravity formulated in a 5-D space-time-temperature framework, in which the variation in temperature is recast as a variation in the 5-D metric. Previous work on MBG has shown that it can reproduce galaxy rotation curves, radial acceleration relation and the weak gravitational lensing of the bullet cluster, without the inclusion of dark matter. In this work we show that MBG can reproduce cosmic inflation, and in the process, analyze fundamental relations between interaction, time and gravity. To analyze cosmic inflation using interacting massless scalar fields, we first analyze theoretically a hypothetical universe with a single massive particle, or a collection of non-interacting massive particles. A quantitative relation between time and interaction is developed using Quantum Field Theory (QFT), which suggests that the notion of time becomes ill-defined for such a universe. The mass terms in MBG and General Relativity cause a discrepancy with the QFT results. An interacting massless scalar field then becomes a necessity to resolve the issue at the onset of inflation. However, the entropic terms in the MBG field equations are seen to be consistent with the QFT results and further accelerate inflation. The slow-roll condition is shown to be a natural consequence of the Euler-Lagrange equations of motion governing the massless scalar field in 5-D space-time-temperature, during the early phase of inflation. Finally, the MBG field equations are solved in the context of a Friedmann metric, leading to inflation. The matter era is also investigated.

Many combinatorial optimization problems admit a maximin fairness variant, where the aim is to find a distribution over possible solutions which maximizes an expected worst-case outcome. However, the support for an optimal distribution may be exponential, which can be intractable to represent in the worst case. To this end, we propose a quantum based approach to solving distribution optimization problems. Expanding on work analyzing the Dynamical Lie Algebras of the Quantum Approximate Optimization Algorithm (QAOA), we show that with a finite number of layers, a QAOA ansatz can be constructed to capture any distribution over bitstrings. We show that the resulting circuit is able to effectively solve the Fair Cut Cover, a fair interpretation of the classical Fractional Cut Cover Problem. In addition, we show that our algorithm is provably better than classical approximations on certain graph structures and empirically outperforms these classical algorithms on tested instances.

Boson sampling is a leading candidate for demonstrating quantum advantage in photonic systems. Despite significant experimental and theoretical progress, a characterization of its output statistics remains incomplete. This is especially true beyond the dilute regime, where photon collisions and bunching become significant. The associated saturated regime, characterized by mode number growing linearly with photon number, or more generally sub-quadratically, is precisely the regime of greatest experimental interest. As a consequence, anti-concentration of the output distribution--a key ingredient in hardness arguments--remains poorly understood in boson sampling. In this work, we leverage representation-theoretic tools to address this gap, obtaining closed-form expressions for second moments of generic particle-number-preserving bosonic observables. We express these quantities in terms of Hilbert-Schmidt norms of projections onto irreducible components of the operator space and show that these projection norms admit compact analytical expressions by exploiting the underlying symmetry structure. Focusing on Fock state output probabilities, we further establish anti-concentration beyond the dilute regime. Together with recent complexity-theoretic results, our findings strengthen hardness guarantees for boson sampling in experimentally interesting settings.

In the literature, there are two differing definitions of contextuality: Kochen and Specker's, and Spekkens' (or ``generalised''). However, researchers using one of these definitions rarely consider the other, meaning comparative analysis of these two notions is rare. In this paper, we advance the idea that Kochen-Specker contextuality provides a generalisation of the idea of system being fundamentally nonclassical, while Spekkens' noncontextuality provides a generalisation of the idea of a system being classical. This allows us to reconcile the two approaches, as different stages in a hierarchy of classicality/nonclassicality.

We analyze data from a standard 3D general-relativistic magnetohydrodynamics (GRMHD) simulation, focusing on equatorial slices in order to examine the details and the evolution of the azimuthal structure of the accreting matter. During flux eruption events, the non-axisymmetric features of the equatorial inner accretion disk are considerably enhanced, with this enhancement being more prominent close to the black hole. Our analysis of the azimuthal structure of the equatorial accretion disk finds that the matter distribution in the vicinity of the horizon is dominated by low azimuthal mode numbers, specifically by the $m = 2$, and $m = 1$ modes, indicating that the non-axisymmetry of the disk during flux eruption events is enhanced due to the emergence of features with a large angular size on the equatorial plane. Our results suggest that the morphology of the equatorial accretion flow close to the black hole is mainly determined by the formation and motion of vertical magnetic flux bundles. These bundles are formed when the initially horizontal magnetic field reconnects into a vertical configuration, effectively detaching from the black hole horizon. This reconnection occurs in a low-density, highly magnetized region on the equatorial plane that expands over time as more field lines undergo vertical reconfiguration. The resulting vertical flux tubes, filled with low-density plasma, are then transported outwards due to magnetic buoyancy. Our results present a detailed quantitative description of the morphology of MADs and of its evolution during flux eruptions, complemented by a description of the physical process by which excess magnetic flux is detached from the black hole, vertically reconfigured, and expelled.

Exploiting enhanced sensitivity of a system in the vicinity of a phase transition boundary, critical quantum metrology to date still suffers from gap-closure related bottleneck effects, namely, critical slowing down of the sensing dynamics and a drastic shrinking of the parameter sensing window. To alleviate the said bottleneck inherent to any homogeneous lattice used for sensing, here we propose to leverage the intrinsic hopping inhomogeneity arising from bosonic ladder-operator matrix elements in Fock-space lattices (FSLs). Specifically, using a two-mode Jaynes--Cummings-type model, we show that the sensing parameter can be imprinted onto a topological zero-energy mode of the FSL. The key system parameters thus become cell dependent, effectively tracing out a curve in a topological phase diagram. Cell-dependent criticality emerges when this curve crosses or approaches a topological phase boundary, without globally tuning the lattice close to criticality. An external control parameter reshapes this curve, continuously tuning the scaling of the quantum Fisher information from the standard to the Heisenberg scaling while maintaining broad sensing coverage and a reduced gap cost. Furthermore, a local photon-number measurement on a single cavity saturates the quantum Fisher information. These results identify FSLs as a scalable and practical route to criticality-based quantum metrology.

The ongoing development of hardware that is capable of reliably executing general quantum algorithms requires quantum error-correcting codes that are both practical for realisation and rapidly reduce logical error rates as they are scaled up. Here we introduce the dynamic compass code, a code that can be implemented with a modest footprint on the heavy-hex lattice while also demonstrating a threshold. The dynamic code is obtained by choosing a novel measurement schedule for the syndrome extraction circuit of the heavy-hex subsystem code. We numerically evaluate its performance and observe that different choices of schedule can provide a trade-off in protection against logical errors in the $X$ vs $Z$ basis. We also demonstrate that this new measurement schedule provides the code with a threshold for stability experiments. We finally show how the dynamic compass code could be used for fault-tolerant logic by illustrating lattice surgery between code patches.

Measuring stochastic signals ("noise metrology") constitutes a central task in quantum sensing and the characterization of open quantum systems. Here we establish ultimate precision bounds for multiparameter estimation of stochastic signals encoded through Markovian Lindblad dynamics, allowing for arbitrary quantum control and noiseless ancillae. Although Markovianity enforces standard-quantum-limit scaling with sensing time $T$, our bounds reveal Heisenberg-type scaling in the number of dissipative channels, $R$: when the stochastic signal exhibits high-rank correlations across the $R$ channels and the probe is entangled, the average variance (per parameter) scales no better than $Ω(1/(TR^2))$. For collective $k$-body dissipation, $R=Θ(N^k)$, signifying super-Heisenberg scaling with the system size $N$. We further show that, when the unknown parameters enter through the dissipative eigenrates, a Rapid Prepare-and-Measure (RPM) protocol that tracks many distinct quantum jumps in parallel attains these limits. In this regime, the estimation problem reduces to a multi-Poisson counting model, providing a conceptually clean route to optimal quantum noise metrology. We illustrate the breadth of the framework with applications to networked noise metrology, collective many-body dissipation, learning Pauli noise, and subdiffraction quantum imaging.

To produce an operable quantum computer that is made with imperfect hardware, we must design and test scalable quantum error correcting codes that are suited for the devices we can build and, in unison, develop decoding strategies that accommodate device-specific noise characteristics. Here, we introduce the \emph{dynamic compass code}, a subsystem code with a novel syndrome extraction cycle, that has a competitive threshold while making efficient use of qubits arranged on a heavy-hex lattice. We use a superconducting qubit array to implement a distance-5 instance of this code, and demonstrate how detailed noise characterisation can boost decoder performance to yield significant improvements in logical error rates. We perform averaged circuit eigenvalue sampling (ACES) to acquire detailed context-dependent error information on all elements of the syndrome extraction process. Furthermore, we leverage soft information produced from measurement devices to augment the decoder with measurement error information and detect leakage errors for exclusion through post-selection. Our noise-informed approach yields up to 38.3\% improvement in the logical error rate of a distance-5 implementation of the dynamic compass code in experiment.

We consider Einstein gravity coupled to two Maxwell fields and one dilatonic scalar, and construct spherically-symmetric and static black holes that are charged under both Maxwell fields in general $D$ dimensions. We find that for suitable dilaton couplings, the equations of motion can be cast into one-dimensional Toda equations of all rank-2 Lie groups. We devise a brute-force approach to obtain the most general but remarkably elegant solutions to the Toda equations. This allows us to construct exact black holes associated with all the rank-2 Lie groups. The $B_2$ and $G_2$ Toda black holes are new. We study their thermodynamics and verify explicitly an earlier claim in the literature that all these thermodynamic quantities can be derived without having to solve for these black hole solutions.

Gravitational waves provide a powerful probe of both the astrophysical processes driving black hole mergers and the dynamics of the Universe, but these measurements rely on accurately inferring the unknown underlying population. We perform an agnostic reconstruction of the primary mass distribution using B-splines, characterising the emergence of structure with increasing model complexity. Using the latest gravitational-wave transient catalog, GWTC-4.0, we identify multiple mass features and find evidence suggesting a logarithmic hierarchy in the population. We show that this structure directly impacts measurements of the Hubble constant, primarily through features at the population boundaries. Finally, we introduce an approach that isolates a subpopulation of low-mass events to mitigate modelling systematics, providing a promising path toward robust population-based cosmology with future datasets.

Non-invertible symmetries of a quantum field theory (QFT) are a natural generalization of unitary symmetries, but in which the product of operators does not satisfy a group multiplication law. We show that such symmetry operations on states define a collection of quantum gates for a parallel quantum computation scheme that includes post-selection / projection as a gate. Structures such as gate complexity and more geometric complexity measures generalize to this setting. We provide a class of distance / distinguishability measures that extend the standard notion of distance for Lie groups to both continuous and discrete non-invertible symmetries, as well as more general linear combinations of unitary quantum gates. We illustrate these considerations by computing the distance between non-invertible symmetries in some 4D and 2D QFTs. We find that the simple objects of a symmetry category can be highly complex computationally.

Non-Clifford gates, used to generate quantum magic, are essential for universal quantum computation. We show that non-Clifford gates arise naturally from path integrals in topological quantum field theories, where their magic-generating properties are determined by the algebraic data of the theory. In Chern-Simons theory, we construct the Ising interaction gate, whose generator is prepared by path integration over simple three-boundary manifolds, and show that it produces non-local magic away from discrete Clifford points. We show that the Toffoli gate is obstructed in $SU(2)_1$ by the $\mathbb{Z}_2$ fusion structure, while $SU(2)_3$ is the minimal theory supporting the required conditional logic, given the density of the mapping class group in the projective unitary group on the manifold boundary. Finally, we demonstrate that the T gate arises as a path integral in Dijkgraaf-Witten theory, with gauge group $\mathbb{Z}_4$, where a single Dehn twist on the boundary torus produces the gate without approximation. These results show that topological path integrals construct gates in multiple levels of the Clifford hierarchy, and across distinct classes of field theories, with implications for topological quantum computing.

We study the growth of entanglement and circuit complexity in random passive linear optical networks as a function of the circuit depth. For entanglement dynamics, we start with an initial Gaussian state with all $n$ modes squeezed. For random brickwall circuits, we show that entanglement, as measured by the Rényi-2 entropy, grows at most diffusively as a function of the depth. In the other direction, for arbitrary circuit geometries we prove bounds on depths which ensure the average subsystem entanglement reaches within a constant factor of the maximum value in all subsystems, and bounds which ensure closeness of the random linear optical unitary to a Haar random unitary in $L^2$ Wasserstein distance. We also consider robust circuit complexity for random one-dimensional brickwall circuits, as measured by the minimum number of gates required in any circuit that approximately implements the linear optical unitary. Viewing this as a function of the number of modes and the circuit depth, we show the robust circuit complexity for random one-dimensional brickwall circuits scales at most diffusively in the depth with high probability. The corresponding Gaussian unitary $\tilde{\mathcal U}$ for the approximate implementation retains high output fidelity $|\langleψ|\mathcal U^\dagger \tilde{\mathcal U}|ψ\rangle|^2$ for pure states $|ψ\rangle$ with constrained expected photon-number.

Gravitational waveform templates are a key ingredient for the detection and characterization of gravitational waves emitted by compact binary mergers in the universe. These templates must be physically accurate and extensive, but also highly computationally efficient, two requirements that are often in tension. One solution to this problem is the development of surrogate models, which are fast, data-driven models trained to predict the output of a slower, physically realistic waveform model. In this article we build on existing work to incorporate machine learning techniques into the conventional reduced order surrogate framework, with a focus on extending coverage to waveform models that describe generically precessing quasicircular binaries. In particular, we present SEOBNRv5PHM_NNSur7dq10, a reduced order neural network surrogate of the SEOBNRv5PHM waveform model, valid up to mass ratios 1:10 for precessing quasicircular binary black hole systems with arbitrary spin magnitudes and orientations. The faithfulness of the surrogate to SEOBNRv5PHM is validated, and the surrogate is successfully applied to Bayesian parameter inference using both real and injected gravitational wave data. The surrogate is approximately 5 times faster than SEOBNRv5PHM when evaluating a single waveform on a CPU, and nearly 1000 times faster per-waveform when amortizing the cost over large waveform batches on a GPU.

Qubit loss is a major source of error in quantum computation, as it invalidates the algebraic structure of the standard stabilizer formalism for quantum error-correcting codes. On the one hand, it complicates decoding; on the other hand, it introduces stochastic flicker patterns in stabilizers as a hallmark of qubit loss. Here, we develop an artificial-intelligence-enabled decoder based on a spatiotemporal Graph Neural Network (STGNN) architecture to extract spatial and temporal correlations from syndrome histories. Our decoder performs a dual-head task, simultaneously correcting standard Pauli errors and identifying the locations of qubit loss. Our decoder achieves significantly higher logical accuracy than both the traditional minimum-weight perfect matching (MWPM) algorithm and even delayed-erasure MWPM decoders that use qubit loss information from the final round as input. Our decoder can also identify more than 90% of loss locations after accumulating stabilizer measurements over the subsequent ten rounds, thereby facilitating qubit reinitialization, for instance, via the continuous loading technique on the atom array platform. For both tasks, our STGNN performs nearly identically to a modified version of AlphaQubit, but it employs a parallel input structure, giving it an advantage in inference time over modified AlphaQubit's recurrent input structure. This work provides a robust and scalable framework for correcting qubit loss errors, paving the way for more efficient fault-tolerant quantum computation.

Quantum coherence plays a central role in Grover's search algorithm. We study the Tsallis relative $α$ entropy of coherence dynamics of the evolved state in Grover's search algorithm. We prove that the Tsallis relative $α$ entropy of coherence decreases with the increase of the success probability, and derive the complementarity relations between the coherence and the success probability. We show that the operator coherence of the first $H^{\otimes n}$ relies on the size of the database $N$, the success probability and the target states. Moreover, we illustrate the relationships between coherence and entanglement of the superposition state of targets, as well as the production and deletion of coherence in Grover iterations.

