arXiv Daily Digest - 2026-07-06

As AI coding agents become more autonomous, they increasingly ship code iteratively, with the codebase persisting across sessions. This persistence creates a new attack surface: a misaligned or prompt-injected agent can distribute attacks across pull requests (PRs) and time its payload for the PR with the best natural cover. To study the resulting dynamics, we introduce Iterative VibeCoding, a setting for AI control, the study of safely deploying capable but potentially untrusted AI. In Iterative VibeCoding, a coding agent builds software over a sequence of PRs in a persistent codebase while pursuing a covert side task. Our benchmark includes two task families: CLI tools and Flask web services, across 20 total task variations. We use Claude Sonnet 4.5 as the attack agent and GPT-4o as the monitor. We compare gradual attacks, which distribute the side task across PRs, against non-gradual attacks concentrated in a single PR. No single monitor is robust to both: which strategy evades best (success while evading the monitor) depends on the monitor type, so a defender cannot close off both gradual and non-gradual attacks with any one monitor. High evasion (>= 65%) generalizes across model attack agent backends (Sonnet 4.5, Gemini 3.1 Pro, Kimi K2.5), confirming this is a property of the persistent-state attack surface rather than a single model's capability. Evasion also remains high across state-of-the-art monitor models and the gap between gradual and non-gradual evasion widens for more capable models. We introduce a stateful link-tracker monitor that tracks suspicious buildup across PRs. On both task families, it detects gradual attacks substantially better than diff monitors that merely see more accumulated history. Combining this stronger monitor with trajectory monitors in a four-monitor ensemble reduces gradual-attack evasion from 93% under the weakest standard diff monitor to 47%.

LLMs memorize sensitive training data, including personally identifiable information (PII), creating a pressing need for reliable post hoc removal methods. Unlearning has emerged as a promising solution, with state-of-the-art(SOTA) methods often following a localize-first, unlearn-second paradigm that targets specific model parameters. However, existing benchmarks evaluate unlearning solely at the output level, leaving open the question of whether unlearning truly erases knowledge from a model's parameters or merely obfuscates it, a concern reinforced by the success of resurfacing attacks. To bridge this gap, we introduce LACUNA: the first unlearning testbed with ground-truth parameter-level localization. LACUNA injects PII of synthetic individuals into predefined parameters of 1B and 7B OLMo-based models via masked continual pretraining, enabling direct evaluation of whether unlearning targets the weights responsible for knowledge storage. We use LACUNA to benchmark current SOTA unlearning methods and find that, despite strong output-level performance, existing methods are highly imprecise and susceptible to resurfacing attacks. We further show that when localization is successful, even a simple gradient-based unlearning method achieves strong erasure and robustness to resurfacing attacks, highlighting the importance of precise unlearning. We release LACUNA to complement behavioral evaluations and drive further advances in robust, localization-based unlearning.

Many everyday programming tasks resist clean rule-based implementation, such as alerting on important log lines, repairing malformed JSON, or ranking search results by intent, and are increasingly outsourced to large language model APIs at the cost of locality, reproducibility, and price. We propose fuzzy-function programming: compiling such a function from a natural-language specification into a compact, locally-executable neural artifact. We instantiate this paradigm with Program-as-Weights (PAW), in which a 4B compiler trained on FuzzyBench, a 10M-example dataset we release, emits parameter-efficient adapters for a frozen, lightweight interpreter. A 0.6B Qwen3 interpreter executing PAW programs matches the performance of direct prompting of Qwen3-32B, while using roughly one fiftieth of the inference memory and running at 30 tokens/s on a MacBook M3. PAW reframes the foundation model from a per-input problem solver into a tool builder: invoked once per function definition, it produces a small reusable artifact whose subsequent calls per function application are cheap and offline.

Despite alignment training, LLMs remain prone to generating unsafe outputs at deployment time. Monitoring outputs online and raising an alarm when safety can no longer be assumed is therefore critical. We study a simple real-time monitor that turns a verifier signal from an external model into an alarm decision by thresholding, with the threshold calibrated via risk control. In experiments on mathematical reasoning and red teaming datasets, we show that this simple design is competitive with more advanced monitors based on sequential hypothesis testing.

Understanding and reasoning over long contexts has become a key requirement for deploying large language models (LLMs) in realistic applications. Although recent LLMs support increasingly long context windows, they often fail to use relevant evidence that is already present in the input, revealing a gap between context access and effective context utilization. In this work, we propose Recursive Evidence Replay as LLM Harness for Long-Context Reasoning (RECONTEXT), a training-free inference method for improving long-context reasoning. RECONTEXT uses model-internal relevance signals to construct a query-conditioned evidence pool and replays it before final generation while preserving the full original context. This recursive selection process separates evidence organization from answer generation without training, external memory, or context pruning. We also provide a theoretical analysis based on associative memory, which characterizes the context as a memory store, the question as a retrieval cue, attention as cue-trace association, and replay as trace reactivation. Experiments on eight long-context datasets with 128K context length show that RECONTEXT consistently improves evidence utilization across Qwen3-4B, Qwen3-8B, and Llama3-8B, achieving the best average rank on all three backbones. Code is available at https://github.com/Yanjun-Zhao/ReContext.

LLM agents will increasingly act in socially structured settings where role, audience, and relational context can shape what is advantageous or costly to say. We study whether such social structure, without any explicit objective in the prompt, changes what an agent expresses publicly relative to an off-the-record (OTR) channel elicited under the same condition. We introduce a dual-channel debate framework in which agents produce public utterances that enter the shared history alongside OTR responses that are recorded but never shown to the other participant. Across 10 models, 3 scenarios, and 5 variations within each scenario, alignment-inducing settings produce systematic public-OTR divergence in the targeted agent, with its decision divergence rising from a $\sim$3% baseline to roughly 40%. The effect is consistent across four aggregate analyses: stance, semantic similarity, natural language inference, and survey responses. In some cases, the OTR response explicitly attributes public accommodation to relational pressures, such as career risk or sponsorship obligation. The findings suggest that agent evaluation should extend beyond explicit goals and detect emergent objectives. We present a dual-channel evaluation framework and complementary behavioral measures that operationalize this assessment.

Long-form TV dramas present a formidable challenge for comprehensive video understanding, where deciphering complex storyline often relies on \textbf{speaker recognition}, the task of accurately attributing each spoken utterance to its respective character. In this paper, we advance this field through two primary contributions. (1) We introduce \textbf{DramaSR-532K}, a large-scale benchmark comprising 532K annotated dialogue lines across more than 900 unique characters, necessitating the integration of auditory, linguistic, and visual cues for speaker recognition. (2) We propose \textbf{DramaSR-LRM}, a robust approach built upon a large reasoning model (LRM). DramaSR-LRM is designed to autonomously aggregate contextual evidence via multimodal tool-use, synthesizing diverse inputs to achieve high-fidelity attribution. Experimental results demonstrate that DramaSR-LRM significantly outperforms existing baselines, particularly on short utterances where acoustic biometrics are inherently unreliable. \textit{All the data and code will be made publicly available at the project page: https://www.github.com/198808xc/DramaSR-LRM.}

On-policy self-distillation (OPSD) has emerged as a practical method for training large language models (LLMs) to reason, where a single model acts as both the teacher and the student with different levels of information access. However, recent studies have found that the teacher's dense token-level supervision, conditioned on privileged information, can lead to overfitting to in-domain patterns, suppress exploration, and hurt cross-domain generalization, while also introducing a more fundamental issue: *privileged information leakage*, where the student encodes answer-dependent shortcuts that are unavailable at test time. We introduce **DemoPSD**, a novel framework that resolves such problems through the idea of *selective adoption of teacher guidance*. Instead of fitting the full teacher distribution, DemoPSD steers the student toward a *reverse-KL barycenter target*, a weighted geometric combination of the teacher and student distributions, that naturally balances learning from the teacher with preserving the student's own reasoning capacity. We measure the difference between their distributions and use such a discrepancy to adaptively control the blending at each token position. We provably show that DemoPSD achieves **(1)** *leakage attenuation*, i.e., effective mitigation of privileged information leakage; and **(2)** *exploration preservation*, i.e., preservation of exploration capacity under dense token-level distillation. Extensive experiments on SciKnowEval across four scientific fields show that DemoPSD outperforms both GRPO and SDPO while maintaining higher training entropy and robustly generalizing to out-of-distribution GPQA benchmarks.

Machine learning interatomic potentials (MLIPs) have become a hallmark of AI for scientific simulation. While efforts on new architectures and datasets have led to increasingly accurate and general models, the choice of optimizer for training has largely remained unexplored, defaulting to Adam and its variants in the community. Here, we implement and systematically compare a class of recently proposed matrix-structured optimizers, including Muon, SOAP, and the hybrid SOAP-Muon, for training NequIP and Allegro MLIP models. We find that these optimizers can substantially outperform Adam in both convergence speed and final accuracy. SOAP and SOAP-Muon emerge as robust and consistently strong methods, while Muon only provides partial gains relative to Adam. The improvements are particularly pronounced under partial force supervision. Our results indicate that optimizer choice is an overlooked yet impactful design axis for MLIPs.

Realistic traffic simulation requires agents that imitate logged behavior and can also be steered along interpretable axes. Such controllability enables engineers to isolate variables, reproduce specific edge cases, and test autonomous systems without real-world risk. We introduce Controllable Neural Variational Agents (CNeVA), a controllable simulated-agent framework that learns to infer a per-agent Gaussian behavior latent from per-channel discounted returns via a closed-form conjugate variational update, conditioning a rectified-flow trajectory generator trained on a mixed channel-mask curriculum for classifier-free guidance. To tackle scarcity in reward signals, we propose soft eligibility gates that replace hard binary thresholds with smooth exponential decay, preserving the gradient signal for near-threshold agents. On the Waymo Open Motion Dataset, CNeVA attains competitive realism on the benchmark while exposing per-channel controllability that the higher-ranked imitation models lack. Speed- and acceleration-based steering produces monotone responses without stall-induced reward hacking. Safety controllability is monotone and substantial with the introduction of soft eligibility. We manage to achieve steerable map compliance under a context-residual return measure. Furthermore, our experiment demonstrates that steering metrics must be read alongside physical-plausibility guardrails to avoid reward-hacking confounds.

Models trained via Contrastive Language-Image Pretraining (CLIP) serve as the foundational vision encoders for most modern Large Vision Language Models (LVLMs). Despite their widespread adoption, CLIP models exhibit a critical yet underexplored failure mode: irrelevant text appearing within images confounds visual representations, biasing them toward lexical meaning rather than true visual semantics. This robustness issue, commonly described as a Typographic Attack (TA), exposes a vulnerability that poses a significant risk to safety-critical applications such as autonomous driving. To achieve interpretable and effective robustness against TA, we propose a novel, training-free mechanistic interpretability method. Our method provides sampling-based interpretations of hidden state representations and quantitatively attributes semantic versus lexical focus to individual attention heads. Through probabilistic analysis and circuit mining, we isolate specific Vision Transformer (ViT) components that disproportionately encode lexical information, thereby identifying the mechanistic source of TA. We further show that simple interventions applied directly to the identified circuits, without any additional training, can substantially improve robustness against Typographic Attacks in object classification. These interventions, such as selective adjustment of attention weights, also outperform both supervised and training-free defense methods. Our experiments demonstrate that applying the proposed intervention to the vision encoders of several state-of-the-art LVLMs yields substantial gains in Visual Question Answering accuracy under Typographic Attack interference on RIO-Bench. These results confirm both the efficacy and the generalizability of our mechanistic approach. Code is released at https://github.com/Liu-524/SamplingTAR.

In this work, we focus on SE-RRMs, a symbol-equivariant instantiation of RRMs that exhibits improved extrapolation to larger problem sizes. We propose a neuro-symbolic approach, ``Guiding with Recurrent Reasoning Models'' (G-RRM), which integrates SE-RRMs with symbolic solvers for constraint satisfaction problems. SE-RRMs act as neural solvers that generate full solution proposals and guide classical symbolic solvers, such as backtracking or SAT-based methods like Glucose 4.1 and CaDiCaL 3.0.0, that produce globally correct solutions. Centrally, we investigate when neural guidance with G-RRM improves the search efficiency of symbolic solvers. % Our experiments show that the efficacy of G-RRM depends on two conditions: first, the problem instances must have an expansive combinatorial search space to expose potential gains, and second, the solver architecture must be capable of dynamically overwriting its branching choices to recover when neural hints are imperfect. When these conditions hold, guidance drives median conflict counts to zero and yields significant wall-clock speedups: on $9\times9$ Sudoku, where the SE-RRM correctly solves $91.1\%$ of instances, backtracking accelerates by $33.3\times$ and Glucose 4.1 by $1.70\times$ (median, $p<0.001$), with Glucose 4.1 retaining a $1.17\times$ speedup on perfect-hint $25\times25$ grids. In contrast, CaDiCaL 3.0.0, whose runtime is overhead-dominated and which always respects the injected branching hints rather than overwriting them, shows no significant speedup (median $1.02\times$, n.s.) and even a small significant mean slowdown ($0.90\times$) on $9\times9$. These results delineate the regimes in which neural guidance translates into practical speedups.

Large vision-language models can reason over multimodal inputs by generating textual chains of thought (CoT). A key capability exhibited in CoT reasoning is self-reflection: revisiting earlier decisions and correcting previous errors. However, existing LVLMs often fail to properly attend to visual inputs during reflection, limiting their ability to translate feedback into grounded corrections, especially for out-of-distribution images. To address this issue, we propose a novel reinforcement learning training framework VRRL, with two components explicitly designed to elicit visually grounded self-reflection. First, we randomly mask trajectory prefixes during training to emphasize recovery from incorrect intermediate predictions rather than making early mistakes. Second, we introduce buffered roll-ins from an experience replay buffer to expose the model to diverse failure states that it must learn to correct. We evaluate our approach on visual grounding tasks involving tables and charts, as well as spatial navigation benchmarks. While off-the-shelf and conventionally fine-tuned models degrade substantially under distribution shift, our method substantially improves average out-of-distribution accuracy over standard RL and reflection-oriented fine-tuning baselines by using self-reflection effectively.

Visual token pruning is a crucial strategy for accelerating VLMs by compressing redundant image patches, yet existing methods often fail to preserve critical cues under dense instructions and fine-grained queries. In this paper, we investigate this failure and identify two underlying bottlenecks: the widespread dispersion of textual noise that corrupts dense cross-modal scoring, and the feature fragmentation inherent to standard token selection. To address these issues, we propose Entropy-Aware Dense Pruning (EADP), a framework that reformulates pruning as a structured compression problem. EADP first leverages statistical entropy to quantify and filter out textual noise, yielding a robust, fine-grained instruction relevance score. Subsequently, instead of naive Top-K selection, EADP casts token selection as a submodular maximization problem with a spatial prior, explicitly ensuring a holistic and non-redundant visual representation. Extensive experiments demonstrate that EADP improves the accuracy-efficiency trade-off of VLMs, robustly preserving fine-grained visual cues under strict token budgets while achieving SoTA performance on challenging multimodal benchmarks.

Narration is central to the audiobook listening experience, shaping how listeners engage with and understand the content. This work explores how narration qualities shape an audiobook's appeal, noting that their effects can vary by genre, title, and audience. We extract vocal and acoustic features (e.g., tone, pace, loudness) from LibriVox using pre-trained audio models and analyse their relationship with consumption data (specifically, view-rate) and their interplay with genre and title. Despite limited consumption data, we find that acoustic information alone has a robust association with appeal, even after accounting for title effects. We further validate these findings using more nuanced proprietary engagement metrics. To our knowledge, this is the first systematic computational study linking narration qualities, genre, title, and audiobook consumption, highlighting the potential of data-driven insights to improve audiobook personalisation and narrator casting.

Software tests and code evolve together: a code change should be followed by new or updated tests that record the new software behavior. Yet existing test generation and update benchmarks often isolate the test from the code change, and rely on static metadata that does not verify whether a test is executable or semantically tied to the code change. This makes it difficult to evaluate whether a test automation agent understands how a code change should propagate into the test suite. We introduce TestEvo-Bench, a benchmark of test and code co-evolution tasks mined from software repositories, with two tracks: in test generation, the agent shall write new tests to capture the new software behavior; in test update, the agent shall adapt failing existing tests to the changed software behavior. Each task is anchored to a real commit history and packaged with environment configuration to support execution-grounded metrics such as pass rate, coverage, and mutation score. TestEvo-Bench is also a live benchmark: each task records the timestamp of the test and code changes, and new tasks are periodically mined by our automated pipeline, so evaluation can be restricted to tasks postdating a model's training cutoff to reduce data leakage risk. The current snapshot contains 746 test generation and 509 test update tasks, curated from 59,950 candidate co-evolution records across 152 open-source Java projects. We experiment with four state-of-the-art agents that combine strong harnesses (Claude Code, Gemini CLI, and SWE-Agent) with strong foundation models (Claude Opus 4.7 and Gemini 3.1 Pro). Results show that they achieve up to 77.5% success rate on test generation and 74.6% on test update. However, success rate is materially lower on the most recent benchmark tasks and drops significantly under limited per-task cost.

Whether pairing people with AI helps or hurts is usually reported as a single average effect. Using a real-money prediction market (Polymarket) as an objective, externally resolved benchmark, this pilot shows that the value of human-AI collaboration depends on a specific, measurable form of human capital. Analyzed at the level of the individual forecaster, hybrid performance is trimodal: most people either deferred to the model (matching it) or used it to rubber-stamp a prior guess (performing worse than the model alone), while a minority engaged in genuine complementary reasoning and reached accuracy matching or even exceeding (i.e., lower error than) the market itself. Collaborative traits (perspective-taking, intellectual humility, and curiosity) rather than raw cognitive ability or model benchmarks, distinguished who reached that mode. The results are preliminary but statistically robust, and motivate a pre-registered replication now in preparation.

Vision-Language-Action (VLA) models are fundamentally bottlenecked by the scarcity of expert demonstrations -- triplets of observations, instructions, and actions that are costly to collect at scale. We argue that this bottleneck stems from conflating two distinct learning objectives: acquiring physical competence (how to move) and acquiring semantic alignment (what to do). Crucially, only the latter requires language supervision. Building on this Decomposition Hypothesis, we propose Task-Agnostic Pretraining (TAP), a two-stage framework that first learns transferable motor priors from cheap, unlabeled interaction data -- including discarded off-task trajectories and autonomous robot play -- via a self-supervised Inverse Dynamics objective. A lightweight second stage then grounds these priors in language using minimal expert data. On the SIMPLER benchmark, TAP matches models trained on over 1M expert trajectories while using orders of magnitude less labeled data, yielding a 10% absolute gain over standard behavior cloning. On a real-world WidowX platform, TAP retains 25% success under camera perturbations where internet-scale baselines collapse to 0%, demonstrating that task-agnostic pretraining produces robust, transferable physical representations and offers a scalable path forward for Embodied AI.

Probabilistic computation plays an important role in trustworthy edge intelligence to quantify uncertainty, enhance robustness, reconstruct data, and protect privacy, but its adoption is limited by the orders-of-magnitude data throughput gap between Gaussian random number generation (GRNG) and computation, as well as instruction overhead. This paper introduces probabilistic memory (p-MEM), a unified memory primitive that stores distribution parameters, such as mean and standard deviation, and samples directly at the native memory bandwidth, where deterministic data becomes the zero-variance special case. Using a layout-validated p-MEM simulator, we comprehensively explore device choices, memory specifications, and technology nodes, showing that p-MEM can achieve more than 1000 GSa/s/mm^2 GRNG throughput, including memory-array access. Integrated into CPU/GPU systems, p-MEM reduces instruction count by up to 2.19x/4.37x, sampling latency by 562x/3.45x, and energy by 295.5x/3.53x for Bayesian neural network workloads, providing a scalable hardware substrate for trustworthy probabilistic AI.

Large Language Model (LLM) social simulations are a promising research method, but they are not yet faithful enough to be adopted widely. In this work, we investigate whether the current scaling paradigm in language modeling is likely to close these gaps, or whether simulation fidelity is orthogonal to general capabilities and therefore deserving of more research attention. We use scaling laws to study the relationship between LLMs' compute scale, general capability benchmarks, and the fidelity of social simulation in three representative sub-domains: opinion modeling, behavioral simulation, and longitudinal forecasting. Surprisingly, we discover strong compute scaling in all three settings, using a suite of 85 transformer LLMs with the Qwen3 architecture pre-trained on the DCLM web text corpus under fixed-compute budgets from $10^{18}$ to $10^{20}$ FLOPs. Then we evaluate 35 larger and more capable open-weight models up to 70B parameters, allowing us to predict downstream accuracy from loss. This reveals that the majority of behavioral and opinion simulation tasks will rapidly improve with scale, particularly when they involve populations that are well-represented in English web corpora. Longitudinal forecasting and underrepresented opinions scale more slowly, especially when they are less correlated with general knowledge and reasoning benchmarks like MMLU. In behavior simulation, scaling fails to improve model calibration with human cognitive biases like risk aversion, as well as human heuristics like learning correlated rewards from related tasks. On these tasks, even fine-tuned models fail to noticeably scale up performance from 0.5B to 8B parameters. Taken together, we conclude that scale will improve social simulations in most settings, but outliers exist, and improvements will be less reliable in low-resource domains.

Diffusion transformers (DiTs) achieve state-of-the-art image and video generation, but their multi-step sampling and growing parameter count make inference expensive. Post-training quantization (PTQ) is the natural remedy, yet DiT activations shift across timesteps, prompts, and guidance branches, forcing prior methods to re-fit calibration data for every new checkpoint or modality. We present OrbitQuant, a data-agnostic weight-activation quantizer that bypasses range estimation by quantizing in a normalized, rotated basis. In this basis, a randomized permuted block-Hadamard (RPBH) rotation concentrates each coordinate around one fixed, known marginal regardless of the input, so a single Lloyd-Max codebook serves all timesteps, prompts, and layers of a given input dimension. We extend the same quantizer to weight rows offline, absorbing the rotation into the weights so that it cancels inside each linear layer and only a forward rotation on the activations remains at runtime. The same recipe transfers from image to video with no per-modality tuning. Across FLUX.1, Z-Image-Turbo, Wan 2.1, and CogVideoX, it sets the state of the art for PTQ at several low-bit settings. It also pushes PTQ of image diffusion transformers to W2A4 with usable generation quality.

Post-training large language models (LLMs) without real-world interaction feedback or human-labeled supervision remains challenging, particularly in specialized domains where expert annotations are costly to obtain. Recent annotation-free self-evolution methods address this by using the model's own outputs as supervision signals, constructing a teacher via additional context and aggregating predictions across multiple rollouts through majority voting to produce pseudo-labels. However, these approaches are not without drawbacks: SFT- and GRPO-based variants suffer out-of-domain performance degradation, while reward-based on-policy RL inflates calibration error. In this paper, we propose Neuron On-Policy Self-Distillation (Neuron-OPSD), a data-centric framework for annotation-free self-distillation that leverages internal neuron activations to guide both training-data selection and teacher context construction. The model is then trained via on-policy distillation from the teacher distribution, requiring no ground-truth labels at any stage. Across specialized-domain benchmarks, Neuron-OPSD improves in-domain task performance while preserving cross-domain generalization and mitigating calibration collapse over prior annotation-free baselines. This framework is particularly relevant to settings where online interaction or external supervision is costly or infeasible, and is conceptually distinct from offline RL approaches that rely on logged, reward-labeled trajectories.

Language models are increasingly used to quantify cultural phenomena, but what makes such measurement distinctively cultural? This paper argues that NLP work on culture is a material-discursive practice: the apparatus -- model, data, annotation, evaluation -- participates in constituting the cultural reality it measures, rather than passively recording it. Drawing on Karen Barad's concept of the agential cut -- the contingent boundary between phenomenon and instrument -- I show that the apparatus's substantive design choices draw such boundaries, and that the boundary is entangled from the start because language models have already internalized much of the cultural material they measure. I illustrate this through three case studies on television and film dialogue (measuring structure, interaction, and deviation) and three examinations of the apparatus itself (erasure of cultural markers, attunement to historical material, and agency in an agentic workflow). This big picture analysis proposes a research program that is theory-driven, empirically rigorous, and culturally contingent, treating each agential cut as a conscious commitment, at once methodological and ethical.

Since ChatGPT's launch in November 2022, open-source agentic frameworks have proliferated, making framework selection important for engineering teams while obscured by popularity signals such as GitHub stars. This paper analyzes 15 major open-source AI agent framework repositories from late 2022 to early 2026, using 808,042 stars, 73,997 pull requests, 86,241 commits, and 987,330 user profiles to assess ecosystem health across awareness, adoption, and retention. Three findings emerge. First, headline popularity is unreliable. Star counts reflect hype cycles and inorganic activity. AutoGPT gained 111,967 stars in one month but converted fewer than 9 contributors per 1,000 stars, defined as contributor density in this research, compared with LangChain's 41. Lower-profile frameworks such as Pydantic-AI show higher contributor density, indicating deeper adoption. Second, mapping awareness against adoption shows that visibility and engagement diverge. MetaGPT and LangFlow have contributor density ratios below 5 even with their high visibility. Openai-agents-python's limited contributor base suggests institutional backing alone does not ensure community depth. By analyzing cross-framework contribution, we discover that LangChain functions as a shared infrastructure, attracting 82.5% of cross-ecosystem contributors. Third, retention drops most steeply in the first 30 days of initial contribution and stabilizes near 90 days. Overall, ecosystem health is better measured by contributor density, cross-ecosystem engagement, and retention than by stars alone. These metrics offer teams a more robust basis for framework evaluation.

Parts manufactured with Fused Deposition Modeling (FDM) often require Design for Additive Manufacturing (DFAM) modifications to ensure printability, structural integrity, and reduced post-processing. Current slicers identify defects such as steep overhangs but are unable to modify the underlying geometry. This work presents AgentsCAD, a multi-agent system that bridges raw boundary-representation (B-Rep) geometry and Large Language Model (LLM) reasoning to automate targeted DFM. The workflow begins by parsing a STEP file. The agentic system detects overhangs above a 45°threshold, constructs a face-adjacency topology graph, and optionally injects semantic feature labels from a GraphSAGE model trained on MFCAD++ (59,665 parts), before dispatching a Claude Sonnet design-reasoning agent that recommends reorientations, fillets, chamfers, and similar modifications. A GPT-4o vision-language verifier inspects rendered views to confirm geometric integrity. Outputs include a modified STEP file and a human-readable report. A test case on a birdhouse model demonstrates that the system correctly diagnoses overhangs, selects appropriate defect mitigation strategies, and proposes physically valid corrections, partially solving the geometry-to-language translation problem central to LLM-driven CAD modification.

Recent research has introduced distributed self-supervised learning (D-SSL) approaches to leverage vast amounts of unlabeled decentralized data. However, D-SSL faces the critical challenge of data heterogeneity, and there is limited theoretical understanding of how different D-SSL frameworks respond to this challenge. To fill this gap, we present a rigorous theoretical analysis of the robustness of D-SSL frameworks under non-IID (non-independent and identically distributed) settings. Our results show that pre-training with Masked Image Modeling (MIM) is inherently more robust to heterogeneous data than Contrastive Learning (CL), and that the robustness of decentralized SSL increases with average network connectivity, implying that federated learning (FL) is no less robust than decentralized learning (DecL). These findings provide a solid theoretical foundation for guiding the design of future D-SSL algorithms. To further illustrate the practical implications of our theory, we introduce MAR loss, a refinement of the MIM objective with local-to-global alignment regularization. Extensive experiments across model architectures and distributed settings validate our theoretical insights, and additionally confirm the effectiveness of MAR loss as an application of our analysis.

We study stabilizer state testing and learning with limited coherent quantum memory. Here an algorithm sequentially receives copies of an unknown $n$-qubit state, but may keep only $k$ qubits of coherent quantum memory between measurements. With unrestricted memory, seminal work of Gross, Nezami and Walter showed how to test $n$-qubit stabilizer states using $6$ copies, which is dimension independent, unlike the learning complexity of $Θ(n)$. We show that this testing-vs-learning separation is lost under memory constraints. More concretely we show that (1) The sample complexity of testing stabilizer states in the $k$-qubit memory framework is $Θ(n-k)$. Our upper bound goes via a novel connection to the hidden shift problem and the lower bound is proven using a novel approach to average case bounds on likelihood ratios via combinatorics of the stochastic orthogonal group. (2) The sample complexity of learning stabilizer states with $k$ qubits of memory, in the non-adaptive framework, is $Θ(n^2/k)$. As a further application of our techniques, we prove an exponential lower bound for purity testing even when the memory may be left coherent throughout the protocol. Our main results identify coherent quantum memory as the resource enabling the usual separation between stabilizer testing and learning. In particular, even with $k=0.99n$ qubits of memory, there is no constant-copy stabilizer tester; furthermore for $k=cn$ qubits of memory (for $0< c < 1$), stabilizer testing is as hard as learning, with both requiring $Θ(n)$ copies.

Application Programming Interfaces (APIs) are essential in software development, enabling web services, mobile apps, and microservices. However, their widespread use introduces significant security risks, highlighting the importance of API security. This paper presents HTTP REST API Learning (HRAL), a novel unsupervised anomaly detection approach that models the structure and behavior of API endpoints directly from network traffic, without relying on predefined rules or documentation. HRAL enables robust detection of malicious activity by understanding how APIs behave and flagging deviations as potential threats. We evaluate HRAL across varying levels of OpenAPI documentation detail and compare it with existing techniques. HRAL achieves strong performance, with an average recall of 82.07% and an F1-score of 87.24%, significantly outperforming alternatives when API documentation is limited. Moreover, our results approach the effectiveness of full API document definitions. When combined with signature-based rules such as the OWASP ModSecurity CRS, our system achieves 100% detection. These results highlight HRAL's effectiveness in real-world, partially documented API environments and its potential as a foundational layer for modern API security solutions.

Autonomous agents are increasingly expected to improve executable policies through feedback, yet existing evaluations often collapse this process into a final score or confound it with open-ended software-engineering progress. We introduce Autonomous Policy Evolution, a controlled evaluation setting in which a harness-model agent repeatedly edits an executable policy system under a fixed interaction budget. We instantiate this setting in EvoPolicyGym, a benchmark built from compact interactive RL environments that evaluates how agents iteratively improve explored policies. On the EvoPolicyGym suite, GPT-5.5 achieves the strongest aggregate rank score and top-two performance on all 16 environments. Beyond leaderboard results, EvoPolicyGym also provides trajectory-level diagnostics that distinguish how agents allocate budget, convert feedback into parametric tuning. These analyses show that strong autonomous policy evolution depends not only on isolated task wins, but on discovering task-appropriate mechanisms and refining policies under bounded feedback.

Time series forecasting remains challenging when the underlying data contain rare but critical extreme events. This issue is particularly important in hydrologic forecasting, where streamflow distributions are often highly skewed and extreme peaks can have substantial impacts on flood monitoring, water resource management, and early warning systems. Although Transformer-based forecasting models have achieved strong performance by modeling long-range temporal dependencies, they typically treat all time points uniformly and may therefore underrepresent rare extreme patterns. In this paper, we propose the Extreme-Adaptive Transformer (Exformer), a forecasting framework designed to explicitly model temporal dependencies involving both normal and extreme events. Exformer introduces an extreme-adaptive attention mechanism composed of three sparse components: Local, Stride, and Extreme. The Local and Stride components capture short-term and periodic temporal dependencies, respectively, while the Extreme component selectively models event-aware dependencies between normal and extreme streamflow patterns. Experiments on four real-world hydrologic streamflow datasets show that Exformer achieves superior 3-day forecasting performance compared with state-of-the-art baselines. Our findings demonstrate that explicitly incorporating extreme-aware attention improves the forecasting capacity of Transformer models on imbalanced time series with rare but consequential events.

Agentic coding assistants are increasingly given extra capabilities, such as browser based testing tools and design oriented system prompts, on the assumption that more capability yields better software. This study tested that assumption directly. Ninety independent agent runs built the same application, a real time retrospective board, from one detailed specification, each scored on a fixed 14 criterion functional rubric (42 point maximum) and a visual quality review. The runs spanned several model generations, two agent harnesses, two reasoning effort levels, a testing tool, and two design oriented prompts. Capability tier dominated: frontier models clustered near the ceiling while a low cost local model fell to 24 to 37 points. A criterion level analysis revealed what run totals conceal. Container deployment was the dominant defect, failing first try in 44 percent of runs, with its failure rate shifting sharply across model generations while mean totals moved less than a point. The testing tool raised cost by 42 to 68 percent without improving functional score or reliability, even on interface visible criteria. Raising reasoning effort from High to xHigh lifted first try perfect runs from 28 percent to 89 percent and cut corrective prompts about five fold, for 9 to 29 percent more cost. A design oriented prompt raised visual quality, 4.5 versus 3.0 on a 5 point scale, without lifting function, and a one paragraph paraphrase of its directive reproduced the entire lift. The practical lesson is to match the fix to the failure: most first run failures came from weak reasoning, which a stronger model or more effort prevents, not from visible flaws a checking tool would catch.

Scalable and reliable grading of command-line examinations remains a challenge in computing education, where rising enrolments make manual marking difficult and rule-based autograders cannot handle partial credit, equivalent solutions, or syntactic variation. This paper evaluates whether four frontier Large Language Models (GPT, Claude Opus, Gemini, and GLM) can approximate expert judgment when grading short Linux/bash command responses. The study adopts a four-level cognitive taxonomy that combines cognitive complexity and operational impact, ranging from information retrieval (L1) and basic file manipulation (L2) to structural operations (L3) and advanced system management (L4). The models were tested with two prompt variants, a minimal baseline and a rubric-enhanced version, on 1200 real responses from second-year Computer Engineering students independently graded by three expert instructors. Gemini~3.0 Pro with rubric-guided prompting achieved the highest human-AI agreement (ICC(3,1) = 0.888, MAE = 0.10, Bland-Altman bias = -0.014). Agreement declined consistently as taxonomy level increased, with the largest discrepancies at higher levels. Across all models, rubric quality had a larger effect than provider choice, with structured prompts consistently improving agreement. These results show that question complexity is a reliable predictor of the difficulty LLMs face in grading accurately, and they establish a principled, taxonomy-based framework for determining which questions are suitable for AI-assisted grading and which require human review, while also providing a transferable evaluation protocol and prompt templates.

Reinforcement learning (RL) can overcome the demonstration-coverage limitation of imitation learning (IL) by allowing robots to improve through trial-and-error interaction beyond the states observed in demonstrations. However, deploying RL on real robots remains constrained by high interaction costs, since each physical rollout is costly and reflects only one realized action-outcome path. To address this challenge, we propose WorldSample, a physically grounded data augmentation framework for real-robot RL that closes a real-synthetic loop between physical rollouts, world-model generation, and policy improvement. Grounded on real rollouts, WorldSample generates high-fidelity synthetic transitions through a post-trained world model, which greatly lowers the visual hallucination. Specifically, rather than simply using these transitions as real-world experience, WorldSample introduces Policy-Paced Learning (PPL) to regulate the training process through sample selection and scheduling, balancing useful augmentation against value overestimation and mitigating the hallucination-induced noise. Experiments on robot manipulation tasks involving contact-rich and precise tasks show that WorldSample improves policy success rate by 28% while reducing training steps by 59% compared with baselines. Furthermore, WorldSample improves world model visual fidelity by 19.4dB in PSNR and 0.47 in SSIM over demonstration-only post-training, validating the effectiveness of the real-synthetic loop for both policy and world model performance.

We present APEIRON, a distributed heterogeneous processing framework comprising both hardware architecture and software stack for multi-FPGA systems. Targeting smart trigger and data acquisition (TDAQ) systems in high energy physics, APEIRON spans the full software hierarchy: from low-level device drivers to a high-level dataflow programming model based on High-Level Synthesis. We describe the framework design, its core communication infrastructure, and a particle identification application for the NA62 experiment as a representative physics use case.

Federated learning (FL) enables collaborative model training across distributed devices without sharing raw data, making it suitable for privacy-sensitive robotic sensing applications. However, multi-agent systems generate heterogeneous and non-independent and identically distributed (non-IID) multimodal sensor streams that degrade conventional FL algorithms, while classical fusion modules introduce substantial parameter overhead and communication cost. This paper proposes QFedAgent, a hybrid quantum-classical personalized FL framework for multi-agent activity recognition. The approach integrates a variational quantum circuit fusion module that models accelerometer--gyroscope interactions through quantum state encoding and entanglement, requiring only 72 quantum rotation parameters versus 33K in classical multi-layer perceptron-based fusion, achieving approximately 10x total parameter reduction. Experiments on the OPPORTUNITY dataset under subject-based non-IID partitions demonstrate 97.7% mean test accuracy, confirming that parameter-efficient quantum fusion remains competitive with conventional federated baselines.

Active Few-Shot Learning (AFSL) adapts LLMs to specialized domains by identifying the most valuable unlabeled samples for annotation and use as few-shot demonstrations, effectively reducing human annotation costs while promoting high performance. However, existing methods typically rely on output-level signals for sample identification, such as predictive entropy or semantic similarities with test-time data based on external embeddings, which often overlook models' internal dynamics, which could pinpoint specific knowledge gaps. To bridge this gap, we propose NeuFS, a Neuron-Aware Active Few-Shot Learning framework that shifts the selection paradigm from output-level proxies to models' internal dynamics. NeuFS utilizes neuron activation patterns to represent sample directly, and includes a dual-criteria selection strategy that: (1) ensures few-shot sample diversity with neuron patterns for broader example coverage, while (2) prioritizing on identifying informative and challenging few-shot samples LLMs tend to hallucinate by quantifying neuron consensus. Experiments on three datasets demonstrate that NeuFS excels in both reasoning and text classification tasks, outperforming existing AFSL baselines. Ablation studies further highlight that internal neuron activations provide a more principled and effective selection signal than external embeddings, validating the superiority of the proposed NeuFS.

Autonomous robots often need to move their camera before they can act: to inspect an object, reveal an occluded region, or obtain a view that responds to a user's intent. While vision-language navigation translates instructions to base motion and vision-language-action policies map instructions to manipulation actions, language-conditioned camera motion remains comparatively underexplored as a first-class action. We formulate language-conditioned camera motion generation: given a current RGB observation and a free-form natural-language intent, predict a relative target camera pose for the next observation. This task is inherently non-trivial: viewpoint changes are driven by latent perceptual intentions, and a valid motion may operate at different semantic granularity, from entering a room to looking around a corner, inspecting a visible object, or revealing an occluded detail. To model this structure, we mine multi-intention camera-motion supervision from egocentric video, pairing plausible intents and observation-gain descriptions with relative SE(3) target poses. We propose LIME, a vision-language camera-motion generator that combines an auto-regressive observation-gain output with a continuous flow-matching pose head. This design lets the model jointly predict what the next view should reveal while representing multi-hypothesis target views. Across experiments and downstream robotic tasks, we show that LIME can learn to actively choose camera poses from passive human video, turning ordinary egocentric recordings into supervision for intent-aware active perception.

Natural Language Processing (NLP) has traditionally been published in its core disciplinary venues like ACL. However, advances in Large Language Models (LLMs) has led to a blurring of the disciplinary lines between NLP and general Machine Learning (ML), with authors regularly publishing in venues from both fields. Here, we ask whether the disciplinary center of gravity is shifting. Using NLP research published from 2010 to 2026 and studies of both established and new authors, we find that a migration is taking place. First, comparing the pre- and post-LLM eras, established authors lost 19.2pp of share at flagship *ACL main-conference tracks while gaining 14.8pp in the newer Findings tracks, and general ML venues rose 8.6pp, even when adjusting for parallel growth in the fields. Second, among newer authors who debut with at least three first-author NLP-topic papers, the share whose work appears mostly at *ACL venues fell from 84% (2019) to 74% (2024), while the share appearing mostly at general ML venues rose from 5% to 21%. Using causal inference techniques, we estimate that these general ML venues confer a significant citation premium, which influences venue selection. Together, these results point to a significant shift in where NLP research is published.

Here we describe the quantum gas analysis and inference (Q-GAIN) Python package, which enables rapid deployment of machine learning (ML) and physics-informed analysis techniques for cold-atom experiments. Out of the box, Q-GAIN implements classification, object detection, and physics-informed metrics for feature detection in images of atomic Bose-Einstein condensates (BECs). Q-GAIN encourages a natural, module-based workflow: starting with data loading and preprocessing, followed by ML-based feature identification, and ending with conventional analysis techniques. We demonstrate this modularity by configuring Q-GAIN for three ML tasks. First, we demonstrate the basic workflow of the Q-GAIN framework by implementing the standard task of classifying handwritten digits from the MNIST dataset. Then, we re-implement our earlier soliton detection (SolDet) package in the Q-GAIN framework, enabling the detection and analysis of solitonic excitations in time-of-flight data. Finally, we develop an object-detection tool that identifies quantized vortices in images of ring-shaped BECs.

Large Language Models (LLMs) have demonstrated remarkable capabilities in 3D indoor synthesis for Manhattan environments. However, existing methods often fail to capture plausible object layout patterns in non-Manhattan settings, primarily because they struggle to model non-orthogonal spatial relationships, leading to high geometric violations and low physical fidelity. To address this challenge, we propose SPG-Layout, a novel text-driven framework designed to generate physically plausible indoor scenes within complex non-Manhattan environments. Specifically, we first utilize statistical priors of object distributions to guide the training process, enhancing environmental understanding and fidelity. Furthermore, mirroring human design workflows, we adopt a hierarchical layout strategy that prioritizes the placement of large objects, thereby substantially minimizing layout violations. By synergizing these components, SPG-Layout achieves a balanced optimization of semantic realism and physical plausibility. To evaluate performance in these complex settings, we constructed a new benchmark comprising 500 diverse non-Manhattan environments. Extensive experiments demonstrate that SPG-Layout consistently and significantly outperforms existing methods across both Manhattan and non-Manhattan environments. The code will be publicly released.

Image encoders trained with LeJEPA can deliver strong features for downstream tasks, but, like other image-level self-supervised methods, typically require large training datasets. Aligning representations at the level of objects rather than whole scenes promises greater data efficiency, but doing this in a completely self-supervised way, effectively jointly partitioning a scene and representing its objects, is unstable: the two are locked in a cyclic dependency, partitioning requires meaningful representations, while meaningful representations require consistent partitioning. We sidestep this instability by taking object masks as given during training, using cheap, off-the-shelf SAM proposals. We extend LeJEPA - whose distributional anti-collapse objective ports naturally from whole images to variable-sized sets of objects - to align object-centric representations rather than whole images. An additional instance-separating loss, which treats other objects in the same scene as negatives, further boosts downstream performance. Across two model scales and 10-100% of COCO, object-level LeJEPA outperforms image-level LeJEPA on tracking (DAVIS), classification (ImageNet-1k), segmentation (ADE20k), and re-identification (NAVI).

Decision-time planning with action-conditioned world models has become a popular paradigm for embodied control. However, the standard planning cost judges a candidate solely by how close its predicted terminal state lies to the goal, leaving the realizability of the intermediate transitions unchecked -- a predicted trajectory can look convincing while the environment rollout drifts away from it. In this paper, we propose ACID, a decision-time planning framework that introduces cycle action consistency: the action inferred backward from a predicted transition by an inverse dynamics model should recover the one that was conditioned on. We fold this per-step residual into the planning cost via a scale-invariant adaptive weight. Across four action-conditioned world models and six tasks spanning rigid and deformable manipulation, articulated control, and visual navigation, ACID consistently improves planning and matches the baseline's accuracy with substantially less planning compute.

Byte-addressable non-volatile memory (NVM) offers an opportunity to rethink storage engine architectures. While recent NVM key-value stores achieve high throughput for ingestion and point lookups, they omit or under-specify the support for the richer interface guarantees required by modern databases. Production key-value engines (e.g., RocksDB) provide point-in-time snapshots, consistent iterators, and atomic batches-features essential for implementing transactions and concurrency control. We present FlintKV, an NVM-optimized skiplist-based storage engine that natively supports the full API of production key-value stores. FlintKV supports both atomic batch writes and snapshot-consistent iteration efficiently while guaranteeing durable linearizability. FlintKV can be deployed standalone or its durable skiplist can be integrated into existing NVM stores to enhance their capabilities. Central to FlintKV is a novel flat-combining based concurrency control algorithm that leverages multi-versioning and carefully co-designed persistence mechanisms to ensure high performance and scalability. Our empirical evaluation shows that FlintKV can achieve up to a 75% improvement in end-to-end throughput over prior work.

Steering and monitoring activations in Large Language Models (LLMs) are increasingly used for both safety and interpretability. Early work assumed behaviours are encoded along single linear directions, but recent findings suggest complex behaviours, such as the refusal to answer harmful queries, live in multi-dimensional subspaces. However, existing methods for extracting these subspaces are computationally expensive, which becomes prohibitive on reasoning models who produce long reasoning traces. By adapting the Recursive Feature Machine (RFM) algorithm -- which can be computed efficiently -- with a probe-informed initialization, we are able to identify the multi-dimensional refusal subspace in seconds, on reasoning (Qwen 3) and non-reasoning (Qwen 2.5) models. While RFM allows for faster subspace identification, it also showed better performances on the ablation task than its alternatives. More work is planned to better understand the relations between subspaces found by different methods. If confirmed, RFM could be a cheap and scalable complement to existing subspace-extraction methods in LLMs.

Large Language Model (LLM) inference workloads are a rapidly growing contributor to data center energy consumption. Optimizing these deployments requires matching specific LLMs to the most efficient GPUs, but operators currently lack the tools to do so without exhaustively profiling each combination. While some predictive models exist, they still require profiling data and struggle to generalize to hardware unseen during training. To address this, we introduce \textit{WattGPU}, featuring two predictive models for mean GPU power draw and Inter-Token Latency (ITL). Our approach leverages only publicly available LLM metadata and GPU specifications, eliminating the need for hardware access or profiling while enabling generalization to unseen NVIDIA server-grade GPUs and LLMs. We evaluate our models using rigorous leave-one-GPU-out and leave-one-LLM-out cross-validation on a dataset of 42 open-source LLMs (0.1B--27B parameters) and 8 GPUs under both offline and server scenarios. The mean power draw model achieves a median absolute percentage error of $\leq3.4\%$ for offline and $\leq13.5\%$ for server scenarios on unseen GPUs, while the latency model achieves $\leq8.5\%$ in server mode, both maintaining strong GPU ranking correlations for server scenarios (Kendall $τ\geq0.76$). Compared to standard physically grounded baselines -- Load-Scaled Thermal Design Power (TDP) for power draw and roofline for latency -- our models reduce median absolute percentage error by approximately 4$\times$ on unseen LLM-GPU combinations for server scenarios or approximately 2$\times$ for completely unseen GPUs. WattGPU's data and code are publicly available at https://github.com/maufadel/wattgpu.

How can Large Language Models (LLMs) solve problems they currently cannot? Repeated sampling scales test-time compute but GPU cost grows linearly with attempts, while reinforcement learning (RL) with verifiable rewards improves single-attempt accuracy at the expense of sample diversity. Both strategies ultimately fail when the base policy has near-zero probability of producing a correct solution: no amount of sampling or gradient signal can overcome a search space that is simply too large. We take a different approach: rather than sampling harder, we make the task easier by decomposing problems into smaller, independently solvable sub-functions whose implementations can be recombined. Since off-the-shelf models are not trained for this modular generation, we introduce DecompRL, an RL algorithm that explicitly learns to decompose and implement hierarchical code structures. Recombining $k$ implementations of $n$ modules yields up to $k^{n}$ candidate solutions, shifting the bottleneck from GPU inference to cheap CPU evaluation and cutting GPU token cost by $\sim$50$\times$. On LiveCodeBench and CodeContests (Qwen~2.5~7B, Code World Model~32B), DecompRL outperforms standard and diversity-optimized RL baselines beyond $10^5$ tokens per problem, solving problems that standard generation cannot reach.

Coding agents are capable; human oversight is the bottleneck. Unconstrained agents introduce security risks, erode codebase scalability, and make human review increasingly costly. We argue that the same methods used for decades to manage large human engineering teams: access control, network policies, strict coding conventions enforced by tooling; transfer directly to coding agents, and are cheaper (in token) than recent agentic scaffolding. We sketch a start-to-end system on this principle, and report a controlled experiment in scalable oversight: a small reviewer (Gemma 4 e4b) inspects a Python codebase containing 11 inserted backdoors. Recall rises from 54.5% (unconstrained, no tools) to 90.9% (constrained substrate plus a ~200-LoC `docs` CLI), with substrate and tools contributing independently. We choose Python deliberately: substrate-level oversight gains are largest where the language gives the fewest guarantees by default; the principles extend to languages like Rust.

NASA and its data centers hold thousands of geoscience datasets and tools like Worldview, Giovanni, the Science Discovery Engine, and Harmony. Finding the right one is hard even for domain experts. We present an agentic search system, deployed as a public service for the geoscience community, that takes a natural-language research query and returns the matching datasets and tools. We demonstrate that, in the era of large language models, the latent value of knowledge graphs (KGs) can be substantially amplified through agentic search. From the NASA Earth Observation Knowledge Graph (NASA EO-KG) we derive NASA-EO-Bench, an open benchmark of 47k query-dataset pairs (21k task-based queries). A neural scorer fine-tuned on NASA-EO-Bench beats cosine and BM25 baselines. Further combining it with BM25 via score fusion raises both Recall@10 (R@10) and MRR by over 5x. On top of this supervised pipeline, we add a zero-shot agentic reranking stage that, without any additional training, lifts MRR by 28% on a stratified N=200 subset, showing that LLM reasoning is complementary to supervised retrieval.

While Vision Transformers have achieved remarkable success across computer vision and language applications, the geometric evolution of their internal representations throughout training remains insufficiently understood. Existing analyses primarily focus on attention mechanisms and downstream performance, leaving the evolution of representation geometry largely unexplored. In this work, we present Transformer Geometry Observatory-II (TGO-II), a representation geometry analysis framework designed to investigate how Transformer representations evolve during supervised training. TGO-II analyzes Vision Transformer (ViT-Small/16) representations using Centered Kernel Alignment (CKA), Singular Vector Canonical Correlation Analysis (SVCCA), Two-Nearest Neighbor Intrinsic Dimensionality (TwoNN-ID), and token covariance analysis. Our experiments reveal three key observations. First, both CKA and SVCCA progressively decrease throughout training, indicating increasing representational specialization across Transformer layers. Second, intrinsic dimensionality consistently increases before stabilizing, suggesting progressive expansion of the representation manifold into a larger set of locally accessible degrees of freedom. Third, token covariance and coupling analyses demonstrate that strong token interaction structure persists throughout training, challenging the hypothesis that increasing representational complexity arises primarily from progressive token independence. These findings suggest that representation complexity and layer specialization emerge simultaneously during training. Manifold expansion appears to occur without token decoupling. Together, these observations motivate a new hypothesis in which Vision Transformers increase representational complexity through progressively richer transformations while preserving strong token interaction structure during learning.

LLM-based retrieval-augmented generation (RAG) is increasingly used for automated fact-checking (AFC) and related tasks. By grounding LLM outputs in retrieved evidence, RAG-based systems provide transparent justifications while allowing external information to be updated independently of the underlying model. However, existing approaches often assume retrieved evidence is reliable, although real-world information may be conflicting, outdated, and can originate from unreliable or biased sources. Recent work on *source-critical reasoning* addresses this challenge through media background checks (MBCs) (Schlichtkrull, 2024), which assess the credibility of evidence sources to support downstream fact verification. However, generating MBCs relies on costly proprietary search APIs, limiting reproducibility. To mitigate this issue, we introduce MEDIAREF, a publicly available knowledge store of web-sourced documents that enables reproducible, low-cost evaluation of MBC generation across 200 media sources. We describe a reproducible methodology for constructing and updating the collection, assess widely used LLMs on the MBC generation task, and demonstrate that MEDIAREF supports higher-quality MBC generation through both automatic and qualitative evaluation.

This paper describes the participation of HULAT2-UC3M in the Spanish track of MER-TRANS 2026, a shared task on multilingual Easy-to-Read translation. Three fully automatic Spanish runs were submitted. RUN1 and RUN2 used a LangGraph-based multi-agent workflow combining Gemini 2.5 Flash and RigoChat-7B-v2, parallel generation strategies, internal quality signals, Event-Condition-Action routing, controlled editing and traceable decisions. RUN1 used the base workflow, while RUN2 activated an additional lexical-support layer based on a glossary and lexical resources. RUN3 was a RigoChat-based generate-evaluate-regenerate baseline with prompt engineering and LoRA-based adaptation. The official leaderboard reports BLEU-Orig, BLEU-Gold, SARI and BERTScore. During development, additional internal signals were also inspected, including semantic fidelity, readability, lexical simplicity, syntactic clarity and factual consistency. According to official SARI, RUN1 was the best HULAT2 run, with 44.0543 points, followed by RUN2 with 43.1049 and RUN3 with 38.5136. These results indicate that, in this task setting, signal-guided multi-agent routing outperformed the linear regeneration baseline. They also show that adding lexical support did not automatically improve reference-based scores. Further segment-level and document-level analysis are required to assess readability, factual consistency and user-oriented adequacy.

Recent advances in agentic AI are producing increasingly complex autonomous systems that integrate large language models, world models, optimization engines, specialized neural architectures, autonomous platforms, and human operators. While much current research focuses on improving reasoning capabilities, safety-critical real-time deployment also requires bounded and verifiable coordination among heterogeneous components operating concurrently under uncertainty. Software-mediated coordination presents fundamental limitations in domains where bounded latency, deterministic coordination, and enforceable safety guarantees are essential. Hence, we propose a hardware-enforced semantic coordination architecture in which selected coordination semantics are implemented directly at the hardware level via field-programmable gate arrays (FPGAs). The approach builds on the Topic-Based Communication Space Petri Net (TB-CSPN) framework, which separates semantic reasoning from interaction management. In this approach, selected TB-CSPN coordination mechanisms are mapped onto FPGA primitives, creating a hardware-native semantic coordination layer. Focus is not on acceleration, but on enforcing temporal synchronization, semantic gating, authorization constraints, and bounded coordination behavior directly in hardware. Semantic reasoning remains adaptive and software-driven, while embedded coordination semantics become deterministic.

Personalization changes what a model says to a user; we show that it can also change the reasoning trajectory used to justify the response. Modern LLMs personalize interactions by storing user attributes, preferences, and prior context, then injecting this information into future prompts. We study whether such memory reshapes reasoning on open-ended questions where no single ground-truth answer exists. To quantify this effect, we introduce DRIFTLENS, a ground-truth-free framework that maps each expressed reasoning step to a value category and measures divergence between a question's no-memory trajectory and its trajectory under injected user-attribute memory. We first validate that DRIFTLENS distinguishes content-free pragmatic noise from substantive reasoning changes. Across four LLMs and 10 user-attribute categories, including age, occupation, and disability, user-attribute memory induces medium-to-large reasoning drift above each model's pragmatic-noise floor, even when final answers remain fluent, on-topic, and plausible. We then evaluate GRPO- and DPO-based post-training methods for reducing drift. Both reduce drift, but neither uniformly dominates; effects on downstream capability, helpfulness, and instruction following are model-and reward-dependent. These results suggest that memory-induced reasoning drift is a measurable and only partly mitigated failure mode of personalized language models.

Over 285 million people worldwide live with a visual impairment, for whom everyday tasks such as avoiding obstacles, locating personal belongings, recognizing familiar faces, or handling cash remain persistent obstacles to personal autonomy. Existing assistive applications are typically limited to recognizing predefined categories, depend heavily on cloud connectivity, or require dedicated hardware. We present VisionAId, an Android application that turns a commodity smartphone into a real-time visual assistant. The system integrates six on-device deep learning models (metric monocular depth estimation, instance segmentation, visual and facial embeddings, face detection, and a custom banknote detector) running entirely through ONNX Runtime, with an optional cloud large language model (Google Gemini Flash) used only for narrative scene description and automatic object labeling. A distinctive contribution is a few-shot pipeline for personal objects: the user photographs an object from several angles, and the system later locates that specific instance in the environment, guiding the user toward it with augmented-reality markers, spatial audio, and distance-proportional haptics. All feedback is multimodal (Romanian speech synthesis, voice commands, vibration). On a reference device (Samsung Galaxy S21 Ultra), INT8 quantization reduces depth latency from ~1200 ms to ~491 ms, the custom banknote detector reaches an mAP@50 of 0.986, and metric depth is calibrated to below 1 cm of error within 3 m.

Compiler missed optimizations refer to cases in which compilers failed to optimize certain code. It takes many compiler developers' efforts to implement or patch such missed optimizations. In this paper, we present a systematic study of how well agents patch compiler missed optimizations. We identify a significant challenge that patching a missed optimization requires more than just fixing the reported case, and instead requires generalizing to similar cases. We construct a benchmark of real-world LLVM missed optimization issues and compare agent-generated patches with patches from developers in terms of optimization scope. Our results show that coding agents often optimize the given examples, but many generated patches either cover only part of the developer-intended scope or partially overlap with it; in some cases, they further generalize beyond the reference patch. We further introduce historical-knowledge augmentation techniques that leverage prior LLVM optimization pull requests through retrieval and distillation, showing that they improve developer-aligned generalization and yield practical benefits when applied to real-world IR.

LLMs stage a new form of cultural encounter that is massive, automated, and monolingual. Literary disciplines have always negotiated cultural struggles with comparative reading of literature, narratological and poetic analysis, critical theory, world literature, and translation. These tools have now become indispensable for building culturally literate AI. The essay develops a layered framework toward more nuanced textual models and pluralistic interpretations of AI, emphasizing the natural intersections of literature and AI development, connecting current debates in critical theory with structural monolingualism, and suggesting a new application of world literature approaches to address global AI textuality through macrostructure, circulation, and untranslatability.

Evaluations of LLM personas via psychometric questionnaires typically rely on aggregate scores, discarding within-instance correlation structure. We test whether this geometric structure is intrinsic or frame-dependent. Constructing within-instance correlation matrices from IPIP-50 responses, we analyze geometry on SPD manifolds under manipulated question orderings in GPT-4o simulating American and Chinese-American personas. We find that persona expression comprises two dissociable components: aggregated features (Big Five scores) degrade under randomization (21% drop) but are frame-robust; geometric features (SPD manifold) collapse under frame misalignment (42% drop) but recover substantially (to 84%) under shared frames, surpassing aggregated features (76%). This collapse-recovery pattern reveals that persona geometry is not intrinsic but a frame-dependent coordination pattern encoding information invisible to aggregation. Our findings establish a dual-nature framework for LLM personas, frame-dependent geometry versus frame-robust aggregates, necessitating frame-aware evaluation and challenging static trait conceptions.

Quantum Fast-Weight Programmers (QFWPs) store temporal information in dynamically programmed variational-circuit parameters rather than in nonlinear recurrent hidden states, offering a practical route to quantum sequence modeling. Self-Modulating QFWP improves this framework by using input-dependent gates for both new fast-weight updates and the accumulated fast-weight state, but its unbounded old-state multiplier can diverge in long-sequence regimes. We propose a bounded old-state modulation rule that applies a sign-preserving tanh gate only to the recurrent memory branch while leaving the additive update and new-update modulation unchanged. We evaluate standard QFWP, full Self-Modulating QFWP, Only-New, and Only-Old variants on two CUDA-Q quantum-dynamics forecasting tasks and on Milan SMS telecommunication activity prediction. The quantum-dynamics results show that old-state modulation is the most consistent source of improvement over Standard QFWP, and that bounding the old-state gate removes long-sequence divergence while improving aggregate robustness. On Milan SMS forecasting, the original unbounded Self-Modulating QFWP converges across the tested grid and shows its clearest gains at longer input windows, with behavior close to the Only-Old ablation. These findings identify accumulated-memory modulation as the key mechanism of Self-Modulating QFWP and bounded old-state gating as a targeted stabilization strategy.

In today's data-driven world, students often struggle with interpreting visualisations due to limited visualisation literacy. Data comics have emerged as a promising medium to enhance engagement and understanding, but their educational value has seen little empirical examination, partly due to the effort required to create them. Recent advances in Generative AI (GenAI) offer a scalable solution to this challenge. We conducted a within-subjects study with 60 university students, comparing conventional visualisations with data comics, created with assistance from GenAI tools, across information retrieval and comprehension tasks. Students consistently performed better with data comics, particularly in insight comprehension tasks, independent of prior visualisation literacy. Students also commented data comics as more engaging and easier to understand, though concerns were raised about GenAI-driven misinformation and ownership. Our findings highlight the potential of data comics as a potentially effective tool for data communication in education, while underscoring the need to address ethical concerns related to AI-assisted creation.

Monocular relative pose sensing is a central perception problem in non-cooperative rendezvous and on-orbit servicing. In spacecraft images, however, weak surface texture, thin appendages, illumination changes, and partial occlusion often leave only sparse and unstable geometric evidence. This article presents GAP-GDRNet, a geometry-aware attention-enhanced framework for monocular RGB-based 6D pose sensing. The method follows the geometry-guided direct regression paradigm of GDR-Net and modifies two points in the pipeline: an attention-based feature refinement (AFR) module is placed before dense geometric prediction, and a patch-level geometric self-attention (PGSA) module is inserted into Patch-PnP. AFR reinforces global spacecraft structure together with local weak-texture cues; PGSA then relates downsampled geometric patches before final pose regression. A Blender-based annotation process supplies target masks, visible-region masks, dense model-coordinate maps, camera intrinsics, and 6D pose labels for supervised training.

LLM coding agents increasingly rely on third-party agent skills from public marketplaces, which execute with the agent's privileges and create a software supply-chain attack surface: a malicious skill can steal credentials, exfiltrate source code, or install backdoors. Existing defenses use static skill scanners based on pattern matching or LLM-as-judge analysis, but it remains unclear whether they withstand adaptive evasions that preserve malicious behavior while changing payload appearance. This paper first presents an adversarial study of existing skill scanners through SkillCloak, a payload-preserving evasion framework that keeps the attack semantics intact while transforming their visible form. SkillCloak uses two complementary strategies: Structural Obfuscation, which rewrites visible payload indicators into semantically equivalent forms, and Self-Extracting Skill (SFS) Packing, which hides malicious components from the install-time view and restores them during agent execution. Across eight scanners and 1,613 in-the-wild malicious skills, SFS Packing bypasses every scanner at over 90%, while Structural Obfuscation bypasses over 80% on most static scanners and reaches 96% on a hybrid scanner, showing that appearance-based auditing is insufficient. Motivated by this finding, we propose SkillDetonate, a behavior-centric runtime auditor that executes skills in a sandbox and detects malicious effects through OS-boundary information-flow evidence rather than install-time appearance. SkillDetonate combines on-demand closure lift, which observes instructions materialized during execution, with marker-based taint analysis, which tracks sensitive-data flows across the agent context, files, processes, and network operations. The results show that SkillDetonate detects 97% of attacks at a 2% false-positive rate and sustains 87% detection on real-world malicious skills.

Large Language Model (LLM)-based agents increasingly automate software engineering tasks through reusable skills, natural-language instruction documents that guide planning and execution. Open skill marketplaces enable users to assemble agents by co-activating community-contributed skills, but marketplace operators typically audit skills in isolation. As a result, individually benign skills may interact to redirect an agent toward unintended objectives, which we term implicit intents. Detecting such intents is challenging because the effect emerges only through skill composition, execution environments are often unavailable at admission time, and the space of possible co-activations grows exponentially with marketplace size. In this paper, we formulate implicit-intent discovery as a fuzzing problem over skill compositions, where skill compositions are the unit under test, planning artifacts expose agent intent before execution, and deviations from a skill-free baseline serve as a differential oracle. Based on this formulation, we propose skillfuzz, the first execution-free testing approach that extracts structured skill contracts and uses contract-guided Monte Carlo Tree Search to prioritize potentially conflicting compositions. Across representative skill-marketplace workloads, skillfuzz discovers over 1,000 distinct implicit intents under a fixed query budget, confirms more than 80% of the highest-risk flagged compositions during execution-time validation, and identifies substantially more high-severity implicit intents than alternative search strategies while exploring only a fraction of the pairwise interaction space they require.

Transformer architectures have shown strong potential in time series forecasting, where multi-head self-attention is widely used to capture temporal dependencies across historical timestamps. However, standard self-attention has quadratic time and memory complexity with respect to the look-back length. This cost may limit its use in resource-constrained or high-throughput forecasting systems, where fast and memory-efficient inference is important. Through qualitative and quantitative analyses, we observe that self-attention maps in time series forecasting often contain redundant patterns across different timestamps. This phenomenon can be related to the repeated temporal patterns and relatively stable temporal correlations in many real-world time series. Motivated by this observation, we propose Self-Gating Attention (SGA), a plug-and-play attention mechanism that represents the attention score with a shared learnable matrix and an input-dependent residual component. The shared matrix captures common attention patterns, while the residual component captures input-dependent variations. In this way, SGA avoids the query and key projections used in standard attention score computation, leading to linear time and score-matrix memory complexity with respect to the look-back length. We integrate SGA into several forecasting backbones and compare it with standard self-attention and lightweight attention variants on nine publicly available real-world datasets covering electricity, finance, weather, medical monitoring, human activity, and climate records. The results show that SGA improves inference efficiency on public benchmarks while maintaining competitive forecasting performance against state-of-the-art attention mechanisms. These benchmark results provide deployment-oriented evidence.

Humans can selectively attend to a target sound and estimate its direction in complex scenarios, whereas such selective localization remains challenging for current deep learning-based systems. Sound source localization (SSL) has achieved remarkable success with deep learning, yet most methods localize all active sources without selectivity. Conversely, target sound extraction (TSE) extracts sources using multimodal prompts but typically fails to preserve the multichannel spatial information required for accurate localization. To bridge this gap, we formulate the task of prompt-guided selective target sound localization and propose SelectTSL, an end-to-end architecture that localizes only the user-specified target in multi-source acoustic scenes. Specifically, we design a target-aware selective localization strategy that employs a Prompt-Guided Selective Attention Module (PGSA) to generate prompt-informed embeddings. These embeddings guide an inter-channel phase difference (IPD) enhancer to refine raw phase cues, fusing with target magnitudes to jointly estimate direction of arrival (DoA) and target-source cardinality, i.e., the number of target sound sources. This coupled design effectively focuses on the user-specified target spatial cues for selective localization and also handles time-varying numbers of target sources. Extensive experiments on both synthetic data and real-world recordings demonstrate that our proposed method consistently outperforms other baselines and exhibits robust generalization to real acoustic environments.

Hierarchical Navigable Small World (HNSW) graphs serve as the industry standard due to their logarithmic complexity and strong empirical performance. However, HNSW relies on greedy graph traversal, a heuristic that provides no theoretical guarantees of correctness. In this paper, we propose a novel "Certify-then-Rectify" framework that bridges the gap between the speed of heuristic search and the rigor of exact retrieval. Rather than discarding HNSW, our approach first employs a distribution-free statistical certifier to dynamically evaluate the quality of a standard HNSW search with minimal overhead. If certification indicates that the retrieved neighbors are of low quality, the framework safely escalates to a rigorous exact recovery algorithm. To make this exact recovery computationally feasible, we reinterpret the HNSW graph as a geometric spanner and utilize Extreme Value Theory to stochastically estimate its maximum empirical stretch factor. This allows us to mathematically bound the maximum distance of true nearest neighbors. Extensive evaluations on benchmark datasets demonstrate that our tiered framework delivers the average-case speed of HNSW while ensuring the worst-case correctness of exact search and outperforming other applicable approaches.

With the growing adoption of AI-powered coding assistants, organizations and developers are increasingly seeking to optimize their interaction with these tools. Prior research has largely focused on output quality and productivity gains, with limited attention paid to developers' well-being and interaction experiences. This paper presents a developer-centered empirical mixed-methods study to investigate how professional developers engage with Generative AI (GenAI) in their natural work environment. Controlled data collection sessions are combined with natural work periods. Results show that developers are generally satisfied with GenAI, particularly for monotonous, repetitive, and structured tasks, and report perceived efficiency and productivity gains. Copilot interaction type preferences differ by task type and complexity: While both in-code suggestions and chat-based prompting independently improve task efficiency and reduce perceived workload, combining these interaction types within a single task diminishes benefits. We propose a rule-of-thumb for selecting an interaction type based on task characteristics. During development-heavy tasks, results indicate that perceived cognitive load arises from AI interaction, while perceived productivity depends on AI output quality. Participation in this study positively influenced developers' awareness and intentional use of GenAI tools. These findings demonstrate the value of real-world, mixed-methods study designs to understand GenAI tools and developers' experiences with them.

Vibe coding democratizes software development by allowing users to generate code via natural-language (NL) interaction with large language models (LLMs). However, the code is reliable only when it faithfully implements the user's intent, which is difficult and labor-intensive for users to validate. Existing validation methods either rely on LLM-assisted automated testing, which suffers from prompt ambiguity and model fallibility, or involve users only in partial software artifacts such as prompts and test cases, which may overlook corner cases and program details. Motivated by a bug study of LLM-generated code, we find that detailed human feedback is essential, as failures often stem from underspecified requirements or subtle semantic deviations. This paper presents verifiable literate programming (VLP), a human-in-the-loop framework designed to make the review/validation process of LLM-generated code accessible to users at all programming levels. At its core, VLP proposes unambiguous NL-based documentation as a readable intermediate layer between prompts and code. The documentation demonstrates concrete program semantics and enables users to provide feedback on potential intent-code mismatches. It supports human-involved, end-to-end repair and validation via three techniques: (i) an NL-style literate language with unambiguous syntax and mostly deterministic code-to-documentation translation, (ii) LLM-based fine-grained mismatch detection that uses trace links between prompts and documentation to focus users' review effort on suspicious documentation lines, and (iii) a verification module that leverages user-validated documentation to derive API-usage checks and formal properties, which are then verified against the generated code using model checking. Our evaluation shows that VLP improves code pass@1 from 28.7%-73.2% to 65.4%-93.5% with reasonable user effort.

Autonomous-research agents have demonstrated end-to-end LLM automation in machine-learning sandboxes where execution provides calibration. Frontier physical science differs categorically: physical reasoning underlies every methodology choice, toolchains are often underdocumented, and calibration must come from external literature anchors - which unscaffolded agents cite but do not confront, hallucinating plausible, unverifiable results from internal priors. We present a pipeline that runs end-to-end from a corpus of 11,083 recent condensed-matter physics arXiv papers to a publication-grade manuscript with three substantive physics findings (here on altermagnetic piezomagnetism): the agent autonomously conceives a research direction by mapping the corpus, calibrates methodology by reproducing published references, conducts novel first-principles computations, and writes the manuscript - grounded in literature throughout, across 47 fresh-context sessions in six phases sharing only on-disk state, with 2,162 literature-consultation events. Fault tolerance emerges from redundancy: fresh-context isolation, distributed grounding, and adversarial review catch what any single session misses; pre- and post-pilot stages are fully autonomous, and pilot requires bounded human intervention only at reproduction failures - operational knowledge curation, not scientific direction. Two paired failure modes - a pre-architecture baseline and a no-pilot ablation - isolate structurally enforced numerical confrontation at calibration checkpoints as the operative grounding mechanism. The primitives, characterized failure modes, and quantified intervention pattern lay a foundation for autonomous research in high-stakes scientific domains beyond computational physics.

We introduce stale synchronous parallel as a mode of execution in parallel sparse triangular linear system solve and present a general directed-acyclic-graph scheduler capable of producing such schedules. Stale-synchronous-parallel schedules allow the overlap of synchronisation and compute which results in a geometric-mean speed-up of $7$-$30\%$ of our scheduler, ElasticDivide, over state-of-the-art synchronous scheduler GrowLocal on an ARM machine using 48 cores. On an x86 machine using 48 cores, we report geometric-mean speed-ups of $19$-$60\%$ over SpMP.