It is commonly accepted that the results of measurements simultaneously realized over two entangled subsystems are statistically correlated instantaneously regardless of the distance between them. In accordance with Bell theorem, everything happens in such measurements as if there was a correlation propagating at infinite speed between the two subsystems.These correlations have been so far verified experimentally up to a distance of 1200 km. We discuss the interest and feasibility of extending this distance to 390,000 km, thus gaining a factor of 300. The idea is to install one of the polarimeters on the Moon, with the other on Earth. Such an experiment would provide a new test of Quantum Physics and allow to put higher constraints on alternative theories and interpretations. We also discuss the possibility to violate Bell inequalities beyond Earth-Moon distance.

The high-mass ($M \gtrsim 30 M_\odot$) black holes (BHs) from the gravitational-wave (GW) observations of LIGO and Virgo came as a surprise to many astronomers. While the collapse of metal-poor massive stars could produce such BHs, gravitational lensing has been invoked to explain their high masses. Broadhurst, Diego, and Smoot (henceforth BDS) argued that the mass distribution of BHs in coalescing binaries is very similar to that of the galactic BHs, and the inferred high masses are the result of neglecting the lensing magnification. They also proposed a redshift distribution of binary BH (BBH) mergers to explain the observed LIGO-Virgo mass distribution. We ask whether such a model is consistent with different aspects of the GW observations: 1) the observed number of BBH mergers, 2) the distribution of their redshifted total mass and apparent luminosity distance, 3) the non-detection of strongly lensed events, and 4) the non-observation of the stochastic GW background. By simulating lensed BBH mergers with the BDS model and comparing them with observations, we conclude that no choice of BDS model parameters is consistent with all aspects of the observations. Lensing magnification is not a viable explanation for the high-mass BHs discovered by LIGO and Virgo.

This is a review of key aspects of a model presented at the Lake Como School: Dark Matter, Dark Energy, and the Cosmological Tensions, June, 2025. The associated publication can be found at: A Schlatter and R E Kastner 2023 J. Phys. Commun. 7 065009

We compute the transmission properties of electromagnetic (EM), gravitational wave (GW), and static gravitational perturbations through geometric throats in spherically symmetric spacetimes. On the ultrastatic Ellis-Bronnikov background, decomposition of the four-dimensional Maxwell equations into vector spherical harmonics yields an effective Schrödinger problem with centrifugal barrier $V_\ell^{(\mathrm{EM})}=\ell(\ell+1)/(σ^2+r_0^2)$ peaked at the throat. For the lowest physical EM mode ($\ell=1$), frequencies below the barrier-top frequency $ω_{\max}=\sqrt{2}/r_0$ are strongly suppressed by sub-barrier tunnelling. Gravitational wave perturbations ($\ell\ge 2$) see a qualitatively similar barrier and are likewise strongly suppressed below their respective barrier-top frequencies. By contrast, the static gravitational monopole ($\ell=0$), governed by the linearised Einstein equations on the same background, satisfies the source-free conservation law $(a^2Φ')'=0$ with no potential barrier, yielding the exact solution $Φ\propto\arctan(σ/r_0)$. We extend these results to a one-parameter family of throat geometries with varying profile shapes, and to a reflected-Schwarzschild (Damour-Solodukhin-type) wormhole, demonstrating that the qualitative asymmetry\emdash strong sub-barrier suppression for all propagating radiation ($\ell\ge 1$) versus polynomial attenuation for the static monopole ($\ell=0$)\emdash is universal for static, spherically symmetric throats. Numerov integration, WKB estimates, and exact analytical solutions are compared throughout. The results establish a structural constraint-wave asymmetry arising from the multipole decomposition of the field equations, independent of the matter content sourcing the geometry, on a fixed background.

Consciousness and quantum mechanics are among the most puzzling phenomena studied in the sciences. Some scholars suggest they are related, though others think this claim commits a "minimization of mystery" fallacy. The aim of this programmatic paper is to draw attention to a less widely discussed parallel between consciousness and quantum mechanics: both challenge the classical objectivist worldview of science. Under certain assumptions, they are each in tension with a package of metaphysical theses -- "non-relationalism", "non-fragmentation", and "one world" -- that jointly make up that worldview. This points to three distinct non-objectivist responses: the "relationalist", "fragmentalist", and "many-subjective-worlds" ones. We will map out their pros and cons.

We present a set of numerical-relativity binary neutron star merger simulations incorporating muons and muonic reactions for two baseline baryonic equations-of-state. In order to investigate the possible impact of muons and muonic weak reactions, we treat neutrinos with a gray (energy-independent) truncated moments scheme and an implicit-explicit time integrator. Newly computed neutrino rates are employed within the full kinematics approach for a set of relevant reactions, and pair-processes are modeled via opacities computed using reaction kernels, that allow a consistent treatment of neutrino interaction rates. We find that equilibration between matter and radiation is successfully captured by a novel two timescales approach. Of astrophysical interest is the general agreement between our muonic and non-muonic results regarding the remnant evolution, disk and outflow properties. Average electron fractions, asymptotic velocities and temperatures are different by less than $\sim 6\%$, while the main impact of muons is a reduction in ejecta masses by at most $\sim 17\%$. Therefore, based on our findings, accounting for the presence of muons and muonic reactions might result much less severe consequences regarding nucleosynthetic yields and electromagnetic counterparts than previously reported in the literature.

We show that averaging matter dynamics over stochastic gravitational fluctuations gives rise to a complex velocity field \(η_μ = π_μ - i u_μ\) living as a section of the pullback bundle \(E = π_{2}^{*}(T^{*}M)\to \mathcal{C}\times M\). We prove that \(η_μ\) is isomorphic, via the Schrödinger representation, to the symmetric logarithmic derivative (SLD) operator \(L_μ\) on the Hilbert space \(\mathcal{H}_{x} = L^{2}(\mathcal{C})\), up to a trace-zero projection. This isomorphism \(\widetilde{\mathcal{T}}:Γ(E / \sim)\to Γ(\mathcal{L})\) is a bundle isomorphism preserving the flat \(U(1)\) connection (proved in \cite{meza2026topological}) and the quantum Fisher metric. The quantum Fisher information metric \(g_{μν}^{\mathrm{FS}}\) is expressed directly in terms of \(η_μ\) as \(g_{μν}^{\mathrm{FS}} = - \frac{4m^{2}}{\hbar^{2}}\mathrm{Re}\langle (η_μ - \langle η_μ\rangle)(η_ν - \langle η_ν\rangle)\rangle_{\mathcal{P}}\). The holonomy of \(η_μ\) is quantized, leading to topological phases observable in atom interferometry.

When quantum emitters couple indistinguishably to light, they can synchronize into a collective light matter system with radiative properties profoundly different from those of independent particles. To date, the resulting collective effects have largely been confined to point like or homogeneous ensembles. Here, we open access to a qualitatively new collective regime by realizing geometrically ordered, spatially extended atom arrays with subwavelength spacing. This establishes a fundamentally new platform in which collective emission is no longer confined to a single Dicke mode but instead emerges from an ordered network of photon mediated interactions. We find that 2D atom arrays undergo strong super and subradiant emission. Despite subwavelength spacing, we achieve site resolved imaging and directly observe the buildup of spatial correlations, demonstrating the transformation of cooperative decay into a strongly correlated many-body process. We observe extensive scaling of superradiance, uncover superradiant revivals, and reveal the ferromagnetic nature of superradiance and the antiferromagnetic nature of subradiance. Our results realize a novel programmable platform for exploring and utilizing dissipative many-body quantum physics, opening new possibilities for photon capture, storage, and atom photon entanglement.

The topology of cosmic reionization, the sizes, shapes, and connectivity of ionized bubbles is a primary observable of next-generation 21\,cm experiments. We show that this topology is sensitive to the microphysics of dark matter. Self-interacting dark matter (SIDM), with cross-sections $σ/m\sim 1$--$10\;\mathrm{cm^2/g}$ motivated by small-scale structure anomalies, reduces halo gas binding energies and increases the duty cycle of ionizing-photon escape. At fixed global neutral fraction $\bar{x}_{\rm HI}$, this reshapes the source population from rare, very bright emitters to more numerous, moderate emitters, producing qualitatively different ionization morphology. We decompose the effect into two scale-dependent levers: a $2$--$3\%$ emissivity-weighted bias shift at $k\lesssim 0.1\;h/\mathrm{Mpc}$, and a factor $2$--$4$ shot-noise suppression at $k\sim 0.1$--$1\;h/\mathrm{Mpc}$. A halo-by-halo semi-numerical simulation at $128^3$ resolution confirms a $\sim 60$--$70\%$ increase in the Euler characteristic of the ionization field for $σ/m \gtrsim 2\;\mathrm{cm^2/g}$, detected at $3.8σ$ across ten independent realizations. A blowout model connecting the binding-energy reduction to the duty cycle through the ISM column density distribution yields a detection threshold at $σ/m \sim 1$--$2\;\mathrm{cm^2/g}$. The signal exceeds the CDM baryonic uncertainty band and is robust to the functional form of the emissivity parametrization. The signal persists even if gravitational heating offsets $50$--$75\%$ of the blowout enhancement, and is not diluted by unresolved low-mass sources. Velocity-dependent SIDM produces a qualitatively distinct opposite-sign bias shift. These predictions are testable with SKA1-Low, establishing reionization as a new arena for probing dark matter models complementary to dwarf galaxies and galaxy clusters.

Neutrino self-interactions delay the onset of free-streaming in the early universe, leaving distinct, scale-dependent signatures on the matter power spectrum. We investigate these signatures in post-reionization 21-cm intensity mapping and the Lyman-$α$ (Ly$α$) forest at redshifts $z \sim 2$--$3.5$, and forecast the constraints achievable with upcoming surveys using Fisher matrix analysis. Modeling neutrino self-interactions through an effective four-fermion parameterization with coupling $G_{\rm eff}$, we compute modifications to the Ly$α$ and 21-cm auto- and cross-power spectra for both strongly interacting (SI$_ν$, $\log_{10}G_{\mathrm{eff}} = -1.77$) and moderately interacting (MI$_ν$, $\log_{10}G_{\mathrm{eff}} = -5$) scenarios. We then combine these with forecasts for a representative next-generation cosmic microwave background (CMB) mission to evaluate the capabilities of SKA1-Mid and PUMA. We find that the Ly$α$--21-cm cross-correlation provides a systematics-resilient probe of the interaction signal, and decisively breaks the degeneracy between the primordial scalar power spectrum amplitude ($A_s$) and $G_{\rm eff}$ that limits CMB only analysis, particularly for the SI$_ν$ mode. Furthermore, the CMB+PUMA combination emerges as the optimal survey configuration for both regimes, reaching 1$σ$ constraints of $\mathcal{O}(10^{-3})$ on $σ(\log_{10}G_{\rm eff})$ for the SI$_ν$ mode and $\mathcal{O}(10^{-2})$ for the MI$_ν$ mode. Compared to the CMB-only baseline, this represents an improvement of approximately one order of magnitude for the SI$_ν$ mode, and nearly two orders of magnitude for the MI$_ν$ mode. We show that this conclusion holds uniformly over the full range of coupling strengths from $\log_{10}G_{\rm eff} = -6$ to $-1.77$.

We present a lattice QCD study of dilepton production in charmonium transitions, specifically focusing on the $1^{+-} \to 0^{-+}$ and $1^{++} \to 1^{--}$ processes: $h_c \to η_c \ell^+ \ell^-$ and $χ_{c1} \to J/ψ\ell^+ \ell^-$, where $\ell = e, μ$. The relevant hadronic matrix elements are computed using gauge field configurations generated by the Extended Twisted Mass Collaboration with $N_f = 2+1+1$ dynamical Wilson--Clover twisted-mass fermions at four lattice spacings. Simulations are performed at physical dynamical $u$, $d$, $s$, and $c$ quark masses, except for the coarsest lattice, where the lightest sea quark mass corresponds to a slightly heavier pion mass. A controlled continuum extrapolation is carried out. In the continuum limit for the $h_c$ decays, we obtain $Γ(h_c \to η_c e^+ e^-) = 5.45(19)~\mathrm{keV}$, and $Γ(h_c \to η_c μ^+ μ^-) = 0.635(22)~\mathrm{keV}$. For the $χ_{c1}$ decays, we find: $Γ(χ_{c1} \to J/ψe^+ e^-)= 2.869(90)~\mathrm{keV}$, and $Γ(χ_{c1} \to J/ψμ^+ μ^-) = 0.1993(72)~\mathrm{keV}$. Our results for the $χ_{c1}$ decays show good compatibility with experimental data. However, our prediction for the $h_c \to η_c e^+ e^- $ decay rate is approximately $3σ$ larger than the BESIII result. We also present predictions for the differential decay widths as functions of the dilepton invariant mass, $q^2$, and for angular observables sensitive to longitudinal transition form factors, which are inaccessible in radiative decays with real photon emission. These results constitute the first fully dynamical lattice QCD predictions for dilepton decay rates in $h_c$ and $χ_{c1}$ charmonium transitions, including their differential distributions and angular observables. They provide benchmark predictions for future experimental studies.

If the Universe underwent a cosmic phase transition, it may have left behind a network of cosmic strings. When these strings arise from the breaking of a gauge symmetry, their decay produces a significant stochastic background of gravitational waves. In contrast, if they originate from the breaking of a global symmetry, their decay predominantly yields Nambu-Goldstone bosons, which can persist as dark matter or dark radiation. In this work, we assess the detectability of this particle spectrum using a range of cosmological probes. We employ semi-numerical methods to estimate the resulting energy density and compute the associated matter power spectrum. We then compare these predictions with observations of the cosmic microwave background, Lyman-$α$ forest, large-scale structure surveys, and the UV luminosity function, thereby deriving constraints on the Nambu-Goldstone boson mass and the symmetry-breaking scale. Finally, we present projections for the sensitivity of upcoming cosmic microwave background missions.

We develop a primal bootstrap framework for effective field theories in the presence of a graviton pole, based on finite-resolution sampling rather than smearing, while also allowing direct control over the number of subtractions. We show that this approach reproduces the known projective bounds obtained from smearing in $D{\ge}6$, while yielding slightly stronger bounds in $D{=}5$. This method also makes it straightforward to impose linearized unitarity directly and provides an access to the extremal spectra. Applying the method to crossing-symmetric dispersion relations, we derive new non-projective bounds that fix the overall scale of the EFT couplings. In $D{=}5$, for example, we find that $\frac{M}{M_{\rm P}}{\lesssim}7.8$, showing that the EFT cutoff cannot be taken parametrically larger than the Planck scale. At the extremal values of the couplings, the spectra exhibit a surprising structure: for projective bounds, they exhibit peaks around quadratic Regge-like trajectories, while for the non-projective bounds they form sharp quadratic bands. In the latter case, the leading coefficients further display an inverse-quadratic dependence on the band number.

Rare and very rare decays of third-generation particles, including $b$-hadrons and $τ$ leptons, provide sensitive probes of physics beyond the Standard Model (SM). Unlike direct searches limited by collider energies, they probe new physics at much higher energy scales. Many of these decays have SM-predicted branching fractions below the sensitivity of current detectors. These proceedings report on recent LHCb searches, including several first searches and results setting the most stringent limits to date. In particular, searches for $b \to s τ^+τ^-$, $b \to s τ^\pm e^\mp$, $b \to s μ^\pm e^\mp$, and $τ^- \to μ^-μ^+μ^-$ are presented, alongside searches for lepton-number-violating processes and loop-suppressed annihilation decays.

Interacting fermions exhibit a rich landscape of surface defects and associated critical phenomena. We investigate novel surface critical behavior in the three-dimensional Gross-Neveu-Yukawa model. For a class of defect renormalization group flows, we obtain exact infrared solutions and show how fermionic anomalies are encoded in the resulting surface dynamics. We further uncover emergent topological and geometric structures in the defect coupling space, and comment on their relation to a defect analogue of the CFT distance conjecture.