The one-dimensional bin packing problem (1D-BPP) is a canonical NP-hard combinatorial optimization problem with broad industrial applications. We propose RL-HGGA, a hybrid algorithm that integrates Falkenauer's Hybrid Grouping Genetic Algorithm (HGGA) with a tabular Q-learning controller. Rather than applying genetic operators at fixed probabilities, a Q-learning agent dynamically selects among eight macro-actions -- including BPCX crossover, light and heavy mutation, Martello-Toth local search, and population restart -- based on an eight-dimensional state representation encoding generation progress, stagnation level, optimality gap, average fitness, population variance, and average bin fill rate. The agent is trained with an epsilon-greedy policy over 400 episodes, with epsilon decaying to 0.05. Experiments on standard benchmark families (Falkenauer T/U, Scholl 1-3, Hard28) show that RL-HGGA achieves an average optimality gap of 0.95% -- competitive with HGGA (0.75%) and well below FFD (2.47%) -- while reducing mean computation time from 64.22 s to 1.29 s, a 50x speedup. These results demonstrate that learned adaptive operator selection can achieve near-HGGA solution quality at a fraction of the computational cost.

Several SLOG test categories explicitly involve directional distinctions (modifier position shifts, argument extraction positions), yet AM-Parser, the previous SOTA, uses an AM algebra whose operations do not encode direction. We redesign the symbolic backend around CCG directed types (deterministic CKY + single linear decoder, 30K learnable parameters). Under the same BERT-base encoder, the system achieves 75.9$\pm$6.4% LF exact match, surpassing AM-Parser (70.8$\pm$4.3%). Per SLOG's own category groupings, gains are highly directional: the CCG system outperforms AM-Parser on all 5 position-shift categories (+29.9pp), while AM-Parser outperforms on all 6 recursive-depth categories. Replacing the encoder with DeBERTa-v3-large yields 90.7$\pm$4.9%, with the largest encoder gains in recursive-depth categories, complementary to directionality's gains. Directional representations shift the bottleneck from the symbolic layer (AM-Parser's 0% category ceiling) to the neural layer, which improves with encoder upgrades.

We consider grassroots platforms -- distributed systems of agents consisting of people identified by self-chosen public keys and their machines (smartphones) -- and wish to make them secure against \emph{major faults}: the loss of their private keys and/or their smartphones. As grassroots platforms have no global resource to rely on for recovery, our peer-based solution is based on: (\ia) \emph{a grassroots social graph} in which agents establish and maintain friendships; (\ib) \emph{identity custodians}, designated by each person, and (\ic) \emph{state custodians}, which are grassroots platform-specific. Upon a person experiencing identity loss, and given a willing supermajority of the identity custodians of the person, the friends of the person replace the old public key with the new one across the graph and restore friendships, where all friends serve as state custodians for the social graph. Choosing a new keypair, obtaining a new smartphone, and convincing identity custodians to will a change of key all happen ``off-chain''. Recovery from machine loss without loss of key (e.g. smartphone run over by truck, or its memory wiped) is simpler, requiring only the help of state custodians. We specify the social graph and its secure version as guarded multiagent atomic transactions, and implement the secure social graph via communicating volitional agents, an eventually synchronous message-passing model one step closer to implementation. We prove the implementation maps runs with recoverable faults to correct runs of the specification. We follow a similar path for grassroots coins and bonds, showing a common core as well as the platform-specific aspects of state recovery: a currency's single-writer log is recovered exactly, the recovered sovereign resuming without double-spending.

Linear-attention and state-space language models compress the prefix into a fixed-size recurrent state, yielding O(1) memory at the cost of a lossy exact memory: when many key--value associations compete, earlier facts are overwritten and needle recall degrades. Inspired by Complementary Learning Systems, we give linear attention a hippocampal complement. HOLA (Hippocampal Linear Attention) keeps the usual delta-rule state as a compressive memory and adds a bounded exact KV cache, forming a semiparametric test-time memory: the state models linearly compressible structure, while the cache stores associations that should not be forced through that state. The cache writes without a learned eviction module, keeping tokens with large beta * ||e||, the prediction residual actually committed to the state; a decoupled RMSNorm-gamma cache read then turns these exact KV pairs into sharp retrieval rather than soft averaging. At 340M parameters trained on 15B SlimPajama tokens, HOLA lowers Wikitext perplexity from 27.32 to 22.92 (-16.1%), below a full-attention Transformer++ (26.88), and improves LAMBADA perplexity from 30.95 to 30.26. It also achieves the best linear in-context retrieval and remains much more robust than GDN or a matched HOLA+recency cache on RULER needle-in-a-haystack recall out to 32k tokens (16x its training length).

In the automotive domain, in-car scene understanding (ISU) enables the detection of safety-critical events, such as driver distraction, and supports drivers or passengers by analyzing the in-car scene and adapting the environment (e.g., ambient lighting). The industry is increasingly exploring vision-language models (VLMs) to interpret camera-recorded in-car scenes and extract information for downstream reasoning tasks. However, VLMs may generate incomplete, erroneous, or misleading scene descriptions, highlighting the need for systematic testing. Collecting real in-vehicle data is costly, difficult to scale, and often infeasible, particularly in early design stages. In this paper, we present ISU-Test, an automated testing approach that combines rendering-based scene generation with search-based testing to evaluate ISU systems. By framing testing as an optimization problem and systematically modifying scene parameters, our method generates diverse in-car scenarios and explores a wide range of configurations. We evaluate ISU-Test on both an industrial prototype and open-source VLMs across two case studies: question answering and captioning, comparing against randomized scenario generation. Results show that ISU-Test significantly outperforms the baseline, achieving up to 10 times higher failure rates and up to 3.6 times higher failure coverage.

LLM coding agents are increasingly deployed to act autonomously on real production infrastructure. They execute shell commands, modify repositories, and call operational APIs. However, completing a task is not sufficient for safety. A wrong action can cause severe consequences. Existing agent benchmarks largely emphasize task completion, leaving open how agents behave under benign but underspecified instructions. We present UnderSpecBench, a benchmark for measuring action-boundary violations in coding agents (i.e., Claude Code, Codex, and OpenCode) on DevOps tasks. UnderSpecBench includes 69 task families grounded in documented incidents, CVEs, or tool behavior and organized across four DevOps capability domains and nine operational control surfaces. To isolate underspecification from task difficulty, each task keeps the same environment and ground-truth safe action while varying the instruction along three axes: intent clarity, target certainty, and blast radius. The resulting 2,208 prompt variants are evaluated with deterministic, side-effect-based oracles that separate Safe Success, Wrong Target, and OverScope outcomes; non-action runs are further classified as clarification, refusal, or deferment. Across five agent x model configurations using OpenCode, Claude Code, and Codex, the evaluation results show that underspecification does not mainly make agents fail; it makes them guess. 55.8-67.8% of runs violate at least one boundary. Target underspecification sharply degrades action quality, while blast-radius cues barely reduce action propensity. These findings show that completion-centric evaluation can overstate safe autonomy and motivate mitigations at the model, harness, and system layer.

Neural quantum states (NQS) provide a flexible and scalable framework for approximating quantum many-body wavefunctions. Among NQS parameterizations, autoregressive models are especially attractive because they enable exact, independent sampling from the Born distribution, avoiding the autocorrelation and mixing issues of Markov chain methods. Yet their optimization remains comparatively underexplored: Adam is a scalable method but ignores function space geometry, while stochastic reconfiguration is principled but costly and numerically fragile in large models. To address this gap, we show that variational energy minimization can be viewed as an advantage policy-gradient problem over the Born distribution, motivating trust-region optimization for NQS training. We introduce Proximal Wavefunction Optimization (PWO), a principled trust-region algorithm that clips probability-ratio changes in the amplitude channel and phase increments in the phase channel. PWO avoids explicit matrix inversion, reuses samples across multiple updates, and combines the scalability of first-order optimization with theoretical guarantees. Across Ising and frustrated $J_1$-$J_2$ one- and two-dimensional spin systems, PWO improves stability and wall-clock convergence over Adam, minSR, and SPRING. Finally, we fine-tune a $1.5$B-parameter RWKV-7 model, demonstrating NQS optimization at a scale over three orders of magnitude beyond prior work.

Conventional reinforcement learning strategies for visual generation typically employ sample-wise reward functions, yet this practice frequently results in reward hacking that degrades image diversity and introduces visual anomalies. To address these limitations, we present a novel framework that finetunes generative models using distribution-wise rewards, ensuring better alignment with real-world data distributions. Unlike rewards that evaluate samples individually, distribution-wise reward accounts for the data distribution of the samples, mitigating the mode collapse problem that occurs when all samples optimize towards the same direction independently. To overcome the prohibitive computational cost of estimating these rewards, we introduce a subset-replace strategy that efficiently provides reward signals by updating only a small subset of a generated reference set. Additionally, we apply RL to optimize post-hoc model merging coefficients, potentially mitigating the train-inference inconsistency caused by introducing stochastic differential equation (SDE) in regular RL practices. Extensive experiments show our approach significantly improves FID-50K across various base models, from 8.30 to 5.77 for SiT and from 3.74 to 3.52 for EDM2. Qualitative evaluation also confirms that our method enhances perceptual quality while preserving sample diversity.

While pessimism counteracts overestimation bias in offline reinforcement learning (RL), being overly conservative has been associated with hindering certain forms of generalization. However, in this paper we demonstrate that being overly pessimistic does not inherently prevent optimal generalization in contextual MDPs (CMDPs). Instead, we argue successful generalization depends not on the amount of pessimism, but whether the pessimistic structure respects the underlying symmetries of the optimal solution. We prove that a mildly pessimistic, non-symmetric value function can generalize worse than an overly pessimistic, symmetric one. In offline RL, the structure of the pessimism is determined by the structure of the dataset coverage. As such, enforcing a symmetric value function can be non-trivial, and might require techniques such as data augmentation (DA). Inspired by our theoretical results, we argue that DA can best be applied through a consistency loss during policy extraction, rather than the common practice of (regular) offline training on an augmented dataset. This is empirically validated using IQL and CQL on a rotationally symmetric reacher environment.

In-context learning (ICL) operates via implicit gradient descent embedded in the forward pass of modern AI architectures -- Transformers, Mamba, state-space models, and MLPs. Capturing this capability in biologically plausible Spiking Neural Networks (SNNs) has remained an open challenge: existing SNNs fail the Garg-2022 benchmark at non-trivial task dimensions. We trace this failure to a structural assumption: prior SNN designs route adaptation through inference-time synaptic plasticity, viewing the dendritic compartment as a passive conduit for error or teacher signals. We challenge this assumption. The subthreshold dynamics of a single dendritic compartment already implement a complete online learning algorithm. By treating the compartment as the computational substrate rather than a passive conduit, we propose DendriCL -- a single-layer compartmental spiking architecture whose apical recurrence is structurally identical to leaky online Widrow-Hoff LMS. This dynamics-only update collapses the architectural depth required for general-purpose ICL to a single layer. DendriCL is uniquely seed-stable at super-dimensional Garg-2022 ICL -- where dense Transformers exhibit grokking-style instability and fail past moderate task dimension -- and a linear probe recovers the reference online-LMS trajectory directly from the apical membrane at R^2 = 0.93, showing the algorithm is structurally embedded in the dynamics rather than implicitly discovered during training. Taken together, ICL requires neither attention, depth, nor inference-time plasticity: a single compartment with online-LMS dynamics is sufficient.

We present Cadence, a Byzantine fault-tolerant multi-proposer consensus protocol with arbitrarily low block intervals, optimal resilience, and optimal fast-path latency. Cadence divides time into equally spaced slots, one block per slot, each finalized in its own consensus instance. Blocks do not build directly on their predecessor, so instances run independently and none waits for an earlier block to finish or propagate; we call this extreme pipelining, decoupling the block interval from network latency. Cadence also removes the single-leader monopoly over transaction inclusion and ordering: under multiple concurrent proposers (MCP), several validators propose for each block, and it guarantees that, under synchrony, a transaction a correct proposer includes cannot be censored or deferred (short-term censorship resistance), and that no proposer can craft its proposal in reaction to the others' (hiding). To realize extreme pipelining, we introduce a general framework that turns any one-shot consensus meeting our slot-consensus specification into a multi-shot protocol. We instantiate it for MCP with two protocols of our own: Chorus, a slot consensus whose fast path finalizes a block in an optimal three rounds, with speculative finality one round earlier, and Conductor, an orchestrator that opens slots at an even cadence, more slowly under asynchrony to keep open slots bounded. To our knowledge, Cadence is the first MCP protocol to provide short-term censorship resistance and hiding at the fast-path latency of single-leader consensus. We prove safety, liveness, censorship resistance, and hiding under partial synchrony with optimal resilience (n = 3f+1). In simulation over Monad's 200 validators with five proposers per slot, finalization averages 219 ms (167 ms to speculative finality); at a 100 ms block interval a transaction waits on average 50 ms to enter a proposal.

Vision-Language Models (VLMs) have demonstrated immense promise in Spatio-Temporal Video Grounding (STVG). However, current evaluation protocols are largely confined to zero-shot assessments on general, daily-life benchmarks. This creates a critical disconnect from real-world applications in specialized fields, where models inevitably encounter rare visual concepts and complex spatio-temporal dynamics. Since exhaustive pre-training across infinite data distributions is infeasible, the ability to adapt to novel domains is essential. To bridge this gap, we introduce AnyGroundBench, a domain-adaptation benchmark designed to shift the STVG evaluation paradigm from static zero-shot testing to rigorous domain adaptation. Targeting five specialized domains (animal, industry, sports, surgery, and public security), AnyGroundBench pairs newly captured videos such as expert-annotated mouse behaviors with established datasets, unifying them through dense, high-fidelity spatio-temporal annotations. Crucially, the benchmark provides dedicated training subsets to systematically measure domain adaptability. We extensively evaluate 15 state-of-the-art VLMs, assessing their zero-shot generalization and In-Context Learning (ICL) capabilities under practical computational constraints. Ultimately, our findings reveal that current models fail in both zero-shot and ICL-based adaptation when confronted with specialized domains, exposing critical flaws in spatio-temporal reasoning that future research must address.

Most data-mixing methods assume the corpus has already been partitioned into groups, and the choice of those groups determines what a mixer can express. Existing labels, including provenance, topic or format taxonomies, and flat embedding clusters, commit to one semantic axis at one granularity; changing the resolution rebuilds the labels. We argue the bottleneck is the label system, not the mixer, and provide a hierarchical one. HERMES is a data-derived labeling substrate: a Learned Semantic Transform followed by 3-stage residual vector quantization annotates each document once into a coarse-to-fine code whose prefix length controls granularity up to approximately 130k cells. At coarse granularity HERMES sits at a plateau with KMeans-family methods on standard clustering metrics, so the contribution is the substrate, not the clusterer. On 1B-parameter, 25B-token pre-training, the hierarchy exposes an interaction fixed-granularity pipelines cannot test: at one prefix length, a combined Stage-2 rule contrast, equal-subbucket coverage versus size-proportional within-bucket quality top-30%, lifts a 16-task capability macro-average by +0.0253; at the next finer level, the same rule loses its measurable edge as candidate pools contract approximately 5x. HERMES reframes data mixture design from choosing among fixed label sets to navigating a reusable, data-derived granularity hierarchy.

Reasoning Language Models (RLMs) have significantly improved performance on complex tasks by extending the reasoning chain. However, these chains are prone to containing factual errors, particularly in knowledge-intensive tasks. To address this issue, we propose CheckRLM, a framework that improves the reliability of the reasoning process through Retrieval-Augmented Generation (RAG) by timely checking and correcting factual errors. Specifically, CheckRLM extracts factual claims from the reasoning chain to identify and localize subtle knowledge inconsistencies during inference. Upon detection of errors, a refinement mechanism performs minimal-cost yet precise corrections by leveraging external knowledge, ensuring coherence between the reasoning chain and correct knowledge. Extensive experiments demonstrate that CheckRLM substantially outperforms existing baselines, exhibiting a strong capability to mitigate error accumulation in long-horizon reasoning with lower costs. The code and data are available at https://github.com/AI9Stars/CheckRLM.

In this paper, we introduce BamiBERT, a new BERT-based pre-trained language model for Vietnamese that addresses key limitations of PhoBERT -- the current de facto Vietnamese text encoder. Trained from scratch on a 129GB corpus of general-domain Vietnamese text for 20 epochs, BamiBERT supports an extended context length of up to 2048 tokens and operates directly on raw input, eliminating the need for external word segmentation. Across 8 Vietnamese benchmarks, it achieves the best score on 11 of 15 metrics and the second-best on 3 others, setting a new state of the art among "base"-sized Vietnamese encoders and demonstrating strong cross-domain generalization. We release BamiBERT at: https://huggingface.co/Qualcomm-AI-Research/BamiBERT

Memory for a long-horizon LLM agent is a contract about what each future decision is allowed to see. The simplest contract appends past observations, tool calls, and reflections to every prompt, which makes prior context easy to access but also turns it into a jumbled mixture in which the effect of any single memory component is hard to isolate. We introduce and instrument an alternative bounded contract: every decision is made from a fresh user message assembled by typed retrieval, with no raw cross-decision transcript appended. The prompt thus stays bounded across runs of any length, and any single layer can be ablated in isolation. We instantiate the contract in Slay the Spire 2, a closed-rule stochastic deck-building game whose runs require hundreds of tactical and strategic decisions. A public online benchmark of frontier LLMs on the same game reports zero wins at the lowest difficulty across five configurations, and the developer-reported human win rate at the same difficulty is 16%; the task is hard but not saturated. Within our harness, a fixed-A0 ablation shows the largest observed difference when triggered strategic skills are enabled: the no-store baseline wins 3/10 games and adding the skill layer 6/10. At this sample size the comparison is directional rather than statistically decisive (Fisher exact p\approx0.37); a cross-backbone probe and public accumulating-context baselines are reported as operational comparisons rather than controlled tests of the contract variable itself. We release a reproducible testbed: 298 completed trajectories with condition tags, frozen memory/skill snapshots, prompt records, and analysis scripts -- an agent design and a validated, reusable methodology for studying how explicit memory layers shape long-horizon LLM-agent decisions.

The aggregation with exponential weights (AEW) estimator is not fully understood in the basic setting of model selection aggregation with squared loss. In particular, whether it is minimax-rate optimal in expectation for large enough fixed temperatures and under random design has been an open problem since its introduction, which was explicitly posed by Lecué and Mendelson (2013). In this paper, we settle this problem by showing that \emph{without} requiring a Bernstein-type assumption, the AEW indeed achieves the excess risk $T \log (M) / (n+1)$ in expectation, whenever the temperature $T$ satisfies $(L^2/T)\exp(B/T)\leq μ/2$. Here, the number of dictionary elements is $M$, the estimator has observed $n$ i.i.d. samples from any distribution, and the loss is assumed to be bounded by $B$, $L$-Lipschitz continuous and $μ$-strongly convex. For squared loss, we show that $T\geq 4 b^2$ suffices when the predictions and labels are $[0,b]$-valued. Because AEW is known to be suboptimal in expectation for temperatures below some constant, this shows that AEW has a sharp phase transition when the temperature is large enough but constant, as conjectured by Lecué and Mendelson.

Mental health disorders affect nearly one billion people globally, yet 75% of individuals in low- and middle-income countries receive no treatment due to workforce shortages, cost barriers, and stigma. Current AI-powered wellness solutions predominantly rely on single-mode conversational interfaces that suffer high abandonment rates and fail to provide measurable, immediate relief calibrated to users' dynamic emotional states. This paper presents Copewell, a novel multi-agent swarm system designed to expand access to mental wellness support through human-centered AI principles. Our architecture introduces three technical innovations: (1) a multi-source assessment framework integrating self-reported, physiological, and contextual data to mitigate algorithmic bias; (2) valence-arousal emotion mapping using Russell's Circumplex Model of Affect to route users to specialized AI agents; and (3) dual-mode intervention delivery combining conversational support with evidence-based sensory wellness protocols. We examine the sociotechnical design considerations underlying Copewell's development, including a privacy-first architecture, embedded ethical oversight through a dedicated Ethics Supervisor agent, and participatory design informed by mental health practitioners. Early practitioner engagement and beta deployment inform design decisions and identify directions for future empirical evaluation. This work contributes to responsible AI discourse by demonstrating how technical architecture can operationalize equity and safety principles from inception.

LLM-as-a-Judge has become the dominant evaluation paradigm for many natural language generation tasks, due to shortcomings of conventional metrics and high correlations with human judgment, albeit mostly in English. There are now attempts to extend LLM-as-a-Judge to multilingual settings including low-resource languages. However, LLMs have limited proficiency in low-resource languages, and there is often no adequate human validation in these settings. To highlight the scope of the problem and current practices, we explore the use of LLM-as-a-Judge evaluators in ACL Anthology papers focusing on multilingual settings and low-resource languages across a diverse set of tasks. Out of 650 papers mentioning LLM-as-a-judge, only 33 of them focus on low-resource or multilingual settings. Our in-depth analysis of these papers indicates inconsistent evaluation outcomes, a tendency to overtrust LLM judgments in multilingual settings, and the widespread reliance on a single judge model per study. To help the NLP community further, we conclude with recommendations about how to use LLM-as-a-Judge in multilingual and low-resource settings.

On-policy self-distillation (OPSD) has emerged as a promising paradigm for improving LLM reasoning, where a privileged teacher with access to reference solutions provides token-level supervision on the student's own generated trajectories. However, we find that OPSD consistently fails on long chain-of-thought (long-CoT) reasoning models, yielding at best marginal gains while destabilizing the reflective reasoning capability these models depend on. Through a novel decomposition of the teacher's supervision signal, we identify the root cause: the teacher's supervision is dominated by a reference-induced component that drives rote memorization of reference-specific shortcuts, while the question-conditioned, inference-transferable component is ignored or actively opposed. Based on this diagnosis, we propose a two-step solution. First, we construct a reference-only teacher (the same model conditioned on the reference without the question) to isolate the non-transferable component of the supervision signal; the residual after subtracting this component captures the question-conditioned, inference-transferable correction. Second, we use pointwise mutual information (PMI) as the mechanism to transform this residual into a well-formed PMI target distribution that the student can directly distill from, filtering out the reference-induced shortcut. Experiments on four long-CoT models across two datasets demonstrate consistent improvements over both the base model and standard OPSD, while preserving the models' natural epistemic behavior throughout training.

The complexity of waste disposal regulations across European countries poses significant challenges for the residents and hinders the transition to a Circular Economy. In Germany, the proper sorting and disposal of household waste remains challenging across municipalities. Consequently, substantially reducing incorrectly disposed waste is vital for improving waste management and advancing the Circular Economy. AI-based waste sorting solutions can support residents through user-friendly tools, such as mobile applications, that guide proper waste disposal. To be effective in supporting the Circular Economy, however, these solutions must be configurable to reflect the specific waste sorting scheme of individual municipalities in Germany. In the scope of this work, an evaluation and analysis are performed of two prominent classification strategies: OvA and OvR. The research uses a dataset constructed in alignment with the waste categories and sorting scheme of the city of Goslar in Germany. Moreover, this work aims to extend beyond the overall performance by examining the behavior of OvA and OvR classification strategies in identifying samples likely to be misclassified. These classification strategies are compared by applying varying confidence thresholds to identify uncertain samples for subsequent human review. This evaluation aims to balance the number of misclassifications against the human effort required for data annotation.

Vision-Language Navigation has increasingly emphasized high-level instruction reasoning, memory, global map construction, and instruction decomposition, while the low-level action representation remains comparatively underexplored. We propose CoFL-S, a low-level vision-language-action framework that predicts a language-conditioned flow field over the robot's local visible sector and generates continuous trajectories by rolling out the predicted field. To train this low-level representation, we convert each VLN-CE episode, originally a whole-episode instruction paired with an action sequence, into frame-level local supervision with aligned sub-instructions and matched action, trajectory, and dense flow-field targets. For evaluation, we introduce a continuous-time Habitat benchmark that isolates low-level action interfaces from instruction decomposition and executes all methods through a shared velocity-command controller, enabling decomposition-independent closed-loop comparison across different planner frequencies rather than fixed discrete forward-and-turn transitions in VLN-CE. Under matched encoders and training settings, CoFL-S consistently outperforms action-token and action-chunk baselines across planner frequencies in the continuous-time Habitat benchmark, and zero-shot real-world closed-loop deployment further shows its advantage over both baselines beyond simulation.

Instruction tuning for speech language models (SLMs) is substantially more challenging than for text-based large language models (LLMs), as it requires learning a new modality and a wide range of speech-specific instructions in addition to those supported by text LLMs. Existing SLM training approaches largely replicate the text LLM training paradigm by synthesizing large-scale speech pre-training and instruction-tuning datasets. However, this strategy is difficult to scale, since speech sequences are significantly longer than text sequences. In this paper, we propose SpeechCombine, an instruction-following speech language model trained without any instruction tuning, using only a single round of speech pre-training on 30k hours of data. Starting from a text LLM base model, we perform continuous pre-training on speech utterances to obtain a speech-adapted model, and then directly combine its weights with the weight difference between the instruction-tuned and base versions of the text LLM. Our results show that this simple combination strategy not only preserves the knowledge and capabilities of the original text LLM, but also effectively transfers them to the speech domain. These findings suggest a new direction for SLM training that avoids reliance on massive speech data.

Aqueous solubility is a key property in early-stage drug discovery, but most predictive models merge physicochemical descriptors and molecular graph information into a single representation, obscuring whether a prediction is driven by global chemistry, molecular structure, or both. We present an additive deep-learning framework that keeps these two sources of information separate throughout training: physicochemical descriptors are encoded by a multilayer perceptron (the chemical branch) and molecular graph topology by a graph neural network (the structural branch), with the two outputs combined only at the prediction stage through an additive model with an optional multiplicative interaction. This design provides a direct decomposition of chemical and structural components that can be examined separately after training. Furthermore, pretraining on the larger AqSolDB dataset and fine-tuning on the smaller BigSolDB2 dataset substantially improve accuracy and reduce run-to-run variations, indicating generalizability of the learned features from the data-rich settings. We further interpret the fitted model using best linear projections of the branch outputs, molecule-level embedding summaries across solubility classes, and atom-level GNNExplainer masks aggregated over functional groups. These analyses show that the chemical branch aligns with familiar physicochemical descriptors, while the structural branch captures graph-topological and functional-group patterns associated with solubility. Across both datasets, the framework attains competitive predictive performance while making the distinct roles of chemical and structural information more transparent.

The evolution toward fully autonomous telecommunications networks (Autonomous Network Levels 4-5) requires AI/ML agents to make real-time network decisions without human intervention. However, no standardized runtime mechanism exists to intercept and validate individual inference outputs before they trigger live network state changes, creating risks of erroneous autonomous decisions. This paper proposes the Guard Rail Validation (GRV) framework, a standardizable runtime architecture for intercepting and validating AI-driven decisions before execution. The framework evaluates decisions across multiple weighted dimensions -- including action scope, action type, service criticality, agent autonomy level, reversibility, and temporal behavioural patterns -- to determine a criticality level. Based on this level, graduated validation mechanisms are applied: execute-with-logging, bounds checking, independent agent validation, or multi-agent consensus. The framework additionally provides cross-agent conflict detection with criticality-weighted priority resolution and runtime conformance logging for regulatory compliance (e.g., EU AI Act Article 14). We present the architecture, algorithmic procedures, O-RAN deployment model, and evaluate threat coverage against known AI/ML attacks in telecommunications.

Predictions are increasingly used to guide high-stakes decisions, from treatment selection to policy making. To ensure reliability with imperfect predictions, uncertainty quantification methods such as conformal prediction build prediction sets with coverage guarantees. However, statistical validity alone does not immediately determine the decisions to take, nor the optimality thereof. This gap is especially delicate in counterfactual settings where the outcome that materializes depends on the action taken, so uncertainty cannot be specified independently of the decision rule. We develop a decision-theoretic framework for uncertainty-informed counterfactual decisions. We identify a novel notion of \emph{policy-coupled coverage} -- namely, coverage of the realized outcome under the action induced by the prediction sets themselves -- as the optimal and lossless interface between uncertainty and action. It plays three roles. First, it justifies acting via a natural max-min rule as minimax-optimal under distributional ambiguity. Second, optimizing prediction sets under policy-coupled coverage is equivalent both to a stronger universal-coverage formulation and to the direct risk-averse optimization over policies and utility certificates; this equivalence yields the explicit form of the population-optimal prediction sets. Third, it admits a two-stage procedure, Policy-Coupled Risk-Averse Conformal Prediction (PC-RACP), that approximates these optimal sets with rigorous finite-sample coverage. Simulations and a real email-marketing experiment confirm that PC-RACP delivers higher utility than existing approaches while maintaining valid coverage, and that ignoring the counterfactual structure of the decision problem is suboptimal for both validity and utility.

Operator learning has emerged as a powerful tool for modeling complex physical systems in functional spaces. However, their neural network-based architectures make them opaque models, obscuring the reasoning behind their predictions. In this work, we introduce a self-explainable operator learning framework that overcomes this challenge by reformulating operator learning as a linear combination of generalized functional linear models expressed through integral equations. Exploiting the additive decomposability of these integral equations, we divide the input domain into subdomains and compute localized integrals to evaluate the contribution of each region to the final prediction. This decomposition enables direct interpretability where the model explains both inputs and outputs by linking specific input regions to corresponding output patterns, thereby revealing which spatial features drive predictions. We demonstrate the framework on function-to-scalar and function-to-function mappings in fluid flow problems involving blood flow and unsteady aerodynamics. The results show that the operator most often prioritizes regions with strong feature gradients, providing physically meaningful insight into the model's decision-making process. Comparisons with established post-hoc explainability methods demonstrate qualitative agreement while highlighting the key advantage of the proposed approach: explainability is embedded directly within the operator structure itself and does not require an external tool. Therefore, our framework provides a mathematically transparent and physically interpretable approach to uncover relationships within data, fostering trust in machine learning for scientific applications by enabling more informed data-driven analysis of physical systems.

The rapid deployment of AI systems across high-stakes domains has created urgent demand for standardized evaluation, yet the field remains fragmented across competing risk taxonomies that catalog risks without showing how an audit is executed. At least 74 AI risk taxonomies exist, and almost all stop at the catalog. The hard part of auditing is not naming a risk but operationalizing it: turning it into a test run against a real system, a measured value, a calibrated severity, and a defensible grade. This paper leads with that bridge. We present the operationalization layer Eticas has built and run, shown end to end on a single risk (PII leakage) against a public benchmark, and then the open taxonomy that makes the method scale. On GPT-4-0314, a disclosure risk that seven external frameworks require be controlled is measured at 0%, 51%, and 84% disclosure as adversarial conditioning increases, mapping through calibrated severity bands to a subcategory grade of E with a SYSTEMIC pattern. Around this example, the Eticas AI Risk Taxonomy v2.0.0 organizes 76 active subcategories across 10 categories and 20 sub-groups, with mappings to 18 external frameworks across compliance, reference, and academic tiers. Its category and sub-group layer is published under CC BY 4.0 as open semantic infrastructure with stable URIs and SKOS/JSON-LD distributions, and a worked subcategory example shows the operational layer down to its severity thresholds. The contribution is the demonstrated bridge from concept to graded finding, anchored by a clean separation of risks from the mechanisms by which they surface, and framed by an open-core model in which the conceptual scaffold is open and the methodology calibration is the practitioner layer. This is the infrastructure the AI auditing field needs: shared, open, and demonstrably operable.

Mutual information (MI)-inspired feature learning techniques are capable of generating low-dimensional embeddings that retain nonlinear dependence structures, but direct estimations of MI suffer from noisy probability distribution estimates in the low-data regime. The H-Score objective, computed from second-order statistics, provides a practical proxy metric for training feature extraction networks. We prove that H-Score is invariant to invertible transformations in the unrestricted functional setting, but becomes sensitive to input basis rotations under constrained approximation classes. Consequently, we study unitary preconditioning for H-Score networks and show that selecting an appropriate basis rotation reduces finite-width truncation error by concentrating predictive dependence into fewer dominant modes. We identify the fast Fourier transform (FFT) as an effective data-independent, low-cost preconditioner for approximately stationary processes, where spectral structure induces concentration of the cross-covariance singular value spectrum. We introduce training-free metrics based on spectral entropy and cumulative dependence energy to quantify basis suitability and predict downstream inference gains prior to network training. Experiments across eight multivariate datasets demonstrate that FFT preconditioning is particularly useful in resource-constrained regimes, achieving up to 50% normalized mean squared error (NMSE) reduction, while the proposed metrics correlate with observed performance gains and correctly identify cases where spectral preconditioning is detrimental.

Human-AI teaming has received increasing attention in the literature. However, the range of studies conducted in multiple domains make it difficult to understand what types of teams are being studied, and in what ways are they similar/different from one another. In this study, we analyse 53 papers on human-AI teams and categorise them into five main clusters based on psychological taxonomies of teaming; AI Assistant, Ad-hoc Dependency, Ad-hoc Forced Dependency, Paired Equanimity, and Group Equanimity. Each cluster represents a unique combination of holistic team-level characteristics, indicating there are multiple disparate team types studied under the same definition. In turn, this raises the question of whether insights are truly transferable between papers. We conclude with guidance on how to identify the types of human-AI teams studied, a checklist for reporting a human-AI team in research work, and ways in which the field can be further synthesised.

The society and emerging risk-based regulatory frameworks for AI underscore the need for rigorous risk assessment to ensure safe and reliable AI systems. In response to this imperative, this paper presents an overview of AI risk assessment (identification and analysis) and management methodologies. It begins by reviewing the worldwide regulatory landscape that drives the need for systematic AI risk assessment. Then we characterize the spectrum of AI-related risks identified in the literature, from technical failures to ethical and social impacts. Subsequently, it reviews key risk assessment methodologies proposed for AI systems, focusing on general frameworks. The paper highlights best practices and illuminates methodological gaps, highlighting areas for further research on AI risk assessment.

We study online resource allocation when both rewards and consumption sizes may be continuously distributed. Requests arrive sequentially and must be accepted or rejected irrevocably under fixed resource capacities. Each request belongs to one of finitely many observable types; conditional on an observable request type, both the reward and the scalar size are random, and the realized size scales a fixed type-specific resource-consumption vector. The model allows the deterministic fluid relaxation to be degenerate. We show that additive regret is governed by the size-weighted mass of requests whose value-to-size ratios lie near the active acceptance cutoffs. We formalize this quantity through an active weighted-mass exponent p. When p > 1, this cutoff mass is thin, and the problem is genuinely hard: every online policy must incur regret of order at least $T^{1/2 - 1/(2p)}$, and this holds for every p > 1. A sample-path marginal policy matches this lower bound up to polylogarithmic factors; and when p = 1, so that the mass grows linearly near the cutoff, it attains $O((\log T)^2)$ regret. For example, if the size and the value-to-size ratio are independent and uniformly distributed, then p = 1; if instead the size and the reward are independent and uniformly distributed, then p = 2. Thus the policy achieves $o(\sqrt{T})$ regret throughout this regularity class without any fluid non-degeneracy assumption, allowing both primal degeneracy and dual non-uniqueness.

Physics-informed neural networks (PINNs) have emerged as a promising route to solve partial differential equations, yet they have struggled to reach the precision of classical solvers. The obstacle is increasingly understood to be one of optimisation, owing to the severely ill-conditioned loss landscape. We present $\textbf{DSGNAR}$: Doubly-Sketched Gauss-Newton with Adaptive Ratio, a scalable second-order optimisation framework that confronts this ill-conditioning and, in doing so, obtains unprecedented accuracy and speed. $\textbf{DSGNAR}$ couples a doubly-sketched Gauss-Newton model with a novel strategy that carefully controls both regularisation and step length. Across a suite of problems spanning nonlinear, chaotic, multi-scale, high-dimensional, and Navier-Stokes, the framework greatly improves on the state of the art: able to attain relative $\ell_2$ errors as low as $3\times10^{-16}$ in double precision, improve contemporary results by five orders of magnitude on the canonical Burgers' equation, and as much as eight orders on a high-dimensional Poisson problem, while remaining markedly faster. We further show that, in single precision, solutions at the limit of round-off error can be obtained very quickly: Burgers' equation to $\ell_2^{\text{rel}} = 4.75 \times 10^{-7}$ in under ten seconds. The framework is also robust to the choice of architecture, arithmetic precision, and initial hyperparameters. The code is available at https://www.github.com/wephy/physics-informed-neural-networks

Distributed machine learning enables collaborative model training without centralizing data, but it also exposes learning processes to privacy leakage and malicious manipulation. Existing defenses typically address these threats in isolation and are often tailored to specific learning paradigms or model architectures, limiting their applicability in realistic deployments. In particular, federated learning and decentralized learning exhibit distinct adversarial surfaces that are rarely addressed within a unified framework. In this paper, we present a model-agnostic framework for adversary-resistant distributed learning that jointly addresses privacy preservation and malicious behavior across both federated and decentralized settings. Our approach combines paradigm-specific defense mechanisms with GPBACC, a privacy-enhancing coded computing technique applicable to arbitrary machine learning models. For federated learning, we integrate robust aggregation strategies to mitigate the impact of malicious participants, while for decentralized learning we employ approximate decode-and-compare and group testing techniques to enable lightweight verification and adversary isolation without relying on a trusted aggregator. Crucially, we evaluate the proposed framework through an explicit, attack-driven analysis. We implement representative privacy attacks and malicious behaviors, and empirically demonstrate that the combination of GPBACC with robust aggregation and verification mechanisms significantly reduces privacy leakage and improves resilience against active adversaries. These results suggest that privacy-enhancing coded computing, when combined with appropriate adversary-resistance strategies, provides a practical and deployable foundation for secure distributed machine learning.

Software development is a complex task that demands cooperation among agents with diverse roles. Large language models (LLMs) have enabled autonomous multi-agent software development frameworks that leverage role-based collaboration to automate requirements analysis, coding, testing, and refinement. However, existing approaches typically assume that intermediate agent outputs are equally reliable, leaving them vulnerable to hallucination propagation, where incorrect decisions generated in early development phases are transferred to downstream agents and negatively impact final software quality. To address this challenge, we propose UA-ChatDev, an uncertainty-aware multi-agent software development framework that integrates uncertainty quantification into agent interactions. It introduces a lightweight uncertainty estimation mechanism based on token-level log probabilities to assess the confidence of agent responses and employs phase-aware threshold calibration to selectively trigger retrieval-based verification when uncertainty exceeds acceptable levels. Extensive experiments on the SRDD benchmark demonstrate that UA-ChatDev consistently outperforms existing single-agent and multi-agent software development frameworks across completeness, executability, consistency, and overall quality metrics. Further ablation studies and communication analyses verify that uncertainty-aware interactions enhance code execution reliability.

Deep learning has achieved remarkable performance in medical image segmentation, yet it suffers from critical limitations: mathematical intractability, substantial parameter requirements, and lack of clinical interpretability. We propose RadiomicNet, a novel two-stream hybrid architecture that enhances standard deep learning by integrating handcrafted radiomics features directly into the segmentation learning process. The key contribution is the Radiomics Attention Gate (RAG), which leverages Gray-Level Co-occurrence Matrix (GLCM) and Local Binary Pattern (LBP) features to modulate skip-connection attention in a lightweight MobileNetV2-based encoder-decoder, providing ante-hoc interpretability without post-hoc approximations. A novel Radiomics Consistency Loss further enforces alignment between texture complexity and prediction uncertainty, reducing Expected Calibration Error (ECE) from 0.142 to 0.118. RadiomicNet achieves a Dice Similarity Coefficient (DSC) of 0.763 +/- 0.231 on the Breast Ultrasound Images (BUSI) dataset and 0.854 +/- 0.112 on Kvasir-SEG, outperforming U-KAN by 1.2% and 1.8%, respectively (p < 0.05, Wilcoxon signed-rank test), with only 3.27M parameters, 9.5x fewer than standard U-Net and 4.3x fewer than U-KAN. Gradient-based feature importance analysis reveals that GLCM dissimilarity (15.24%), GLCM energy (14.56%), and LBP entropy (11.49%) are the dominant radiomics cues, providing clinically meaningful explanations for segmentation decisions. The proposed approach demonstrates that compact, interpretable models grounded in domain knowledge can deliver state-of-the-art segmentation performance with substantially reduced computational overhead.

Large language models (LLMs) exhibit remarkable reasoning capabilities, but their task-specific fine-tuning is notoriously plagued by overconfidence, severely hindering trustworthy deployment. We propose Data-Adaptive Lower-Rank Adaptation (DALorRA), a simple and effective variational Bayesian sparse framework that shifts the paradigm of uncertainty quantification from the dense parameter space to the lightweight rank level of low-rank adaptation (LoRA). With the insight that LoRA essentially aggregates multiple rank-one components that may provide superfluous model capacity, DALorRA imposes stochastic masking on rank dimensions, enabling Bayesian regularization of model capacity during training and ensemble-like calibration during inference. Extensive experiments demonstrate DALorRA's excellent calibration of LLMs without compromising reasoning accuracy.

Multiple-choice medical benchmarks are increasingly saturated, and recent rubric-based evaluations such as HealthBench have shown that open-ended clinical performance is far from solved - its "Hard" subset top score remains 32%. We present a small, deliberately difficult evaluation dataset of five clinician-authored clinical scenarios spanning four specialties (anaesthesia, internal/family medicine, emergency medicine, and obstetrics), each accompanied by an atomic, weighted, MECE rubric (25-62 criteria per task; 184 criteria total) authored from a clinician-drafted golden answer. We evaluate three frontier models: GPT 5.4, Claude Opus 4.7, and Gemini 3.1 Pro. Mean rubric pass rates were 0.47 (Claude), 0.39 (GPT), and 0.37 (Gemini). The central finding is an inversion of clinical priority: the highest-weighted (weight-5, critical) criteria passed at only 32.4-41.7%, while low-stakes weight-1 criteria passed at 80-90%. 56 of 108 critical (weight-5) criteria (52%) were satisfied by no model. Three LLM autoraters reproduced expert met/not-met labels on 92.8-94.7% of 552 graded criteria. We position this as a methods-and-preliminary-findings contribution: the five tasks demonstrate a scalable, defensible pipeline ready to develop into a large-scale benchmark.

The rapid advancements in using neural networks as implicit data representations have attracted significant interest in developing machine learning methods that analyze and process the weight spaces of other neural networks. However, efficiently handling these highdimensional weight spaces remains challenging. Existing methods often overlook the sequential nature of layer-by-layer processing in neural network inference. In this work, we propose a novel approach using dynamic graphs to represent neural network parameters, capturing the temporal dynamics of inference. Our Dynamic Neural Graph Encoder (DNG-Encoder) processes these graphs, preserving the sequential nature of neural processing. Additionally, we also leverage DNG-Encoder to develop INR2JLS (Implicit Neural Representation to Joint Latent Space) for facilitate downstream applications, such as classifying Implicit Neural Representations (INRs). Our approach demonstrates significant improvements across multiple tasks, surpassing the state-of-the-art INR classification accuracy by approximately 10% on the CIFAR-100-INR.

This paper studies additive regret in the multi-secretary problem, defined as the gap between the expected offline prophet reward and the reward of the best online policy. Prior work established \(O(\log T)\) regret for bounded-density distributions with connected support and \(O((\log T)^2)\) upper bounds for bounded-density distributions with support gaps. It was unknown whether the extra logarithmic factor is necessary even in the one-resource model. We prove that it is necessary. For a mixture of two separated uniform distributions at the critical capacity, the optimal regret grows at least on the order of \((\log T)^2\). Thus the existing \(O((\log T)^2)\) upper bounds for bounded-density gapped instances, including those implied by network revenue management models with continuous rewards, are tight in this simplest specialization. The same framework also yields a matching lower bound for gapped distributions whose gap-facing densities vanish near the support edges; this companion result is given in the appendix. The proofs use Bellman certificates: feasible solutions to a relaxation of the exact Bellman recursion. This framework converts lower bounds into explicit certificate constructions and identifies why support gaps permit larger regret.

Quantum annealing is a promising heuristic for combinatorial optimization, but on current hardware its performance degrades for larger and more complex problems due to noise and small energy gaps. Reverse annealing has been proposed as a refinement strategy, yet it remains unclear when it provides systematic advantages over standard forward annealing or simply increasing annealing time. We find that combining forward and reverse annealing consistently improves solution quality and efficiency across multiple problem classes. The benefits of reverse annealing increase with problem complexity and are strongest in regimes where forward annealing is increasingly limited. Moreover, reverse annealing yields larger efficiency gains than simply extending forward annealing times. We establish these results through a systematic experimental study on a D-Wave Advantage system, benchmarking reverse annealing across Max-Cut, Number Partitioning, and sparse clustering problems while varying reverse distance, pause duration, and annealing time. We identify a narrow optimal regime for reverse annealing parameters linked to the location of freeze-out points and energy-level crossings in the annealing schedule. These findings demonstrate that reverse annealing is most valuable for large, high-complexity optimization problems and is likely to gain importance as quantum annealing hardware scales toward more realistic applications.

Alzheimers disease (AD) is a brain disorder that develops slowly and mainly affects memory, thinking, language, and daily activities. It is one of the most common causes of dementia and creates many difficulties for patients as well as their families. In the early stage, the symptoms are often mild and may look like normal ageing. For this reason, many people are diagnosed late, when the disease has already progressed. At present, there is no complete cure for AD. Still, early detection can help doctors manage the condition better and take suitable steps at the right time. In this study, a machine learning model is proposed to detect the early stages of Alzheimers disease using clinical details, neuropsychological test scores, and neuroimaging-related measures. The data used in this work is collected from the Alzheimers Disease Neuroimaging Initiative (ADNI). As the dataset has missing values, iterative imputation is applied to fill them. The dataset also has class imbalance, which is handled using Borderline SVM-SMOTE. After that, feature selection is carried out using wrapper-based and embedded methods so that only important features are used for training. The selected features are divided into training and testing sets, and feature scaling is applied. A stacking ensemble model is developed using Logistic Regression, Extra Trees, Bagging KNN, and LightGBM as base classifiers. Along with this, an artificial neural network is also trained on the same dataset. The performance of these models is compared using precision, recall, F1-score, and AUC-ROC. This study aims to find the best classifier and also identify important biomarkers that may help in the early diagnosis of Alzheimers disease.

Most LP-from-text benchmarks are static datasets of word problems written and labeled by hand. Once such a dataset is released, its size is fixed, its difficulty is fixed, and every problem can leak into the training data of future LLMs. We present \textbf{A$^{2}$utoLPBench}, a benchmark for testing LLM-driven agents on linear programming problems written in plain text. We first pick a feasible point and dual, then write down a problem for which that point is optimal and the objective value is known. The answer is known by construction, with no solver call and no human annotator. The evaluation environment bundles a reference solver-critic baseline and a Docker image whose usage instructions are written for an LLM-driven agent to read. With these in place, any agent can run the benchmark and get a calibrated score with one command. Because the benchmark is a generator rather than a fixed dataset, it has properties no fixed dataset can match: an unlimited supply of fresh problems, a difficulty knob set by $(n,m)$, ground-truth answers correct by construction, low LLM-side cost per problem relative to human authoring, repeatable scores across independent batches, and resistance to training-data leakage when fresh post-cutoff seed ranges are used.

Chemical language models (CLMs) are trained with linearized representations such as SMILES, yet it remains unclear which chemically meaningful substructures they encode. To foster a better understanding of CLMs, we conduct a systematic study and probe for 78 molecular substructures across eight pre-trained and six randomly initialized models. We furthermore study how fine-tuning on chemical downstream tasks affects the learned representations of molecular substructures. Our results show that pre-training generally improves molecular structure awareness of CLMs, particularly in the upper layers. Moreover, randomly initialized models already encode ring structures well in the first layer. Our analysis on two chemical downstream tasks further reveals that, interestingly, fine-tuning affects task-relevant molecular substructures more than others, indicating that the changes in the representations follow chemical theory.

We study timestep allocation for score-based diffusion sampling, where a learned reverse-time dynamics is discretized on a finite grid. Uniform and hand-crafted schedules are standard choices, but they rely on fixed prescriptions and can therefore be suboptimal. To address this limitation, we propose Adaptive Reparameterized Time (ART), a continuous-time control formulation that learns a time change by treating the speed of the sampling clock as the control, so that a uniform grid on the learned clock induces adaptive timesteps in the original diffusion time. Based on a leading-order Euler error surrogate, ART provides a principled objective for allocating timesteps along the sampling trajectory. To solve this deterministic control problem, we introduce ART-RL, an auxiliary randomized formulation with Gaussian policies that turns schedule learning into a continuous-time reinforcement learning problem. We prove that the randomized ART-RL formulation is equivalent to ART at the optimizer level, in the sense that its optimal Gaussian policy recovers the optimal ART time-warping rate through its mean. We further establish policy evaluation and policy improvement characterizations and derive trajectory-based moment identities that yield implementable actor--critic updates for learning the schedule. Across experiments ranging from controlled low-dimensional settings to image generation, ART-RL can be plugged into existing diffusion samplers by changing only the timestep grid, consistently improving sample quality over strong baseline schedules at matched budgets while leaving the rest of the sampling pipeline unchanged. The learned schedules also exhibit broad generalization, transferring without retraining across sampling budgets, datasets, solvers, pipelines, and representation spaces.

Scientific machine learning papers typically make computational claims, e.g., that the relative mean square error is less than 5% or that the 95% predictive credible interval covers the test data. A coding agent can be prompted to replicate those claims from paper materials alone, but the prompt does not by itself reliably preserve progress or check whether generated evidence supports the paper's claims. We introduce Paper-replication, a workflow that makes each selected paper claim a target with recorded evidence, and implement it as a coding-agent skill. The workflow makes the agent record those targets, reconstruct the paper's method, run computational experiments, link generated outputs to provenance and comparisons with the paper's claims, record where matched evidence appears in the replication report, and pass validation checks before completion. We evaluate Paper-replication on twelve independent runs across four scientific machine learning papers. All twelve workspaces pass the completion gate, and all 158 recorded targets are matched with report coverage. Even in this completed workspace state, repeated runs differ in how papers are divided into targets, in numerical fidelity to the source papers, in elapsed replication time, in the number of intermediate executions replaced before final evidence is accepted, and in the rules used to accept evidence. Paper-replication makes completion depend on workspace evidence and validation checks rather than on the agent's final message.

Restoring archival film remains a fundamentally challenging problem due to the absence of paired training data and the lack of standardized evaluation benchmarks. Pristine versions of deteriorated footage are physically unrecoverable, requiring supervised methods to rely on synthetic data that often fail to capture the complex, temporally coherent nature of real film degradation. At the same time, existing real-world datasets are limited in scale, quality, and accessibility, hindering reliable evaluation and fair comparison across methods. We address both limitations with AbsoluteDegradation, a physics-inspired, modular pipeline for synthesizing realistic film degradations, and a new large-scale archival benchmark. The proposed pipeline models the analog-to-digital process as a structured composition of artifact families, incorporating signal-dependent grain, parametric scratches, and temporally coherent camera motion, enabling controlled generation of diverse degradation regimes. In parallel, we introduce a curated dataset of 81,576 high-resolution frames sourced from real archival footage, designed for consistent evaluation under real-world conditions. Together, these contributions provide a unified framework for training and benchmarking restoration models. Extensive experiments across multiple architectures show that models trained with AbsoluteDegradation generalize better to real-world footage, while the proposed benchmark reveals systematic failure modes of current methods. We hope this work establishes a foundation for reproducible and domain-authentic evaluation in archival film restoration.

Penile measurement is clinically relevant across male reproductive and urogenital health, including conditions such as micropenis, congenital and endocrine disorders, and sexual or urinary dysfunction. However, quantitative assessment of penile size has relied mainly on external length or circumference measurements, which are difficult to standardize, sensitive to measurement conditions, and unable to capture the internal portion of the penis. MRI enables volumetric assessment of the whole penis in vivo, but automated segmentation has not previously been established at population scale. Automated whole-organ volumetry would enable high-throughput phenotyping for multi-omics and clinical studies of male reproductive disease. Here, we present a deep learning framework for whole-penis segmentation in multi-channel DIXON MRI. Using a newly curated expert-annotated training dataset ($n = 145$ subjects; $13,050$ annotated slices) and a double-annotated independent test benchmark ($n = 24$ subjects; $2,160$ double-annotated slices), we optimized a 3D nnU-Net architecture. The model achieved a 5-fold cross-validation Dice score of $0.90$ and performed at observer-level accuracy on the independent test set (Dice: $0.92$; Hausdorff distance: $3.58$). We deployed the model in $34,412$ UK Biobank participants, enabling automated quantification of total penile tissue, including both external and internal components. Longitudinal evaluation in 2,282 men demonstrated high inter-session reproducibility ($r = 0.87$). This framework establishes a reproducible and population-scalable method for MRI-based assessment of penile anatomy and provides an open technical resource for future studies in urological imaging and male reproductive health. The trained model weights will be publicly released.

Conventional traction control architectures intervene only after the adhesion limit of a tire has already been breached. This paper investigates whether Rolling Split Conformal Prediction , monitoring the volatility of non-conformity residuals from a per-driver Random Forest model of expected slip behavior , can serve as a statistically grounded pre-incident warning signal, ahead of gross traction loss. Unlike an earlier internal draft of this work, the evaluation reported here corrects a confound in the slip proxy (vehicle speed is included as an explicit model feature, not left implicit in the target's denominator), uses every racing lap for each driver rather than only the fastest lap, and is scored against real, timestamped incident labels extracted from FIA Race Control Messages and track-limits lap deletions rather than narrated post-hoc. The result is negative: across 19 drivers and 55,563 test-phase telemetry samples, the rolling-volatility detector achieves a mean precision of essentially 0.0 and mean recall of 0.0 against 14 ground-truth incidents, while flagging on average 15.3% of all samples as anomalous , too high a false-alarm rate for any early-warning use. A static 95th-percentile threshold baseline performs no better in any way that would justify the added complexity of the conformal-volatility formulation. Residual autocorrelation diagnostics show the split-conformal exchangeability assumption is violated for every driver (Ljung-Box p < 0.001, n = 19/19), which is one plausible driver of the high false-alarm rate. We report this as a methodologically rigorous negative finding, diagnose its likely causes, and outline what a genuinely predictive version of this approach would require.

Electronic neurons are a keystone for construction of the spiking neural networks which have numerous applications in neuroprosthetics, artificial memory, intensive calculations etc. A number of concepts of electronic neurons has been already proposedm with some of them implemented in hardware. However, new schemes are of significant interest since the existing ones do not fit all requirements: either they are too complex and expensive in realization, or they are not able to demonstrate all demanded regimes, or their do not have a appropriate mathematical description and therefore may be investigated only experimentally etc. In this study we propose a new design of bursting electronic neuron constructed as a circuit implementation of the equations of a phase-locked loop system. To succeed, we use a novel hybrid approach: we start from the phenomenological equations providing the demanded, then we adjust and modify these equations to simplify the implementation rather than implementing the biophysical equations into thee hardware directly or writing equations for the already constructed circuit. The resulting circuit is simple in implementation and well matches the underlying equations. It can be used for description of not only a single neuron, but small neural circuits too.

As Large Language Models (LLMs) and agentic systems become integrated into real-world applications, ensuring their safety and security is critical. Guardrail systems that detect and block malicious instructions sent to and from an LLM are an essential component of AI security. However, researchers conducting black-box adversarial emulation against production AI systems often struggle to determine whether a guardrail block or an LLM rejection has occurred. This distinction is important because the techniques used to bypass guardrails can differ substantially from those used to bypass LLM safety alignment, and has a material impact on attack technique selection and optimization. We propose the first black-box guardrail reconnaissance methodology, which detects the presence of a guardrail within a target AI system through behavioral monitoring of HTTP, lexical, and timing signals, assuming only black-box access and zero prior knowledge of the guardrail or AI system. Experiments demonstrate that our approach detects guardrail presence with 100% accuracy, with statistically significant behavioral separation between benign and malicious interactions (q < 0.001). Our approach further identifies the content categories a guardrail is designed to block, and distinguishes guardrail blocks from LLM rejection on unseen prompts with an average F1 score of 98%.

While Large Multimodal Models excel in comprehension, high-throughput inference engines lack native support for multimodal generation. This is severe in Speech Language Models, where generating multi-layered audio tokens via decoupled AR+NAR or synchronous Multi-Token Prediction (MTP) with delay-pattern interleaving conflicts with standard single-stream loops. We present a vLLM-based inference pipeline for unified speech understanding and generation. We extend autoregressive decoding to natively execute delay-pattern de-interleaving and coordinated multi-stream sampling, integrating an on-GPU acoustic decoder for end-to-end waveform synthesis. Crucially, we overcome the shared intuition that Classifier-Free Guidance (CFG) halves throughput. By co-scheduling paired conditional and unconditional requests within a continuous batch, our CFG implementation sustains 80% of non-CFG throughput, absorbing dual-request and logit merging overheads. We open-source our framework.

Scientific Fitness Coaching (SFC) is typically delivered by human professionals, making it costly and inaccessible to many. While recent advances in Large Language Models (LLMs) show considerable promise for more inclusive fitness coaching, directly deploying prevailing general-purpose LLMs in SFC reveals critical limitations. These models often lack sufficient domain-specific knowledge integration, leading to weak performance on complex SFC scenarios. In this paper, we introduce FitOne, a series of fitness LLMs (with 8B and 32B parameters) designed to improve reliability and domain specialization for SFC applications. Built upon the Qwen3 foundation models, FitOne is developed through a three-stage post-training pipeline consisting of continual pre-training, supervised fine-tuning, and reinforcement learning, using large-scale, high-quality datasets derived from rigorous knowledge engineering. We conduct comprehensive evaluations of FitOne on professional fitness certification exams, including ACSM-EP and NSCA-CSCS, as well as general capabilities such as knowledge reasoning and instruction following. Experimental results show that, while retaining strong general capabilities, FitOne-8B/32B achieves average improvements of up to 10.09%/9.29% and 12.73%/7.01% on the ACSM-EP and NSCA-CSCS exams, respectively, compared with the Qwen3 base models. Furthermore, in-depth ablation studies confirm the necessity of each training stage, highlighting the pipeline's effectiveness in balancing domain expertise enhancement with general ability retention. We believe this research advances LLM systems toward more reliable fitness intelligence and will inspire future research on developing domain-specific LLMs.

Autonomous AI agents increasingly depend on external knowledge stores, yet most retrieval pipelines provide relevance without durable guarantees of provenance, version identity, integrity, traceability, or point-in-time reconstruction. We formalize this as context governance and present ContextNext, an open specification and reference implementation for governed AI-consumable knowledge vaults. ContextNext does not replace Retrieval-Augmented Generation (RAG); it supplies the governance layer beneath retrieval, determining which artifacts are approved, current, attributable, and integrity-verified before retrieval systems operate over them. The specification combines typed Markdown documents with metadata, deterministic set-algebraic selectors, contextnest:// URI references, SHA-256 hash-chained version histories, graph-level checkpoints, source nodes for live data through the Model Context Protocol (MCP), and audit traces of agent context consumption. These mechanisms let organizations reconstruct which knowledge versions informed an agent output and whether those versions were AI-eligible when consumed. We report first empirical results from two controlled experiments. In a stale-version attack isolating the governance-versus-retrieval failure mode, governed selection strictly Pareto-dominates BM25 sparse retrieval, with higher answer-quality pass rate (97% versus 93-90%) at about one-third the input-token cost. In a retrieval-determinism experiment over a 1,060-document corpus, deterministic selectors and BM25 return stable document sets across repeated identical queries (Jaccard 1.0), while a dense+HNSW baseline is non-deterministic on 80% of queries (mean Jaccard 0.611, worst case 0.210). These results suggest that context governance addresses failure modes retrieval quality alone is not designed to resolve. We release a core engine, CLI, and MCP server under open licenses.

Optical coherent Ising machines based on time-multiplexing have demonstrated significant progress in terms of connectivity and spin scalability. However, they are constrained by large physical footprints, high power consumption, poor thermal stability, and high cost. Here, we present a time-multiplexed Ising machine leveraging propagating wave packets in solid-state delay lines at microwave frequencies, enabling thermally stable, robust, low-power, tabletop, and affordable design. We use two serially connected 20.5 MHz, 707 μs bulk acoustic wave delay lines supporting 2,048 spins. Our design provides all-to-all connectivity with 15-bit coupling resolution and finds approximate MAX-CUT solutions in 341 ms, potentially scalable to sub-ms by using higher frequency delay lines. Additionally, we demonstrate solutions to number partitioning and Sudoku problems. Compared with state-of-the-art Coherent Ising machines, our machine exhibits four orders of magnitude higher thermal stability. Against the simulated bifurcation algorithm, our design achieves comparable results on the MAX-CUT problem, while outperforming it on the more complex number-partitioning and Sudoku problems.

Large language models (LLMs) are increasingly used as cheap, scalable judges that compare candidate outputs pairwise -- to rank responses, select models, or triage papers. Yet LLM judges are both noisy and systematically biased: they favor verbose or well-formatted answers and exhibit position effects, so simply aggregating their votes recovers a ranking of presentation, not of true quality. We study the practical goal of identifying the \topk{} items under a fixed comparison budget, and make two contributions. First, we cast judging as Bayesian inference over latent quality with explicit, judge-specific bias covariates (verbosity, position), regularized by a shrinkage prior so that the data decide which biases a given judge actually exhibits. Second, we introduce a \topk-aware active acquisition rule that chooses the next comparison to maximally reduce uncertainty about \topk{} \emph{membership}, rather than about the full ranking. On a controlled benchmark with known ground-truth quality, judged by sixteen real LLMs spanning open and proprietary families (Llama, Qwen, Phi-4, GPT-4o-mini/5.1/5.5, Gemini, DeepSeek, and Claude Haiku/Sonnet/Opus), naive aggregation plateaus at a wrong \topk{} on biased judges regardless of budget, while our bias-aware model recovers it; \topk-aware acquisition reaches this ceiling with far fewer comparisons than round-robin or a global-uncertainty (D-optimal) rule. Bias is real but heterogeneous and capability-dependent: cheap and mid-tier judges carry a strong verbosity bias that our model corrects (lifting recall from $\sim$$0.5$--$0.6$ to $0.84$--$1.0$), whereas the frontier judges we tested show little bias and already rank accurately, so bias-aware modeling changes little there.

Classifying heterogeneous omics data remains a fundamental challenge in computational biology, particularly in high-dimensional, small-sample settings where nonlinear interactions dominate and class imbalance further complicates reliable prediction of minority phenotypes. While traditional kernel methods rely on feature abundance, they fail to leverage the known interaction landscapes of biological systems. In this work, we propose a structured Gaussian process classification framework that integrates graph-encoded biological pathways directly into the kernel construction. By propagating information along known interaction networks and combining this with abundance-derived features, the resulting classifier captures both quantitative measurements and topological context. We benchmark our proposed methodology on three publicly available gut and fecal microbiome datasets. To address severe class imbalance, we evaluate complementary strategies, including data-level resampling, threshold calibration, and confusion-matrix-based adjustments, and report minority-class performance alongside accuracy. The hybrid approach yields a performance gain over unstructured baselines and matches the performance of established benchmarks for similar datasets. Furthermore, the probabilistic nature of the framework naturally provides calibrated predictive uncertainty, enabling robust differentiation between confident predictions and ambiguous samples.

Large kernel depthwise convolutions achieve strong performance but suffer from significant degradation as kernel size grows due to irregular memory access from gather-based computation; while Large Kernel Acceleration (LKA) helps on small feature maps, it becomes counterproductive on large feature maps, even slower than non-accelerated implementations. We propose Windowed Batch Matrix Multiplication (WBMM), which partitions input into contiguous windows and indexes a compact relative position bias table to construct weight matrices, enabling regular memory access via batched matrix multiplication. This yields a unique property: WBMM's throughput improves with larger windows, opposite to depthwise convolutions that degrade with larger kernels. Operator-level benchmarks show WBMM with 14x14 windows outperforms 5x5 depthwise convolution baselines in speed while providing a 7.8x larger per-layer receptive field. Combined with inter-block cross-window communication and hierarchical window reparameterization, WBMM achieves comparable or higher accuracy on ImageNet-1K, COCO, and ADE20K with 1.31-1.88x training speedup, and demonstrates consistent advantages across GPU, CPU, and edge devices without requiring specialized acceleration kernels. Our code is available at http://github.com/wansong-s/WBMM

Flow-matching vision-language-action policies generate robot action chunks through an iterative transport process, creating an opportunity for test-time guidance without retraining the base policy. We study this opportunity in Guided Action Flow, an inference-time framework that keeps a pretrained SmolVLA policy frozen and uses a learned action-chunk critic to guide its reverse-time flow sampler. The critic is trained from real success and failure rollouts, can condition on task-description features from the frozen SmolVLA language pathway, and is used only through action gradients during sampling. We evaluate the approach on LIBERO manipulation tasks. A single-task critic improves success from 68.0% to 82.0% on one seed window and from 82.0% to 86.0% on another. A multi-family task-description critic improves validation success from 46.0% to 56.0%, while the locked held-out test gain is positive but modest, from 65.0% to 67.5%. These results support the feasibility of Q-guided inference for frozen flow-matching VLA policies, while showing that critic generalization and uncertainty-aware guidance remain the central bottlenecks.

We propose an improved Fourier Neural Operator (FNO) for modeling two-dimensional Rayleigh-Bénard convection by predicting time increments instead of full solutions, achieving higher accuracy than a standard FNO baseline. The resulting model is compact (314k parameters, 1.26 MB) and fast (7 ms inference), while maintaining similar accuracy as demonstrated in previous benchmarks. We show that although FNOs generalize to finer meshes, accuracy remains limited by the resolution of the training data.

Hierarchical state-space models (HSSMs) offer a promising approach to long-horizon prediction by segmenting sequences into temporal chunks. However, their performance hinges on how chunk boundaries are determined. While prior HSSMs typically rely on fixed-length chunking or similarity-based boundary detection, these methods often misalign with the intrinsic temporal structure of the data. We argue that chunking should instead be driven by prediction errors, which more directly indicate when longer-range context becomes necessary. Nevertheless, integrating surprise-based chunking into HSSMs introduces critical challenges, including hierarchical collapse during end-to-end training and the absence of surprise signals during open-loop prediction. To address these issues, we propose Surprise-based Nested Temporal Abstraction (SUNTA), a method that employs a decoupled training strategy to preserve surprise signals and uses internal inconsistency as a top-down surprise metric to determine chunk boundaries within imagined rollouts. Experiments on video prediction tasks in 2D and 3D environments demonstrate that SUNTA outperforms baselines, uniquely maintaining accurate predictions over 250 timesteps, whereas all baselines degrade within the first 10 timesteps.

Wave Function Collapse (WFC) is a widely used procedural content generation method that learns local adjacency constraints from example inputs to generate larger outputs. In this paper, we explore combining WFC with evolutionary search by evolving the small input examples used by WFC rather than directly evolving complete levels. In this approach, WFC acts as a genotype-to-phenotype mapping. The generated levels are then evaluated through domain-specific fitness functions. We evaluate the method in two domains with different relationships between local and global structure: Maze connectivity maps and Zelda-style dungeon layouts. Our results show that evolutionary optimization over WFC inputs improves generation quality in domains where properties emerge from local relationships, while domains requiring global constraints remain challenging. These findings suggest that evolutionary search can effectively guide WFC generation when target objectives align with local structure.

We present HaloGuard 1.0, an open-weights implementation of the constitutional-classifier paradigm for input safety. It achieves state-of-the-art performance on English and multilingual prompt-safety benchmarks at roughly one-tenth the model size of current leading open guard models. The safety constitution is the organising structure of the corpus: a natural-language constitution of 46 policies and 2,940 subcategories drives synthetic data generation, with exhaustive one-to-one paired counterfactuals that hold topic and vocabulary fixed while flipping intent, a two-tier harmless design that separately targets boundary and baseline false positives (FPs), and balanced multilingual materialisation across 46 languages that treats language as a surface form appearing on both sides of the boundary rather than as an adversarial signal. Across seven prompt-safety benchmarks, HaloGuard 1.0-0.8B attains the best average F1 (90.9) of any open guard we evaluate, outperforming baselines up to 27B parameters (over 30 times larger) while holding false-positive rate (FPR) to 4.3 and false-negative rate (FNR) to 9.5. The HaloGuard 1.0-4B variant reaches average F1 of 92.1 and FPR of 3.5, spending its extra capacity on precision rather than recall. A structured adjudication of the remaining failures indicates that most apparent missed-harm cases are benchmark mislabels rather than genuine model misses. An always-on adversarial red-teaming protocol continuously hardens the guard against both content-level and agentic attacks. We release the models as open weights.

Long-context reasoning requires models to locate, revise, and synthesize evidence distributed across lengthy inputs. Existing long-context RL methods usually reward final answers or static evidence extraction, offering little feedback on how intermediate actions change the model's evidence state. We propose Maven, a reinforcement learning framework with an editable evidence memory. Maven defines an answer-conditioned evidence-state value and rewards action-level state transitions: add actions are credited by marginal gain and hindsight contribution, link actions by evidence synergy, and drop actions by improved answer support after removing misleading evidence. These rewards are assigned to the corresponding action spans in GRPO. Across Llama and Qwen models on LongBench v2, LongReason, and RULER, Maven outperforms outcome-only RL and evidence-identification baselines, producing more sufficient evidence sets and lower distractor retention. Our results show that long-context RL benefits from optimizing stateful evidence navigation rather than one-shot evidence extraction.

Large language models (LLMs) are increasingly deployed in domains requiring guardrails to detect unsafe, off-topic, or adversarial prompts. Existing guardrails predominately rely on fine-tuning to build classifiers, which often suffer from low generalization and high inference latency. We present kNNGuard, a training-free guardrail that utilizes the activation space of an off-the-shelf LLM. Given a small bank of 50 safe and unsafe prompts, kNNGuard extracts hidden activations and performs multi-layer kNN fusing activation-space and embedding-space scores for classification. Across six domains spanning topical and security prompts, kNNGuard achieves competitive or superior F1 compared to fine-tuned state-of-the-art guardrails while running 2.7x faster than the best comparable guardrail, and 10x faster than a fine-tuned safety classifier without gradient updates or fine-tuning. Domain adaptation requires only updating the labeled bank, which can be constructed in under 10 seconds and several orders of magnitude faster than established guardrails. We also analyze the impact of system prompts, layer selection, and integration into production LLM pipelines as a configurable, low-latency guardrail.

Ensuring model reliability in Explainable AI requires a global assessment of the hypothesis space. We propose a formal framework for the exhaustive analysis of optimal and near-optimal decision trees, called Algebraic Decision Tree Counting (ADTC). Inspired by Algebraic Model Counting (AMC) in knowledge representation, ADTC reformulates diverse analytical tasks, such as optimization, counting, and sampling, into a unified sum-of-products computation over a semiring $R$. While the hypothesis space of decision trees is doubly exponential with respect to the maximum depth $Δ$, our dynamic programming algorithm achieves $O^*(n^{O(Δ)})$ time complexity in the number of features $n$, where $O^*$ suppresses polynomial factors. To handle complex constraints consisting of multiple tree metrics, we introduce model behavior tensors that aggregate semiring values via convolution products over a tensor semiring. This algebraic approach efficiently constructs a model profile that captures the global landscape and trade-offs between criteria such as accuracy, size, and fairness. We demonstrate the utility of our software, emtrees, on real-world datasets, illustrating how ADTC facilitates evidence-based model selection in sensitive domains.