We present an updated overview of the symmetry preserving Contact Interaction model in hadronic physics, developed a little over a decade ago to describe the mass spectrum and internal structure of mesons and diquarks composed of light and heavy quarks. Over the years, the Contact Interaction has evolved into a framework capable of treating both ground and excited states, providing a simple yet consistent approach to nonperturbative QCD. In this review, we examine the mass spectrum and elastic form factors of forty mesons with different spins and parities, together with their corresponding diquark partners. Importantly, we update the comparison of Contact Interaction predictions using recent results from the literature, offering a fresh perspective on the model's performance, strengths, and limitations. The analysis presented here refines previous conclusions and supports the Contact Interaction as a practical tool for hadron structure studies, with potential applications to baryons and multiquark states. We also present comparisons with other theoretical models and approaches, including lattice quantum chromodynamics, and comment on future prospects in view of ongoing and planned hadron structure experimental programs. In particular, forthcoming measurements at FAIR, together with future studies at Jefferson Lab and the Electron Ion Collider, are expected to provide key insights into hadron structure, with FAIR offering indirect constraints via hadron spectroscopy, hadronic interactions, and in-medium properties, while high-precision data on meson structure and form factors from Jefferson Lab and the Electron Ion Collider will provide valuable benchmarks to confront Contact Interaction based predictions.

In this note we explore monodromy defects for non-invertible symmetries in Maxwell theory, exploiting the conformal mapping to $AdS_{3} \times S^{1}$. With this approach we recover the spectrum of the defect conformal primaries. We also dedicate some time discussing the behaviour of Wilson/'t Hooft lines in the presence of such a monodromy defect, and highlight the following aspects of their behaviour: i) the lines can terminate on the defect, ii) lines of the unit electric (magnetic) charge may seize to be indecomposable, and can be represented as integer powers of some more elementary lines, and iii) they behave as topological objects when brought close to the defect, and this behaviour is governed by a Chern-Simons theory.

Searches at high object masses probe both resonant production of new particles and nonresonant distortions of Standard Model spectra. This contribution follows the material presented in the Moriond Electroweak 2026 talk and summarizes recent CMS results in this regime: the Run~2 combination of heavy vector boson searches, the Run~3 search for $W^\prime \to \ell ν$, the Run~2 dijet angular analysis, and searches for pair-produced dijet resonances in inclusive and $b$-tagged final states. No significant deviation from the SM expectation is observed, and the new results extend the sensitivity of CMS to multi-TeV scales in several benchmark scenarios.

The decay $B^0 \to Λ_c^+ \barΛ_c^- K_S^0$ is studied at LHCb for the first time using proton-proton collision data recorded by the LHCb experiment at a center-of-mass energy of $\sqrt{s} = 13$ TeV, corresponding to an integrated luminosity of 5.4 fb$^{-1}$. The branching ratio relative to the decay $B^+ \to Λ_c^+ \barΛ_c^- K^+$ is measured to be $$ \frac{{\cal B}(B^0 \to Λ_c^+ \barΛ_c^- K_S^0)}{{\cal B}(B^+ \to Λ_c^+ \barΛ_c^- K^+)} = 0.53 \pm 0.05 \pm 0.05, $$ where the first uncertainty is statistical and the second is systematic. Evidence is found for contributions from two resonant states, $Ξ_c(2923)^+$ and $Ξ_c(2939)^+$, in the $Λ_c^+ K_S^0$ system. The two states show a significance of $3.9σ$ relative to the nonresonant hypothesis. These two $Ξ_c^+$ states are consistent with being the isospin partners of the states observed in $Λ_c^+ K^-$ system.

The Wilson line defect half-indices for 3d $\mathcal{N}=2$ gauge theories with boundary confining phases admit a formulation in terms of the Askey-Wilson type moments. In the dual Landau-Ginzburg description the dual line operators can be realized as vortex line defects which induce singular behavior of chiral multiplets associated with the minimal monopole operators, together with additional one-dimensional degrees of freedom. By incorporating such a singular structure as an effective spin shift into the index computation, we obtain exact closed-form expressions for the line defect half-indices which are Askey-Wilson type moments.

In this paper, we analyze the $1 \rightarrow 4$ decay channel for the production of doubly heavy quarkonium, $(b\bar{c})$ or $(c\bar{c})$, via top-quark decays, $t \to (b\bar{c}) + c + c + \bar{s}$ and $t \to (c\bar{c}) + b + c + \bar{s}$, within the framework of nonrelativistic QCD (NRQCD). The dominant contributions are considered in color-singlet S-wave states, i.e., $(b\bar{c})[^1S_0]$, $(b\bar{c})[^3S_1]$, $(c\bar{c})[^1S_0]$, and $(c\bar{c})[^3S_1]$. Our calculations show that the decay widths for $\bar{B_{c}}$, $\bar{B_{c}^{*}}$, $J/ψ$ and $η_{c}$ production are 0.2251, 0.3099, 0.0537 and 0.0555 MeV, respectively, resulting in ${\cal O}(10^{4}\text{--}10^{6})$ level of $\bar{B}_c^{(*)}$ events and ${\cal O}(10^{3}\text{--}10^{5})$ level of charmonium produced at LHC per year. In particular, we find that the dominant contribution to $J/ψ$ and $η_{c}$ production via top-quark decays arises from this decay channel proposed in this work. Moreover, this multi-body top-quark decay process can serve as a sensitive probe for validating the narrow-width approximation (NWA). Finally, we provide a detailed analysis of theoretical uncertainties and differential distributions to facilitate the corresponding experimental searches. The production of a hadron associated with three quarks contains rich physical information, providing new insights for the LHC to study $B_c$ mesons and charmonia.

We combine the Glauber--Lachs formula from quantum optics and the two-component picture for pion production to analyze data on two- and three-pion Bose--Einstein correlation at 7 TeV from the LHCb Collaboration. For the pion exchange function $E_{\rm 2B}$, we chose a dipole form and an inverse one-and-a-half pole form. The extensions are computed in the configuration space of 4-dimensional Euclidean space ($ξ=\sqrt{|\bm r_1-\bm r_2|^2+(t_1-t_2)^2}$).

We study the linear cosmological evolution of inelastic self-interacting dark matter in a two-component dark sector with a small mass splitting, assuming thermal initial conditions for the two species. We derive the coupled background and perturbation equations for inelastic conversion between the two species, considering both Power-law and Low-velocity saturation cross sections. Exothermic conversion injects kinetic energy into the light component, generating pressure support that suppresses small-scale structure and produces dark acoustic oscillations in the matter power spectrum. The resulting cutoff at scale $k > 1\,h\,\mathrm{Mpc}^{-1}$ depends on the normalization and velocity dependence of the cross section, the dark matter mass and the mass splitting. Using linear power spectra computed with a modified Boltzmann solver, we apply recast constraints from Lyman-$α$ forest data and high-redshift UV luminosity functions, finding non-monotonic but closed exclusion regions driven by the competition between efficient conversion and rapid depletion of the heavy component. These results show that the internal thermodynamics of a secluded multi-component dark sector can leave observable imprints on structure formation, providing a complementary probe of dark matter beyond Standard Model interactions.

In the present work we investigate the phenomenological implications of a vanishing effective Majorana neutrino mass within a $3+1$ neutrino framework adding a eV-scale sterile neutrino beside three active neutrino states in light of latest cosmology driven bounds on sum of neutrino masses ($\sum_{i}m_i$). We explore the parameter space where the destructive interference between active and sterile states leads to vanishing amplitude, $M_{ee}$, of neutrinoless double beta ($0νββ$) decay. The allowed parameter space has been identified and predictions have been obtained taking into account the latest Planck and DESI+CMB bound on $\sum_{i}m_i$. We find that these bounds restrict the sterile mixing angle $θ_{14}$ and the lightest active neutrino mass. Furthermore, we incorporate the refined precision data from JUNO experiment regarding solar oscillation parameters ($θ_{12}, Δm_{21}^2$). We find that the sterile neutrino parameters like $θ_{14}$ may not be sensitive to the JUNO precision measurements as the constraint imposed by precise $θ_{12}$ is washed out by new cancellations driven through additional CP violating phases leading to vanishing $|M_{ee}|$.

Quasi-primary correlators in two-dimensional conformal field theories deformed simultaneously by $T\bar T$ and root-$T\bar T$ are studied. A path-integral formulation motivated by the geometric realization of the combined deformation is used to develop a geometric framework for evaluating the deformed correlators. Within this framework, the two-point function is obtained to all orders in the $T\bar T$ coupling and to leading order in the root-$T\bar T$ coupling, while the leading correction to the three-point function is computed. It is further shown that the deformed two-point correlator admits a kernel representation as a weighted average of undeformed CFT correlators over conformal dimensions. This representation is derived explicitly for both the pure $T\bar T$ deformation and the combined flow. In this way, the mixed $T\bar T$/root-$T\bar T$ deformation is incorporated into the geometric description of irrelevant deformations, and the structure of local correlators beyond the pure $T\bar T$ case is characterized more explicitly.

We study the ODE/IM correspondence between the linear problem associated with the supersymmetric affine Toda field equation for the twisted affine Lie superalgebra $C (2)^{(2)} = \mathfrak{osp} (2 | 2)^{(2)}$ and two-dimensional $\mathcal{N} = 1$ superconformal field theories (SCFTs). On the ODE side, we introduce a boundary condition more suitable for the conformal limit and the subsequent WKB analysis and diagonalize the resulting Lax operator. This leads to a WKB expansion from which we extract the WKB periods and non-local conserved quantities up to tenth order. On the IM side, we compute the eigenvalues of the local integrals of motion on the cylinder in both the Neveu-Schwarz and Ramond sectors of 2d $\mathcal{N} = 1$ SCFTs. We then compare the two sides and verify, up to sixth order, that the WKB periods coincide with the eigenvalues of the local integrals of motion for highest-weight states in the Neveu-Schwarz sector.

We review the recent progress made with regard to the hadronic light-by-light (HLbL) contribution to the Standard Model prediction of the muon anomalous magnetic moment and how well this compares with predictions from holographic QCD models, which had predicted larger contributions from axial vector mesons and short-distance constraints than the White Paper of 2020. A new holographic prediction concerns tensor-meson contributions, which in holographic QCD play a significant role in short-distance constraints beyond the Melnikov-Vainshtein constraint. When matching also the symmetric longitudinal short-distance constraint, the resulting prediction for the tensor-meson transition form factors agree well with available singly virtual data, but lead to different results than the traditional quark-model ansatz and a sizable positive contribution that could explain the remaining current tension between lattice and data-driven results for the HLbL contribution.

We study the coupled cosmological evolution of primordial black holes (PBHs) and radiation in the Arkani-Hamed-Dimopoulos-Dvali (ADD) framework with $n$ large extra dimensions and a fundamental gravity scale $M_\star$ at the TeV scale. For PBHs with horizon radius smaller than the compactification scale, the higher-dimensional geometry implies a larger horizon size at fixed mass and therefore a suppressed Hawking temperature. As a result, radiation accretion can overcome evaporation in the early Universe and drive a ``runaway'' phase of rapid mass growth. By numerically solving the coupled mass and energy-density evolution equations, we show that for $n \geq 2$ initially microscopic PBHs with initial mass $M_i \gtrsim 10^{12}\,$g can grow by many orders of magnitude and potentially reach macroscopic, even solar-mass, scales by matter-radiation equality. We determine the critical initial abundance $β_{\rm crit}$ required for PBHs to account for the observed dark matter density and find that extra dimensions dramatically lower this threshold, allowing viable scenarios with $β_{\rm crit}\sim 10^{-44}$. This identifies a previously unexplored region of parameter space in which the dark matter abundance is achieved through dynamical mass growth rather than large initial collapse fractions.

We investigate near-threshold $J/ψ$ photoproduction off the nucleon, focusing on hadronic rescattering effects induced by open-charm meson-baryon intermediate states. Beyond the conventional Pomeron-exchange mechanism, the $\bar D^0Λ_c^+$ and $\bar D^{*0} Λ_c^+$ channels are incorporated within an effective Lagrangian framework. The relevant production amplitudes are evaluated at tree level from $t$-, $s$-, and $u$-channel diagrams in a gauge-invariant manner. The resulting total and $t$-dependent differential cross sections are compared with recent near-threshold data from the GlueX, $J/ψ$-007, and CLAS experiments at Jefferson Lab. We find that the open-charm rescattering contributions significantly improve the description of the data, particularly at large momentum transfer, and naturally generate cusp-like structures near the $\bar D^0 Λ_c^+$ and $\bar D^{*0} Λ_c^+$ thresholds in the GlueX data. We further present predictions for the associated open-charm processes $γp \to \bar D^{(*)0} Λ_c^+$, whose cross sections are estimated to be of the order of 5 nb.

In the 4FGL-DR4 point-source catalog of the Large Area Telescope (LAT) onboard NASA's Fermi Gamma-ray Observatory (Fermi-LAT), around a third of the sources are still unidentified (unIDs). In this work, we perform a detailed study of one of them, namely 4FGL J2112.5-3043. Only gamma-ray emission has been detected from this unidentified source, with no counterpart observed at any other wavelength as of today. Together with its high detection significance, this makes 4FGL J2112.5-3043 a particularly compelling target for further investigation. The results of our spectral and spatial analyses show that the source photon spectrum is better described with a subexponential cutoff power-law spectral model, with no significant flux variability over time, and a morphology consistent with being a point-like source. We investigate and discuss the characterized emission within the context of both conventional and exotic astrophysics, namely a pulsar origin or potential dark matter (DM) annihilations in a nearby Galactic subhalo. Although our results are inconclusive and neither confirm a DM origin nor firmly establish an astrophysical nature, we find a spectral preference for the $b\bar{b}$ and $c\bar{c}$ DM annihilation channels over a pulsar origin, thus making this unID a particularly intriguing candidate for next multiwavelength observations.

A review of recent multiboson and vector boson scattering (VBS) measurements from the ATLAS and CMS Collaborations at the LHC is presented. Results are reported from precision diboson cross-section measurements, novel CP-sensitive and polarisation observables in $Wγ$ production, VBS observations in semileptonic and fully leptonic final states including the first measurements at $\sqrt{s}$ = 13.6 TeV, and observations of triboson processes. These results constitute a comprehensive test of the electroweak gauge sector of the Standard Model, and provide stringent constraints on anomalous gauge couplings in the effective field theory framework.

We extend the OPE-based renormalization algorithm to composite operators with operator mixing, focusing on scalar operators in $φ^4$ and $φ^3$ models. Using the OPE of operators with a fundamental field, we show that the $Z$-factors of these composite operators are determined by OPE coefficients of lower-dimensional traceless symmetric tensor operators, and establish a recursive renormalization framework. We report the five-loop anomalous dimensions for operators with $Δ\le5$ in the $φ^4$ model and the two-loop anomalous dimensions for operators with $Δ\le10$ in the $φ^3$ model. These results further demonstrate the versatility and efficiency of the OPE-based algorithm.

We present the first comprehensive study of the rare Dalitz decay $K^*(892) \rightarrow K \ell^+ \ell^- (\ell = e, μ)$, providing a prediction for the branching fraction and the dilepton mass spectrum. This decay involves the emission of a virtual photon which converts into a lepton pair, offering a probe of the transition form factor $F_{K^*K}(q^2)$ and underlying meson structure. Using a single pole approximation for the form factor, we present the calculation of the branching fraction for this rare decay channel for the first time. Furthermore, we also investigate the potential to search for a light $A^\prime$ boson (dark photon) appearing as a narrow resonance in the dilepton spectrum, and discuss the experimental sensitivity and new physics opportunities at the dedicated BESIII experiment. Our results establish $K^*(892) \rightarrow K \ell^+ \ell^- (l = e, μ)$ as a new laboratory for hadronic structure and dark-sector searches.