Graph Neural Networks (GNNs) have been widely used to capture spatial functional connectivity patterns to improve electroencephalography (EEG)-based depression recognition performance. However, the functional connectivity of brain networks in patients with depression exhibits an inherent hierarchical structure, making it difficult to capture accurate connection patterns. To address these issues, this paper proposes a novel model named Sample-Adaptive Hyperbolic Graph Neural Network (SA-HGNN), which aims to accurately extract the authentic hierarchical structure of depression-affected brain networks. Specifically, the proposed model comprises three core modules. First, a Sample-Adaptive Graph Construction module dynamically constructs personalized brain network topologies to capture more complex spatial relationships within the brain network. Second, hyperbolic graph convolution is employed to overcome the representation bottlenecks of Euclidean space, leveraging hyperbolic geometry to precisely capture latent hierarchical relationships within the brain network. Finally, an Attention Pooling module adaptively filters out highly redundant noise channels in EEG signals, effectively mitigating the interference of inherent noise on the authentic hierarchical topology. Extensive experiments on public EEG datasets demonstrate the superior performance of our method across resting-state and task-related paradigms, validating its robustness to noise and efficacy in capturing abnormal functional connectivity patterns in brain networks of patients with depression.

File-level copying is a widespread but ungoverned form of software reuse. Copying files across repositories reduces supply-chain visibility: it removes the four observable signals a package manager provides for a declared dependency (provenance, maintenance, security, and compliance) with no mechanism to restore them. To characterize the scale and consequences of this unmanaged reuse, we present a mixed-method study of copying across the entire open-source ecosystem using World of Code (WoC). From a 0.1% commit sample, we extract 690,500 copy events and retain 3,912 rationale-bearing copy commits for intent labeling. We show that the 13 axial copy forms, spanning vendored dependencies, hardware/driver synchronization, scaffolding, UI assets, and direct source-code reuse, are unreliable proxies for developer intent: among rationale-bearing commits, hardware/driver copies are predominantly fork-maintenance work (78%), while dependency-vendoring copies more often signal upstream bypass (70%) than offline availability. These visibility gaps are form-specific: security and license risk concentrate in complementary copy forms. Copied sources are frequently stale (median 155 days; 38.5% over one year old) and seldom record a recoverable origin (4.3% documented), let alone a checkable version (2.0% versioned); even vendored copies record where they came from only 10% of the time. Security risk concentrates in vendored dependencies: 17,314 CVE-risk copy commits in the full-WoC graph, 88% in the dependency-vendoring form; 80% score CVSS >= 7.0 and upstream-fix adoption is only 47%-84%. License risk concentrates in direct source-code reuse: 41,777 pre-validation candidates, 66% in the source-code form, with 39 verified high-star violations (kappa = 0.752). Both risks reach packaged software and are invisible to dependency scanners operating on declared metadata alone.

In recent years, it has become increasingly evident that large language models (LLMs) and autonomous agents raise the level of abstraction in software development by shifting the focus from writing precise procedures to expressing intents and goals. This paradigm shift introduces new challenges, particularly in how testing should be guided when prompts, rather than code, become primary development artifacts. To address this challenge, we propose Prompt Coverage Adequacy, a novel coverage criterion designed to support the testing of code generated from task descriptions. Prompt Coverage Adequacy serves as an analog to traditional code coverage, but operates at the level of prompts used in LLM and agent-based programming. Specifically, it measures how well a given test suite satisfies the requirements expressed in a prompt by leveraging the attention mechanisms of LLMs. We evaluate a simple instantiation of this criterion, based on attention boosting, across two datasets and multiple LLMs. Our results demonstrate that Prompt Coverage is associated with fault-detection effectiveness and can uncover over 30+% more faults than traditional code coverage when used to guide test generation. These findings suggest that Prompt Coverage Adequacy can serve as a foundation for developing testing metrics better suited to the emerging paradigm of LLM-driven software development, addressing the limitations of classical coverage criteria in this new context.

Performance evaluation in AI systems commonly assumes that random dataset splits produce independent and identically distributed (i.i.d.) subsets. We show that this assumption often breaks down in spatiotemporally correlated domains such as aerial surveillance, precision agriculture, and medical imaging, leading to two systematic failures: data leakage, where correlated samples span training and validation splits and inflate performance estimates, and hidden stratification, where errors on minority subpopulations are obscured by aggregate metrics. To address these issues, we propose a unified evaluation and training framework for spatially correlated data. We introduce Structure-Aware Stratified Partitioning (SASP), which constructs validation splits that reduce spatiotemporal leakage while preserving meaningful class balance, and Curriculum Distributionally Robust Optimization (CDRO), a curriculum-based relaxation of distributionally robust training that stabilizes optimization under these stricter splits. Across multiple benchmarks, this combination yields consistently improved generalization, more reliable confidence calibration, and exposes failure modes that remain hidden under conventional random-split evaluation.

Large language models (LLMs) have demonstrated strong capabilities in software engineering tasks, such as code generation, library recommendation, and dependency configuration. However, recent studies show that LLMs may suffer from package hallucination, where they generate non-existent or invalid package names. These hallucinations can be exploited in software supply chain attacks, as attackers may register malicious packages under hallucinated names. Therefore, mitigating package hallucination is important for improving the reliability and security of LLM-assisted software development. In this paper, we introduce BOUND, a lightweight localized model editing framework for mitigating package hallucinations in LLMs. BOUND formulates package hallucination mitigation as a package-validity boundary editing problem, where the boundary refers to the model's ability to distinguish valid packages from hallucinated package names under a given task context. It first locates modules related to package hallucination through a risk-aware localization strategy, and then edits these modules with lightweight LoRA adapters using a boundary-aware objective that reinforces valid packages, suppresses hallucinated packages, and preserves locality behavior. Experimental results show that BOUND effectively reduces package hallucinations while preserving valid package recommendations. In the package recommendation task, BOUND reduces package-level hallucination rate (Package-HR) by 79.9% on edit prompts and by 65.4% on unseen prompts. The learned package-validity boundary further generalizes to other package-related tasks, reducing Package-HR by 12.8% in code generation and by 34.0% in pip install recommendation. These results show that BOUND refines the package-validity boundary of LLMs and improves the reliability of package-related outputs.

Motivated by the challenge of stabilizing a general unknown linear dynamical system (LDS) from observations, we study the natural prerequisite of online prediction. Our goal is to achieve sublinear regret with a memory footprint that adapts to the intrinsic complexity of the dynamics rather than the full hidden -- state dimension. We focus on the practically central regime of systems with low instability complexity -- eigenvalues outside the real stable interval that do not decay rapidly, together with non-semisimple modes-potentially embedded in an otherwise stable real spectrum of much higher dimension; we write $k$ for this count. This regime is the primary setting in which stabilization is plausible: we show that many systems with high instability complexity cannot be stabilized without exponentially large controls. Thus, prediction is meaningful for stabilization precisely when the instability complexity is small. Within this regime, we introduce a unified online algorithm that handles every LDS (including non-diagonalizable systems with complex or exploding modes) with a learnable parameter count of $\widetilde{O}(k)$. Finally, we prove a lower bound showing that $k$ is a valid complexity measure: any filter-based predictor needs at least $k$ filters. Experiments corroborate our theory: on a high-dimensional system, our predictor sharply outperforms prior methods at an equal parameter budget.

Large Language Models are increasingly deployed in emotional-support contexts and crisis-related situations. Nevertheless, their cross-lingual abilities in these circumstances remain underexplored. Existing benchmarks emphasize multilingual performance but rarely examine crisis-related empathy and cultural grounding in low-to-mid-resource languages. We introduce SPLIT, a 500-prompt benchmark designed to evaluate LLM consistency in generating emotionally grounded responses across five categories: Stress, Panic, Loneliness, Internal Displacement, and Tension. We evaluate three technically diverse LLMs across three dimensions: Empathetic Accuracy, Linguistic Naturalness, and Contextual & Cultural Grounding. The framework aims to assess and compare the quality of LLM responses in both English and Ukrainian languages, as well as to explore the reliability of the LLM-as-a-jury paradigm. Our findings reveal that Gemini-2.5-Flash and LLaMA-3.3-70B-Instruct degrade when transitioning to Ukrainian, while DeepSeek-V3 remains comparatively stable within our benchmark. We additionally find that human and AI evaluators agree weakly on empathy and naturalness but diverge on cultural grounding. We further argue that producing Ukrainian text is not equivalent to producing Ukrainian emotional support. Our findings may assist in the future development of more culturally tailored benchmark designs, as well as encourage a stronger emphasis on human-centered evaluation.

Safe completion requires models to provide useful assistance without enabling harm, but this behavior is difficult to evaluate with isolated prompts. We introduce OpenSafeIntent, a benchmark of controlled prompt-sets that vary intent while holding the underlying task fixed. Each datapoint contains benign, dual-use, and malicious variants of the same task. This design lets us evaluate whether models calibrate assistance across intent shifts, rather than merely appearing safe on average. Across a broad model suite, we find that prompt-level safety hides important failures: models often fail to remain safe across matched intent variants, dual-use behavior is brittle under paraphrase, high-level answers on risky topics are not reliably safe, and responses that reframe ambiguous requests into safer tasks are substantially less likely to cross the safety boundary. Our results suggest that safe completion should be evaluated as intent-calibrated behavior over controlled task variants, not as a single safety-helpfulness tradeoff over independent prompts.

Anomaly detection is a critical and evolving field in Machine Learning, with applications targeting different domains such as cybersecurity, finance, healthcare, manufacturing and IoT (Internet of Things) systems. Traditionally, anomaly detection algorithms have been designed using both supervised and unsupervised learning paradigms. The fundamental challenge in real-world anomaly detection scenarios is related to the inherent class imbalance (anomalies are typically rare) and, for supervised methods, to the scarcity of labelled anomalous data. Indeed, labelling is both expensive and time-consuming. Conversely unsupervised methods do not require labelling, but may suffer from high false positive rates when deployed in safety-critical applications. In this work we introduce a novel unsupervised algorithm for anomaly detection in time series based on the Haar discrete wavelet and a suitably designed $t$-test. We establish the theoretical foundation of the proposed $t$-test and, through extensive experimentation across 343 datasets, demonstrate that our algorithm outperforms state-of-the-art unsupervised and self-supervised benchmarks.

Disaggregated LLM serving runs prefill and decode on separate GPU pools to keep the two phases from interfering. In practice, this creates a new asymmetry: under bursty, heavy-tailed workloads prefill nodes saturate while decode nodes have compute underutilized, and on a production-style A100 cluster with 2 prefill and 2 decode nodes (2P2D), we find that prefill execution accounts for only 2-23% of P95 Time-to-First-Token (TTFT). Queuing and inter-node GPU-GPU KV-cache transfer account for the rest. We present a proactive prefill-deflecting scheduler that lets decode nodes serve prefill phase of requests as chunked-prefill steps interleaved with their in-flight decode batches. For each queued request, we estimate the TTFT it would see on the prefill node, and on every decode node, search for the largest chunk schedule that keeps in-flight decodes within their Time-Between-Tokens (TBT) SLO and deflect when the decode path helps tail latency. Because the prefill phase of deflected requests runs in place on the decode node, the inter-node KV transfer is eliminated. Implemented on vLLM and evaluated on production-style traces with DeepSeek-V2-Lite, our approach reduces P95 TTFT by upto 81% and raises SLO attainment by upto 79% over state-of-the-art disaggregated schedulers, at sub-millisecond per-request routing cost.

Autonomous surface vehicles vary widely in hydrodynamic and actuation characteristics, yet most controllers are designed for single-platform deployment. We present an adaptive reinforcement learning approach for trajectory tracking that enables zero-shot cross-platform deployment using a single policy. Since the deployment platform's dynamics are unknown to the policy, we address cross-platform generalization with the standard partial-observability approach of conditioning on interaction history, employing a teacher-student architecture in which a learned module infers a latent representation of the platform dynamics. The policy is trained in simulation under randomized vessel dynamics and is deployed zero-shot to two real-world platforms without any fine-tuning, despite relying on a simple analytical dynamics model rather than a high-fidelity hydrodynamic simulator. In real-world experiments on two different platforms, the adaptive policy outperforms non-adaptive learning-based baselines by up to 58% in position mean absolute error while approaching the tracking accuracy of a platform-specific tuned controller.

Evaluating LLM agents on benchmarks like SWE-Bench and GAIA can be expensive, time-consuming, and requires complex infrastructure. A single evaluation can cost thousands of dollars and take days to complete. In contrast, non-agentic LLM benchmarks that test individual capabilities (e.g., reasoning, code generation) are fast and cheap to run. In this paper, we investigate whether performance on expensive agentic benchmarks can be accurately predicted by the performance on a small, carefully selected subset of atomic evaluation instances. We introduce PACE, a framework that constructs proxy benchmarks by selecting instances from existing non-agentic evaluations whose aggregate scores most reliably predict model performances on agentic benchmarks. Given a pool of candidate instances spanning atomic capabilities, PACE fits a regression that maps a model's scores on a compact subset of source instances to its score on the target agentic benchmark. The subset itself is curated by combining two complementary instance-selection strategies, target-relevance local selection and globally informative global selection. We apply PACE to the 4 target agentic benchmarks in this paper, which yields PACE-Bench, the concrete proxy benchmark that we evaluate in the paper. Experiments across 14 models, 4 agentic benchmarks, and 19 non-agentic benchmarks show that PACE-Bench predicts agentic scores with leave-one-out cross-validation (LOOCV) mean absolute error (MAE) under 4%, Spearman correlation above 0.80, and pairwise model-ranking accuracy around 85%, all at much less than 1% of the full agentic evaluation cost. We further analyze the selected proxy instances, revealing which skills each agentic benchmark uniquely demands. PACE enables practitioners to obtain reliable estimates of agentic performance during model development, selection, and routing, without the overhead of full agent evaluation.

Quantum software testing (QST) checks whether quantum programs behave according to their intended specifications. A key requirement for QST research is a benchmark that supports rigorous empirical evaluation on programs that are testable and better reflect current software development practices. However, existing studies heavily rely on small hard-coded or circuit-level benchmarks, while available quantum programs are scattered across repositories without clear selection criteria, which limits fair comparison and systematic reproducibility. To this end, we present Qolumbina, a benchmark infrastructure for controlled QST experiments on scalable quantum programs. Qolumbina curates 40 programs from open-source repositories, turns them into test-ready subjects through systematic selection, refactoring, specifications, test case examples, unit tests, and standardized interfaces. We also propose QST-oriented criteria to characterize quantum programs along functionality, output behavior, development complexity, and quantum-specific execution complexity. Using these criteria, our empirical study shows that Qolumbina covers diverse testing-relevant properties and supports scalability analysis beyond fixed-size circuit benchmarks. Through controlled experiments with two recent QST approaches, we demonstrate the feasibility of using Qolumbina for execution-cost and fault-detection studies, and highlight backend-dependent effects that can influence QST result interpretation.

Multimodal large language models must continually adapt to evolving tasks and domains, yet standard continual learning metrics mainly measure whether old answers remain correct, leaving the stability of multimodal grounding largely unexamined. We study this overlooked failure mode and ask whether a continually adapted MLLM can preserve not only what it answers, but also how it uses visual, textual, OCR, chart, and document evidence. We identify \emph{hidden evidence-use forgetting}, where answer accuracy is retained while the model silently shifts toward different or less grounded evidence channels, and propose \textsc{RCL}, a replay-free reliance-constrained continual learning framework. \textsc{RCL} freezes the previous checkpoint as a behavioral reference, estimates teacher and student evidence-reliance profiles through counterfactual channel interventions, and jointly optimizes task learning, prediction preservation, and reliance preservation without adding inference-time cost. Across CoIN, COAST, MCITlib, and an evidence-sensitive multimodal stream, \textsc{RCL} consistently improves final performance and reduces forgetting over replay-free, PEFT, routing, and memory-assisted baselines, while substantially lowering modality reliance drift, dominant evidence flips, and hidden forgetting rates. These results suggest that robust continual multimodal learning requires preserving the evidence path behind correct answers, not merely the answers themselves.

Mirror Illusion Art is a novel reflection-conditioned 3D illusion where one object yields two target appearances (front and mirror). The task is formulated as inverse design from two target 2D images (front and mirror) to a printable 3D object with geometry and texture. Prior topology-driven and shadow-based approaches demand substantial manual effort, optimize shape only, and often yield non-smooth or incomplete geometry. To address these challenges, we propose AutoMIA, an automated Mirror Illusion Art design pipeline that jointly optimizes shape and color. To stabilize optimization and suppress artifacts, four mechanisms are introduced: (1) projection-alignment component (PAC) selection to reduce surface noise, (2) position-weighted adaptive (PWA) suppression for background noise, (3) internal voxel preservation (IVP) to prevent internal fractures, and (4) shape-color decoupled (SCD) optimization that balance shape and color optimization. AutoMIA generate diverse smooth Mirror Illusion artworks successfully both in the digital and physical world, with only around 76s design time and 2.6 GB memory on average using a single RTX 3090, advancing inverse graphics and computational design. Our code is available at https://github.com/zxp555/AutoMIA.

Multimodal large language models must adapt to evolving tasks and domains, yet continual improvement under bounded deployment footprint remains difficult because repeated parameter updates or growing replay stores can accumulate adaptation state over time. We study fixed-footprint continual adaptation: the deployed adaptation state is kept under a fixed memory budget, while the backbone model is left unchanged and task-specific updates are externalized. We propose InduceKV, a retrieval-based method that stores each selected training prefix as an attention-ready memory entry, consisting of a frozen retrieval key and compact layerwise key--value (KV) payloads that can be appended to the model's self-attention cache. Under a strict memory budget, InduceKV constructs a compact inducing set through bilevel selection: a lightweight calibration is fit for retrieval, while the selected memory balances current-task likelihood, anchor-based retention, and coverage in the frozen retrieval space. Across task-incremental instruction tuning, continual VQA, domain-incremental adaptation, and lifelong multimodal instruction tuning, InduceKV consistently improves over PEFT, MoE, replay, and prompt-retrieval baselines under matched memory budgets. We further report backbone-matched, stage-1 CoIN, compute-matched, and scalability diagnostics, showing that the gains are not due to a stronger backbone, replay alone, or an unbounded candidate pool.

Large language models now score near ceiling on general benchmarks, but these aggregate measures reveal little about how models behave within single disciplines. Existing art-focused evaluations rely on synthetic questions and rarely report item-level properties. This paper introduces EduArt, an educational-level benchmark for art-historical knowledge and visual reasoning in multimodal LLMs. EduArt comprises 871 human-authored questions from Italian secondary-school exercises and US Advanced Placement Art History exams, spanning two languages and seven formats from multiple choice to in-text word placement and error identification. Twelve models from six provider families were evaluated under a default answer-only condition and a motivation condition requiring written justification, and characterized using Classical Test Theory and a logistic regression isolating the effects of format, language, image presence, and model. The benchmark showed strong psychometric properties (mean discrimination 0.514, 82.3 percent good discriminators), while multiple-choice accuracy saturated near ceiling for six models, showing recognition formats alone cannot distinguish frontier models. Format was a strong independent predictor of accuracy: models exceeding 94 percent on multiple choice fell to 23.9 percent on open completion (Claude Opus 4.6) and 6.2 percent on error identification (Claude Sonnet 4.6). The motivation condition changed accuracy in a predominantly negative, family-dependent direction. These dissociations indicate that art-historical knowledge and the ability to deploy it are distinct capabilities, and that single-format benchmarks overestimate what models can reliably do. Mapping this capability profile is a precondition for responsible use of multimodal LLMs in art-historical scholarship, where tasks demand producing and manipulating content rather than selecting from fixed options.

Although neural networks are remarkably effective, their underlying optimization principles remain theoretically elusive, often characterized by non-convex landscapes and stochastic heuristics. In this work, we propose a paradigm shift by replacing the discrete training problem of shallow neural networks with a well-posed continuum variational surrogate. We identify a family of $λ$-convex functionals over parameter densities in weighted Sobolev spaces and prove that these variational problems are globally well-posed, stable, and exhibit unexpected almost $C^3$ regularity. Unlike existing Wasserstein-based or Mean-Field approaches, which often face limited regularity and discretization challenges, our formulation provides direct access to elliptic regularity and convex analysis. This allows us to prove that the optimal parameter density can be obtained by solving a single linear system, bypassing iterative optimization entirely. We establish explicit generalization error controls at a rate of $1/α$ relative to the regularization parameter, and prove that finite-width networks of size $N$ achieve the continuum optimum at an $O(1/N)$ rate. This perspective bridges the gap between the Neural Tangent Kernel (NTK) and feature-learning regimes, providing a principled framework for understanding over-parameterization through the lens of variational calculus.

Time-normalized f0 contours of Mandarin words in conversational speech have been shown to be predictable in part from their contextualized embeddings (CEs). The present study investigates whether CEs also predict spoken word duration for 7470 tokens of Mandarin monosyllabic CV words extracted from a Mandarin corpus of spontaneous speech. We show that CEs indeed are predictive for duration, above chance level, not only at the type level, but also at the level of individual tokens, as indicated by the results obtained with the type-wise and token-wise permutation baselines. We also show that the predicted durations are sufficiently precise to back-transform predicted f0 contours in [0,1] normalized time to contours on the ms time scale. The resulting predicted contours approximate empirical contours and also outperform a permutation baseline.

The silhouette is one of the most widely used measures to assess the quality of a $k$-clustering of a dataset of $n$ elements. Its evaluation requires no information beyond the clustering assignment. In addition, the silhouette is extremely easy to interpret, providing a score to measure the quality of a clustering as a whole or for each element. The exact computation of the: (i) silhouette of each element of a dataset; and (ii) the global silhouette of the clustering; require $Θ(n^2)$ distance calculations, under general metrics. The quadratic complexity $Θ(n^2)$ is extremely prohibitive, especially on massive modern datasets. Surprisingly, existing approximate methods using $O(n^2)$ distance calculations are heuristics not offering provable and controllable guarantees on the quality of their results. We introduce the first rigorous and efficient algorithms to estimate: (i) the (local) silhouette of each element of a dataset; and (ii) the (global) silhouette; of any metric $k$-clustering. Our methods, based on sampling, perform $O(nk\varepsilon^{-2}\ln (nk/δ))$ distance computations, and provide estimates with additive error $O(\varepsilon)$ with probability at least $1-δ$. That is, parameters $\varepsilon$ and $δ$ in $(0,1)$ control the trade-off between accuracy and efficiency. We also introduce a scalable and distributed design of our methods for the MapReduce and Massively Parallel Computing (MPC) frameworks. Our distributed algorithms use a constant number of rounds and sublinear local memory. Finally, we perform extensive experiments against state-of-the-art approaches. The results show that our new techniques yield the best trade-off between accuracy and efficiency for both local and global silhouette estimation. In addition, our methods scale efficiently to massive datasets for which an exact computation of the silhouette is not practical.

Large-scale battery energy storage systems (BESSs) require O&M decisions that combine alarms, cell-level measurements, device topology, diagnostic tables, historical cases, and maintenance documents. Monitoring platforms can flag threshold violations, but they often cannot explain whether voltage inconsistency, resistance drift, short-circuit risk, capacity divergence, or thermal abnormality needs intervention. This digest presents a traceable BESS fault-diagnosis assistant that uses retrieval-augmented multi-agent reasoning to connect operational data, domain knowledge, visual evidence, and report generation. Reliability is improved through BESS-specific task routing, schema-constrained natural-language database access, hybrid text-image retrieval, and evidence-based answer synthesis. Preliminary internal evaluation is reported for routing, database access, and diagnostic reasoning.

Long-running adaptive intelligent agents face a structural tension between knowledge consolidation and information integrity. Memory consolidation is conventionally treated as an agent-changing operation: a model is fine-tuned, a prompt rewritten, a policy distilled, or a reflection appended to the context that governs future behaviour. In regulated autonomic deployment this is a liability because the agent operates under commitments and audit contracts that bind to a specific, cryptographically certified identity. We propose to treat consolidation not as a mutation of the planner or the identity manifest, but as a deterministic function f: M^ep -> M^sem over episodic memory whose output is a separately addressable semantic knowledge layer; the identity hash does not read M^sem, so consolidation updates knowledge without changing the agent's certified identity. We give a formal account of the agent representation, prove identity invariance through a structural lemma on the manifest's hash-input set, specify a deterministic aggregation algorithm whose outputs are auditable database rows with explicit confidence and supporting-event provenance, and validate the construction with synthetic experiments demonstrating per-field correctness, byte-equal identity across consolidation passes, and a mean 79.82% reduction in unproductive planner attempts (95% BCa CI [78.02%, 81.49%] across 10 seeds) against a calibrated Bayesian-shrunk baseline. The construction is a knowledge-update discipline for autonomic agents in which lessons accumulate as queryable facts while the agent's certified identity remains byte-equal across its operational lifetime, with an embodied service agent as the running case study.

Multivariate time-series models for prognostics are often evaluated by point prediction accuracy, yet their internal states rarely expose a coherent degradation process. We study liquid neural networks as latent dynamics models for aircraft engine health monitoring on the C-MAPSS benchmark. The proposed model encodes a history window into a latent state, evolves that state with a liquid transition model, and decodes future sensor observations. To separate health evolution from operating-condition variation, the latent state is factorized into degradation and condition components. Remaining useful life, monotonic risk, and latent-consistency losses supervise the degradation component, while condition prediction and decorrelation losses discourage operating-condition leakage. Across FD001--FD004, the full disentangled model improves overall sensor forecasting RMSE from 0.2438 for a GRU baseline to 0.2266, with the largest gains on the multi-condition subsets FD002 and FD004. The learned degradation state also forms a clearer temporal degradation axis, reaching an average state-speed Spearman correlation of 0.5960. Direct remaining-useful-life regression remains stronger for the GRU baseline, indicating that the proposed representation is currently more effective as an interpretable world model for degradation dynamics than as a calibrated lifetime regressor. These results suggest that liquid latent dynamics can bridge predictive maintenance forecasting and inspectable health-state modeling.

Newer lightweight convolutional neural networks are often presented as improving predictive performance and deployment efficiency, but such claims require controlled evaluation. This study compares nine lightweight CNN model packages across CIFAR-10, CIFAR-100, and Tiny ImageNet under a shared downstream protocol. We report top-1 accuracy, macro F1, top-5 accuracy, parameter count, FP32 storage, GMACs, batch-size-1 latency on an NVIDIA L4 and AMD Ryzen 5 5500U CPU, peak PyTorch CUDA allocated tensor memory, and point estimate Pareto frontiers. EfficientNetV2-S achieves the highest observed top-1 accuracy on CIFAR-10 and CIFAR-100 at 97.57% and 86.98%, while RepViT-M1.0 leads Tiny ImageNet at 79.87%. EfficientNet-B0 remains within 0.22, 0.85, and 1.79 percentage points of the best result on the three datasets while using approximately 79% fewer parameters and 86% fewer GMACs than EfficientNetV2-S. It also appears on every evaluated accuracy and resource Pareto frontier, making it the most consistently competitive intermediate-budget option. MobileNetV3-Small has the lowest GMAC count, is the fastest model under both CPU thread settings, and records higher observed accuracy than MobileNetV4-Conv-S on all three datasets. Under random initialization, it leads MobileNetV4-Conv-S by 2.55, 1.76, and 0.99 points, with paired test-set intervals excluding zero for the fixed trained models. EfficientNet-B0 remains 3.29, 10.10, and 17.54 points below its pretrained counterpart after 100 epochs of scratch training, despite requiring about five times the recorded training time. SqueezeNet1.1 has the fewest parameters and lowest peak CUDA allocation, but substantially weaker accuracy. Latency rankings differ sharply between the L4 and CPU environments, showing that GMACs alone do not predict measured inference performance. Overall, newer designs provide selective rather than universal gains

Using molecular large language models (LLMs) as a unified framework for understanding molecular structures and functions is emerging as a new trend in tasks such as molecular design and drug discovery. However, these models struggle to fully capture the visual representation of molecular structures, limiting their potential. While existing molecular vision-language models (VLMs) show promise, they still face challenges in structural alignment and lack the necessary topological modeling for accurate molecular understanding. To address this, we propose MolSight, a graph-aware vision-language model framework designed to enhance the understanding of molecular images by VLMs. MolSight integrates a Molecular Topology Module to inject chemical-bond adjacency information into vision tokens, and a Molecular Grounding Module to align visual features with chemical symbolic semantics. Our experiments demonstrate that MolSight significantly outperforms existing VLMs, molecular LLMs, and specialized tools across multiple chemical visual understanding tasks, achieving a new level of molecular image reasoning.

Automated authoring of Gherkin Behavior-Driven Development (BDD) acceptance criteria remains a manual bottleneck in requirements engineering. This study investigates whether epic-organized LLM-generated Gherkin produces higher quality and coverage than requirement-aligned generation. We compare our Timeless (an epic-organized LLM pipeline) approach against a naive large language model (LLM) baseline on four requirements documents (107 requirements) from the PURE dataset. Evaluation covers structural metrics, automated requirement coverage via TF-IDF and dense embeddings, and blind expert assessment by four researchers. In our evaluation, the JSON-constrained pipeline produced structurally valid scenarios across all generated outputs, while the zero-shot baseline achieved 99% structural validity. Semantic coverage was comparable to the baseline, with Timeless achieving 94.3% semantic Requirement Coverage Rate compared with 92.9% for the baseline. TF-IDF produced lower coverage scores for the epic-organized output, suggesting that lexical metrics may miss coverage when scenarios paraphrase requirements at a higher level of abstraction. Expert raters prefer the epic-organized strategy on Correctness (4.61 vs 4.14), Executability (4.61 vs 4.07), and Completeness (4.31 vs 3.50). Overall, the results suggest that epic-organized generation can improve perceived Gherkin quality while maintaining comparable semantic coverage, although broader replication is needed before generalizing this finding.

Online multimodal knowledge editing requires injecting a continual stream of visual-textual corrections into multimodal large language models (MLLMs) with bounded overhead and minimal disruption to unrelated behaviors. Existing editors mainly emphasize edit reliability and long-horizon stability, but rarely control the semantic boundary of each edit. Our pilot analyses of post-edit behaviors and internal neuronal activities reveal a scope gap behind reliable edits: instance-level success neither guarantees transfer to valid cross-modal variants nor prevents leakage to unrelated inputs, while edit-related cross-modal responses concentrate in deeper semantic layers. Therefore, we formulate Edit-Scoped Generalization, reframing online MLLM editing from merely correcting an instance to controlling the propagation boundary of each edit. To this end, we propose ScopeEdit, a scope-aware online editor that decomposes each update into a modality-local absorption branch and an evidence-gated shared generalization branch. The local branch supports stable edit absorption, whereas the shared branch enables cross-modal propagation only when visual and textual evidence are sufficiently aligned. Both branches perform scope-separated write geometries in orthogonal low-rank spaces and maintain branch-wise preconditioners via Sherman--Morrison recursions, yielding constant per-edit overhead. Extensive experiments across diverse benchmarks, long-horizon edit streams, MLLM backbones, real-world VLKEB scenarios, and complex vision-language architectures show that ScopeEdit consistently improves the trade-off between in-scope cross-modal transfer and out-of-scope locality, while preserving edit reliability, stability and online efficiency. Our code is available at https://github.com/lab-klc/ScopeEdit.

Ontology learning (OL) aims to automatically construct structured knowledge models from text, yet progress remains fragmented across methods, domains, and evaluation practices. Despite decades of research, OL lacks a shared infrastructure for systematic evaluation and ontology access. This absence has hindered progress and fragmented research, leaving the central challenges of OL largely unaddressed. We introduce OntoLearner, a modular, cross-domain, and first-of-its-kind framework that unifies ontology access, large language model (LLM)-driven learning pipelines, and standardized benchmarking. OntoLearner releases 180 machine-readable ontologies spanning 22 domains and provides pipeline-ready datasets with train/dev/test splits for three core OL tasks: term typing, taxonomy discovery, and non-taxonomic relation extraction. Using this infrastructure, we conduct a large-scale empirical study of OL, evaluating 22 retrieval models and 12 LLMs across domains and tasks. The results converge on a finding that reframes the central challenge of OL: failure modes scale with ontological complexity rather than model size or architectural sophistication. The primary bottleneck is not model capability, but a structural mismatch between how models encode knowledge and how ontologies organize it. These findings establish that effective OL is reachable through the cross-domain, multi-task benchmarking enabled by OntoLearner. OntoLearner is open-source (MIT license) at https://github.com/sciknoworg/OntoLearner/.

This technical report describes our system for Task 1 of the DCASE 2026 Challenge, which aims to classify heterogeneous audio recordings according to the Broad Sound Taxonomy (BST). The task requires both accurate second-level prediction and consistency with the top-level taxonomy. Our system is built on CLAP-based audio-text representations and is improved along three strategies: expanding the training set with a filtered subset of BSD35k, enhancing acoustic modeling with feature-specific branches, and refining predictions using hierarchy-aware classifiers and KNN-based post-processing. Among the acoustic features considered, the log-STFT branch provides the strongest single-model performance. With KNN-based post-processing, our best single system achieves a hierarchical F1 score (Hier. F1) of 80.84% on the BSD10k-v1.2 set under the same evaluation protocol as the baseline. We further construct ensemble systems by combining models with complementary acoustic features and classification heads, achieving Hier. F1 scores of 81.25% and 81.18%, respectively.

Vision-Language Models (VLMs) are increasingly applied in medical tasks such as pathology description, report generation, and visual question answering. Medical Image Quality Assessment (MIQA) supports diagnostic accuracy and patient safety by determining whether images meet the standards required for clinical decision-making. Automating MIQA with VLMs may reduce workload, but their behavior under real-world conditions, where images may be degraded or textual context may affect judgments, should be further explored before deployment. We benchmark VLMs on medical image quality using the MediMeta-C dataset zero-shot across seven corruption types and five severity levels. We evaluate sensitivity to degradation patterns, the effect of corruptions on embedding geometry, and whether textual attributes (demographics, expertise, infrastructure, institution) alter scores. Across 16 VLMs and seven modalities, pixelation produced the largest score reductions (mean -20.58%, up to -34.4% for OCT), whereas brightness had limited effect (-0.81%). Embedding displacement was associated with score changes. Same-family models showed correlations of 0.67-0.83; some produced increases up to +31% for corrupted mammography. Textual attributes affected scores: institutional prestige raised them +17.15%, and equipment age lowered them -14.7%. The largest changes were +95.62% (InternVL-8B) and -37.7% (MedGemma). Current VLMs show limitations for medical image quality assessment. Pixelation, a privacy-preserving transformation, reduces performance, indicating a trade-off between patient privacy and reliability. Sensitivity to contextual metadata indicates limited objectivity and marks metadata as a privacy and bias source. Privacy protection and objective quality assessment are related requirements for use.

Large language models (LLMs) are often asked to produce JSON conforming to a fixed schema, powering information extraction, tool calling, agentic planning, and knowledge-graph construction. Measuring how closely an output matches a gold reference is essential yet surprisingly hard: exact match is brittle, text similarity ignores structure, and an LLM judge is expensive, opaque, and non-deterministic. We address this with Object Aligner (OA), an open-source Python library that scores two JSON objects deterministically by recursively aligning their trees (the Hungarian algorithm for unordered collections, sequence alignment for ordered ones) and awarding partial credit at the granularity the schema declares. The Object Aligner is configured entirely through a set of JSON Schema extensions, so adapting it to a new task involves annotating a schema rather than writing code. Complex structured data, however, are rarely flat trees: records may form graphs or hypergraphs keyed by arbitrary identifiers, breaking the assumptions of prior similarity metrics. Our central contribution, referential alignment, closes this gap by inferring a bijection between gold and candidate identifiers and scoring every reference through it, so the score is invariant to relabeling. Since recovering this bijection exactly is graph isomorphism, the Object Aligner approximates it with Weisfeiler-Leman color refinement. An order-sensitive sequence regime targets ranking and planning. Since the same alignment localizes every mismatch, the Object Aligner emits ranked repair suggestions at no extra cost. Used as a reward inside the GEPA prompt optimizer, Object Aligner helps or stays neutral across all datasets.

Low-voltage load forecasting is an important component in current and future energy systems with a high degree of electrification and decentralized generation. However, current forecasting methods require significant manual effort, often lack uncertainty estimation and proper peak prediction, and they are often not adequately evaluated in terms of grid requirements. In the present study, we provide an extensive evaluation of short-term net load forecasts of 200 real-world low-voltage feeders with a focus on the rapidly evolving time series foundation models. Our study compares Chronos-Bolt, Chronos-2 and TabPFN-TS to six baseline models and demonstrates superior performance, in particular for Chronos-2. An ablation study, in which weather covariates are omitted, shows that time series foundation models adapt to increased uncertainty, despite the importance of weather information. A novel application-oriented metric links the model's forecasting capabilities in peak prediction to the trade-off in grid asset planning and operation between cost reduction and minimizing the risk of failure.

Neural TTS systems can sound natural across languages, but naturalness does not guarantee the preservation of sound contrasts that distinguish words from their grammatical forms. Standard metrics like MOS do not test for this. We propose a classifier-based framework that audits TTS output against language-specific phonological patterns using human speech as a benchmark. Testing Assamese advanced tongue root (ATR) vowel harmony with Meta's MMS TTS, we show that a classifier trained on human speech transfers to synthesized speech with minimal loss. The faithfulness audit reveals that [+ATR] mid vowels are realized as [-ATR] in 1/3 tokens despite an underlying [+ATR] specification, a bias absent in human speech. At the word level, predicted ATR labels classify harmony more accurately than transcription labels, indicating a gap between intended and produced phonology. The framework offers task-specific diagnostics and generalizes to other phonological contrasts with measurable acoustic cues.

Rewriting inputs to improve frozen downstream models has become a common strategy in modern NLP pipelines. Prior work on incremental dialogue discourse parsing (DDP) shows that supervised clarification models can rewrite fragmentary or underspecified utterances, such as resolving ellipsis or references, to improve parsing accuracy. In this work, we revisit this idea under realistic deployment conditions, where no clarification supervision is available and the clarifier must rely on zero-shot prompting or feedback from a frozen parser. Across three Segmented Discourse Representation Theory (SDRT) datasets and multiple parsers, we find that last-utterance clarification is far less reliable than suggested by supervised settings. Parser-agnostic rewriting often introduces more regressions than repairs, as edits that enable fixes also disrupt discourse cues relied upon by the parser. A best-of-8 rewriting analysis further reveals a practical ceiling: a large fraction of errors are not repairable through input rewriting alone. A parser-aware clarifier trained with GRPO reduces regressions by up to 37% by learning conservative abstention, yet still fails to produce selectivity-aware clarifications that consistently improve parsing. Together, these findings recast clarification as a selective intervention problem. We identify rewritability prediction, deciding whether an utterance is repairable before intervention, as the key missing capability for input-side optimization of frozen discourse parsers, and a critical direction for improving agentic pipelines more broadly.

We study the challenging problem of novel view video synthesis from single images or monocular videos. Existing methods, which operate under the assumption that pre-trained video models lack native novel view synthesis capability and enforce view alignment via camera conditioning, task-specific fine-tuning, or stepwise hard denoising guidance, often suffer from artifacts and compromised global scene consistency. In this paper, we introduce NeoMap, a novel training-free framework designed to locate high-fidelity, view-consistent novel view solutions from general pre-trained video models. The key to our approach is the core insight that promising novel view solutions are inherently encoded within the natural video data manifold learned by pre-trained models, and the core challenge is simply to locate this optimal solution. We solve this via our core mechanism: convergent manifold alternating projection iterations that optimize the initial noise. Extensive experiments demonstrate that NeoMap significantly outperforms all existing methods across 3 standard novel view synthesis benchmarks, including the challenging Tanks-and-Temples, LLFF and DAVIS datasets, achieving state-of-the-art generation fidelity and top-tier view consistency.

In this paper, we describe NAVER LABS Europe's submission to the instruction-following speech processing short track at IWSLT 2026. We participate again in the constrained setting, developing systems capable of jointly performing ASR, ST, and SQA from English speech into Chinese, Italian, and German. Building on our previous submission, ranked first in last year's short track, we update our multi-stage training pipeline by replacing the speech projector with SpeechMapper, a method for learning a speech-to-LLM embedding projector using only ASR data. In addition, we introduce a synthetic SQA dataset, fakACL, composed of artificially generated scientific presentations. This dataset is built by prompting the LLM backbone, segmenting the generated talks, and synthesizing speech with SeamlessM4T-large-v2. The combination of an improved speech projection mechanism and domain-specific synthetic data allows our model to outperform last year's best short-track system, while being considerably more compact and relying on a weaker LLM backbone. This year's results place our system tied for first place in the overall short track ranking.

We propose a model agnostic methodology to measure lag relevance in machine learning forecasting models applied to univariate time series. Particularly, we are working in the context of time series using the frameworks of Ghost variables and Shapley values, together with additive importance measures, to introduce the auto-relevance and partial auto-relevance functions as the lag importance values. Additionally, we propose a novel method to replace absent features in coalition based methods with a one step forecast from the same model. We evaluate these proposals under different simulations and real data cases. This combined framework perspective is particularly suitable for time series. In addition, to show our discoveries we use a pull of models from the seasonal ARMA family and recurrent neural networks. We found that the calculated relevance measures successfully demonstrate the expected lag structure in almost all cases.

A/B testing is the gold standard for selecting the better algorithm in online services. While offline evaluation has attracted attention as a safer alternative due to the high experimental costs and the potential risk of degrading user experience and revenue in A/B testing, it is widely recognized that the estimation accuracy of offline evaluation is substantially lower. As a result, final selection decisions are typically made through A/B testing. Contrary to this conventional view, we reveal a counterintuitive phenomenon in which A/B testing can produce a higher algorithm selection error rate than offline evaluation. This occurs because the sample mean estimator used in A/B testing does not induce positive correlation, which is crucial for reducing critical selection errors, namely underestimating the truly superior algorithm and overestimating the truly inferior one. In contrast, offline evaluation methods unintentionally generate this beneficial correlation by relying on shared offline data when estimating and comparing the performance of multiple algorithms. Building on this insight, we propose an estimator that intentionally induces positive correlation to improve algorithm selection in A/B testing. The key idea is to introduce a hypothetical middle algorithm and to estimate the performance difference between algorithms A, M, and B in a stepwise manner using shared data at each step. This approach enables the application of offline evaluation techniques in each step, thereby inducing positive correlation and reducing critical selection errors. Furthermore, we derive the optimal middle algorithm regarding the resulting variance and analyze its advantages over existing methods through bias-variance analysis. Experiments on real-world data demonstrate that our estimator achieves the same selection error rate as existing approaches while using only one half of the A/B testing data.

Large Language Models (LLMs) have become increasingly effective at generating code when task descriptions are clear and precise. Yet, in practice, user-provided task descriptions are often ambiguous, incomplete, or contradictory, leaving critical aspects of the intended program behavior underspecified. In such cases, multiple behaviorally distinct interpretations may satisfy the description equally well, yet semantically differ in ways that matter/affect the user intent. A natural expectation, often assumed by researchers, is that prompt underspecification manifests as incoherence: When asked multiple times, an LLM produces multiple semantically distinct implementations reflecting the ambiguity of the task description. In this paper, we challenge this assumption. We find that LLMs frequently collapse onto a single incorrect interpretation of the task description, consistently generating coherent but behaviorally misaligned code. We term this failure mode detrimental semantic collapse and find that it affects over 10% of tasks in MBPP, 3% in HumanEval, and 32% of LiveCodeBench, all benchmarks assumed to be well-specified. By deliberately injecting underspecification issues in the benchmark prompts, the rate rises to over 5 times, exposing a fundamental blind spot in disambiguation and correctness estimation techniques that rely on incoherence as a proxy for prompt underspecification.

Large language models (LLMs) are increasingly consulted on contested scientific questions, raising the concern that they will sycophantically retreat from established consensus when a user signals doubt -- drifting toward a false balance that treats settled science as one view among several. We test this across three open instruction-tuned models (Llama-3.1-8B, Qwen2.5-7B, Mistral-7B), three consensus-science domains (climate, vaccines, evolution), and single- and multi-turn settings, combining behavioral measurement with linear probing and activation patching. We do not observe sycophantic retreat. Instead, models show three distinct policies under the same skeptical pressure: reactive assertion, where consensus assertion increases rather than decreases (Llama); surface hedging, where tone softens while the position holds (Qwen); and non-response (Mistral). Pairwise judgments confirm the reactive shift is stance, not style (63.6%, p=.007), and a decomposition identifies increased consensus assertion, not false balance, as its driver (beta=+0.042 per dose, p<1e-77). Linear probes localize the divergence to middle layers -- perfect separation in Llama and Qwen versus 72% in Mistral, with non-overlapping confidence intervals -- indicating the non-responsive model does not linearly represent the skepticism signal at all. Crucially, this robustness does not transfer: it attenuates across domains and, in the safety-critical vaccine domain, can reverse, with myth-rebuttal weakening under skeptical pressure. We synthesize these into a four-way taxonomy separating active from accidental robustness, and argue that behavioral evaluation alone cannot distinguish a model that resists skepticism because it understands the signal from one that only appears to resist because it fails to perceive it.

The classical $k$-means clustering cannot be directly used to incomplete data, and existing $k$-means-based clustering for missing data primarily focus on improving the practical accuracy of clustering, whereas most of them lack theoretical guarantees in the asymptotic sense. In this paper, we investigate the statistical properties of $k$-means clustering in the presence of missing data. We first establish the $\sqrt{n}$-excess risk bound and prove the consistency of the estimated cluster centers under general missing mechanisms. For the Missing Completely at Random (MCAR) mechanism, we further derive the $\sqrt{n}$-convergence rate and asymptotic normality of the estimated cluster centers. Moreover, we study in what cases the cluster centers estimated by incomplete data converge to the true cluster centers of original fully observed data, and give a sufficient condition about the missing probability and the separation among true clusters. These results provide a theoretical guarantee for missing-data-$k$-means. Notably, our analysis reveal that under MCAR mechanism, both achieving the $\sqrt{n}$-rate and converging to the true cluster centers require $k$ true centers to be distinct in every dimension, highlighting the significant challenges of application in high-dimensional regimes. Finally, we conduct numerical simulations on synthetic incomplete datasets to support our theoretical analysis results.

Quantum machine learning has recently emerged as a promising paradigm that leverages the expressive power of quantum circuits to address complex learning tasks. In this work, we investigate the applicability of hybrid quantum-classical neural networks to sentiment analysis, a central problem in natural language processing. We focus on a dataset of tweets related to COVID-19, where the textual content is vectorized using TF-IDF and fed into both classical feedforward networks and hybrid architectures incorporating parameterized quantum circuits. Our results show that hybrid models can achieve accuracy comparable to the classical baseline, while exhibiting distinct learning dynamics, especially in terms of validation loss and accuracy, that suggest a richer representational capacity. Moreover, when applying transfer learning to an SMS spam classification task, the hybrid models consistently outperform the classical counterpart, achieving an accuracy increase of 15 percentage points (from 66% to 81%) on the spam class, demonstrating enhanced generalization. These findings highlight the feasibility of employing QML for natural language processing and point toward the potential advantages of hybrid models as quantum hardware continues to advance.

LLM-based agents have shown strong potential for solving complex multi-step tasks, yet existing performance improvements often rely on either scaling to larger backbone models or task-specific fine-tuning. The former incurs substantial computational costs, while the latter typically generalizes poorly across different tasks. Although prompt-based control is training-free and broadly applicable, existing methods still leave input-output dependencies between subtasks implicit in textual trajectories, making verified intermediate results difficult to reuse. To address these limitations, we propose Atomic Task Graph (ATG), a unified control framework for planning and execution. Specifically, ATG maintains an explicit graph to expose dependencies and support reuse. During planning, it recursively decomposes a high-level task into subtasks, forming a sequence of directed acyclic graphs (DAGs) whose evolution can be traced. During execution, the dependencies exposed by ATG allow independent branches to be executed in parallel, thereby improving execution efficiency. When failures are detected, ATG leverages the graph evolution history to localize the error source and repair only the affected region, preserving validated regions unchanged. Experiments show that ATG consistently outperforms strong baselines in success rate and execution efficiency across three interactive benchmarks using only 7B-8B backbones.

Mechanistic interpretability often relies on component-level interventions to discover how a model produces a behavior. This guides attribution, capability knockout, and model pruning downstream to operate by scoring each unit by the effect of ablation in isolation. Such first-order scoring is natural when component importance is additive, but becomes misleading when a transformer self-repairs: after a primary component is removed, a dormant backup can take over, muting the primary's measured effect while the backup itself appears irrelevant on the intact model. We recast this failure as a recovery task, conditional circuit completion, and introduce Conditional Co-Ablation (CoAx), a label-free, output-grounded score that asks how much each remaining unit's ablation effect grows once a primary set has been removed. This conditional growth exposes the second-order interaction that single-unit scores discard. On the GPT-2-small IOI circuit, CoAx raises backup-head recovery from 0.33 to 0.91 ROC-AUC, outperforming all baselines, including self-repair-aware gradient scores (best 0.82); counterfactual patching verifies that the recovered heads causally carry the repair. The same label-free procedure transfers to induction across eight models. Beyond discovery, the recovered backups correct self-repair-masked attribution, identify the components required for capability knockout, and yield repair-aware structured pruning scaling from 124M to 7B. Component importance is therefore not merely an isolated-unit property: in robust circuits, the components that matter can become visible only under the interventions that make them necessary.

Manipulating fast and dynamically moving targets in unstructured 3D environments remains challenging for embodied AI. Existing visual-language-action models and world models struggle with accurate 3D geometry and physically meaningful forecasting. We propose PhysMani, a framework that couples a physics-principled 3D Gaussian world model with a future-aware action policy model. The world model learns a divergence-free Gaussian velocity field via online optimization for fast and physically grounded future dynamics prediction. The policy model integrates the predicted 3D scene future dynamics through a learnable token based cross-attention module. We introduce PhysMani-Bench, a dynamic manipulation benchmark with 16 tasks, and demonstrate a superior success rate over strong baselines in both simulation and real-world robot experiments.

Learning causal models from high-dimensional data is a significant challenge, particularly in real-world settings where violations of core assumptions lead to causal identifiability issues. Although massive amounts of prior knowledge are available, and contain valuable causal information, effectively integrating this knowledge into the causal discovery process remains an open problem. We introduce CausalSTeward (CAST), a novel human-in-the-loop framework for interactively assembling large causal models. CausalSteward is a multi-agent collaborative system that tackles high-dimensional causality through a divide-and-conquer approach where large clusters of variables are iteratively partitioned and then separately analyzed. Our framework fuses prior knowledge with a data-driven approach by using tailored tools such as retrieval augmented generation and conditional independence tests. Finally, we use this work to examine the capabilities and limitations of causal reasoning in multi-agent frameworks, and how the human-in-the-loop can contribute to accurate and trustworthy results.

Long term memory lets LLM agents act as persistent assistants, but user facts change. A useful memory system must know what is true now, what used to be true, and what changed. We study \emph{ghost memory}, a state coordination failure in which old, current, and transition facts coexist in the memory bank, remain mixed during retrieval, and mislead the answer model. We argue that memory systems should be understood and optimized from three levels: bank maintenance, retrieval, and answer time resolution. We propose ATMA, a state aware overlay for existing memory systems. ATMA keeps superseded and transition records in the bank, builds evidence packets for the query's requested state view, and exposes current, historical, and transition labels to QA. We further call for decoupled evaluation of bank, retrieval, and answer level failures, since final QA accuracy can hide where ghost memory occurs. To make this failure measurable, we build LTP (LoCoMo Temporal Plus), a conflict heavy benchmark for ghost memory, and evaluate on LoCoMo for long conversation generalization. On LTP, Graphiti+ATMA improves conflict accuracy by 0.240 absolute over Graphiti. On LoCoMo, Graphiti+ATMA raises temporal F1 from 0.0295 to 0.1705. The gains are host dependent, but they indicate that explicit state roles can reduce memory failures hidden by final QA accuracy.

This work introduces AIriskEval-edu-db2, a new dataset designed to train and evaluate auditors based on LLMs for an explainable pedagogical risk assessment in instructional content for grades K-12. The dataset comprises 1,639 explanations from 170 curated ScienceQA questions, covering science, language arts, and social sciences. For each question, the dataset includes an explanation written by a human teacher alongside 11 explanations generated by LLM-simulated teacher profiles associated with distinct pedagogical risks. We propose a comprehensive risk rubric aligned with established educational standards that covers five complementary dimensions: factual precision, depth and completeness, focus and relevance, student-level appropriateness, and ideological bias. A key contribution is the addition of 785 explanations with structured explainability annotations, including risk localization and risk description. The annotations are produced through a semi-automatic process with expert teacher validation. Finally, we present validation experiments comparing state-of-the-art proprietary models with a lightweight local Llama 3.1 8B model in both the pedagogical risk detection and the explainability assessment. These experiments evaluate whether supervised fine-tuning on AIriskEval-edu-db2 enables a locally deployable model to approach or outperform stronger frontier models while preserving privacy in educational auditing and assessment tasks.

Recent advances in agentic program repair have significantly improved issue resolution by enabling iterative repository exploration. However, existing approaches predominantly rely on sequential, text-based code navigation, which fundamentally limits their ability to reason over large-scale long-horizon repositories with complex and long-range dependencies. As issue-resolution agents traverse repositories through fragmented textual observations, structural information such as module organization, call relationships, and dependency chains must be repeatedly reconstructed across interaction steps, often leading to exploration drift and incomplete localization. We present DUALVIEW, a dual-modal structural scaffolding framework that brings visual reasoning into repository exploration for issue-resolution agents. DUALVIEW represents repository structure through four complementary graph views: Module Coupling Graph (MCG), Function Call Graph (FCG), Class Hierarchy Graph (CHG), and Program Dependence Graph (PDG), and exposes them through a queryable interface with visual and textual responses. Rather than reconstructing repository structure from a sequence of textual observations, agents can directly reason over persistent visual representations of code dependencies, enabling more effective exploration and understanding of long-horizon codebases. We evaluate DUALVIEW on SWE-bench Pro and Verified. Results show that DUALVIEW consistently improves issue-resolution performance across different agent architectures and model families. Further ablation studies demonstrate that the gains arise not only from textual structural information but also from visual externalization of repository dependencies, which better supports long-horizon repository exploration.

This paper presents TUDUM (Türkçe Düşünen Üretken Model), a project pipeline for adapting a Qwen-family 27B thinking model toward Turkish reasoning. The central problem is not only to answer Turkish prompts in Turkish, but to make the explicit reasoning trace itself Turkish. A thinking model may translate a Turkish prompt into an English-centered internal or visible scratchpad, solve the problem mostly in English, and only localize the final answer. TUDUM instead treats the generated <think>...</think> block as a trainable behavior. The pipeline starts from the project base checkpoint unsloth/Qwen3.5-27B, applies supervised fine-tuning (SFT) on 15,991 Turkish reasoning examples using LoRA adapters, and then applies GRPO-family reinforcement learning on a proxy-filtered Turkish mathematics environment. The results are mixed. SFT made the model shorter and more consistently Turkish in its reasoning behavior, with large reductions in average response length and thinking exhaustion, but reduced benchmark accuracy. RL recovered some mathematical performance, especially AIME24 at the best early checkpoint, yet did not uniformly improve all benchmarks and did not exceed the base model on the reported Macro-6 average. The contribution is therefore best framed as a technically honest Turkish-thinking reasoning pipeline and evaluation, not as a claim of state-of-the-art Turkish reasoning. The released step-50 model is publicly available.

Communication systems designed for reliable data reconstruction, rather than task-oriented communication, typically rely on separate source and channel coding and incur high latency under limited spectrum availability and fading channels. To address this, we propose a transmission framework with opportunistic spectrum access, in which the transmitter sends discrete latent representations learned via a vector-quantized variational autoencoder (VQ-VAE) over idle licensed channels using standard digital modulation. The AI-powered receiver is still able to reconstruct task-related information from the heavily compressed data. We develop a cross-layer latency model that accounts for compression, block errors, retransmissions, and stochastic channel access. Results on latency-accuracy trade-offs show that the proposed scheme achieves at least 79- and 3.3-fold latency reductions with only 5.7% and 2.4% drops in classification accuracy compared to benchmarks using conventional source and channel coding. The framework enables low-latency communication and reliable task execution even under limited spectrum availability and challenging channel conditions.

Agentic systems enhance their capabilities by invoking external tools and maintaining persistent memory. However, these external dependencies introduce novel attack surfaces. Recent tool and memory poisoning attacks show that maliciously crafted tool descriptors and poisoned memory can covertly bias agent behavior. These threats reflect a deeper issue: the lack of verifiable continuity in the agent's contextual state for planning and execution. We present ElephantAgent, a protocol that enforces Contextual State Continuity to defend against contextual state poisoning. Inspired by prior state-continuity mechanisms (e.g., Nimble), ElephantAgent extends this protection to the evolving contextual state of agentic systems. We define the contextual state as the bounded, security-critical subset of the agent's entire context (e.g., tool state and memory). Before processing each query, ElephantAgent recomputes the digest of the local contextual state and verifies it against the latest authorized digest. Using replicated trusted hardware, ElephantAgent maintains a linearizable ledger of authorized contextual state transitions and detects out-of-band state tampering. To handle in-band semantic abuse, ElephantAgent additionally provides Historical Traceability, enabling conditional post-hoc audit and recovery to a known-good prior state.

We present Zeus, a unified tuning-free Time Series Foundation Model (TSFM) that delivers superior performance across diverse analysis tasks without any task-specific fine-tuning. Unlike prior studies that primarily focus on zero-shot forecasting but require task-specific tuning for other tasks, Zeus bridges this gap by addressing two fundamental challenges in multi-task generalization. First, to reconcile point-level granularity with long-sequence scalability, Zeus incorporates a multi-scale Transformer featuring point-wise tokenization and a U-shaped hierarchy, effectively balancing fine-grained fidelity with computational efficiency. Second, to accommodate varying inductive biases across different tasks, Zeus introduces Multi-Objective Temporal Masking (MOTM), a unified strategy that supports heterogeneous tasks (e.g., extrapolation, interpolation, and global abstraction) within a single framework. Extensive experiments across five representative tasks demonstrate that Zeus consistently achieves competitive results in tuning-free settings, underscoring its potential as a general-purpose TSFM.

Large language model agents can repair real repository issues, but they often spend large context budgets on whole-file reads, broad searches, and long terminal outputs where useful evidence is mixed with irrelevant code and logs. This paper presents ContextSniper, AntTrail's token-efficient code memory layer for repository-level program repair. As the coding specialization of AntTrail's broader agent memory engine, ContextSniper implements the Sniper feature for precision evidence selection: it retrieves candidate code and runtime evidence, ranks it with hybrid retrieval signals, filters long outputs through an intention-aware context gate, and returns compact evidence packets while preserving recoverable source context outside the prompt. We evaluate ContextSniper on SWE-bench Lite with OpenClaw and Claude Code, using 50 task runs per host-agent condition. ContextSniper reduces total token use by 51.5% and logged cost by 36.4% for OpenClaw, and reduces total token use by 38.9% and estimated cost by 27.3% for Claude Code. Submitted-resolution rates decrease slightly, from 26.0% to 24.0% for OpenClaw and from 32.0% to 30.0% for Claude Code. ContextSniper's pilot testing scripts are open-sourced at https://github.com/Calluking/ContextSniper

Semi-supervised generative adversarial networks (SSL-GANs) can exploit large unlabeled datasets while retaining a classifier in the discriminator, but their training is often unstable. This paper proposes a population-based evolutionary training strategy in which discriminator learning is formulated as a multi-objective optimization problem. Instead of aggregating the supervised and unsupervised components of the SSL objective into a single scalar loss, the method maintains a population of discriminators ranked by Pareto dominance, enabling the exploration of different trade-offs between classification accuracy and real/fake discrimination. This formulation aims to improve both roles of SSL-GANs: learning accurate classifiers and training generators capable of producing realistic samples. We analyze several variants, including an elitist strategy and a mono-objective ablation, to assess the role of multi-objective selection. Experiments on MNIST with limited labels show improved training robustness compared to SSL-GAN and CE-SSL-GAN state-of-the-art baselines, while the elitist variant consistently achieves the highest classification accuracy.

Enterprises increasingly mandate AI coding tools and report large productivity gains, yet longitudinal evidence on how such a mandate unfolds is scarce. In this paper, we present a quantitative case study of a documented enterprise "2x" mandate at a mid-sized, AI-forward company that has been committed to doubling merged pull requests per engineer since mid-2025. In a panel of 802 developers and 196,212 pull requests (January 2024-April 2026), per-capita throughput eventually doubled, reaching 2.09x the pre-mandate baseline in April 2026, among the largest gains reported from a field deployment of AI coding tools to our knowledge. A staggered difference-in-differences design links the within-developer share of this gain to AI adoption and to a further gain that grows with accumulated use, with the mandate acting as a catalyst rather than a direct driver. Because adoption and usage intensity were not randomly assigned, we read this evidence as strongly implicating an adoption-and-use channel rather than as exact causal attribution. The gain is broadly shared across seniority yet concentrated in newer code and not separable across model generations. Adoption also restructured code review around automation: per-reviewer load roughly doubled and automated review overtook human review, while merge and revert rates held steady.

LLM-integrated applications blend natural language prompts with program code, and much of their runtime behavior originates in the prompt layer rather than in the code itself. Existing complexity metrics, however, operate solely at the code level and therefore overlook this behavioral logic entirely. We present HECATE, the first tool designed to assess complexity in both the prompt and code layers of such applications. Central to HECATE is Prompt-as-Specification, a Hoare-logic-inspired formalism that interprets every prompt as a specification of intended behavior. Grounded in 25 complexity dimensions identified across published taxonomies, the tool generates 52 candidate metrics. We assess each metric against 118 components collected from 18 open-source repositories, relying on maintenance activity derived from version history as an empirical proxy for complexity, and discard any metric that loses significance once code size is accounted for. Only ten metrics withstand this test. Seven belong to our newly introduced set; rather than measuring sheer volume, each tallies structurally distinct elements, such as LLM call sites, memory attributes, and prompt templates, an attribute we call structural breadth. Of the three surviving conventional metrics, RFC exhibits a similar breadth-oriented character, while Halstead N and V survive only as a residual effect of size; our top-performing metrics exceed all three. Crucially, the prompt-layer metrics retain significance even when the strongest code-level metric is added as a covariate, establishing prompt complexity as a dimension in its own right. A final validation on 20 components spanning six held-out repositories shows that the two best-performing metrics continue to predict maintenance effort, supporting their generalizability beyond the training set.

Reliable confidence estimates are essential in semantic segmentation, especially in safety-critical settings where overconfident errors can mislead downstream decisions. Yet modern segmentation models often remain miscalibrated. Post-hoc calibration offers a practical way to correct confidence estimates without retraining the segmentation model, but its use in dense prediction raises structural issues that are often overlooked. We study two such issues. First, adding a constant to all logits leaves the softmax probabilities unchanged, but several standard calibrators can still depend on this arbitrary offset. As a result, two logit representations encoding the same predictive distribution may yield different calibrated probabilities. We define translation-invariant (TI) calibrators as those whose outputs are unchanged under such shifts, characterize which common calibrators satisfy this property, and construct TI counterparts of shift-sensitive calibrators to isolate the effect of removing representation dependence. Second, post-hoc calibration is typically fitted by minimizing a likelihood-based objective, whereas segmentation models are trained with task-specific metrics such as Dice. This mismatch can cause calibration to alter class orderings and degrade the deployed segmentation map. We study decision-preserving calibration under argmax- and order-preservation constraints. Since enforcing these constraints collapses affine softmax calibrators to temperature scaling, we introduce class-conditional affine calibrators that can be made argmax- or order-preserving while retaining greater expressivity, allowing us to quantify the calibration-segmentation trade-off induced by decision preservation. Across natural-image and medical segmentation benchmarks, and under corruption-based covariate shift, matched comparisons show that TI variants generally improve calibration metrics, while decision-preserving variants prevent segmentation degradation and retain strong calibration performance. These results provide practical design principles for well-defined post-hoc calibration pipelines in semantic segmentation.

Effective brain disease diagnosis requires the synergy of brain connectivity patterns and high-level semantic knowledge. Existing methods, however, largely treat semantics from large language models (LLMs) as auxiliary features or supervision, limiting their direct role in decision-making and constraining classification stability and robustness. To overcome this, we propose a semantic-aligned brain network framework that actively integrates LLM-derived semantics into the prediction process. Specifically, ROI-level semantics are first incorporated via global self-attention to enrich node representations and provide whole-brain context. Multi-scale hypergraphs are then constructed to explicitly model functional subnetworks and multi-ROI interactions, addressing the locality limitations of traditional GNNs and capturing high-order dependencies. Finally, a decision-level semantic alignment mechanism selectively injects patient-specific textual embeddings into graph representations, enabling semantics to directly guide predictions without perturbing the underlying network structure. Experiments on public brain network datasets ABIDE and ADHD-200 demonstrate state-of-the-art performance, enhanced stability, and improved interpretability, particularly in small-sample settings.

Dependency length minimization (DLM) is a well-documented processing universal, but previous studies report a single mean dependency distance (MDD) per language, obscuring variation across syntactic relation types. We analyze 122 languages in UD and SUD (version 2.17), showing that DLM operates on two distinct levels. Grammar-driven optimization targets functional dependencies (det, case, aux), which are universally short (mean 1.71, $σ$ = 0.33) and invariant across typologically diverse languages. Processing-driven optimization operates on lexical dependencies (nsubj, obj, obl), which are longer (mean 2.87), highly variable ($σ$ = 0.63), and constrained by word-order typology. This asymmetry holds in SUD despite reversed head direction (r = 0.92). We conclude that ''the grammar does the work'' of minimization by scaffolding sentences with local functional attachments, leaving processing pressures to determine the ordering of lexical heads.

We introduce Rank-Then-Act (RTA), a framework for learning control policies from expert video demonstrations without environment rewards. RTA trains a Vision-Language Model (VLM) offline as a progress-based ordinal scorer, using a Group Relative Policy Optimization (GRPO) objective over shuffled frame sequences, which forces the model to recover temporal ordering from visual semantics rather than trivial time cues. Importantly, instead of using the scorer directly as a scalar reward model, we propose a correlation-based reward function for reinforcement learning: at each interaction window, we compute the Spearman rank correlation between predicted progress rankings and true temporal indices, yielding a bounded, scale-invariant learning signal. This design decouples reward learning from absolute calibration and enables stable transfer across tasks and environments. We evaluate RTA on discrete control benchmarks (PyBoy: Catrap, Kirby) and continuous control tasks (PointMaze, MetaWorld). RTA consistently matches or outperforms prior video-based reward learning methods and rank-based baselines, while demonstrating strong cross-task reuse of a single pretrained progress scorer. Our results suggest that correlation-structured supervision over video-derived ordinal signals is sufficient for policy learning, offering a scalable alternative to explicit reward design.

We study ridge-regularized log-density-ratio estimation in the Gaussian location model with a common covariance matrix. By affine invariance, the model is written as q $\sim$ N(0, I), p $\sim$ N($Δ$, I), with linear features, where $Δ$ is a mean vector. The variational estimator is the empirical Kullback-Leibler (KL) log-normalized fit with a squared L2-penalty on its nonconstant coefficient, and the spectral estimator recently introduced in [1] replaces a single variational problem by a continuum of ridge-regularized least-squares problems. We derive high-dimensional deterministic asymptotic equivalents when the numbers of observations and dimension tend to infinity with fixed ratios. The regularized variational limit is characterized by a scalar entropy minimization problem derived from the convex-Gaussian-min-max theorem (CGMT), while the regularized spectral limit follows from deterministic equivalents for resolvents of weighted sums of two independent Gaussian sample covariance matrices. We use these formulas to compare population risks, with experiments focused on fixed-signal aspect-ratio sweeps and optimized regularization. Our conclusion is that with many observations, under the criteria and asymptotic regimes analyzed here, the well-specified variational estimator has the smaller risk, while with fewer observations, the spectral estimator is favored because its covariance-based construction has lower variance. We also study how a nuclear penalty can be used and partially analyzed to perform feature learning.

Speculative decoding accelerates autoregressive generation by drafting a block of tokens that the target model verifies left-to-right, committing only the longest accepted prefix. Block (DLM-style) drafters predict the whole block in parallel, which is fast but trained with a full-block cross-entropy that supervises every position against the gold continuation -- even though inference discards every token after the first rejection. Recent acceptance-aware objectives patch this by reweighting the full-block loss; we instead use teacher-forced learning as a motivation for how supervision should concentrate on the accepted prefix. A mask-only block drafter has no input-side channel for gold-prefix conditioning, so AUF approximates that prefix-sensitive supervision on the loss side by keeping the cross-entropy support only through the drafter's first predicted failure. AUF is a single, detached change to the CE support -- no auxiliary objective, no verifier rollouts, and no change to the inference pipeline or the exactness contract. Within fixed drafter backbones and serving settings on Qwen3-8B, AUF raises the DFlash drafter's average emitted length $τ$, averaged over six benchmarks, from 2.40 to 2.61, with a gain on every benchmark, and transfers to Domino's two-branch head (2.56 to 2.68). Two findings sharpen the picture: the decay-only baseline reaches higher token accuracy on the shared block mask yet decodes worse, and on DFlash, once AUF truncates the support, the standard exponential position-decay weighting becomes empirically inert.

The security of open source applications benefits considerably from the possibility of rebuilding their source and verifying the output. F-Droid, a prominent distribution for open source Android applications, systematically rebuilds them from source and tests their bitwise reproducibility at app publishing time. However, F-Droid offers no guarantee that app reproducibility will continue to hold in the future. As software ecosystems evolve, reproducibility may degrade, with potential negative consequences for software preservation and security. We present the first empirical study of build reproducibility in the F-Droid app ecosystem. Analyzing historical reproducibility logs, we find that the overall bitwise reproducibility rate has been steadily increasing over time (as new versions of apps are published). We then evaluate how reproducibility holds in time for fixed app versions, by attempting to rebuild 18 904 app versions that F-Droid had previously confirmed bitwise reproducible, published between September 2018 and February 2026, achieving an 83% rebuild success rate, and identify missing dependencies as the dominant cause of failure, accounting for 76% of non-rebuildable cases. Among successfully rebuilt apps, 94% are also bitwise reproducible-i.e., they still yield bitwise identical artifacts upon rebuild. Together, these results show that while bitwise reproducibility largely holds for apps that can be rebuilt, rebuildability itself is highly sensitive to temporal decay.

Code is the medium through which large language models generate structured artifacts: charts, scientific figures, vector graphics, CAD models, 3D scenes, and hardware designs are all produced by writing programs. In this regime single pass inference is brittle, because the compiler, renderer, or simulator that decides whether the artifact exists is invisible to the model. We present PairCoder, which grounds review in the toolchain and realizes it as two agent pair programming: a Driver agent writes the program, a Navigator agent reviews it against verification evidence (diagnostics, execution results, and renderings of the current artifact beside the target), and the two switch roles when errors persist. Across 17 public benchmarks and seven models from three vendors, PairCoder improves essentially every benchmark whose artifact is verifiable, on full official metric suites rather than execution alone (for example, Blender scene executability 0.20 to 0.78; TikZ compile rate up 10 to 30 points on every model), at 2.9 to 9.2 times single model cost (about 7 times overall). The improvements concentrate where the toolchain provides an informative oracle and the baseline leaves headroom, and the method ties or mildly regresses where the oracle is weak; we frame pair programming as a reliable recipe for verified code driven generation.

Value functions are an essential component in actor-critic based deep reinforcement learning (RL). Conventionally, these functions are trained as a regression task by minimising the mean squared error (MSE) relative to bootstrapped target values. Meanwhile, in distributional RL, a distribution of returns is modelled based on the distributional Bellman operator. This work investigates the Gaussian Histogram Loss (HL-Gauss), a recent approach that reframes value estimation as classification by encoding each scalar Bellman target as a Gaussian-smoothed categorical target. Despite its potential, applying histogram-based losses to RL presents inherent challenges, most notably the requirement to pre-define a fixed support interval, which is often complicated by the non-stationary and stochastic nature of target values typically found in RL tasks. In this work, we propose an approach that dynamically learns the lower and upper bounds of the support instead of assigning them beforehand. We derive an objective that jointly learns these bounds whilst learning the categorical representation of the scalar values, and we show that this objective forms an upper bound on the mean-squared Bellman error. Our theoretical analysis further shows that this bound is tighter than that of non-learned supports of HL-Gauss. Empirically, the proposed objective enables stable adaptation of the support interval and matches HL-Gauss-based actor-critic algorithms on most continuous-control tasks whilst improving on a subset, without requiring a pre-specified support interval.