BESIII has accumulated 4.5 fb$^{-1}$ of $e^+e^-$ collision data in the 4.6 to 4.7 GeV energy range, corresponding to the world's largest sample of $Λ_c^+\barΛ_c^-$ pairs. This paper summarizes recent BESIII results on charmed-baryon decays, including the observation of the rare semi-leptonic decay $Λ_c^+\to ne^+ν_e$ using a Graph Neural Network, the first measurement of the decay asymmetry in the pure $W$-exchange decay $Λ_c^+\toΞ^0K^+$, and branching fraction measurements of the inclusive decays $Λ_c^+\to Xe^+ν_e$ and $\barΛ_c^-\to \bar{n}X$. We also report partial wave analyses of $Λ_c^+\toΛπ^+π^0$ and $Λ_c^+\toΛπ^+η$, measurements of Cabibbo-suppressed decays such as $Λ_c^+\to pπ^0$, and studies of $K_S^0-K_L^0$ asymmetries in $Λ_c^+$ decays.

We prove existence, uniqueness and structure results for complete noncompact 7-dimensional G2-holonomy metrics with ALC (asymptotically locally conical) asymptotics. We regard such spaces as G2-analogues of ALF gravitational instantons in 4-dimensional hyperkähler geometry. Our main results include the existence of a G2-analogue of the Atiyah-Hitchin metric in 4-dimensional hyperkähler geometry, the existence of a good moduli theory for ALC G2-holonomy metrics and rigidity results for ALC G2-metrics in terms of the symmetries of their asymptotic model. The analytic toolkit needed to prove all these results is a robust Fredholm theory for the natural geometric linear elliptic operators on ALC spaces. We provide a self-contained derivation of this Fredholm theory for arbitrary Riemannian manifolds with ALC asymptotics. Since our ALC Fredholm theory does not rely on imposing any holonomy reduction or curvature conditions it may also be of utility beyond the setting of ALC special holonomy metrics. As one such application of our general Fredholm theory we prove some Hodge-theoretic results on general ALC spaces.

Freeze-in of multi-component dark sectors is governed not only by the interaction with the thermal plasma, but also by their internal dynamics. Full thermalisation within the dark sector is not guaranteed, raising the question of impact of departures from local thermal equilibrium onto the evolution and ultimately relic abundance and momentum distribution of dark matter. In this work we explore this question in a minimal two-scalar model, which can give rise to observable signatures in indirect detection and long-lived particle searches at forward physics experiments. Focusing on the phenomenologically viable regions, we analyse the impact of non-thermal evolution on the dark matter abundance, finding deviations of up to an order of magnitude between the full phase-space treatment and the traditional number-density approach. Our results highlight the importance of phase-space level computation for accurate freeze-in predictions and further motivate dedicated numerical tools for studying the evolution of multi-component dark sectors at the phase space level.

We study the parametrization of the energy-momentum tensor for the case of a proton in momentum space in terms of gravitational transverse momentum-dependent distributions (TMDs). These gravitational TMDs are investigated with the inclusion of higher-twist contributions to predict the mechanical properties, specifically the transverse pressure and shear force distributions, along with the polarization-dependent $Π^q_S$ and $Π^q_A$ terms. The corresponding distributions are computed individually for both $u$ and $d$ quark flavors. The calculations have been performed in the light-cone framework using the spectator diquark model. A strong binding contribution to the transverse pressure is observed in the low-momentum space for both quark flavors of the proton.

The standard cosmological paradigm assumes that the inflaton field becomes dynamically negligible during the post-reheating evolution of the Universe. We demonstrate that this assumption fails for a broad class of inflationary models where the potential behaves as a monomial form $V(φ) \propto φ^k$ (with $k \ge 4$) around the minimum. In such scenarios, the effective inflaton mass depends on the field amplitude and vanishes asymptotically as the Universe expands. This vanishing-mass mechanism renders the inflaton kinematically accessible to the thermal plasma long after reheating, facilitating the regeneration of inflaton quanta through 1-to-2 decays and 2-to-2 scatterings of bath particles. This mechanism is quite generic and the coupling responsible for reheating can be constrained if the inflaton is overproduced, while the inflaton quanta can constitute dark matter in specific scenarios. Furthermore, if reheating occurs via the Standard Model Higgs portal, the process can be further constrained by big bang nucleosynthesis, cosmic microwave background, and colliders such as the LHC. This mechanism provides a new framework for probing post-inflationary reheating.

In this paper, we construct eight pairs of hexaquark currents to search the $Λ_cΣ_c$ and $\barΛ_cΣ_c$ dibaryon states via QCD sum rules. We show that the two currents of each pair are equivalent and we choose one of them to calculate the masses and pole residues of ground states. For either $Λ_cΣ_c$ or $\barΛ_cΣ_c$, the $J^P$ of the considered hexaquark currents are $0^-$, $0^+$, $1^+$ and $1^-$, respectively. We found three possible molecular states, they are $Λ_cΣ_c$ dibaryon with the $J^P=1^+$ and $\barΛ_cΣ_c$ dibaryons with the $J^P=0^-$ and $1^-$. The other five are unlikely to form the bound dibaryon states, and we assign them as the resonance states.

We propose an effective Hamiltonian formulation of quantum field theories using a Daubechies wavelet basis in position space. Combined with flow-equation methods of the similarity renormalization group (SRG), this approach provides an efficient framework for analyzing quantum field theories by reducing the dimensionality of the Hamiltonian and systematically decoupling degrees of freedom across scales. As an application, the free scalar field theory has been reformulated within this framework to calculate the low-lying energy spectrum of the theory. These basis elements are known to transform the free scalar field theory into a theory of coupled localized oscillators, each of which is labeled by a location and a resolution index. In this representation, the Hamiltonian is naturally organized into fixed-resolution blocks, alongside blocks associated with the interactions between different resolutions. To decouple the different resolution modes and obtain a block diagonalized Hamiltonian with each block associated with a fixed resolution, the flow equation approach of SRG is applied. Finally, we demonstrate that with increasing resolution, the low-energy spectrum can be extracted from the effective lowest-resolution block of the Hamiltonian, leading to a significant reduction in computational cost.

In low-energy supergravity treatment of type IIB superstring on general D-instanton wormhole profiles in the bulk, we obtain non-vanishing scalar two-point functions in addition to the vanishing $\langle τ^* τ^* \rangle$ that corresponds to the BPS amplitude detected by two D-instantons at their respective boundaries. This is exploited to show that the modes of broken local supersymmetry in the bulk deliver the fermionic (diagonal) modes on the boundaries through the deformation by the form of current-current two point functions propagating on the tree level cylinder geometry. Our treatment is generalizable to multi D-instanton cases and general Euclidean branes.

Ultralight dark matter searches widely assume that signals are monochromatic, with a single frequency set by the mass. This assumption is generally violated in the presence of field mixing, even when the constituent fields have similar frequencies. Instead, dark matter signals can exhibit a two-timescale structure with intrinsic slow modulation. We demonstrate that mixing between ultralight wave dark matter fields induces a parametric structure, leading to a scenario we refer to as wave-envelope dark matter, in which a slow-beating envelope emerges alongside the primary oscillation. This results in distinctive features such as slow modulation and characteristic sideband structures in the frequency spectrum, beyond the conventional monochromatic expectation. As a representative example, we briefly discuss implications for neutrino observables.

We investigate the behaviour of quantum fields in null-shifted Rindler wedges and analyse the particle spectra perceived by accelerated observers associated with these null deformations. Unlike the standard Unruh effect, our analysis compares two accelerated frames connected by a null displacement. We consider both massive scalar and Dirac fields, constructing their corresponding mode solutions in Rindler coordinates. Using normalised field expansions, we compute the Bogoliubov transformations between modes defined in the two null-shifted wedges. Our results demonstrate a fundamental breakdown of thermality: the presence of mass modifies the mode structure, rendering the characteristic exponential mixing of frequencies absent. This suggests that the massive field remains unexcited on this background, leading to a manifestly nonthermal response. These findings highlight that thermality in accelerated frames depends sensitively on the conformal symmetry of the field, which is broken by the introduction of a mass term.

We develop a rigid $\mathscr{N} =(1,1)$ superspace framework for spatially inhomogeneous impurity deformations in $D=1+1$ dimensions by embedding the impurity profile into a real background superfield (spurion). This spurionic completion provides a manifestly supersymmetric description at the level of the action and offers a systematic route to identify which inhomogeneous backgrounds preserve a nontrivial subset of supercharges. Focusing on static interface-type configurations, we determine the half-BPS condition on the spurion background and the corresponding supersymmetry projector. In the resulting half-BPS sector we derive the associated first-order BPS equation for static bosonic matter configurations and establish an exact Bogomol'nyi completion of the static energy, yielding a sharp bound saturated by BPS solutions. We further comment on how explicit coordinate dependence and derivative-dependent impurity couplings can obstruct the Bogomol'nyi structure, thereby motivating spurionic extensions that retain supersymmetric control over inhomogeneous deformations.

While conventional oscillation experiments measure neutrino mixing parameters with high precision, these measurements are strictly confined to sub-TeV scales. At higher energies, renormalization-group effects can cause these parameters to evolve with the transferred momentum, $Q$. High-energy and ultra-high-energy astrophysical neutrinos, spanning TeV to EeV energies, probe high values of $Q$ unreachable by conventional experiments, offering an unprecedented test of high-energy mixing. We use the flavor composition of these neutrinos -- the relative proportions of $ν_e$, $ν_μ$, and $ν_τ$ -- to constrain this evolution, both phenomenologically and within dimension-6 Standard Model Effective Field Theory. We account for astrophysical uncertainties -- an unavoidable requirement to obtain realistic results, even though this weakens the bounds. Although present IceCube measurements lack the sensitivity to detect this running, we forecast that upcoming multi-detector combinations will place unprecedented bounds on the high-energy evolution of neutrino mixing.

The Light Dark Matter eXperiment (LDMX) is an electron-beam fixed-target experiment primarily designed to achieve world-leading, model-independent sensitivity to sub-GeV dark matter particles. LDMX aims to identify dark sector particle production through the detection of events with substantial missing energy and momentum, a signature of invisible particles escaping detection. Beyond this primary objective, LDMX offers a complementary search strategy for long-lived, visibly decaying particles, such as dark photons and axion-like particles. We present the first detailed evaluation of the ability of LDMX to identify visibly decaying, long-lived particles that couple to electrons using a detailed simulation, based on the Geant4-toolkit, that incorporates realistic detection efficiencies and background levels. We demonstrate that LDMX can achieve a sensitivity that is competitive with other experiments that are currently running. The models explored in this paper are distinct and complementary to those probed in the LDMX flagship missing-momentum analysis. Through searching for both invisible dark matter and visibly decaying long-lived signatures, LDMX will significantly advance the search for light dark matter and provide a broad exploration of the sub-GeV dark sector.

We present a theoretical analysis of the one-loop effective potential of a self-interacting real scalar field in the presence of two parallel conducting plates with geometric roughness. Using WKB methods to evaluate the spectral density of the modified Laplace-Beltrami operator, together with contour integration within a $ζ$-function regularization scheme, we derive analytical expressions for the quantum corrections to the effective potential induced by perturbative boundary roughness and finite temperature. Furthermore, we compute explicit contributions to the Casimir energy and to the topological mass generation associated with the geometry.

We present the implementation of next-to-next-to-leading order (NNLO) electroweak (EW) virtual corrections at next-to-leading logarithmic (NLL) accuracy in the amplitude generator OpenLoops. The implementation covers the automated computation of processes involving massless fermions and transversely polarised vector bosons. For energies above the EW scale, logarithmic EW corrections are strongly enhanced in the tails of kinematic distributions of key LHC processes, reaching several tens of percent at NLO and several percent at NNLO. The two-loop implementation is validated against analytical results from the literature. We present phenomenological results for representative LHC processes and discuss the role of two-loop EW corrections in reducing theoretical uncertainties from missing higher-order contributions.

We investigate fermions on Carrollian manifolds. We complement previous intrinsic analysis by deriving Carrollian fermion actions from a relativistic Dirac theory via a systematic expansion in the speed of light ($c$). We then study relativistic fermions in light-cone coordinates and their connection to Carrollian fermions in one lower dimension. This follows from the recent observation that the Poincaré algebra, written in lightcone coordinates contains (two) co-dimension one Carroll sub-algebras. Our results establish a clear bridge between intrinsic Carrollian constructions, small $c$-expansion and light-cone dynamics. In the process, we understand why Carrollian fermions in $D$-dimensions have features that relate them to relativistic fermions in both $D$ and $(D+1)$ dimensions.

Axion-like particles (ALPs), the QCD axion, and dark photons in the MeV-GeV mass range are motivated by various dark matter models and the strong CP problem, and are ubiquitous in extensions of the Standard Model. A long-standing blind spot for experimental searches is the sub-100 MeV mass range, where the particle lifetime is too long to be constrained by prompt-decay collider searches yet too short to be reached by beam-dump experiments. We investigate and estimate the sensitivity of the Light Dark Matter eXperiment (LDMX) to such axions and dark photons, motivated by the clean environment in which these particles can be produced and by the near-target tracking capabilities of LDMX. With reasonable charged track and momentum reconstruction capabilities, we find that LDMX could close much of this low-mass blind spot for axions and dark photons.

We explore the possibility that neutrinos couple to an interacting sterile sector, providing a novel portal that generalizes the heavy neutral lepton portal to a composite setting. For a low confinement scale, high-energy neutrino beams can disintegrate into collimated sprays of hidden states, referred to as dark jets. This dynamics gives rise to two characteristic signatures in high energy neutrino beams. First, long-lived dark resonances can enhance the neutral-current to charged-current ratio. Second, shorter-lived dark states produced in neutrino neutral currents can produce single or multiple displaced vertices and even emerging jets, depending on the kinematics. These signals probe regions of parameter space beyond existing constraints from meson, electroweak, and Higgs decays, as well as from searches for displaced decays at beam dump experiments. We study these phenomena within broad classes of ultraviolet completions and identify scenarios in which high-energy neutrino beams provide leading sensitivity to neutrino compositeness. Such scenarios generically induce higher-dimensional contact interactions, which we classify and study alongside their complementary experimental signatures. Finally, we outline an experimental program spanning both the intensity and energy frontiers. Near-term neutrino facilities (DUNE, FPF) and running flavor experiments (LHCb, Belle II) can probe neutrino compositeness through neutrino disintegration into dark jets and displaced B-meson decays. Future colliders, particularly the Future Circular Collider (FCC-ee), will ultimately provide the strongest sensitivity to the compositeness scale via displaced Z decays.

The electric and magnetic dipole moments of the electron and of the muon provide stringent tests of the Standard Model and sensitive probes of new physics. By contrast, the corresponding dipole moments of the $τ$ lepton remain weakly constrained. This study explores the potential of future lepton colliders, focusing on the $e^+e^-$ Future Circular Collider and a multi-TeV muon collider, to probe $τ$ dipole moments. We consider multiple channels, including $\ell^+\ell^- \to τ^+τ^-$ ($\ell=e,μ$), associated Higgs production $μ^+μ^- \to τ^+τ^- H$, radiative Higgs decays $H \to τ^+τ^-γ$, and vector-boson scattering $\ell^+\ell^- \to \ell^+\ell^-τ^+τ^-$ and $μ^+μ^- \to \barνντ^+τ^-$. Our results show that these facilities are highly complementary and can extend existing bounds by several orders of magnitude.