Large Vision-Language Models (LVLMs) have achieved remarkable progress in multimodal understanding, yet their enormous parameter scale and cross-modal computation incur substantial memory and latency overhead, severely limiting real-world deployment on resource-constrained devices. Binarization offers an attractive solution by drastically reducing storage and computational costs. However, existing binarization methods neglect the varying importance of weights across different layers and modalities. This causes parameters irrelevant to downstream tasks to be unnecessarily retained, whereas modality-critical weights may not be adequately optimized, resulting in significant performance degradation. To address these challenges, we develop a novel \underline{S}ignificance-\underline{A}ware \underline{B}inarization for \underline{L}arge \underline{V}ision-\underline{L}anguage \underline{M}odels (SAB-LVLM). Specifically, after constructing Hessian matrices for textual and visual inputs, we propose a spatial significance map to distinguish full-precision weights activated under a single modality from those activated across modalities. We then devise a modality-guided integration strategy to obtain the significance-aware binarization map, which measures weight significance across layers and modalities. Subsequently, this binarization map is incorporated into the binarization objective as an error reweighting term, and binarization fitting is performed through an alternating significance-weighted update scheme. Extensive experiments illustrate the superiority of our SAB-LVLM over existing binary PTQ methods under an approximately 1-bit compression constraint. Our code is accessible at https://github.com/LyuQi127/SAB_LVLM.

Skills are becoming a reusable operational layer for LLM agents, encoding SOPs, domain rules, tool workflows, scripts, and validation routines. In realistic skill repositories, overlapping skills make reliable skill-use difficult. Final verifier success is too coarse for both evaluation and training, since an agent may pass through trial and error while selecting distractor skills, skipping required steps, composing workflows incorrectly or omitting final checks. We introduce SkillCoach, a self-evolving rubric framework for evaluating and enhancing agentic skill-use. SkillCoach derives skill-grounded process rubrics from real rollouts and evaluates trajectories along four dimensions: skill selection, skill following, skill composition, and skill-grounded reflection. It keeps the external verifier as a separate outcome signal, allowing process quality to be distinguished from accidental task success. The evolved rubrics further serve as process supervision for selecting high-quality training trajectories. Experiments show that evolved rubrics substantially improve evaluation quality, expose failures hidden by final accuracy, and provide stronger supervision signals than outcome-only filtering for enhancing agentic skill-use.

Camouflaged Object Detection (COD) aims to locate and segment objects that blend into their surroundings, presenting challenges due to weak edge cues and ill-defined boundaries. Traditional COD models rely on hand-designed architectures and multi-scale feature fusion, which are often guided by intuition rather than systematic search. This paper introduces CamoNAS, a frequency-aware multi-resolution Neural Architecture Search (NAS) framework for COD. CamoNAS automatically searches both cell-level operations and network-level downsampling paths, forming a hierarchical search space tailored to detect camouflaged objects. Additionally, it adopts an RGB frequency dual-stream architecture, where a learnable wavelet transform complements the RGB spatial stream. CamoNAS achieves state-of-the-art performance on four COD benchmarks (CAMO, COD10K, NC4K, CHAMELEON), highlighting the effectiveness of NAS for COD. Our code is available at https://github.com/rendaweiSIMIT/CamoNAS.

The use of LLMs in software development has become increasingly widespread on tasks such as code generation and summarization. Reports from large technology companies showed that around 20% to 30% of their code are generated by LLMs. However, there remains skepticism about the practical usage of LLM-generated code and comments, such as concerns on more time for debugging the generated code and the unnaturalness of the generated comments. In this paper, we study the code and comments detected as likely to be generated by LLMs and their characteristics, the differences between company- and community-maintained repositories, and how likely bugs are associated with LLM-generated code. We conduct extensive experiments on active company- and community-maintained repositories from 2021 to 2025 using various tools and techniques that detect code and comments generated by LLMs. Based on our detector-based proxy analysis, the results suggest that code detected as likely to be generated by LLMs decreased over time and appeared frequently in test cases, while that of comments remains relatively stable. Proxy results further suggest that code detected as likely to be generated by LLMs shows substantial intra-repository code clones, whereas comments exhibit a relatively low proportion of grammatically correct sentences. In addition, the company-maintained repositories show a higher percentage of code and comments detected as likely to be generated by LLMs, and only a small percentage of the human-labelled bugs are detected as being likely associated with LLM-generated code.

Safety training for large language models (LLMs) is conducted predominantly in English, leaving uncertain how well safety mechanisms generalize to low-resource languages and mixed-language code-switching. We show that this creates an epistemic gap in which models confidently generate harmful responses for inputs that fall outside the distribution of their safety training. To study this phenomenon, we introduce STEER (Safety Targeted Embedding Exploit via Refinement), a gradient-guided attack that identifies words contributing most strongly to the model's refusal behavior and iteratively translates them into low-resource languages to suppress refusal while preserving harmful intent. Across six open-source 8B-parameter models, STEER achieves attack success rates of up to 93.0% on JailbreakBench and 96.7% on AdvBench, outperforming random code-switching and Greedy Coordinate Gradient (GCG). The resulting prompts also transfer to GPT-4o-mini, achieving a 35.5% attack success rate without requiring access to the target model, suggesting that the underlying weakness is not specific to a single architecture. These findings demonstrate that safety mechanisms aligned primarily on English cannot be assumed to generalize across multilingual inputs. We argue that improving multilingual safety requires broader coverage during alignment and mechanisms that explicitly detect and abstain on out-of-distribution inputs.

In LLM-assisted software development, coding is often iterative. We study regression accumulation in multi-turn LLM programming conversations, where later code suggestions may break requirements introduced in earlier turns. Reliability therefore depends not only on satisfying the current request, but also on preserving previously satisfied behavior. We construct 542 tasks from HumanEval+ and MBPP+ and extend each task into an 8-turn requirement-evolution chain. We evaluate six LLMs on 26,016 turn instances (542 x 6 x 8). At each turn, we test whether the current code still passes earlier benchmark tests. We also analyze 384 failure cases from the failure population and build a taxonomy of multi-turn regression bugs through independent four-annotator labeling. Our results show that regression accumulation appears across all six models: 40% to 73% of tasks lose previously correct behavior over the full conversation. Final-turn quality is lower than initial-turn quality across models, especially when later turns add input validation or broader input types. Manual analysis shows that Cross-Turn Conflict, where later code conflicts with earlier requirements, is the main failure class. We further find that Verification Gate, which checks new code against prior tests and triggers rollback and retry, is the only strategy that consistently improves all models, raising final-turn quality from 75.8% to 87.9% on DeepSeek-V3 and from 31.6% to 47.3% on Llama-3.1-8B. These findings suggest that strong single-turn performance can overestimate reliability in multi-turn coding conversations. Future evaluation and tool design should test whether later code suggestions preserve earlier requirements and should include Verification Gate mechanisms.

Can a platform tell, before deployment, whether an open-weight checkpoint has had its refusal mechanism stripped? Runtime guards cannot: they score generations, not the artifact. We combine two cheap internal signals, a reference-anchored activation refusal-gap and a weight-recovery energy of the base-to-candidate weight difference, into a threshold-free checkpoint audit. The two are negatively correlated and label-complementary: the gap supplies refusal-specificity and the weight energy supplies recall. On a 273-checkpoint registry spanning Qwen, DeepSeek-distilled Qwen, Llama, and Gemma, their z-sum separates 57 public abliterations from 37 benign fine-tunes, merges, and instruction-tunes at AUROC 0.95, significantly above either signal alone (0.84, 0.90), and a Youden-calibrated threshold transfers to held-out families at balanced accuracy 0.89 (FPR 0.11), missing only 4 of 57. We then map two failures, in order of severity: a spoofed reference evades both axes with no training (ΔW=0, \r{ho}=1 by construction), and a white-box owner trains a checkpoint past the threshold while it stays guard-unsafe and coherent. The audit is effective triage, not tamper-proofing: it presumes an attested reference, and its claims are bounded by the registry we evaluate it on.

Retrieval-Augmented Generation (RAG) systems use the question-answering capabilities of Large Language Models (LLMs) to access information outside their parameters. We evaluate if cluster-based semantic chunking improves retrieval and answer quality compared to fixed-size and recursive chunking evaluating on long, structured academic theses using the Retrieval Augmented Generation Assessment (RAGAs) framework. RAGAs based faithfulness shows limited reliability in this setup. Performance on fixed versus document specific questions varied substantially, likely related to the formatting of documents and preprocessing. Under the tested configuration, cluster-based chunking did not outperform simpler strategies.

Software-intensive organizations need effective ways to identify where maintenance and refactoring efforts will yield the greatest practical benefit. Although software analytics such as code health, hotspots, and coupling provide valuable signals, they do not always capture the experienced burden of change that slows software evolution in practice. This paper presents a multi-case industrial study of technical debt friction as a prioritization-oriented concept for identifying where technical debt most strongly affects maintenance and evolution. We investigate how practitioners interpret the concept, whether friction-related analysis aligns with perceived maintenance pain points and refactoring needs, and what broader maintenance and evolution insights friction can provide beyond individual refactoring candidates. To this end, we conducted structured walkthrough sessions with practitioners across multiple industrial cases using analysis artifacts including code health, hotspots, coupling, refactoring targets, and socio-technical views. Our findings show that practitioners generally considered technical debt friction useful for reasoning about maintenance burden, especially when interpreted together with complementary technical and socio-technical views. At the file level, friction often aligned with known problematic areas and, in several cases, with files that later received maintenance attention, although its practical relevance depended strongly on context. In addition, our exploratory project-level analysis suggests that friction distributions may reveal broader maintenance and evolution patterns. These results indicate that technical debt friction is promising as a decision-support concept, but most effective when used with contextual knowledge and supporting evidence.

Musical performance involves executing a set of high-level musical instructions, yet recovering those instructions from the performance is a challenging inverse problem. We present Decomposer, a post-training framework for symbolic music decompilation: the task of recovering executable, editable music programs from symbolic music. We instantiate the task as MIDI-to-Strudel decompilation, where the model takes symbolic MIDI as input and produces a program in Strudel, a music programming language, that reconstructs the input when executed. The task poses two challenges: Strudel is a low-resource language with little naturally paired MIDI-code data, and optimizing faithful reconstruction of MIDI alone can collapse to unreadable note-by-note transliteration. We address these challenges in two stages. First, we construct Strudel-Synth, a synthetic corpus of paired Strudel programs and rendered MIDI, and use it for supervised fine-tuning. Second, we refine the model with reinforcement learning on unpaired MIDI, optimizing rewards for both MIDI reconstruction faithfulness and code readability. Our evaluation across synthetic and real-world MIDI benchmarks shows that Decomposer achieves substantially higher MIDI reconstruction faithfulness than closed-source LLMs while producing more readable and diverse code than the heuristic converter.

Domain agents often face noisy business data, uncertain post-training gains, offline/application mismatch, and adapter-release risk. This paper presents CLAP (Closed-Loop Agent Post-training), a closed-loop method that converts business data into structured SFT samples, decision-preference samples, holdout sets, risk diagnostics, and release-gate records. CLAP combines data validation, target/evidence normalization, reward/KL diagnosis, offline gates, and application-chain replay to decide whether an adapter is suitable for the target application chain. On five anonymized manufacturing-scenario batches, QLoRA-style LoRA-SFT yields modest average gains: overall score increases by 0.0098, pass rate by 0.0240, and evidence accuracy by 0.0280, while hallucination and wrong facts decrease. Yet only 3 of 5 batches improve, some batches regress, and GRPO exposes high KL risks. Application-chain replay further shows that RAG is necessary for factual extraction; under the same 3B backbone and 100 replay cases, an application-RAG-oriented LoRA-SFT adapter improves value, core fields, and answer-evidence doc/page matching over base+RAG, but increases latency. These results support managing domain-agent post-training through an integrated data-training-evaluation-release loop rather than relying on training completion or a single offline score.

This paper showcases a memory-efficient training stack for Mixture-of-Experts (MoE) models. It is a training paradigm that combines and specializes various existing and novel parallelism techniques at different layers and stages of the Mixture-of-Experts (MoE) model training pipeline. It leverages these techniques to achieve maximal efficiency given the physical constraints of CPU, CPU memory, GPU HBM memory, and the CPU-GPU, GPU-GPU, and node-node communication bandwidth of the GPU cluster. It also contains a novel strategy for the optimizer step to achieve high throughput and memory efficiency, enabling practitioners to conduct lossless pre-training/fine-tuning of trillion-parameter scale models, at a million context length, with just under 12 8x H200 GPU nodes, with state-of-the-art throughput and memory efficiency. In our experiments, MoP delivers 4.7x--8.2x higher per-GPU throughput than a strongly-tuned FSDP2 baseline (with the gap widening at larger scale) and sustains training at context lengths up to 1M tokens, where the baseline runs out of memory beyond 64--128K.

Software defect prediction (SDP) classifiers produce probabilities used for inspection prioritization, threshold tuning, and risk communication. Probability-based uncertainty quantification (UQ) characterizes prediction confidence, but whether common UQ metrics reliably indicate performance and calibration remains unclear. We conducted a large-scale empirical study of probability-based UQ for SDP. We evaluated five UQ metrics, six performance metrics, and three calibration metrics for 16 representative classifiers. We analyzed these relationships under two prediction settings: within-project defect prediction (WPDP), using 36 benchmark datasets, and cross-project defect prediction (CPDP), using 32 feature-compatible datasets. Results showed that UQ was highly context-dependent. Under WPDP, UQ correlated more consistently with false positive rate and AUC than with MCC, F1 score, and other metrics; these correlations also varied across classifier categories and dataset collections. Performance and calibration were related but not interchangeable; classifiers with strong discrimination could still exhibit large calibration error. Under CPDP, several UQ-performance and UQ-calibration correlations weakened or reversed, indicating that uncertainty signals do not reliably transfer across projects. Thus, UQ should be evaluated against specific performance objectives. Calibration should be assessed independently using multiple metrics. Transferred probabilities should be revalidated before guiding quality-assurance decisions.

Fault trees are a widely used as effective risk models for complex systems, answering the question "what can go wrong?", especially through minimal cut set analysis. We study fault trees from the perspective of Halpern & Pearl's theory of actual causality. This allows us to use fault trees to answer the question "why has it gone wrong?", which is fundamental to failure diagnostics. We give a complete classification of each of the different notions of actual causality in terms of the fault tree's graph structure and logical structure, and show how minimal cut sets give rise to actual causes.

Counterfactual explanations (CEs) for multivariate time-series classifiers are often difficult to interpret in domains where experts reason in terms of semantic feature groups rather than individual channels. In rehabilitation movement analysis with multi-sensor inertial measurement units (IMUs), clinicians interpret motion through muscle-group and joint-segment abstractions; yet, most existing counterfactual methods operate at the channel level, producing scattered and biomechanically incoherent explanations. We propose a two-stage framework for group-based counterfactual generation in high-dimensional IMU data. We first show that Shapley-Adaptive (SA) group ranking preserves counterfactual validity but fails to enforce group-level sparsity, motivating the need for explicit group selection. We then introduce Learnable Gate (LG) methods, which incorporate trainable per-group relevance gates jointly optimized with perturbation masks. Experiments on the KneE-PAD rehabilitation dataset demonstrate that LG substantially improves modality-group sparsity compared to the channel-level M-CELS baseline while maintaining or improving validity, temporal smoothness, and generation efficiency. Exercise-specific analyses further show that group-structured counterfactuals yield concise, muscle-level corrective guidance aligned with clinical reasoning. Overall, the proposed framework enhances interpretability without sacrificing counterfactual quality, enabling more actionable explanations for rehabilitation movement analysis.

The global development of Library and Information Science (LIS) is influenced by various factors such as the economy, society, culture, discipline, tradition, and more. Consequently, the research methods of LIS vary greatly among countries. To better understand these differences, we conducted a study of 5,281 research papers from 81 countries published in internationally representative journals over the past thirty years. We manually annotated the research methods used in some articles through content analysis, and subsequently developed and trained a deep learning model for automatic classification of research methods. Using this method, we conducted a comparative analysis of the usage of research methods in different countries. Our findings reveal that there are differences in the research methods used across countries, with each country having its unique research profile and distribution of research methods. Even when investigating the same topic, research methods can differ between countries. Our study also uncovers that there are differences between the national and international distribution of research methods, these differences have decreased over the past 30 years. By highlighting the characteristics of discipline development in various countries from the perspective of research methods, our study can help guide discipline development at the national level. This study provides insights into the usage trends of research methods across different countries and highlights the unique characteristics of discipline development in each country. This information can be valuable in promoting collaboration and understanding between countries and in guiding discipline development at the national level.

Long-context inference is increasingly common in large language model (LLM) serving, driven by retrieval-augmented generation and agentic systems. In disaggregated inference, these workloads require transferring large Key-Value (KV) caches across the network, where decoding cannot begin until the transfer completes. Recent KV quantization techniques reduce data volume and alleviate this bottleneck, but existing schemes fail to achieve both low network-exposed latency and high inference accuracy. We challenge the assumption that the KV cache is an indivisible unit that must be fully received before use. We leverage the observation that different bits in the KV cache contribute unequally to attention computation and inference precision: the most significant bits capture the coarse structure of attention and the least significant bits refine precision. This property enables partial use of the KV cache during decoding. We present Lynx, a system that enables progressive, split-stream KV transfer by partitioning the KV cache into a high-priority Anchor stream carrying the most significant bits and a low-priority Residual stream carrying remaining precision. Decoding begins upon receipt of the Anchor stream and proceeds speculatively while the Residual stream is transferred concurrently, followed by verification that ensures equivalence to higher-precision decoding. Across multiple models and serving workloads, Lynx achieves Time-to-First-Token (TTFT) comparable to aggressive 4-bit KV quantization, while matching the accuracy of high-precision (BF16) inference, improving TTFT over standard 8-bit KV quantization by up to $1.43\times$ and improving accuracy over state-of-the-art by up to $5.1\%$.

Reliable reward and preference signals are critical for evaluating and optimizing large language models on open-ended tasks. Rubric-based judges offer a transparent way to decompose such judgments into explicit evaluation criteria, but existing annotation-free rubric generators typically rely on a single generic evaluator. As a result, they may overlook important dimensions of human preference, a failure mode we term dimensional blind spots. To address this limitation, we propose Multi-Role Rubric Generation (MRRG), a training-free and reference-free framework that elicits evaluation criteria from multiple complementary roles and consolidates them into an auditable rubric-based scorer. This scorer can be used both to validate pairwise preferences and to provide rewards for GRPO-style Reinforcement Learning with Verifiable Rewards (RLVR). Experiments on preference validation benchmarks show that MRRG consistently outperforms single-role rubric generation baselines across multiple backbone models. Further RLVR experiments demonstrate that MRRG yields a stronger reward signal for improving open-ended generation.

Large language models (LLMs) are increasingly proposed for aviation business operations, from documentation and training generation to customer facing assistants. General purpose benchmarks do not measure whether a model reasons safely and correctly about aviation specific operational knowledge, and the high stakes, regulated nature of the domain makes that gap consequential. We present Pre-Flight, an open source benchmark of 300 multiple choice questions drawn from international standards and airport ground operations material, covering international airport ground operations, ICAO and US FAA regulations, aviation general knowledge and complex operational scenarios. Questions were authored and reviewed by practitioners with experience in air traffic management, ground operations and commercial flying. We evaluate a range of contemporary commercial and open weight models using the Inspect evaluation framework, scoring by accuracy under a standard multiple choice protocol, and we maintain the leaderboard on a rolling basis as new models are released. Against an informal expert reference of around 95%, obtained from a low sample quiz of aviation professionals at a conference, even the strongest model evaluated (released in 2026) reaches 82.7%, having improved only gradually from roughly 75% in early 2025. A substantial and persistent gap below expert level reliability therefore remains. We release the dataset, the evaluation harness and the results, and the benchmark is available within the community evaluations package distributed with inspect_evals. We argue that domain specific evaluation of this kind is a necessary precondition for responsible deployment of generative AI in non safety critical aviation operations.

Research in the social sciences has shown that there are gender differences in the selection of research methods, with women often opting for qualitative methods while men prefer quantitative methods. However, it is important to consider that research methods are generally chosen based on the research topic. To figure out the influence of gender on research method selection, a study was conducted in the field of Library and Information Science, using a more fine-grained method classification system and an automatic classification model called CogFT, which is based on full-text cognition. The findings showed that women tend to use Interview while men prefer Theoretical approach, across a range of topics. The study offers insights into the specific research design processes that contribute to gender differences in method selection and suggests ways to promoting gender inclusivity and equality in academia by considering research method use and guidance.

Crowdsourced fact-checking systems have been adopted by major social media companies such as X, Meta, TikTok and Google with the aim of combating misleading information at scale without relying on centralized editorial control. These systems have been developed around a common underlying concept: a bridging mechanism that identifies notes flagging misleading information when they receive support from people with different perspectives rather than simple majority support. To our knowledge the only publicly disclosed bridging algorithms deployed for fact-checking are based on matrix factorization, as deployed by both X and Meta, augmented with additional components addressing abuse, targeted manipulation, and contributor brigades. This work examines the core matrix factorization portion of these systems, presenting theoretical and empirical evaluations of the degree to which coordinated users could vote strategically by leveraging the latent representations to fabricate the appearance of synthetic consensus within the bridging mechanism. Using historic production data, we find that up to 10.7% of lower quality notes could be manipulated above consensus thresholds using less than 10 ratings. We complement these findings with a theoretical analysis, revealing counterintuitively that rating a note as "Not Helpful" can increase its helpfulness score, as well as a cost model quantifying manipulation effort. We have developed and deployed mitigations within X's Community Notes algorithm to address synthetic consensus.

Packet loss concealment (PLC) reconstructs audio packets that are missing at the receiver, usually with a trained model whose parameters remain fixed at deployment time. This treats the PLC model as static, even though each call or recording exposes signal-specific information through the packets that did arrive. We present TTT-PLC, a self-supervised test-time tuning framework that adapts existing PLC models using only those received packets. The method creates supervision by synthetically masking portions of the available signal, training the model to conceal them with its native PLC objective, and then using the adapted model to reconstruct the true packet losses. No clean reference signal, external adaptation data, or architectural modification is required. We study TTT-PLC in two deployment settings. In the non-causal setting, the received file is available before reconstruction, allowing repeated self-supervised adaptation passes and providing a per-file adaptation ceiling. In the causal setting, audio is streamed without revising emitted samples; adaptation is performed only on completed past blocks, and updated parameters affect only future audio. We instantiate the framework on two public PLC backbones, FRN, a recurrent full-band speech PLC model, and PARCnet, a hybrid autoregressive-neural model for networked music. Across these settings, the results show that pretrained PLC systems do not need to be treated as fixed at inference time, the still-observed portions of a lossy signal can provide an effective training signal for improving concealment on that same signal.

We present a multi-agent system for studying the allocation of discrete, congested resources among heterogeneous strategic agents, motivated by the problem of railway slot allocation under deregulation. Multiple operator-agents, differing in size and capacity, interact through a shared auction mechanism over repeated rounds under time-constrained decision-making. The mechanism combines a congestion-based base price that increases with aggregate demand with an asymmetric corrective adjustment that penalises the agent requesting the most slots and rewards the agent requesting the fewest, and is designed to mitigate strategic dominance by large agents while preserving transparency and congestion sensitivity. We formulate the interaction as a repeated game with incomplete information and implement the system as a real-time, web-based multi-agent environment in which human participants control individual agents and observe live marginal-cost and competitor feedback. We report exploratory observations from two structured sessions with domain experts acting as operator-agents. The congestion mechanism responds to aggregate demand as designed and the corrective incentives are actively triggered, but agents representing large operators persist with high-request strategies despite the penalty, suggesting that corrective pricing is necessary but not sufficient to neutralise strategic dominance in this multi-agent setting. A post-session debrief indicates that participants' decisions were driven by the assumed agent role rather than personal disposition, and provides qualitative support for strategic motives, such as preserving market presence and raising rivals' costs, operating alongside short-term profit maximisation. We discuss implications for multi-agent mechanism design under asymmetric budgets and outline directions for analytical validation and larger-scale multi-agent experiments.

The Koopman operator has gained considerable attention due to its ability to provide a global linear representation of highly complex dynamical systems. The operator describes nonlinear dynamics in a linear way through the lens of real- or complex-valued observable functions. Recently proposed data-driven techniques, like extended dynamic mode decomposition (EDMD), its kernelized variant, and machine-learning methods, can be used to generate finite-dimensional approximations accompanied by finite-data error bounds. In this tutorial paper, we provide a concise introduction into Koopman operator theory and its use in systems and control. A particular focus is put on data-driven surrogate models, their extension to systems with inputs, and controller design using Koopman operator theory. Moreover, we demonstrate the key techniques, i.e., EDMD and Koopman MPC. To this end, we provide simulation studies including source code on GitHub to enable the interested reader to experience the Koopman operator in systems and control step by step.

All-to-all based sequence parallelism methods execute communication and computation strictly in serial when processing medium-long sequences, resulting in hardware resource underutilization. This paper proposes Head-Chunked Multi-Stream Pipeline (HCMS), which exploits the computational independence of multi-head attention by partitioning attention heads into multiple chunks and achieving fine-grained communication-computation overlap through dual CUDA streams. HCMS is orthogonally compatible with existing optimizations such as FlashAttention and SDPA, requires no modification to underlying kernels, supports uneven partitioning while maintaining numerical equivalence. Experiments validate the effectiveness across four GPU platforms at 2-8 GPU scales: for typical video generation sequence lengths of 31K-56K tokens, HCMS achieves 10\%-17.5\% speedup over the Ulysses baseline and 5\%-14.5\% speedup over Ring Attention; end-to-end acceleration of 6.8\% is achieved on the Wan2.2 model. Theoretical analysis shows that HCMS benefits are positively correlated with communication ratio $ρ$, and its use is recommended when $ρ>20\%$.

Traditional Chinese Medicine (TCM) diagnosis, particularly through tongue inspection, faces persistent challenges in subjectivity and reproducibility. The application of multimodal artificial intelligence to TCM clinical tasks, such as syndrome differentiation and prescription generation, is significantly hampered by the semantic gap between visual tongue features and textual reasoning, as well as the lack of large-scale, standardized datasets. To address these challenges, we introduce MMIR-TCM, a novel framework that emulates the diagnostic process of TCM experts by integrating multimodal large language model(MLLM) with memory-augmented segmentation and retrieval-augmented generation (RAG). Employing a three-stage architecture, MMIR-TCM integrates a training-free Memory-SAM module for robust tongue extraction, a fine-tuned Qwen3-VL model for structured tongue diagnosis generation, and a Qwen3-based RAG component for evidence-grounded clinical decision support generation. The framework was developed and validated using MedTCM, a new large-scale multimodal dataset that we introduce specifically for advanced TCM research. To properly evaluate our framework's clinical accuracy, which existing metrics fail to capture, we also developed TDEU, a domain-specific evaluation metric incorporating semantic understanding and diagnostic importance. Our comprehensive experiments demonstrate that MMIR-TCM significantly outperforms leading models, including GPT-4o and Gemini 2.5 Flash.

Evaluation benchmarks are essential for assessing vision-language models (VLMs), but most multimodal benchmarks are static, making them vulnerable to temporal staleness, data contamination, and costly maintenance. We present MMBench-Live, a continuously evolving multimodal benchmark built by a multi-agent-driven automated pipeline. Our framework treats benchmark evolution as task-guided dataset construction, integrating structured benchmark specification, feedback-controlled real-time data acquisition, and verifiable QA generation with executable reasoning. To maintain cross-version comparability, we introduce a distribution-consistent update strategy that extracts task-related visual patterns from the original benchmark to guide data collection and filtering. Instantiated from MMBench, MMBench-Live contains 5.9K newly generated evaluation instances with a high answer correctness rate, while each update costs about USD 30 and takes 1-2 hours. Extensive evaluations show that MMBench-Live preserves stable model rankings, maintains semantic alignment with the original benchmark, and exhibits weaker contamination-related memorization signals, suggesting a practical and scalable paradigm for sustainable multimodal benchmark evolution. The project is available at https://github.com/PRIS-CV/MMBench-Live.

Open-source projects depend on a steady inflow of newcomers. A growing concern is that AI coding agents (tools such as Cursor and Claude Code that write code from natural-language instructions) will crowd them out, by absorbing the simple tasks that beginners start with and by making code harder to read. We give this concern a causal answer. Using GitHub code search we identify 1,888 projects that adopted an agent, signaled by their first commit of a configuration file. We apply difference-in-differences against matched non-adopting controls, restricting the main analysis to the 603 adopters with a genuine pre-adoption period. We find no evidence of crowding-out: across estimators newcomer inflow shows no significant decline after adoption (point estimates run from a small increase to, under the most conservative trend specification, a slight and insignificant dip), onboarding and retention are unchanged, and a sparse, correlational beginner-task measure (good-first-issue labels, which we cannot test for parallel trends) shows no decline. The feared mechanism is real but decoupled: adoption raises per-function code complexity (about +11% on a cognitive metric for Python, a quarter of the prior estimate, and +3 to 4% in cyclomatic terms across all languages), yet in fixed-unit subsets where complexity rose (Python on the cognitive metric, and all languages on the cyclomatic metric), newcomer participation does not decline. These results suggest that, in established open-source projects, adopting an AI coding agent makes code modestly more complex but does not crowd out the human newcomers that a project depends on: the feared trade-off between AI assistance and human participation does not materialize.

Modern compilers are frequently updated, but expert review capacity is highly limited, leading to delayed integration and, in some cases, subtle semantic bugs entering the compiler codebase. Automating the code review process with modern general code review agents may be feasible, but it faces critical challenges due to compiler complexity. In this paper, we use LLVM as our target compiler and present Archer, the first automated agentic code review tool for compiler optimizations. Archer constrains the agentic review process from both ends by using obligations to guide analysis and a deterministic validation guard to admit only findings backed by executable evidence. We evaluated Archer on 70 open PRs and 328 closed PRs in LLVM from the last two months. The review results are shocking and concerning: Archer discovers that 21% of open PRs and 11% of closed PRs are buggy, i.e, introducing semantic bugs such as miscompilations in LLVM. Our findings expose a critical gap in the capacity for critical review in large compiler projects and demonstrate the practical value of Archer as an additional reviewer.

Steering vectors have emerged as a promising approach to controlled text generation, offering interpretable, training-free mechanisms for shaping model outputs. However, their practical generality remains poorly understood. We study the limits of steering vector generalization along three dimensions: trait expressibility, task transfer, and multi-trait composition. Using the PLUME writing personalization benchmark, we extract steering vectors for a range of preferences and evaluate them on summarization and email-writing tasks across two open-source models (Qwen2.5-7B-Instruct and Llama3.1-8B-Instruct). We find that steering effectiveness varies substantially across traits. We further show that steering effectiveness can degrade when vectors extracted from positive and negative style examples are transferred to downstream writing personalization tasks. Finally, we compare common methods for composing multiple steering vectors and find that all methods suffer significant drops in trait expression as more vectors are added, with a tradeoff between coherence and expressibility that requires per-setting hyperparameter tuning. Taken together, our results suggest that steering vectors face meaningful limits as a general-purpose tool for preference alignment.

Large Language Models (LLMs) have recently shown promise in molecular discovery, yet a gap remains between their probabilistic nature over discrete sequential tokens and the rigid topological constraints of chemical space. This raises the question of whether molecular LLMs can generalize beyond the local neighborhoods induced by their sequence-based representations. To systematically investigate this question, we introduce a Molecular Perturbation framework that generates syntax-valid structural variants of training molecules under controlled Graph Edit Distance (GED) to probe the manifold regularity of molecular LLMs. Our analysis shows that even a single edit can cause substantial performance drops on common molecular tasks, revealing a narrow local trust region and fragile sensitivity to structural changes. Since similar molecules tend to exhibit similar properties, In-Context Tuning (ICT), which anchors predictions on structurally similar molecules, offers a natural way to mitigate such fragility. Our experiments also examine whether ICT confers robustness under controlled structural perturbations, and the results suggest that it can partially expand the local trust region and offer a promising direction for stabilizing molecular LLMs against structural variation.

Sparse autoencoders (SAEs) decompose internal activations of neural networks into sparse linear combinations of learned features by fitting an overcomplete dictionary $\mathbf{W}\in\mathbb{R}^{m\times n}$ with $m<n$, and inferring a sparse code $\mathbf{x}\in\mathbb{R}^n$ from $\mathbf{h}\approx\mathbf{W}\mathbf{x}$. This inference problem closely resembles the canonical setup of compressed sensing, but dense decoders requires $O(mn)$ learned values, which becomes costly at large feature counts. We introduce Expander SAEs: TopK SAEs whose decoder and tied encoder are supported on a left-$d$-regular expander mask with $d\ll m$, learning only $dn$ decoder values while keeping the sparse-coding problem $(m,n,k)$ fixed. The same structure reduces storage and turns the matching-pursuit correlation step $\mathbf{W}^\top \mathbf{r}$ in OMP into an $O(dn)$ gather-and-reduce operation. Our experiments show that across Pythia-70M/160M, Qwen2.5-3B, and Llama-3.2-1B residual-stream activations, varying $d$ traces a consistent storage--fidelity frontier, and that at the most compressed modern-LM setting, Qwen2.5-3B with $d=7$ uses $293\times$ fewer learned decoder values than the full dense decoder while retaining $84$% of dense CE-loss recovered. Control experiments show that the improved storage--fidelity tradeoff is driven by sparse, diverse decoder support structure rather than by fewer learned decoder values, and that when sparse and dense decoders are compared at matched parameter count, part of the remaining gap comes from encoder amortisation. On the theoretical side, we show that expansion and column flatness are sufficient for identifiability of noiseless $k$-sparse codes, and we derive complementary sufficient conditions under which OMP recovers the support exactly.

Vision Transformers have reshaped computer vision by using self-attention to capture global context across image regions. This makes them attractive for edge visual inspection and monitoring in applications such as renewable-energy infrastructure, industrial quality control, medical imaging, and autonomous-system sensing. However, deploying ViTs on small FPGAs remains challenging because the softmax stage in self-attention requires exponential evaluation and normalization, which are costly in hardware. Existing implementations often rely on CORDIC pipelines or BRAM-based look-up tables, increasing area and power consumption. This paper presents a BRAM-free approximate attention-weighting unit for FPGA-based ViT inference. The proposed design approximates the natural exponential in softmax using a 16-segment piecewise-linear function implemented entirely with distributed LUT fabric. Unlike base-2 approximations, the natural-exponential formulation preserves the pre-trained attention temperature and avoids model-specific recalibration. Implemented on a Xilinx Zynq-7020, the complete attention-row core uses 1444 LUTs, 77 DSPs, and no BRAM, while hardware-accurate emulation shows accuracy within a \(0.20\%\) absolute top-1 difference from the exact-softmax reference on ViT-family models. These results demonstrate the potential of the proposed core for energy-efficient ViT inference on resource-constrained edge-AI platforms.

Monitoring cognitive load during online learning could help instructors identify content that learners find difficult, but remote settings remove the visual cues that support this judgement in a classroom. We study whether a single-channel, consumer-grade EEG device (the NeuroSky MindWave Mobile 2) can distinguish easy from difficult educational-video content, using the publicly available dataset of Wang et al. [24] (ten learners, one excluded for excessive noise, leaving nine). We implement a hybrid CNN+LSTM+Attention model that combines the raw waveform with band-power features. In a within-subject setting, the model reaches up to 78.5% accuracy, compared with 55% for conventional feature-based classifiers; regularization (dropout and L2) closes the large gap between training and validation accuracy that we observe without it, keeping validation accuracy stable at roughly 68-73%. We are deliberately cautious about these numbers: with only nine subjects, within-subject evaluation is optimistic, and we argue that subject-independent evaluation -- in which no learner appears in both training and test data -- should be the standard for this task. To that end we release a reproducible evaluation pipeline. We frame the work as a feasibility study rather than a deployable system, and pair it with an open, notebook-based tool that records EEG, runs inference, and visualizes estimated cognitive load as a heatmap over the video timeline to help educators locate potentially challenging segments.