The path-integral approach to the double well has long been limited by the dilute instanton gas approximation. We show that if the finite Euclidean-time structure is taken seriously by using exact saddles, the dilute gas can be sidestepped, allowing the partition function and energy levels to be computed systematically. At each instanton order, the full resurgent structure -- which saddles contribute, what asymptotic growth is expected and how ambiguities cancel -- is encoded in a finite-dimensional Picard--Lefschetz contour integral over the quasi-zero modes with a clear geometric interpretation. Working at finite $T$ is essential: the dilute instanton gas can only access the ground-state splitting, whereas the exact finite-$T$ computation systematically produces the non-perturbative energy splittings for all excited states, including their full dependence on the level number. The key ingredients -- Weierstrass elliptic functions for the saddles, Lamé operators for the fluctuations and Picard--Fuchs equations for the periods -- form a coherent mathematical framework that both overlaps and complements that of Exact WKB.

Higgs final states are prime targets in the search for physics beyond the Standard Model. In the conventional picture, $SU(2)$ symmetry together with the Goldstone Equivalence Theorem correlates Higgs and gauge-boson final states, implying comparable sensitivity in channels such as $hh$, $ZZ$, and $WW$ in searches for heavy resonances. In this work, we identify a mechanism to parametrically violate this expectation. We show that higher-order Higgs couplings can induce an electroweak-symmetry-breaking enhancement that selectively amplifies Higgs-rich final states, allowing them to become the leading discovery channels of new resonances. For scalar resonances, this can make di-Higgs the dominant bosonic signal. For resonance masses higher than a couple of TeV, it also opens resonant tri-Higgs and four-Higgs channels as well-motivated search targets. The same underlying mechanism extends to heavy fermionic and vector resonances, where it can similarly enhance channels such as $ht$, $Zh$, and $γh$. We present this framework in effective field theory, demonstrate possible UV completions, and discuss its implications for collider searches.

In the framework of Soft de Sitter Effective Theory (SdSET), the Fokker-Planck equation for the late-time dynamics of the massless minimally coupled scalar field and its extension to the Kramers-Moyal equation are obtained from operator mixing of composite operators of the effective superhorizon field. We construct the formalism for composite-operator renormalisation, mixing and matching in dimensional regularisation, allowing for computations beyond the leading order. The general formalism is illustrated in free SdSET, which already features non-trivial structures including the well-known diffusion coefficient for stochastic inflation. As explicit examples in the interacting theory, we renormalise the one-loop bispectrum and the two-loop one-point function of the composite operator $\varphi_+^2$, and match them onto their full-theory counterparts. These results allow us to determine the next-to-leading order (two-loop) correction to the diffusion term of the Fokker-Planck equation of stochastic inflation for the first time.

We use the amplitude formulation of the SMEFT to introduce a spurion analysis of the SMEFT low-energy amplitudes in terms of the Higgs VEV. Each SMEFT contact-term is given as a sum of a few spurion structures, whose number depends on the electroweak charges of the external legs. The coefficients of these structures involve singlet combinations of Higgses from higher-order SMEFT contributions. We use this to derive the spurion expansions of the W- and Z-boson masses and mixing, and their three-point couplings to fermions. The textures of these couplings are saturated by the dimension-eight SMEFT. Our analysis can be generalized to higher-point amplitudes and nonzero Yukawa couplings.

Collinear factorization and the leading-twist operator product expansion (OPE) in perturbative QCD express suitably inclusive observables in scale-separated kinematics as composites of perturbative short-distance coefficients with universal long-distance non-perturbative correlators such as parton distribution functions (PDFs), up to controlled power corrections. A persistent structural feature is \emph{presentation non-uniqueness}: coefficients and correlators are not individually physical, but are defined only up to finite factorization-scheme redefinitions induced by collinear subtractions and renormalized-operator mixing. We formalize this redundancy categorically by introducing an \emph{interface algebra object} encoding admissible finite collinear counterterms/mixing kernels and by organizing coefficient data and hadronic data as right/left modules over this algebra in a symmetric monoidal category encoding the chosen recomposition calculus. Our main result, the \emph{Core Representation Theorem}, identifies the universal scheme-invariant carrier: the functor of balanced (scheme-invariant) pairings is represented by the relative tensor product $C\otimes_A f$, which is terminal among all quotients of the naive composite $C\otimes f$ that preserve scheme-invariant semantics. Finally, we show how standard physics inputs (symmetry constraints, locality/OPE, and a stated accuracy truncation) canonically induce the interface algebra and module structures, and we prove a minimal closure principle for completing a generating set of long-distance operators/correlators to an $A$-stable sector.

A search for a resonance heavier than 250 GeV decaying into four leptons via two intermediate bosons is presented. The search uses proton-proton collision data at $\sqrt{s}$ = 13 TeV collected by the CMS experiment, corresponding to an integrated luminosity of 138 fb$^{-1}$. Novel techniques are used to enhance the sensitivity to a collimated pair of dileptons reconstructed as a single merged object, resulting from the decay of an intermediate light boson. No significant excess of data over the background predictions is observed. Upper limits are set on the production cross section for a four-lepton resonance, including the previously unexplored phase space at the LHC with a dilepton mass of 0.4$-$15 GeV.

A partial wave analysis of the process $ψ(2S)\rightarrowγχ_{c1}, χ_{c1}\rightarrowπ^+π^-η^{\prime}$ is performed using $(2712.4\pm14.3)\times10^{6}$ $ψ(2S)$ events collected with the BESIII detector. An isovector state with exotic quantum numbers $J^{PC}=1^{-+}$, denoted as $π_{1}(1600)$, is observed for the first time in the charmonium decay of $χ_{c1}\rightarrowπ_{1}^{\pm}(1600)π^{\mp}$, $π_{1}^{\pm}(1600)\rightarrowπ^{\pm}η^{\prime}$ with a statistical significance over $21σ$. Its mass and width are determined to be $1828 \pm 8 ({\rm stat})^{+11}_{-33}({\rm syst})~\mathrm{MeV}/c^2$ and $638 \pm 26 ({\rm stat})^{+35}_{-86}({\rm syst})~\mathrm{MeV}$, respectively, using a relativistic Breit-Wigner function with a mass-dependent width. The corresponding product of branching fractions is determined to be $\mathcal{B}\left[χ_{c1}\rightarrowπ_{1}(1600)^{\pm}π^{\mp} \right] \times \mathcal{B}\left[π_{1}(1600)^{\pm}\rightarrowπ^{\pm}η^{\prime}\right] = \left( 4.30 \pm 0.14 ({\rm stat})^{+1.04}_{-1.03}({\rm syst})~ \right) \times 10^{-4}$.

We searched for proton decay via $p \to e^{+}π^{0}π^{0}$ and $p \to μ^{+}π^{0}π^{0}$ in 0.401 megaton-years of data collected in all pure water detector phases of Super-Kamiokande (SK) I-V. A theoretical study predicts proton decay rates without assuming a particular grand unified theory and suggests that three-body proton decays involving two pions can have decay rates comparable to those of $p \to e^{+}π^{0}$ and $p \to μ^{+}π^{0}$. This is the first search for proton decay into a charged anti-lepton and two neutral pions in SK. One data candidate event was found for each of the two decay modes, which is consistent with the expected atmospheric neutrino background. We set lower limits on the lifetime of $τ/B(p \to e^{+}π^{0}π^{0}) > 7.2 \times 10^{33}$ years and $τ/B(p \to μ^{+}π^{0}π^{0}) > 4.5 \times 10^{33}$ years at 90 $\%$ confidence level. These limits are more than one order of magnitude higher than those of the previous experiment.

We perform the first amplitude analysis of the singly Cabibbo-suppressed decays $D^+ \to π^+ π^{+(0)} π^{-(0)} η$, using $e^+e^-$ collision data collected with the BESIII detector at the center-of-mass energy of 3.773\,GeV, corresponding to an integrated luminosity of 20.3 $\rm{fb}^{-1}$. The absolute branching fractions of the $D^+ \to π^+ π^+ π^- η$ and $D^+ \to π^+ π^0 π^0 η$ decays are measured to be $(3.20\pm0.06_{\text{stat.}}\pm0.03_{\text{syst.}})\times 10^{-3}$ and $(2.43 \pm 0.11_{\text{stat.}} \pm 0.04_{\text{syst.}}) \times 10^{-3}$, respectively. % , both achieving three times better precision than the current PDG values. The decay process $D^{+}\to a_0(980)^{+}f_0(500)$ is observed for the first time with an unexpectedly large branching fraction. Moreover, we observe the decays $D^+ \to a_0(980)^{+(0)} ρ(770)^{0(+)}$ and measure the ratio $r_{+/0} \equiv \frac{\mathcal{B}(D^+ \to a_0(980)^+ ρ(770)^0)}{\mathcal{B}(D^+ \to a_0(980)^0 ρ(770)^+)}$ for the first time to be $0.55\pm0.08_{\text{stat.}}\pm0.05_{\text{syst.}}$. These results offer a novel insight into our comprehension of the nature of the $a_0(980)$ and $f_0(500)$ states.

We investigate the stellar population properties of pseudo-bulges in barred galaxies drawn from the Sloan Digital Sky Survey (SDSS DR7) to assess how bars regulate central star formation and secular evolution. Our sample comprises barred spiral and barred lenticular (S0) galaxies with reliable spectroscopic indices obtained from multicomponent structural decompositions. Stellar ages and recent star formation are traced using the 4000 Å break strength ($D_{n}(4000)$) and the Balmer absorption index ($Hδ_{A}$), complemented by bulge, bar, and disc colours. Barred spirals show a clear bimodality in $D_{n}(4000)$, with peaks at $D_{n}(4000)\sim1.3$ and $\sim1.8$. Low-$D_{n}(4000)$ pseudo-bulges exhibit strong $Hδ_{A}$ absorption, blue colours, and high specific star-formation rates, indicating young, actively growing centres. High-$D_{n}(4000)$ systems instead show weak $Hδ_{A}$, red colours, and low sSFR, consistent with older, quenched pseudo-bulges. Barred S0s display an old-bulge-dominated distribution, suggesting that gas-poor barred spirals transition into S0s following disc-wide quenching. We also find elevated AGN incidence among old pseudo-bulges. These trends support a scenario in which bars funnel gas inward to build pseudo-bulges and later suppress central star formation by depleting or stabilising the inflow. IFU observations show that bars assemble cold nuclear discs that age and quench over time, while high-redshift imaging confirms that bars are already present at $z\sim4$, implying that this evolutionary cycle operates across cosmic time. The strong correspondence between stellar age, colour, and structure indicates that bar-driven secular evolution governs both the growth and quenching of central components, linking blue barred spirals to red S0 galaxies.

Bars play an integral role in regulating star formation (SF) in spiral galaxies, from triggering central starbursts to driving quenching. The diverse SF morphologies observed in local barred galaxies reflect different evolutionary stages of the bar, motivating studies across these stages. Here we study 12 nearby barred galaxies (z=0.01-0.06) identified as centrally quenched galaxies (having extended star-forming discs but quenched inner regions) by leveraging the differences in SFRs between the MPA-JHU and GSWLC catalogues. However, they exhibit residual central emission in the SDSS 3" fibre spectral region. Emission line analysis shows that this emission originates from either ongoing SF or LINER-like activity, suggesting diverse central ionization mechanisms. Using spatially resolved UV-optical colour maps from SDSS (r-band) and GALEX (FUV and NUV band) imaging data, we find that discs are star-forming and bluer in colour (NUV-r < 4 mag) while the bulge and bar regions are systematically redder (NUV-r > 4 mag) and dominated by older stellar populations. The NUV-r radial colour profiles show a clear transition from red to blue colours at the bar end with a corresponding median stellar age of ~ 1 Gyr. Compared to fully centrally quenched barred galaxies from our earlier work which lack SDSS fibre emission, these galaxies remain systematically bluer at similar radii, despite showing NUV-r > 4 mag inside the bar suggesting an intermediate stage in bar-driven quenching. We also estimate black hole masses associated with kinetic-mode AGN feedback and find them below the threshold (logM_BH < 8.0). Adding this with the presence of pseudo bulges, our results support bars as the primary drivers of quenching, with these galaxies representing an evolutionary phase just before their inner regions are completely quenched.

Binary interactions during the AGB phase can lead to the formation of chemically peculiar stars with overabundances of s-process elements. Only a few hundreds of these stars have been subject to detailed chemical or dynamical studies. This work aims at compiling a systematic sample of s-process-polluted candidates using GALAH DR4. We also want to compare their properties with those of confirmed s-polluted stars to have stronger evidence of their nature. GALAH DR4 uses neural networks and automatic spectral analysis methods as well as data of a lower spectral resolution than normally used to characterise these objects. Because of this, we built a validation sample, for which we obtained UVES@VLT and HERMES@Mercator high-resolution spectra. We compare our stellar parameters and abundances with those of the survey and use this validation to define the thresholds that a star in GALAH DR4 must pass to be flagged as a good s-process-rich candidate. Based on our comparisons, we define thresholds on [s/Fe], [Y/Fe], [Zr/Fe], [Ba/Fe], and [La/Fe]. We identified 1073 stars in GALAH DR4 that are good candidates to be s-process polluted stars, covering a broad parameter space. They share many similarities with the samples of confirmed s-rich stars, especially their ratios of heavy over light s-elements ([hs/ls]), which strengthen our confidence in the purity of the sample. We find that only 7% of the candidates have measured orbital periods and eccentricities, limiting for now a full comparison with confirmed Ba and related stars. However, their binary fraction is, as expected, higher than the one we found for the full GALAH DR4 catalogue. Our sample of candidates is almost five times larger than the number of currently confirmed polluted stars. This and the fact that it has been homogeneously treated by GALAH open very interesting avenues to confront nucleosynthesis and binary evolution models.

We apply Random Forest and XGBoost machine learning algorithms to determine which galaxy properties most effectively predict star formation and quenching in simulated galaxies. Using spatially-resolved data from approximately 63,000 annular bins across 6,189 TNG100 galaxies, we train classification models to predict quenching states and regression models to predict star formation rate surface densities. Despite their different algorithmic approaches, both methods produce consistent feature importance rankings, with XGBoost distributing importance more evenly among correlated features. For central galaxies and high-mass satellites, black hole mass dominates quenching predictions, consistent with quenching via active galactic nuclei (AGN) feedback. Classification of low-mass satellites shows overwhelming importance for halo mass, indicating environmental quenching. Star formation predictions are dominated by local stellar mass surface density across all star-forming galaxy types, confirming that active star formation is a local process while quenching is driven by global properties.

We present a summary of gravitational-wave (GW) follow-up using the Las Cumbres Observatory global network of telescopes during the third (O3) and fourth (O4) observing runs of the GW detectors. As in O2, we implemented the Gehrels et al. 2016 galaxy-targeted strategy. Here we test its efficacy in O3 and O4 and analyze the Las Cumbres Observatory response time and depth for nine GW alerts that showed a possibility of having an electromagnetic counterpart (GW190425, GW190426_152155, S190510g, GW190728_064510, GW190814, S190822c, GW191216_213338, S240422ed and S250206dm). We find that Las Cumbres Observatory is able to begin observations in response to GW alerts within minutes of the alert, with the observations being deep enough to detect possible GW170817-like kilonovae out to a median distance of 250 Mpc. In this sense a global rapid-response network of telescopes like Las Cumbres is an excellent GW follow-up facility. However, the galaxy-targeted follow-up strategy was much less efficient in O3 and O4 than originally predicted, given the larger than assumed GW localizations. We conclude that coordination between various facilities to include both wide-field and rapid-response capabilities is required to achieve efficient and comprehensive follow-up of GW events.

The Compton Spectrometer and Imager is a NASA satellite mission under development that will survey the entire sky in the 0.2-5 MeV range using a wide-field germanium detector array, surrounded on the sides and bottom by active shields (the Anticoincidence Subsystem, ACS). The ACS aims to suppress and monitor background events, as well as detect transient sources, such as Gamma-Ray Bursts (GRBs), through its onboard triggering algorithm. The data related to GRBs are sent to the ground and analyzed by an automated pipeline to localize the GRBs and share their positions with the community. In this work, we present a brief GRB localization method based on ACS data, utilizing deep learning (DL) techniques, which can estimate the 90\% confidence region, including cases where it is split into multiple areas. To address this, we developed a neural network classifier that predicts the GRB location as a probability distribution across the sky map following the HEALPix framework. The distribution can be used to compute the 90\% confidence regions. Future work will compare this DL-based localization approach with classical methods such as $χ^2$ fitting and Maximum Likelihood Estimation.

Omega Centauri is the most massive and chemically complex multi-population globular cluster with a wide metallicity range that has been extensively studied photometrically and spectroscopically. Using the wide metallicity range of omega Cen, HST photometry (F275W, F336W, F435W, F625W), and MUSE spectroscopy ([M/H]), we derive [M/H]- and M_{F625W}-dependent stellar loci to estimate photometric metallicities from HST colors. Our tests yield metallicity precisions of 0.10\,dex for giants and 0.22\,dex for fainter dwarfs. We construct a photometric metallicity catalog from simultaneous F336W, F435W, and F625W observations (plus F275W where available), containing 20,778 giants and 346,546 dwarfs. A subsample of 20,533 giants is used to study the spatial metallicity distribution and gradient. We find no significant metallicity gradient within the half-light radius, consistent with previous work. Moreover, the previously reported ring-like structure is less pronounced in our data, and no physically significant, irregular two-dimensional metallicity pattern is detected, indicating that the stellar subpopulations are well mixed within the half-light radius. Our catalog significantly extends the metallicity sample of omega Cen, and this approach can be applied to other HST data to estimate photometric metallicities.

We map the three-dimensional structure and large-scale kinematics of the young stellar populations in the G352 giant molecular cloud (GMC) complex. In radio and infrared images, G352 appears as long filament extending ~$3^{\circ}$ (~150 pc) parallel to the Galactic midplane. It connects the NGC 6357 and NGC 6334 giant H II regions and the GM1-24 compact H II region. We identify 1727 stellar members of G352 via matching large catalogs of Chandra X-ray point sources and Spitzer mid-infrared excess sources to the Gaia DR3 astrometric catalog. Our catalog of 11,470 X-ray point sources ranks among the three largest contiguous X-ray survey datasets ever assembled for a massive star-forming complex. We revise the mean heliocentric distance of G352 to $1670\pm 80$ pc, with the median parallaxes of seven constituent groups exhibiting a trend toward increasing distance with decreasing Galactic longitude. We identify two foreground stellar groups superimposed on NGC 6357 that may belong to the Sag OB4 association. The three massive clusters in NGC 6357 exhibit peculiar velocities that trail Galactic circular motion by ${\sim}8$ km/s, while the stars associated with NGC 6334 are more consistent with a circular orbit. GM1-24 has a distinct proper motion and smaller parallax compared to NGC 6334. The steep pitch angle of the GMC filament into the sky appears inconsistent with a spiral arm. The various stellar groups are not gravitationally bound to each other, making G352 a proto-OB association.

Variations in the Faraday rotation measure (RM) of repeating fast radio bursts (FRBs) provide critical diagnostics of the dynamically evolving magneto-ionic environments surrounding their progenitors. Sudden, transient ``RM flares'' can trace the passage of discrete magneto-ionic structures, such as stellar coronal mass ejections from the companion or other dense plasma clumps, across the line of sight. However, identifying these rare events is difficult because RM evolution manifests a wide range of complex behaviors, from smooth, long-term trends to chaotic stochasticity, further complicated by highly non-uniform temporal sampling. This complexity makes it a non-trivial challenge to distinguish localized physical flares from intrinsic environmental volatility. We present a generalized algorithmic framework that establishes a statistically robust methodology for the automated detection and characterization of RM flares. By objectively isolating discrete transient perturbations from quiescent backgrounds, this pipeline enables the first uniform census of environmental variability across the FRB population. Applying this framework to 15 repeating FRBs, we find that high-confidence RM flares are remarkably rare, with FRB 20220529A being the only source to exhibit a statistically significant event under standardized parameters. Other active repeaters instead display high-level intrinsic fluctuations or secular evolution. This work provides a rigorous foundation for distinguishing between different modes of local plasma dynamics, offering a crucial diagnostic tool for identifying the diverse progenitor systems and local environments of FRBs.

Mergers and interactions can significantly affect the morphological and dynamical properties of galaxies, however the impact of mergers on turbulence at $z > 1$ has not been observationally constrained. In this work we use the interaction strength parameter $Q_P$ to identify likely interacting and isolated galaxies at cosmic noon ($z \sim 1-2$) within the KMOS\textsuperscript{3D} integral field spectroscopy survey, utilising redshifts from the 3D-HST, CANDELS and UVCANDELS surveys. For $186$ galaxies, we measure deconvolved H$α$ kinematics, including velocity dispersion, using a spatially non-parametric approach to account for observational effects in the dynamically diverse range of galaxies. We compare offsets in H$α$ flux, star formation rate (SFR), dust attenuations, and velocity dispersion of likely interacting galaxies to isolated control galaxies matched in mass and lookback time. We find increased H$α$ fluxes and SFRs in the likely interacting sample at the level of $\sim 0.1$ dex, a similar enhancement to studies of local pairs. In contrast, we find no significant increase in the level of velocity dispersion in interacting galaxies compared to their controls. The lack of increase in dispersion may reflect a combination of physical and observational factors, including limits to increasing turbulent motions in an already turbulent medium and spectral resolution limits.

The X-ray spectra of neutron stars with moderate magnetic fields ($B\sim 10^{12}$ G) in high-mass X-ray binaries (HMXBs) at low X-ray luminosities ($L_\mathrm{X}\lesssim 10^{35}$ erg/s) are characterized by a double humped shape. This shape has been explained either as the radiation from a two-temperature magnetized atmosphere, where thermal radiation dominates at soft X-rays below about 10 keV, and cyclotron radiation with an imprinted cyclotron line dominates at high energies, or by the complex redistribution of primary X-rays in a structured atmosphere. The theoretical explanations of the double humped structure predict the spectra to depend on the magnetic field. We aim to connect the model predictions with observations. We analyzed archival NuSTAR observations of four HMXBs consisting of a neutron star and a Be star (BeXRBs), with known magnetic fields at luminosities low enough to show the characteristic double-hump spectrum. We modeled these spectra empirically and derived a relation between the energy of the intersection of the two humps and the magnetic field strength. In a second step, we tested whether this correlation is supported by fitting synthetic spectra simulated with the physically self-consistent polcap model. We find a linear correlation between the magnetic field strength and the intersection energy for the real BeXRB NuSTAR spectra and polcap-based simulated NuSTAR spectra alike. The effect of the magnetic field on spectral formation results in an observable correlation between the field strength and spectral shape. This derived positive correlation between intersection energy and magnetic field strength also allowed us to roughly estimate the magnetic field strength. Additional observations of XRBs and dedicated modeling efforts will be necessary to determine whether this approach is valid beyond the B-field range that was tested in this work.

We present the study of an X-ray selected sample of active galactic nuclei (AGN) in pairs at projected spatial separations 1 <$ r_p$/kpc < 100 at z < 0.1, using XMM-Newton and Chandra data. The pair sample is derived from an initial pool of approximately 2,000 X-ray-selected AGN, and is composed of both AGN-AGN pairs (so called dual AGN) and AGN-galaxy pairs. From this selection, we find that approximately 10% of AGN reside in pairs, and about 4% are paired with another AGN. We performed a detailed X-ray and SDSS optical spectral analysis for AGN in duals and X-ray analysis for AGN in AGN-galaxy pairs, to characterise their absorption properties and investigate the possible triggering mechanisms. We then investigated how obscuration, luminosity, and Eddington ratio depend on projected separation $r_p$. Amongst all AGN in pairs, we found that ~55% are obscured (with hydrogen column density $N_H$ > $10^{22}$ cm$^{-2}$), amongst which ~6% are Compton-thick ($N_H$ > $10^{24}$ cm$^{-2}$). The fraction of absorbed AGN is significantly higher in late-stage mergers ($r_p$ < 30 kpc) compared to early-stage mergers ($r_p$ > 60 kpc). Amongst the AGN in pairs, we also observed an average excess of AGN pairs with respect to a control sample of inactive galaxies in pairs, and that such excess significantly increases with decreasing $r_p$ only for obscured AGN. Finally, in dual-AGN systems, both the bolometric luminosity and the Eddington ratio of the less massive black hole in the pair increase as the separation decreases. These findings suggest that mergers may have an important role in triggering AGN accretion and activity.

The gravo-turbulent fragmentation of the interstellar medium is expected to create a hierarchical cascade of cloud structures, crossing the scales from core to disk. We aim to predict how the currently observed top-heavy core mass function (CMF) in the massive protocluster W43-MM1 evolves due to core subfragmentation. We used the getsf algorithm to extract sources in five ALMA images of W43-MM1 at 3 mm, with a spatial resolution ranging from 14 kau to 270 au. Then, we applied FAMILY, a graph-theory-based analysis tool, to create and characterize networks of nested sources in W43-MM1. We compared the hierarchical fragmentation cascade of W43-MM1 to those measured in the NGC 2264 protocluster and in synthetic images of an Orion-like protocluster simulated by magneto-hydrodynamical calculations. Assuming self-similarity, we measure a small fractality index of mathcal F3D =1.19+/-0.10 in W43-MM1, which means that, on average, a cloud structure will fragment into only 1.19 fragments each time the physical scale decreases by a factor of two. We estimate an imbalanced mass partition between siblings, with 2/3 of the mass of siblings at a given scale belonging to the dominant sibling. The mass transfer efficiency, computed from one physical scale to another, is high and corresponds to a core formation efficiency (CFE) from 2400 au to 200 au of ~16%. Based on the fractality and efficiency values measured in W43-MM1, the gravo-turbulent model by Thomasson et al. predicts that its fragmentation below ~14 kau is not driven by turbulence but by gravity. Using these parameters and the measured mass partition, we demonstrate that the fragment mass function, from which the the initial mass function (IMF) emerges, has a high-mass end which remains top-heavy. Therefore, core subfragmentation in W43-MM1, and perhaps more broadly in massive Galactic protoclusters, plays a minimal role in the IMF origin.

Energetic particles are ubiquitous in space and astrophysical plasmas, and interplanetary shocks are widely regarded as one of the main particle accelerators in the heliosphere. Indeed, in-situ measurements typically show that energetic particle fluxes peak at the shock, indicating a local acceleration process. Furthermore, the time profile of energetic particle fluxes is highly influenced by particle transport properties upstream and downstream of the shock. By advancing previous numerical test-particle models that simulate the transport of monoenergetic particles around an infinite planar shock, in this work we add the acceleration of such particles via energy gains at each shock crossing, in a first-order Fermi-type mechanism. Moreover, the acceleration of a 70 keV particle population, namely the seed population, is reproduced by integrating a Langevin-type equation upstream and downstream of an infinite planar shock. Particles can diffuse in the simulation box via random "kicks", which belong either to a Gaussian distribution (normal diffusion) or to a Levy distribution (superdiffusion). We perform several simulations by varying the parameters of the model. The particle energy spectra in both diffusive and superdiffusive simulations are in remarkable agreement with the theoretical predictions.The output energetic particle densities have been compared with those observed by the ACE spacecraft during an interplanetary shock crossing on December 14, 2006. We show not only that particle fluxes in different energy bins reproduce very well the observed ones upstream and downstream when superdiffusion is at work, but also that anomalous, superdiffusive transport speeds up the acceleration process and leads to values of particle energies consistent with observations.

High-fidelity galaxy mocks are crucial for validating analysis pipelines and for cosmological inference. In this context, the Science Pipeline at PIC (SciPIC) is a pipeline specifically designed for the fast generation of synthetic galaxy catalogues from the halo properties identified in cosmological simulations. SciPIC delivers galaxy catalogues that aim to reproduce the observed luminosity function and clustering above a given flux detection limit over a wide redshift range. In this work, we introduce SciPICal, an automated pipeline that calibrates the parameters that set the main mock galaxy properties, namely number density, luminosities, colours, and positions. The pipeline is applied to the Euclid Flagship 2 Wide and Deep halo catalogues, specifically built to support the \textit{Euclid} wide and deep surveys. Compared to the recently released Flagship 2 Wide mock, our calibrated version improves the clustering predictions by approximately 50\% based on chi-squared values. Furthermore, we produce the Euclid Deep mock catalogue, which reaches up to $z = 10$ by populating a light-cone and a complementary snapshot at $z = 0$. We validate these catalogues using measurements from spectroscopic and photometric galaxy surveys, as well as with results from a hydrodynamical simulation. The obtained good agreement (within $15\%$ for most of the samples) in the clustering predictions across the different galaxy samples considered, validates our calibration strategy and demonstrates the strong predictive power of the generated mocks. This pipeline will allow us to improve the methodology applied in assigning the galaxy properties and ensures that the galaxy mocks remain up-to-date by incorporating constraints from upcoming observational data in the calibration procedure.

We study global transonic solution for a relativistic, magnetized, viscous advective accretion flow around a rotating black hole, incorporating the effects of mass and angular momentum loss through winds. Our model considers dominant toroidal magnetic fields with synchrotron radiation as the primary cooling mechanism. To self-consistently model mass loss, the mass accretion rate is prescribed to decrease inward as a power-law with disk radius. With this, we solve the governing equations that describe the accretion flows in presence of winds and obtain the flow structure in terms of the inflow parameters (energy $\mathcal{E}$, angular momentum $λ$, plasma-$β$, accretion rate $\dot{m}$, and viscosity $α_{\rm B}$), the wind parameters ($p$, governing mass loss; and $l$, governing angular momentum transport by winds), and the black hole spin ($a_{\rm k}$). Our analysis reveals that winds substantially modify the accretion flow leading to a significant decrease in disk luminosity. We specifically identify global solutions that admit standing shocks and find that winds profoundly alter shock properties, such as the shock radius ($x_{\rm s}$), compression ratio ($R$), and shock strength ($S$). Furthermore, we determine the critical wind parameter $p^{\rm crit}$ beyond which steady shock solutions cease to exist. We demonstrate that increased viscosity and strong angular momentum extraction by winds lead to reduce $p^{\rm crit}$. These findings evidently highlight a complex interplay between viscosity and winds in governing the dynamics of shock formation in accretion disks.

Using the first data release of the Five-hundred-meter Aperture Spherical radio Telescope (FAST) All-Sky HI survey (FASHI), we compile a catalogue of 70 dark galaxy candidates (DGCs) within 50 Mpc. We select DGCs without an identified optical counterpart at a limiting g-band magnitude of ~ 28 mag arcsec^-2 in the DESI Legacy Survey, using both automatic cross-checking with optical catalogues and visual inspection of the colour images. After validating our DGCs, excluding potential spurious detections, issues in the registered position of the HI sources, and possible Radio Frequency Interferences (RFIs), we analyse their distribution over the surveyed sky, HI mass, linewidths, and inferred distance. They appear evenly distributed across the surveyed area, with no apparent bias to isolation. We did not find any DGC within the Local Volume (11 Mpc) in the sky surveyed by this first release of FASHI. We compare the observed properties of DGCs with those of galaxies with optical counterparts, finding that DGCs tend to have higher linewidths for a given HI mass. We discuss our DGCs in light of theoretical works, and compare them with other observational samples from previous HI surveys. This work presents a catalogue of dark galaxy candidates, which can serve as a basis for follow-up studies.

The 2023/24 NICER monitoring campaign of the 7 Crab bright black hole X-ray binary Swift J1727.8-1613 covered the outburst in almost all accretion states. High-quality data are available in the high-Eddington-fraction hard-intermediate state, hard-to-soft transition, the soft state, and the poorly studied back-transition to the dim hard state, making it an ideal dataset to compare the accretion flow at vastly different accretion rates. We apply disk continuum fitting techniques to investigate the evolution of the inner disk radius throughout the outburst. Taking a temperature-dependent color-correction factor into account, we see evolution of the disk inner radius by a factor of a few comparing the hard states to the thermal/soft state. We tentatively detect an onset of disk truncation in the soft-to-hard transition, right after the source leaves the soft state. After accounting for model systematics, we find the disk to be more truncated in the high-luminosity bright hard state compared to the low-luminosity dim hard state.

Ionization balance in the intergalactic medium (IGM) is central to the interpretation of quasar absorption spectra, linking observed ionic columns to the underlying gas density, temperature, metallicity, and ionizing radiation field. Because ionization, recombination, and cooling timescales can be comparable to the timescales over which the ultraviolet background (UVB) and gas thermodynamic state evolve, ion populations may retain a strong memory of their past history. To this end, we present a fast, metals-inclusive, zero-dimensional framework for modeling the redshift evolution of the IGM. The model follows the coupled thermal and ionization evolution of a Lagrangian gas parcel in a redshift-dependent UVB, solving stiff, time-dependent rate equations for H, He, and 107 metal ions while self-consistently evolving the temperature through photoheating and standard cooling processes. We validate the framework against full three-dimensional hydrodynamical non-equilibrium calculations and find that it reproduces the thermal and ionization histories of the IGM with good accuracy over a wide redshift range, including the heating associated with $\rm He_{\,\rm II}$ reionization. As an application, we predict the cosmic $\rm C_{\,\rm IV}$ density parameter, $Ω_{\rm CIV}$, and use it to infer the origin of metal ions in the IGM and the corresponding metallicities from observational measurements, obtaining values broadly consistent with literature constraints. The framework is well suited for rapid parameter studies of how reionization timing, UVB spectral hardness, self-shielding, and UVB inhomogeneity shape the thermal and ionization history of the IGM and the resulting metal-line observables.

In the context of the hierarchical formation of galaxies, we investigated the role played by mergers in shaping the scale relations of galaxies, that is the projections of their Fundamental Plane onto the \IeRe, \IeSig, \MRa\ and \Lsig\ planes. To this aim, we developed a simple model of multiple dry mergers among galaxies by suitably combing the formalism and properties of the so-called infall models of galaxy formation and evolution with the formalism of the scalar Virial Theorem. In this context, we mimicked the hierarchical formation of galaxies and generated simple models of galaxies undergoing a number mergers in the course of their evolution. The results are used to interpret the large scale simulations and the companion scale relations from observational and theoretical perspectives. The aim is to interpret the observational data of the MANGA and WINGS samples and the results of theoretical detailed numerical cosmo-hydro-dynamical simulations, such as Illustris-TNG100. In this context, we derived the above scale relations for our theoretical models and compared them with the observational counterparts from the MANGA and WINGS database, (and indirectly the large scale simulations of Illustris-TNG100). The multiple dry merging mechanism is able to explain all the main characteristics of the observed scale relations of galaxies, such as slopes, scatters, curvatures and zones of exclusion. The distribution of galaxies in these planes is continuously changing across time because of the merging activity and other physical processes, such as star formation, quenching, energy feedback, and so forth.} The precision of the present simple merger theory is comparable with that obtained by the modern cosmo-hydro-dynamical simulations, with the advantage of providing a rapid exploratory response on the consequences engendered by different physical effects.

The cosmic vorticity field, an essential tracer of nonlinear structure formation, has remained observationally inaccessible because transverse galaxy motions are difficult to measure and analytic models struggle to capture shell-crossing. Here we report an empirical reconstruction of this field by applying an artificial intelligence framework trained on simulations of the concordance LambdaCDM model to Sloan Digital Sky Survey galaxies. The recovered three-dimensional velocity and vorticity fields reveal coherent vortical structures, including spiral flows in clusters, filaments, and voids, and the cosmic web inferred from vorticity closely matches that derived from density segmentation. The power spectra of the reconstructed velocity and vorticity fields agree statistically with LambdaCDM predictions, and the inferred velocity field effectively removes redshift-space distortions, yielding an almost isotropic clustering signal. These converging lines of evidence, obtained from an independent perspective, reinforce the concordance cosmological model. By closing a long-standing observational gap, our results highlight the potential of AI-driven reconstruction to access otherwise unobservable quantities and to address fundamental questions in cosmology and galaxy formation.

We present the maximum ejecta mass $(M_{\rm ej})_{\rm max}$ and the maximum ratio of ejecta mass and accreted mass $(M_{\rm ej}/M_{\rm acc})_{\rm max}$ of a nova for various white dwarf (WD) masses ($M_{\rm WD}=0.6$ - 1.38 $M_\odot$) and mass accretion rates ($\dot{M}_{\rm acc}=1\times 10^{-11}$ - $3\times 10^{-7} ~M_\odot$ yr$^{-1}$) based on the energy balance with nuclear burning. These maximum values serve as an upper bound of mass ejection for individual novae. Recently, B. E. Schaefer concluded that the WD masses in the recurrent novae U Sco and T CrB decreased at nova explosions, because the ejected mass is much larger than the accreted mass, i.e., $M_{\rm ej}/M_{\rm acc}= 26$ and $540$, respectively. These values are derived from the orbital period change at the nova explosions. Recurrent novae have been considered to be a progenitor system of Type Ia supernovae (SNe Ia) because their WD masses are now close to, and will possibly grow up to, 1.38 $M_\odot$ at which WDs explode as SNe Ia. From the different view point of energy generation at the thermonuclear runaway, we have obtained the much smaller value of the maximum ratio of $M_{\rm ej}/M_{\rm acc}\lesssim 2.6$ for a $1.37 ~M_\odot$ WD. This conclusion simply means that the nuclear (hydrogen) burning cannot release energy enough to expel such a large ejecta mass as B. E. Schaefer's claims. We also conclude that $(M_{\rm ej}/M_{\rm acc})_{\rm max}$ hardly increases even if we include the effect of frictional mass ejection process in the common envelope phase of a nova.

Understanding the prevalence of disk-like galaxies at very high redshifts is crucial for constraining the early formation of angular momentum-supported structures. The advent of JWST now permits rest-frame UV and optical morphological studies deep into cosmic epochs where disks have traditionally been considered uncommon. We apply an identical denoising VAE-GCNN classification pipeline to multi-filter JWST/NIRCam cutouts in order to obtain homogeneous, morphology-based disk fractions across the sample. Our approach comprises two steps: (i) a U-Net Variational Autoencoder (VAE) is trained to remove astrophysical and instrumental contaminants while preserving intrinsic morphology, and (ii) a rotation - and reflection - equivariant GCNN classifier is applied to the denoised cutouts to distinguish disk-like galaxies from non-disks. We determine the fraction of disk-like galaxies as 0.34 for a sample of JWST 100 galaxies over the redshift range 4 < z < 7.7, also in dependence on the galaxy mass range. Our GCNN-based morphological analysis indicates that disk-like systems constitute a significant fraction of the considered high-redshift population and underscore the importance of such studies for the models of disk formation in the first billion years.

Variability in active galactic nuclei (AGN) probes the physics of accretion onto supermassive black holes. This variability is characterized using metrics derived from the flux distributions of temporally separated epochs. We studied the stability of two variability metrics, the Stetson index "J" and the smoothness "s", against baseline, cadence, and host galaxy contamination. We studied 23 nearby AGNs using Zwicky Transient Facility's Data Release 24. Both metrics are robust to baseline variations of $\sim 2$ years. However, s is sensitive to cadence, showing variations $\gtrsim 40\%$, while J shows minor variations $\lesssim10\%$. We studied the host galaxy impact using Mrk 493 as a representative case. We found that J remains unchanged after host subtraction, while s increases. We concluded that J is a robust tool for characterizing AGN variability, while s should be interpreted with caution.

The James Webb Space Telescope (JWST) has recently discovered a population of compact, red sources at z > 4 known as "Little Red Dots" (LRDs). They are characterized by their V-shaped continuum spectra and prominent broad Balmer emission lines. As their underlying physical nature remains debated and direct study at high-redshift is challenging; therefore, we seek to identify and characterize LRD analogues in the low-redshift universe to constrain their properties and potential evolutionary pathways. We identified five candidates at z = 0.2-0.4 from the Dark Energy Spectroscopic Instrument (DESI) that exhibit spectral energy distributions (SEDs) and broad Balmer emission lines closely resembling their high-redshift counterparts. However, we find significant differences: our low-redshift sample occupies a different region on the Baldwin, Phillips \& Terlevich (BPT) diagram, and their stellar masses are significantly higher, suggesting a more substantial host galaxy contribution. These sources are not necessarily direct local analogues of high-redshift LRDs, but may represent later evolutionary stages of compact, rapidly accreting systems, or systems with related observational properties arising under different physical conditions. This sample provides a valuable laboratory for detailed follow-up studies to elucidate the nature of LRD-like phenomena.

We present a spatially resolved, multiphase study of the outflow in the edge-on starburst galaxy ESO~484-036 from the GECKOS survey, combining VLT/MUSE H$α$ and ALMA CO(1$-$0) observations to analyse the atomic ionised and cold molecular gas. Both show extraplanar emission consistent with a conical outflow. Ionised gas is enclosed by molecular gas, which is detected up to 2.5 kpc from the disc. Molecular gas dominates near the disc, except at the nuclear base, while ionised gas extends beyond 3 kpc. The deprojected outflow velocities are $\lesssim400\ \rm km\ s^{-1}$ in both phases and are consistent with ballistic motion, with some gas possibly falling back onto the disc. We find that the mass outflow rates are in the range of $\dot M_{\rm ion}\sim1-5\ \rm M_\odot\ \rm yr^{-1}$ and $\dot M_{\rm mol}\sim13-54\ \rm M_\odot\ \rm yr^{-1}$, giving mass loading factors of $η_{M\rm, ion}\sim 0.1-0.6$ and $η_{M\rm, mol}\sim 1.5-6.2$. These ranges reflect velocity and geometric uncertainties. Despite the short depletion time ($τ_{\rm dep} = 16-48\rm\ Myr$), the outflow may regulate rather than permanently quench the gas reservoir. Energy loading ($η_E\leq0.16$) and momentum loading ($η_p\lesssim1$) support a purely starburst-driven outflow. Comparing ESO~484-036 with a literature sample, we find a systematic 1~dex shift in mass-loading relations when molecular gas is included. This produces a $\sim3.5$~dex discrepancy with cosmological simulations in $η_{M\rm, mol}/η_{M\rm, ion}$, implying that current models strongly underpredict cold gas production and the role of short-range recycling flows in starburst galaxies.

The James Webb Space Telescope (JWST) revealed a large population of active galactic nuclei (AGN) with redshifts greater than five. We show that if they emit ultrahigh-energy protons with energies up to $\lesssim 10^{19}$ eV, the cosmogenic neutrino production in the high-redshift CMB field yields a neutrino flux with a bump at around 50~PeV. This flux is consistent with the current estimate of neutrino intensity from the IceCube Neutrino Observatory. We argue that the predicted neutrino intensity naturally arises from the average AGN luminosity and number density observed by JWST, without the need for fine-tuning of relevant parameters. Future neutrino observations that confirm the 50-PeV bump and constrain the small-scale anisotropy will infer ultra-high energy cosmic-ray emissions in the early universe.

We present realistic forecasts for the constraining power of the Nancy Grace Roman Space Telescope on fundamental cosmological parameters, with particular emphasis on the absolute neutrino mass scale, using full-shape analyzes of the galaxy power spectrum. We analyze simulated lightcone mock catalogs of H$α$ emission-line galaxies spanning the redshift range $0.5 < z < 2$ over $2400\ \mathrm{deg}^2$, designed to reproduce the expected properties of the Roman High Latitude Wide Area Spectroscopic Survey. We perform parameter inference on the galaxy power spectrum multipoles using two complementary theoretical frameworks: a model-dependent approach based on the Effective Field Theory of Large-Scale Structure (EFT of LSS) within $Λ$CDM, and a model-independent phenomenological approach that makes no assumptions about the background cosmological model. In the $Λ$CDM analysis, we find $m_ν< 0.380(0.162)\ \mathrm{eV}$ at $95(68)\%$ C.L. using Big Bang Nucleosynthesis (BBN) prior and a broad prior on $n_s$, which tightens to $m_ν< 0.276(0.121)\ \mathrm{eV}$ when Planck priors on $ω_b$, $ω_\mathrm{cdm}$, and $n_s$ are added. Our forecasts show that Roman can additionally constrain $H_0$, $Ω_m$, and $σ_8$ with precisions of $1.3\%$, $4.3\%$, and $2.9\%$ in line with Stage IV galaxy survey measurements and forecasts. In the model-independent analysis, we demonstrate that the phenomenological model can robustly recover unbiased measurements of the angular diameter distance, the Hubble parameter, and the growth of structure across all redshift bins, in the same range of scales as the EFT model, and obtain $m_ν< 0.63(0.36)\ \mathrm{eV}$ at $95(68)\%$ C.L. when Planck priors are included.

We present a robust method to empirically infer the timing of the last significant merger in the Milky Way which is tested against fully cosmological models of galaxy formation. We apply it to Milky Way subgiant stars with spectro-photometric ages, finding that the last significant merger (Gaia-Sausage-Enceladus, GSE), occurred $\sim11\,$Gyrs ago. This coincides with the birth of a coeval in-situ group of globular clusters (GCs), which constrains the merger-induced starburst (hereafter {\it Tainá}) to have occurred at $11.2\pm 0.1\,\rm{Gyr}$, the most precise dating of this merger event. The GSE's most metal-rich GCs were also born around this time ($τ=10.9\pm0.1\,\rm{Gyr}$) and likely formed during the merger interaction prior to disruption of the GSE. We argue that $ω$ Centauri is the most likely candidate for the surviving remnant of the GSE, and show that its stellar populations have final ages and metallicities consistent with the GSE GCs together with observational evidence it may have been affected by bar resonances. Furthermore, we argue that the mean metallicity for which stellar orbits transition from halo-like to disc-like kinematics shows an upward inflexion point at $[\rm{Fe/H}]\sim-1.33$, and this sets an upper-limit for the age when the disc was forming. To corroborate this, we identify proto-MW GCs with highly disc-like orbits that formed before the last significant merger (with ages up to $τ=13.0\pm0.5\,\rm{Gyr}$). This places the disc formation time as far back as as $z_{\rm disc\, form}\gtrsim4$.

The guiding center formalism is employed to analyze the motion of a charged relativistic particle in an inhomogeneous magnetic field, subject to magnetic mirroring and energy loss due to cooling. The governing equation for the evolution of the magnetic moment is derived. An example representing a neutron star (pulsar or magnetar) magnetosphere is presented to illustrate typical particle orbits. Notably, radiative losses are most pronounced near a trapped particle's turning point. Depending on the initial particle's pitch angle, energy loss can become catastrophic, resulting in the rapid migration of the particle into the loss cone and subsequent precipitation onto a neutron star. Conversely, particles with a larger pitch angle remain temporarily trapped and form a gradually decaying "cooled-loss-cone" or "funnel'' distribution, characterized by the maximum momentum space particle density being located at the edge of the loss cone. The size of the loss cone is energy-dependent and scales as $α_{c} \propto γ^{3/10}$. Synchrotron losses are strongest in a well-localized region of the magnetosphere, about a few hundred to a thousand star radii under typical pulsar and magnetar conditions. This region is a plausible site for synchrotron radiation originating in the outer magnetosphere, and could also be responsible for non-polar coherent pulsar emission, as well as weak fast radio bursts.

We present detailed, multi-wavelength analysis of GRB 210704A: a Fermi Gamma-ray Burst Monitor discovered and Fermi Large Area Telescope (LAT) detected gamma-ray burst (GRB). The burst is dominated by a short ($\approx 2$ s) pulse followed by weaker, softer emission. We line stack our afterglow spectrum and determine the most likely redshift to be $z = 2.34$. This is corroborated by the photometric redshift of the extended source underlying the GRB. The spectral energy distribution fit parameters, late-time imaging, as well as the GRB's energetics, spectral lag, and location point to a collapsar nature. Follow-up observations reveal excess optical/infrared emission with respect to a standard afterglow, peaking around $T_0 + 7$ d ($2$ d in the rest frame). The excess is extremely luminous ($M_{r} = -22.0$ mag) and rapidly evolving. Strikingly, it resembles the emission seen in recently discovered Einstein Probe fast X-ray transients EP241021a and EP240414a, as well as the population of luminous fast blue optical transients (LFBOTs). This provides a link between these sources and GRBs. Fermi/LAT observations imply a high Lorentz factor, making this a case where LFBOT-like emission is also associated with a powerful successfully launched jet. We model the excess as likely coming from an energetic refreshed shock.

Extragalactic Fast X-ray Transients (EFXTs) represent an emerging class of high-energy phenomena characterized by X-ray outbursts lasting from tens to hundreds of seconds. However, for more than half of the EFXTs, their physical origins remain elusive. In this Letter, we report the discovery of EP250302a, a luminous EFXT detected by the Einstein Probe (EP) at a redshift of $z = 1.131$. The multi-wavelength light curves of EP250302a reveal remarkable temporal features that distinguish it from the previously known EP-detected EFXT population, most notably a needle-like X-ray flare accompanied by smooth optical rebrightening during the afterglow phase. We suggest that the distinct X-ray and optical behaviors constitute the first observed instance of late-time violent collision of two relativistic shells in an EFXT. Drawing on insights from GRB studies, such a collision process strongly indicates the reactivation of a central engine, making EP250302a-like transients a unique laboratory for probing the late-time activity and jet physics of EFXT central engines.

The mean pairwise velocity of massive halos reflects the gravitational force law on cosmic scales. We combine cosmic microwave background intensity maps from the Atacama Cosmology Telescope and a galaxy catalog from the Sloan Digital Sky Survey to estimate the mean pairwise velocity using the kinematic Sunyaev-Zeldovich (kSZ) effect. On scales from 30 -- 230 megaparsecs, we constrain the gravitational acceleration between pairs of halos at separation $r$ to be $g\propto 1/r^n$ with $n=2.1\pm 0.3$, which is consistent with Newtonian gravity in an expanding spacetime (\textit{i.e.}, the standard $Λ$CDM model). This constraint shows agreement with an inverse quadratic radial dependence over the large distances separating galaxy halos, as expected in standard cosmology. Upcoming surveys have the potential to rule out $n = 1$ at $10σ$ significance. Our results establish the kSZ effect as a powerful tool for testing gravity on cosmological scales.

High- to ultrahigh-redshift clustering of halos provides a powerful tool to understand cosmology and galaxy formation. However, theoretical predictions are not firmly established in the first billion years, where current and upcoming surveys are beginning to reach percent-level precision. Here we measure dark matter halo biases at $z=6$ - 19 from simulation data, and find they are $\sim$ 3 - 4$\%$ higher than canonical results calibrated at low $z$. We provide an updated linear-bias fit at these early times, reducing the mean systematic offset to $< 1\%$. These results will enable robust interpretation of early-Universe galaxy clustering from JWST, Roman, and intensity-mapping surveys.

We present the results from rapid-response 1.3 mm observations of GRB 260127A using the Submillimeter Array (SMA). SMA arrived on-source 12.6 minutes after the initial detection by the Neil Gehrels Swift Observatory, representing the earliest millimeter/submillimeter observations of a GRB to date. From these observations, we find a source with flux density $6.9\pm1.7$ mJy, consistent with the X-ray afterglow position but slightly offset from the optical afterglow position (2.7'' offset, with the SMA detection having a 90% confidence radial position uncertainty of 0.9''). Subsequent observations 1.9 days later show no sources of emission, with a $3σ$ upper limit of 0.70 mJy. If the SMA detection is associated with GRB 260127A, we infer that the 1.3 mm light curve for GRB 260127A declined at least as fast as $t^{-0.5}$, suggesting that peak brightness of the event at this wavelength was reached in under a day. We discuss how these findings may be consistent with both forward shock and reverse shock afterglow scenarios, and implications for future millimeter/submillimeter observations of GRBs on these timescales.

We present a 100 ks XRISM Resolve observation of the Ophiuchus cluster that measures turbulence and bulk motion in the wake of the largest radio bubble on the sky. We detect a significant velocity shift of $-80\pm20$ km/s from the cluster centre to the bubble's wake and a clear increase in velocity dispersion from $135\pm10$ km/s to $210\pm20$ km/s. The measured bulk velocity in the wake is low and suggests that the bubble's trajectory is inclined with respect to the line of sight. If we subdivide the bubble's wake, fitting spectra simultaneously with cross-region responses, we find that the velocity shift and dispersion increase are primarily detected in the very centre of the wake. This is consistent with the expected updraft, or `splash', found beneath buoyantly rising radio bubbles. In the cluster's cool core, the turbulent kinetic energy is only 1% of the thermal energy radiated over a cooling timescale of 7 Gyr, and even falls short, by a factor of 5, of the thermal energy radiated over the bubble's rise time. Whilst turbulent energy generated in the large wake region may provide additional heating, this propagates too slowly to prevent rapid cooling across the core. The turbulent-dissipation heating rate is a factor of ~3 below the cooling luminosity. Despite the vast power of the giant radio bubble in the Ophiuchus cluster, the gas motions in the wake are remarkably modest and turbulent-dissipation appears unable to prevent rapid cooling.

Tidal stellar shells and streams are two of the most intriguing low-surface-brightness features within galaxies, consisting of stars accreted from satellite galaxies. A crucial ingredient in determining which type of feature will be formed is the orbit of the satellite galaxy. Additionally, the distribution of stars from these satellite galaxies within the merger remnant and the original location of these stars within the progenitor satellite galaxy provide important clues about the deposition of the stellar component in the resulting galaxy. We utilize the cosmological hydrodynamical simulation Magneticum Pathfinder and expand on the work by Valenzuela & Remus (2024) and Stoiber et al. (2025) to present a case study for the formation of a stream and a shell system. We analyze their orbits and the distributions of stellar particles within their host galaxy and compare them to their initial location within the progenitor satellite galaxy. We find that the orbit of the stream progenitor is more circular than the progenitor of the shell system. The stellar particles of the stream from different initial radii are found at roughly the same distances with respect to the host galaxy. However, the part of the stream visible in mock observations - not hidden by the host galaxy - consists of stars from within the core of the progenitor ($r/r_{1/2} < 1$). On the other hand, the stellar particles of the shell system retain their radial ordering: Stars that were initially at small radii in the satellite galaxy also remain closer to the center of the host galaxy.

The combination of deep imaging data and kinematic measurements in galaxy outskirts promises to reveal extensive insights into the structure and history of individual galaxies. From a census of tidal features around galaxies from the Magneticum simulation, we disentangle the dynamics for a selected stellar stream from the underlying halo by identifying the stream progenitor galaxy. While these dynamics are challenging to measure observationally, we show that they are effectively obtained through planetary nebulae (PNe) as tracers, which we model in the simulation using the PN framework PICS (PNe In Cosmological Simulations). We find that the PNe in the brightest 1.5 mag of their luminosity function are sufficient to recover the underlying stellar dynamics of the massive stream. We thereby establish PNe as an attractive alternative to expensive deep IFU observations, where combining low-surface-brightness observations and PN dynamical measurements will enhance our ability to constrain the gravitational potential of galaxies.

We present a systematic study of gas accretion and angular momentum in the circumgalactic medium (CGM) using high-resolution FIRE cosmological simulations. Our analysis includes halos spanning the critical $\sim 10^{12}\ \mathrm{M}_{\odot}$ scale where several transitions have been identified, including inner CGM virialization, the transition from bursty to steady star formation, and the emergence of thin disks. We find that the temperature of inflowing gas is correlated with the virialization of the inner CGM. CGM inflows are almost entirely cold ($T < 10^5$ K) in pre-virialized halos, while hot inflows ($T > 10^5$ K) dominate in virialized halos. When hot inflows dominate, cooling generally occurs simultaneously with circularization at galaxy radii. The dominance of hot inflows onto massive galaxies persists even at high redshift, where cold streams may coexist. Consistent with previous studies, cold inflows have higher specific angular momentum than dark matter and hot gas. However, in bursty, low-mass galaxies, cold inflows do not circularize prior to star formation, while in steady, massive galaxies, hot inflows circularize, cool, and form stars with disk-like kinematics. We additionally find that in bursty galaxies, accreted gas typically forms stars after residing in the galaxy for less than $\sim 5$ galaxy free-fall times, while in steady galaxies, gas can persist for up to $\sim 25$ free-fall times before forming stars. This highlights a key difference between star formation in bursty galaxies fed by cold accretion and steady equilibrium disks fed by hot accretion.

Filamentary structures in molecular clouds are thought to play a central role in star formation, yet the statistical distribution of their mass per unit length -- and any connection to the stellar initial mass function (IMF) -- remains poorly constrained observationally. Here we present a systematic analysis of the filament linear density function (FLDF) across seven nearby molecular clouds spanning $\sim$100--1000\,pc in distance and a wide range of star-forming environments, from quiescent to massive, using the multiscale extraction method $getsf$. The median linear densities of filaments increase approximately linearly with spatial scale, $\tildeΛ \propto Y$, and the fraction of supercritical filaments varies widely among clouds, from a few per cent to over 50%. Only when integrating over the full hierarchy of spatial scales do the combined linear densities across all clouds yield a FLDF that follows a power law ${\rm d}N/{\rm d}\logΛ\propto Λ^{-α}$ with $α\approx 1.30$--$1.34$, mirroring the Salpeter IMF slope of $1.35$. Our results suggest that the stellar mass spectrum is pre-encoded in the filamentary structure of molecular clouds, providing a direct observational link between the large-scale distribution of interstellar gas and the universal slope of the stellar IMF.

It is difficult to implement numerical simulations on a region extending from the vicinity of a black hole to the Bondi radius. Most previous numerical simulations have primarily concentrated on the region close to the black hole. They found that strong winds can be generated in the hot accretion flows near the black hole, and that radiative cooling significantly affects the strength of these winds. However, the effects of radiative cooling on the production and properties of winds around the Bondi radius remain unclear. In this paper, we perform two-dimensional magnetohydrodynamic simulations to study the impact of radiative cooling on the dynamics and wind production in hot accretion flows around the sub-Bondi radius. As the increase of mass accretion rate, radiative cooling gradually becomes strong, resulting in a reduction in the thickness of the accretion disk (defined as the accretion flows within the density scale height). Based on the Høiland criterion, we find that within the accretion disk, the region of convective stability accounts for $\sim$ 55 - 62 %, and therefore the accretion flows are marginally stable in convective stability. In the runs with weak radiative cooling, the winds play a significant role in the inward decrease of the mass inflow rate. In the runs with strong radiative cooling, the mass outflow rate of winds is significantly reduced and then the inward decrease of the mass inflow rate is mainly attributed to turbulence driven by magnetorotational instability and convection. Radiative cooling has the potential to suppress accretion processes and reduce the power of winds.

We study the evolution of Luminous Compact Blue Galaxies (LCBGs) by making use of H I emission line data provided by the full 856 h COSMOS H I Large Extragalactic Survey (CHILES), which spans a redshift range of $0\leq z\leq 0.48$ within the COSMOS field. We report the results on a cubelet stacking analysis, which we use to estimate the average H I mass evolution of LCBGs in the field up to $z=0.48$. For the stacks that do not show a detection, we report an upper limit estimate of the average H I mass. We also report on two directly detected LCBGs. We find the average H I mass in LCBGs at redshifts $z=0.26$, $z=0.35$ and $z=0.45$ respectively to be $\langle M_{\rm HI}\rangle<4.89\times10^9$ M$_\odot$, $\langle M_{\rm HI}\rangle=(2.49\pm0.75)\times10^9$ M$_\odot$ and $\langle M_{\rm HI}\rangle=(6.44\pm2.71)\times10^9$ M$_\odot$. We see no strong evidence for evolution in the average H I mass over this redshift range, consistent with other recent studies of the evolution of the H I in galaxies at $z<0.5$. On average, LCBGs appear to retain substantial gas reservoirs, with gas fractions staying constant and remaining broadly consistent with those of the larger star-forming population. LCBG gas depletion timescales are nearly an order of magnitude shorter than in normal star-forming galaxies across the studied redshift range, aligning with the period during which their number density drops sharply.

Stellar streams are dynamically fragile structures formed by the tidal disruption of dwarf galaxies and stellar clusters. These objects are valuable tracers of the gravitational potential and accretion history of the Milky Way, and are key probes for the presence and interactions of starless dark matter subhalos. The Jet stream is a $\sim 30^\circ$-long stellar stream that is situated at 30.4 kpc and originates from a disrupted globular cluster. It consists of metal-poor stars that follow a retrograde orbit, reducing the impulse imparted from the Milky Way bar and making it especially sensitive to gravitational perturbations from dark matter subhalos. This paper investigates the known extent of the Jet stream by leveraging photometric metallicities derived from a narrowband filter centered on the Ca II H&K lines at $\sim$3950A on the Dark Energy Camera (DECam), as part of the Mapping the Ancient Galaxy in CaHK (MAGIC) survey. The wide field-of-view of DECam enables the efficient derivation of photometric metallicities for stars across the full extent of the stream, allowing for a metallicity-based selection to identify likely members. We demonstrate the efficacy of photometric metallicities in isolating stream members when used with Gaia DR3 proper motions, identifying a sample of 213 candidate Jet stream member stars. This then allows for the study of stream morphology, through which we identify a clear fanning of the stream toward the end farther from the Milky Way bar. We provide a list of candidate members, enabling spectroscopic follow-up of the Jet stream to facilitate further studies of its dynamics.

Low-ionization metal absorption lines provide a primary probe of cool gas in and around galaxies. We confront observations of metal-line absorption in quasar spectra with predictions from the IllustrisTNG cosmological simulation in order to benchmark how well current galaxy formation models reproduce the observed circumgalactic medium (CGM) and intergalactic medium (IGM) absorption signatures. We implement two ionization prescriptions: a purely collisional model and a model including photo-ionization by a uniform ultraviolet background (UVB). Using a grid-based framework, we compute MgI, MgII and FeII column densities and construct column density probability distribution functions (PDFs) and equivalent width (EW) statistics for comparison with observations. The observational samples considered here are based on the High Resolution Echelle Spectrometer (HIRES), the Ultraviolet and Visual Echelle Spectrograph (UVES), the Sloan Digital Sky Survey (SDSS) and the Dark Energy Spectroscopic Instrument (DESI). The computed PDFs broadly reproduce the observed ones across the sampled column density range of $10^{11.4}\lesssim \text{N}\lesssim 10^{16}\ \rm{cm^{-2}}$, indicating that the simulation captures the dominant physical drivers of low-ionization absorption. We then compute the cosmic incidence of MgII systems, namely the evolution of their number with redshift $d\mathcal{N}/{dz}$. The model that includes UVB accurately produces $d\mathcal{N}/{dz}$ up to equivalent widths (EW) of $\rm W_0^{2796} < 0.6\ \mathring{A}$, consistent with low-density photo-ionized gas in the outer CGM. At high EWs of $\rm W_0^{2796} > 1\ \mathring{A}$ TNG underestimates $d\mathcal{N}/{dz}$ and fails to capture its rise toward $z\sim2$.