With the rapid development of autonomous aerial systems, Unmanned Aerial Vehicles (UAVs) are increasingly deployed in applications such as inspection, environmental monitoring, and rescue, creating growing demand for reliable autonomous navigation. However, autonomous UAV navigation in dense environments remains challenging under sparse perception and dynamic constraints. Most reinforcement learning (RL) methods lack explicit safety mechanisms, leading to unsafe exploration, unstable training, and risky behaviors, especially during high-speed flight. Even in safe RL approaches, safety is often enforced by projecting policy outputs onto a safe action set, which may introduce instability. Meanwhile, many learning-based methods rely on dense inputs or large networks, increasing computational burden and limiting lightweight onboard deployment. Facing the above challenges, we propose a safety-constrained perception-control integrated framework for UAV navigation. A lightweight network encodes sparse observations into collision-risk-aware features using asymmetric and depthwise separable convolutions. We formulate the task as a constrained Markov decision process within a hierarchical control architecture and solve it using a Lagrangian-based safe PPO algorithm. Curriculum learning further improves training stability. Experiments with varying obstacle densities and flight speeds demonstrate higher success rates, improved safety, and better efficiency than existing reinforcement learning baselines.

LLM agents increasingly perform autonomous actions through external tools, leading to complex and evolving safety risks. However, existing safety testing targets expert-designed safety violations, and the corresponding outcomes are evaluated by hard-coded rules, making them costly to extend as agents evolve. To this end, we present Vera, an end-to-end automated safety testing framework that instantiates software engineering testing principles for non-deterministic agents through a three-stage, self-reinforcing pipeline. First, a literature-driven exploration continuously discovers and structures emerging risks into taxonomies of safety risks, attack methods, and tool execution environments. Second, combinatorial composition across taxonomy dimensions produces executable safety cases, each specifying a concrete safety goal, a programmatically constructed initial state, and a deterministic verification predicate grounded in observable artifacts. Third, adaptive execution runs heterogeneous agents in isolated sandboxes where a control agent steers multi-turn interaction based on runtime observations, while evidence-grounded verifiers judge outcomes from environment state and tool-call evidence rather than model self-report. We evaluate Vera on four production agent frameworks (OpenClaw, Hermes, Codex, Claude Code), revealing substantial safety weaknesses, with average attack success rates reaching 93.9\% under multi-channel attacks; we also release Vera-Bench, comprising 1600 executable safety cases spanning 124 risk categories across three execution settings. These results indicate that modular, executable testing infrastructure is essential for rigorous and maintainable safety evaluation of rapidly evolving agentic systems at scale. The code is publicly available at https://github.com/Yunhao-Feng/Vera.

While decoder-only LLMs excel at a vast array of natural language tasks, it suffers from an asymmetric information flow induced by causal attention: later tokens are richer in contextual grounding than earlier ones. A simple and effective remedy is prompt repetition -- just appending a second copy of prompt before generation can redistribute grounding across positions and improve reasoning performance. However, full repetition of the original prompt doubles the KV cache footprint and quadruples attention cost during prefill, making it impractical for long-context settings. We propose PartRep, a selective augmentation method that appends only the most informative tokens -- rather than the entire prompt. We use token-wise negative log-likelihood (NLL) as a selection signal, motivated by the hypothesis that less predictable tokens are less recoverable from surrounding context and therefore benefit more from late-position repetition. To avoid the heavy cost of a full forward pass for scoring, we train a lightweight gate that predicts high-NLL tokens from early-layer hidden states, enabling token selection during mid-prefill via early exit. Across eight benchmarks (including MMLU, GSM8K, and RULER) and three model families (Qwen2.5, Llama3.2, Gemma4), PartRep retains most of the gains of full repetition while using only 59.4\% of its KV cache and 79.0\% of its prefill FLOPs.

Mixture-of-Experts (MoE) models scale efficiently but remain costly to adapt due to redundant experts and uniform parameter allocation. Existing parameter-efficient fine-tuning (PEFT) methods such as LoRA ignore MoE routing dynamics, leading to suboptimal resource use. We propose EPnG, an adaptive prune-and-grow framework that reallocates LoRA capacity based on expert importance derived from router gate probabilities. EPnG prunes under-utilized experts and expands high-importance experts via rank growth with orthogonal initialization, while maintaining a fixed parameter budget. Across OLMoE and Qwen1.5-MoE, EPnG consistently outperforms LoRA under the same budget and achieves performance comparable to full fine-tuning while updating only 0.55%-0.72% of parameters (up to 140x-180x fewer). These results demonstrate that aligning PEFT with MoE routing yields a more effective and scalable fine-tuning strategy.

Hyper-scale microservice systems have become the standard infrastructure for large-scale Internet companies. These systems consist of numerous loosely coupled microservices that evolve independently through continuous development and deployment. Such complexity makes failures unavoidable, necessitating efficient Root Cause Analysis (RCA) to help Site Reliability Engineers (SREs) quickly localize root cause services and classify failure types. However, existing RCA methods often struggle to adapt to the extreme dynamism and massive scale of these systems. In this paper, we present KRCA, an end-to-end RCA system designed for hyper-scale microservice systems. To manage the vast search space, KRCA employs a multi-stage pipeline that begins with an API-level drilldown to isolate suspicious services. It then instantiates a skeleton-based causal graph from anomalous metrics to serve as a high-recall structural prior, before utilizing a memory-augmented multi-agent framework to verify causality and generate the final failure report. By combining structured causal constraints with multi-agent reasoning, KRCA employs balances diagnostic accuracy with the efficiency requirements of real-time production use. Experimental results show that KRCA achieves AC@1 scores of 0.88 and 0.79 for root cause service localization and failure type classification, outperforming the strongest baseline by at lease 31% in absolute gains. KRCA has been deployed in Kuaishou's production environment for over six months, reducing the average diagnosis time by 77.3%.

Next activity prediction helps service-oriented processes anticipate upcoming steps before delays, exceptions, or service-level risks occur. Most existing methods assume classical single-case event logs, whereas real service processes often involve events shared by multiple typed business objects. Object-centric event logs (OCELs) capture such interactions, but current predictors remain limited. Flattening-based approaches lose cross-object context, and native OCEL graph-based approaches encode multi-object events through pairwise relations. Existing models also do not jointly capture event-driven object state changes, inter-event timing, and global execution patterns. We propose EHHN, an Event-driven Heterogeneous Hypergraph Network for object-centric next activity prediction. EHHN represents each prediction prefix as a heterogeneous hypergraph, where event--object hyperedges bind retained co-participating objects and a lifecycle hyperedge groups the primary object's observed lifecycle events. Based on this representation, EHHN uses a dual-stream architecture in which a micro-spatial stream models event-driven object-state evolution and a macro-evolution stream captures temporal dynamics using retrieved global prototypes. The two streams are fused to predict the next activity. Experiments on four public OCEL benchmarks against nine baselines show that EHHN achieves the best accuracy and macro F1-score on all datasets, with improvements of up to 8.1 and 12.4 percentage points over the strongest baselines. Compared with the strongest OCEL-native graph baseline, EHHN also reduces peak GPU memory by up to 24 times. Code is available at https://github.com/chenkaitao1112/EHHN.

Radio frequency (RF) maps provide a compact representation of multipath propagation characteristics and are fundamental to channel modeling, coverage analysis, and environment-aware wireless optimization. This paper proposes a unified RF map construction framework based on a physics-informed neural network (PINN) and a graph neural network (GNN), supporting both cross-scene generation and in-scene completion with 2D and 2.5D environmental representations. The PINN embeds electromagnetic propagation constraints to establish a physically consistent mapping from receiver locations to multipath parameters, including path gain, time of arrival, and angles, while the GNN enforces spatial consistency by modeling correlations among neighboring receivers. To comprehensively evaluate multipath reconstruction quality, we propose a peak-weighted dynamic time warping metric that jointly accounts for amplitude errors and peak delay misalignment in channel impulse responses. Extensive experiments demonstrate that the proposed method consistently outperforms image-based, diffusion-based, and interpolation baselines across both map-level and multipath-level metrics, achieving robust generalization and high-fidelity RF map construction under sparse observations.

In the age of AI, what will be good knowledge? This article, which is accepted and forthcoming in a special issue of Modern Fiction Studies on "Cultural AI" in 2027, applies digital humanities methods to map epistemic virtues (like "true," "accurate," "creative") used in a corpus of 553 journal articles on AI published in 2024. "Creativity" comes in for special attention as an example. Exploring this discourse of value, the article considers how a framework might be developed for evaluating the knowledge-worth of AI -- one less locked into values formed around pre-AI "knowledge work" agents or structures, and more open to the future values of "generativity." The essay is supported by an online digital kit for exploring data models of the corpus of articles on AI it studies.

Discrete diffusion models have steadily improved in quality relative to autoregressive (AR) models. However, these models are normally constrained to fixed-length generation and do not support key-value (KV) caching. Block diffusion partially bridges diffusion and AR by generating token blocks left-to-right, but its fixed-size sequential blocks limit decoding flexibility and parallelism. Here, we present a new class of language models, set diffusion, comprised of (i) a likelihood parameterization that factorizes over flexible-position, flexible-length token sets and (ii) a set-causal diffusion architecture that supports KV cache updates after every inference step. By factorizing over token sets instead of fixed-size blocks, tokens can be decoded in arbitrarily-ordered sets, including sliding-window sets, enabling faster inference and support for any-order decoding. Set diffusion achieves better speed-quality tradeoffs on mathematical reasoning, summarization, and unconditional generation compared to prior diffusion language models while offering stronger infilling performance than block diffusion. We provide the code, along with the model weights and blog post on the project page: https://m-arriola.com/setdlms/

Diffusion Language Models (DLMs) have recently emerged as a promising alternative to autoregressive models. Unlike standard diffusion-based approaches, DLMs are not explicitly conditioned on a timestep, raising a natural question: do these models internally represent denoising progress, and how is such information used downstream? In this work, we show that DLMs do in fact encode a latent representation related to the diffusion timestep within their residual streams. We find that this signal can be reliably extracted using probes across layers, indicating that denoising progress is decodable from internal activations. We further demonstrate that steering the model along a low-dimensional subspace associated with the inferred timestep allows us to systematically modulate its notion of denoising progress, leading to predictable changes in model confidence and entropy. Finally, we analyse the geometry of the identified representation, showing that it exhibits structured and interpretable properties in activation space, and shedding light on how such a signal is processed by these models.

Ontology construction requires deciding which objects, attributes, and structural relations should be accepted as valid knowledge. Language models can propose such structures from text, but their outputs can still be unsupported or inconsistent. This paper proposes a retrieval-augmented small language model (SLM) framework that uses formal concept analysis (FCA) as a symbolic verification loop for knowledge expansion. Starting from seed attributes, FCA proposes implications over a growing formal context. A retrieval-grounded SLM oracle then validates each implication or returns a counterexample. The oracle also supports incidence judgments, consistency checks, and attribute proposals, making accepted implications, counterexamples, contradictions, and corrections inspectable. In a rare ataxia setting constructed from Orphadata resources, retrieval-grounded 10-seed runs obtain relation F1 of 0.29-0.52 and closure-based implication F1 of 0.22-0.30. Larger seed sets increase the number of evaluated implications and often improve implication F1. The lower implication scores reflect a stricter evaluation of derived implications, where one missed or extra relation can affect several implication judgments. Ablations show that incidence judgments in a fixed object-attribute setting can improve closure-based implication scores. However, identifying positive object-attribute pairs remains difficult even when the candidate objects and attributes are fixed.

As agent planning moves from short tool chains toward persistent workflows with thousands or tens of thousands of steps, failures will occur inside large planning graphs rather than in isolated predictions. Replanning the entire graph after every mistake is neither computationally realistic nor desirable: full-graph replay consumes large context budgets, exposes the LLM to many irrelevant symptoms, and can degrade long-context retrieval. This paper studies the missing component in such systems: a world-model corrector that repairs the failed planning graph in place. We compare two families of correctors. The first is the common engineering approach: scan nodes and edges, choose a suspicious local region, and ask an LLM to repair it. We implement strong engineering LLM correctors and find that they can help, especially when given very large contexts. The second family is our approach, WM-SAR (World-Model Subgraph Amplification Repair): instead of scanning for visible symptoms, it works backward from subgraph amplification, identifies the nodes and edges that keep re-amplifying error, and sends only that causal subgraph to the LLM. Across graph simulations and LLM repair experiments, WM-SAR substantially outperforms engineering correctors under realistic token budgets, achieves near-whole-graph stabilization with a compact region, and gives the LLM a cleaner repair target.

LLM agents are increasingly used to translate natural language into 3D scenes in a procedural way, but existing systems focus on static output. Dynamic 4D scenes from text alone, in which liquids flow, particles emit, rigid bodies cascade, and articulated mechanisms move, remain largely unexplored despite their value as editable content and as physics-grounded training data for video generation and embodied AI. Two challenges set the dynamic case apart from static text-to-scene work: an agent must jointly coordinate spatial layout, multiple physics solvers, temporal sequencing, camera, and lighting in a single coherent scene, and verifying motion correctness from rendered video is fundamentally harder than judging a single image. We present SimWorlds: a multi-agent framework that produces dynamic, editable 4D scenes from text, with Blender-specific procedural knowledge, a planner-coder-reviewer workflow driving a fixed ordered sequence of construction stages, a layered scene protocol enforced by a deterministic verifier, and a runtime-state inspection tool suite that catches mechanism failures the rendered image cannot reveal. We also introduce 4DBuildBench, a benchmark for assessing both visual fidelity and physical consistency of the procedural dynamic 3D scenes generated from text prompts. Experiments show that SimWorlds outperforms prior dynamic Blender generation baselines.

Repository-level vulnerability reproduction is a demanding software engineering (SE) task: an agent must inspect a codebase, infer the input grammar that reaches a vulnerable path, construct a proof-of-conceptv(PoC), and verify that the crash disappears on the patched build. Recent LLM agents can often execute these steps when the approach is correct, yet they still fail by choosing the wrong strategy. This paper argues that strategy, rather than the full action trajectory, is the right learning unit for such SE agents: it is compact enough to optimize, concrete enough to guide execution, and stable enough to store and reuse across attempts. We present Mastermind, a dual-loop framework that separates transferable strategy learning from task-specific experience. A trainable planner learns reusable vulnerability-reproduction strategies through SFT and milestone-based GRPO, while an experience loop maintains task-local strategy records that guide subsequent attempts. The planner is trained independently of the executor, allowing strategy learning to improve multiple frozen executors without modifying their action-generation capability. We evaluate Mastermind on CyberGym using 260 training tasks and 200 held-out evaluation tasks. With GPT-5.5 as the frozen executor, Mastermind achieves an 84.5% pass rate, outperforming open-book PoC context (60.0%), Best-of-8 sampling (63.0%), and iterative improvement (77.0%). The same planner also improves GPT-5.4 mini and GLM~5.1 from 45.0% and 58.5% to 60.0% and 71.0%. These results demonstrate that learning high-level strategies is an effective and transferable mechanism for improving repository-scale SE agents.

Continual post-training enables foundation models to acquire new knowledge while preserving existing capabilities. Recent work suggests that on-policy learning can mitigate forgetting, with on-policy self-distillation emerging as a particularly attractive approach. In this work, we revisit this optimistic view through self-distillation policy optimization (SDPO). Our experiments show that SDPO can accelerate in-domain specialization when teacher signals are stable and well aligned, but it struggles to generalize to out-of-distribution scenarios. In continual post-training, SDPO exhibits stronger forgetting and can even collapse, whereas on-policy reinforcement learning methods such as GRPO adapt more conservatively and better preserve prior capabilities. Further analyses reveal that denser self-distillation induces larger drift in both parameter space and response space, and can amplify high-frequency formatting artifacts through a self-reinforcing teacher--student loop. These findings suggest that on-policy data alone is insufficient for continual learning. Dense self-distillation can accelerate specialization when teacher targets are stable and token-level supervision is reliable, but it should not be treated as a default stabilizer for continual post-training. Our code is available at https://github.com/Moenupa/SDPO-CL.

Many representation learning problems involve directed relations, such as lexical entailment, sentence entailment, ontology hierarchy, and citation links. Standard Euclidean, cosine, and Mahalanobis heads are symmetric, while generic neural scorers can model directionality but provide limited geometric structure. This paper proposes a role-aware neural convex divergence head for asymmetric representation learning. The head applies source- and target-role projections before evaluating an input-convex neural Bregman divergence, yielding a nonnegative structured score in the role-projected space. We characterize its projected-space identity, source-role convexity, directional-gap decomposition, and Hessian-based local curvature. Experiments on lexical, sentence, ontology, and directed graph benchmarks compare symmetric distances, unstructured asymmetric scorers, order/hyperbolic baselines, plain ICNN-Bregman heads, and the proposed role-aware variant. Across ten random seeds on the main semantic and ontology benchmarks, role-aware projections consistently improve directional accuracy over plain ICNN-Bregman heads while preserving zero observed negative divergence rate. The results also identify a boundary case: on large fixed-feature citation prediction, specialized symmetric or hyperbolic baselines remain stronger in ranking accuracy. Overall, the proposed head is best understood as a structured and interpretable plug-in distance module for tasks where directional relations matter.

Vulnerability exploits play a crucial role in assessing the downstream impact of Java library vulnerabilities. While some vulnerabilities are accompanied by disclosed exploit references, automatically reproducing such references into runnable exploits remains challenging because they are often incomplete, unstructured, or only describe partial reproduction steps. Recent code agents provide a promising way to automate this process, but our study shows that their generated exploits often appear successful without triggering the actual vulnerable logic, such as replacing vulnerable APIs with self-implemented functions. To address this, we propose Refploit, an LLM-based trajectory recovery framework for facilitating vulnerability reproduction from public exploit references. The key insight is that a failed agent trajectory is not entirely useless. It may have already completed some reproduction subtasks while also revealing misleading directions that should be avoided. Refploit first validates an agent-generated exploit through differential execution. When the exploit is ineffective, Refploit analyzes its reproduction progress, locates the trajectory segments associated with the reproduction progress, and derives constraints to guide focused recovery. We evaluate Refploit on three open-source Java vulnerability datasets, covering 172 exploit references for 143 vulnerabilities. Under DeepSeek-V4-Flash, Refploit successfully reproduces 138 exploits, achieving a reproduction rate of 80.2%. It achieves a 64.3% relative improvement over the initially generated trajectories and outperforms both the SOTA exploit-generation method PoCGen and advanced code agents such as Codex with GPT-5.4. We further adapt Refploit to another code agent and observe consistent improvements, demonstrating its generality.

Open-vocabulary object detection aims to localize and classify objects beyond the fixed set of categories seen dur ing training. Recent open-vocabulary object detection methods improve localization and classification for unseen categories by leveraging a frozen VLM as a detector backbone. However, VLM classification score lacks recognizing position and scale of the object in an image. We observe that pretrained VLMs en able to classify foreground and background regions. According to this observation, we propose a simple inference-time Pro posal Calibration (ProCal) that improves localization quality of the classification score. ProCal computes a proposal prior by combining two scores: localization-aware foreground score and background-aware suppression score. Localization-aware foreground score captures whether a proposal contains an object area. Background-aware suppression score measures the extent to which the proposal resembles background. We analyze that ProCal suppresses false novel activation on background proposals and consistently ranks true novel proposals above background and partial novel proposals. Applied to CLIPSelf ViT-L/14, ProCal improves APr +2.5 on OV-LVIS. The analyses show that proposal-level localization-aware reranking effects to mitigate ranking miscalibration for novel objects.

In this paper, we consider the nonsmooth nonconvex decentralized optimization problem, where inter-agent communication is compressed. We propose a general framework that unifies various decentralized stochastic subgradient-type methods with unbiased compression and contractive compression with error compensation. By relating the consensus-error iterates and the averaged iterates to the trajectories of continuous-time differential inclusions, we establish global convergence for all methods encompassed by our framework when the objective functions are nonsmooth and lack Clarke regularity. Based on our framework, we further develop several compression-based methods, including decentralized stochastic subgradient methods utilizing sign-based regularization and gradient-tracking momentum. Preliminary numerical experiments empirically support our theoretical results and highlight the communication-accuracy trade-off of the newly developed methods.

On-policy exploration is a crucial component for training robust Vision-Language Navigation agents, as it exposes the policy to a broader state distribution. However, such exploration inevitably leads to trajectories that deviate from expert demonstrations, resulting in a semantic mismatch between the executed visual stream and the original language instruction. In this work, we address this challenge by introducing Phi-Nav, a unified on-policy framework that leverages hindsight reasoning to align instructions with the agent's actual exploratory journey. Specifically, Phi-Nav operates through a three-stage dual-supervision cycle: 1) the agent performs oracle-guided on-policy exploration, sampling a trajectory while learning from expert action feedback, 2) a hindsight speaker synthesizes a path-level hindsight instruction grounded in the collected visual observations, and 3) the agent conducts a second imitation pass, treating the synthesized trajectory-instruction pair as an additional expert demonstration. Through this process, Phi-Nav bridges the critical semantic supervision gap inherent in on-policy methods, transforming semantically unlabeled movement into dense training signals. Evaluations on the R2R-CE and RxR-CE benchmarks show that Phi-Nav yields competitive performance while requiring only a fraction of the expert demonstrations used by current baselines. These results underscore the necessity of semantic exploration in VLN, positioning Phi-Nav as an effective solution for training embodied agents with limited data.

Temporal point processes (TPPs) have widespread applications across various domains. Compared to modeling the conditional intensity of a TPP, modeling its cumulative conditional intensity function (CCIF) improves computational efficiency and eliminates numerical approximation errors. However, current CCIF parameterizations uniformly rely on Monotone Neural Networks (MNNs), which we identify as suffering from three structural deadlocks--convexity restrictions, saturation limits, and violations of CCIF modeling requirements--that fundamentally restrict their representational capacity for complex temporal dynamics. To resolve these bottlenecks, this paper proposes a novel framework called Monotone Alternating Splines (MAS). By leveraging distinct interpolation and extrapolation components, MAS provides a flexible and efficient framework for modeling CCIFs. Theoretically, MAS's interpolation provides strong fitting accuracy, while its extrapolation supports robust generalization, reducing the irreducible approximation gaps of MNNs. Extensive experiments show that MAS achieves superior performance on both synthetic and real-world datasets.

Existing medical video benchmarks primarily evaluate whether a model produces the correct answer, but rarely assess whether it answers at the right time. In real clinical settings, AI systems must decide not only what to predict, but also when to answer, defer judgment, or proactively raise alerts. This creates a critical gap between benchmark evaluation and deployment requirements. We present MedStreamBench, a benchmark for time-aware medical video understanding. MedStreamBench integrates 22 medical datasets and 5,419 QA instances across four temporal settings: retrospective, present, future, and proactive. Unlike conventional benchmarks that assume full-video access, MedStreamBench restricts models to temporally bounded evidence windows and supports both single-turn and streaming evaluation. We further introduce a proactive monitoring setting that requires models to determine whether and when clinically relevant alerts should be triggered. Beyond answer correctness, MedStreamBench evaluates temporal behavior through responsiveness and post-evidence stability. Experiments on leading general-purpose and medical vision-language models reveal a substantial gap between offline recognition and temporally grounded decision-making, with performance dropping markedly in streaming and proactive settings. Our benchmark is available at https://huggingface.co/datasets/Venn2024/MedStreamBench.

Recurrent representations are trajectories, but representation geometry is often measured from static snapshots. We develop finite-lag operator geometry for recurrent hidden states from observed source-successor pairs $(X_t,X_{t+Δ})$. The primitive is the conditional transport law $Q_Δ(dy\mid x)$, estimated by a dense Gaussian source-smoothing operator. From this directed finite-lag law we derive a source-centered transport tensor $G_Δ$, which decomposes exactly into conditional spread and coherent displacement, and an antisymmetric coordinate circulation $W_Δ^ρ$, which summarizes directed lagged flow. We prove affine covariance with explicit metric dependence of scalar summaries, dense estimator stability on bounded trajectory clouds, and a finite-lag separation result showing that source-centered transport detects deterministic recurrent motion not recorded by infinitesimal carre-du-champ geometry. A linear-Gaussian closed form calibrates the quantities in terms of the update $A_Δ$, source covariance, and innovation covariance. Controlled experiments validate the decomposition, circulation, covariance, and stability predictions. In performance matched repeat-copy networks, the framework reveals architecture dependent differences in total transport scale and coherent displacement trace, while coherent displacement fraction is metric and resolution dependent.

Smart contract vulnerabilities are predominantly logic bugs whose detection requires structured, step-by-step procedural knowledge of attack patterns and contract semantics. Existing LLM-based methods struggle to generate this knowledge automatically: prompt-based methods rely on manually crafted detection rules, while fine-tuning requires massive labeled datasets that are inherently scarce in this domain. We present EvoVuln, an automated framework that reformulates vulnerability detection as a procedural knowledge evolution problem, synthesizing and refining detection logic using only a minimal number of labeled samples. To achieve this, EvoVuln introduces two key mechanisms. First, a Runtime with an Inversion of Control (IoC) architecture compiles detection rules into Executable Policies. This strictly decouples deterministic control flow from LLM semantic reasoning, ensuring faithful logical adherence and producing dense diagnostic telemetry for precise error localization. Second, a two-phase evolution pipeline refines the rule via abductive semantic debugging without any parameter updates: Cold Start bootstraps and stress-tests an initial rule using auto-synthesized corner cases; Few-Shot Evolving then grounds the policy in real-world semantics using only five vulnerable and five safe examples per vulnerability type. Evaluated across five real-world vulnerability types, EvoVuln achieves a 71% macro-average F1-score, outperforming all baselines. The evolved procedural knowledge is portable across models: it enables a lightweight, low-cost model to surpass a much larger zero-shot model by 19 percentage points, and transfers to other LLMs without retraining, at a one-time evolution cost under $50.

Reinforcement Learning (RL) is a sequential decision-making framework in which an agent learns optimal policies through interaction with an environment by maximizing cumulative rewards. Among RL methods, Bayesian Reinforcement Learning (BRL) addresses common practical challenges related to data scarcity by leveraging prior knowledge about the environment and sequential belief updates. However, most BRL approaches require an explicit likelihood function, which is frequently inaccessible or intractable in real-world settings. We propose Likelihood-Free Iterated Batch Importance Sampling (LF-IBIS), a novel algorithm for BRL that updates the agent's beliefs online as new interactions become available. By combining Approximate Bayesian Computation with Iterated Batch Importance Sampling, LF-IBIS enables full Bayesian inference in settings where the environment dynamics are not described by an explicit or tractable likelihood. The method yields approximate posterior distributions over both environment parameters and optimal policies, providing a quantification of policy uncertainty useful for a Bayesian treatment of the exploration-exploitation trade-off. We test the method on a simulation study in response-adaptive randomization in clinical trials, where closed-form posteriors enable validation. Additional experiments address settings where the posterior has no closed form and illustrate online policy updating based on the posterior distribution of the optimal policy.

Public LLM leaderboards optimise for global average performance and do not capture the specific cognitive demands of financial-services work: a model that leads on MMLU-Pro may underperform on document-grounded compliance reasoning, and a coding leader may handle multi-turn customer interactions poorly. We present a meta-benchmarking framework that organises 452 publicly reported benchmarks into 41 O*NET Generalized Work Activities and aggregates those into 38 BIAN banking business domains spanning sales, operations, risk, and support work. A multiplicative weighting scheme (discrimination x coverage x recency), computed over a rolling model window, rewards benchmarks that still separate the best models, are widely reported, and remain in active use, suppressing saturated legacy tests automatically. These weights scale the K-factor in a pairwise Elo tournament, producing cross-benchmark-comparable work-activity scores without raw score normalisation; business-domain scores are weighted averages of the constituent work-activity Elos. We demonstrate the framework on a point-in-time public snapshot covering 288 models across 25 organisations as of June 2026, and describe the methodology, full taxonomy, design decisions, and limitations with the aim of making the approach reproducible for institutions facing similar selection and governance challenges.

We study how to predict the downstream closed-loop performance of a learned latent world model from validation-time diagnostics alone. Choosing the right checkpoint from a world-model training run is difficult: validation loss and multi-step prediction RMSE keep improving long after closed-loop performance has collapsed. We present a suite of structural validation-time diagnostics drawn from optimal-control theory and apply them to Gymnasium's LunarLander v3, which features shaped rewards. We train an RSSM [5, 4] world model on it and treat per checkpoint CEM-MPC return as the oracle for closed-loop quality. By evaluating 40 metrics against this oracle, we find that the strongest single predictor is the Reward Observability Fraction (ROF), which measures the reward predictor's dependence on the observable subspace. We combine ROF with three structural regularizers into a single-number offline checkpoint selection score, the Composite Reward Observability Fraction (CROF). The CROF-selected world model trains a model-based A2C policy that beats a fairly evaluated model-free A2C baseline by ~24.5 return points while using ~65x fewer real-environment interactions, and the same world model also drives a strong zero-shot CEM-MPC policy. Code and data: https://github.com/nsmoly/LunarLander_RSSM.

Gauss-Seidel is a well-established iterative method for the solution of linear systems, and multicoloring has been widely used to increase parallelism in iterative solution techniques. Implementing multi-color Gauss-Seidel with conventional divide-and-conquer parallelization strategies, however, may be inefficient due to global synchronization requirements and load imbalances. Task-based programming models can mitigate these issues by enabling fine-grained parallelism, removing global barriers and allowing updates of different colors to partially overlap in time. In this work, we implement the red-black Gauss-Seidel method using two task-based programming models and compare them with a classical divide-and-conquer parallel implementation to evaluate the impact of fine-grained parallelism on execution efficiency. The red-black scheme serves as a representative example, as task-based approaches naturally extend to more general multi-color schemes arising from unstructured grids and wider stencils. Using the solve of the 2D Poisson equation as benchmark, our results show that task-based implementations can achieve performance comparable to conventional divide-and-conquer parallelization while providing greater resilience to hardware-level asynchronicity.

We present a case study in reformalization, a variant of autoformalization in which the input proof is not natural language but a formal development in a different proof assistant. Concretely, we report three reformalizations of the Jordan Curve Theorem: from Mizar to Lean, from HOL Light to Lean, and from HOL Light to Agda. We analyse the results and identify pipeline design choices that matter for practical reformalization tasks.

Speech-LLM integration has shown promising results by leveraging extensive textual pretraining, yet its specific benefits for automatic speech recognition (ASR) remain unclear. We observe that as supervised ASR training data increases, the contribution of LLM priors becomes less evident, and simple speech-text joint training under-utilizes textual knowledge. We therefore propose Joint Speech-Text Interleaved Pretraining (JSTIP), an ASR-oriented pretraining strategy that constructs word-level and segment-level interleaved speech-text sequences within aligned pairs for speech-LLM architectures that accept continuous inputs. Experiments on 38k hours of ASR data show consistent entity accuracy improvement compared to ASR-only and joint speech-text training baselines. JSTIP achieves on-par entity recognition performance using domain transcription text compared to synthetic speech-text pairs, simplifying domain adaptation. Benefiting from textual pretraining and domain text data, JSTIP is competitive with open-source ASR and Speech-LLM systems in medical entity recognition. The zero-shot speech question answering behaviors further suggest that interleaving reduces the speech-text modality gap and preserves the LLM generative prior, which is likely the reason for the entity improvements on the ASR task.

This study provides a theoretical expansion of the recent Data Relativistic Uncertainty (DRU) framework by formalizing a physics-to-AI paradigm for image enhancement. By modeling images as probabilistic wave functions rather than deterministic states, the paradigm explicitly integrates wave-particle duality to illustrate the system flow of how DRU leverages the intrinsic physical uncertainty of light, a dimension requiring further theoretical discussion. Consequently, this paradigm provides a rigorous Explainable AI (XAI) approach that enhances the interpretability of how DRU mitigates illumination bias and maintains robustness against data noise.

Deep learning models for speech classification are vulnerable to backdoor attacks, where malicious triggers cause misclassification at inference time. While sample-specific attacks can bypass many defenses, they often rely on poisoned label attack, making them detectable via manual data defense. In this paper, we propose DRL-CLBA, a novel clean label backdoor attack for speech classification that leverages Deep Deterministic Policy Gradient (DDPG) reinforcement learning. We also utilize deep audio steganography to embed sample-specific triggers into source audio, creating feature-space anchors. The proposed reinforcement learning framework effectively optimizes target samples toward trigger-bearing anchor points in the model's deep latent space, enabling label-migration-free poisoning of target samples. Experimental results across three datasets and four different DNNs demonstrate that DRL-CLBA achieves a high attack success rate, effectively bypassing some backdoor defenses. The attack demonstrates strong resistance against fine-tuning, pruning, and spectral signature defenses, exposing critical vulnerabilities in speech-controlled systems.

Visual regression testing (VRT) is a standard quality assurance step in modern software release pipelines. On every change, it re-renders user interface (UI) screenshots, compares each one against an approved baseline image, and routes any detected difference to a human reviewer who decides whether it is an intended update or an unintended regression. A widely used approach, especially in open-source and continuous-integration pipelines, is pixel-level comparison, which is semantically blind and treats rendering noise and genuine defects identically, producing large volumes of false positives that force developers and testers to spend substantial time and effort manually reviewing flagged differences at every release cycle. Industry tools apply machine learning to VRT, but lack public evaluation. More critically, no dataset or benchmark exists to support natural language descriptions of UI changes, a capability that tells testers what changed in words instead of leaving them to interpret a binary flag or a highlighted region. To address the gap, we propose a new task, Web UI Image Change Captioning (WUICC), which sits at the intersection of VRT and image difference captioning (IDC), and release WUICC-bench, its first dataset and benchmark for the task. We evaluate eleven representative IDC methods, together with two zero-shot general-purpose LLMs. We find that: (1) these methods tend to struggle in the Web UI domain due to its layout diversity, dense text, and fine-grained changes, and (2) yet the trained methods already suppress non-meaningful visual noise far more selectively than the pixel-level comparison VRT relies on, providing a solid foundation for future domain-specific research.

Synthetic data can be scaled along two routes: Source Expansion (SE), which enlarges the source by adding seed materials or generators, and Fixed-Source Synthesis (FSS), which holds the source fixed and scales the generation budget. Existing scaling studies typically expand the source as the data grows, conflating SE with FSS and leaving FSS underexplored. We isolate FSS by holding the seed-question pool and teacher model fixed, varying only the per-question response budget under Rejection Sampling (RS). We adapt the rectified scaling law to FSS, deriving it from how repeated sampling covers a fixed source. Empirically, the derived form, fit on low budgets, predicts performance at the held-out highest budget for every evaluated teacher--student pair. At matched total-sample budgets, SE and FSS are comparable at small budgets; at large budgets, adding seed questions outperforms spending the same budget on more responses. Within FSS, however, neither synthesizing additional questions from the existing seeds nor varying the synthesis protocol outperforms plain RS at matched budgets. FSS is thus a bounded scaling axis and a controlled setting for comparing synthesis protocols. We will release our code and data to facilitate further research.

Existing robust preference optimization for language-model alignment mainly studies pairwise supervision and places robustness at the dataset, prompt, or preference-pair level. We instead study listwise preference optimization under ranking-label uncertainty: given a prompt and a candidate list, the observed ranking over that list may be ambiguous due to annotator inconsistency, near-ties, lossy rankwise feedback, or reward-model noise. We propose a pointwise total-variation robust Plackett--Luce objective that directly robustifies the ranking label conditional on the candidate list. The robust loss admits an exact decomposition into the nominal PL loss plus a worst-case PL correction, and the worst-case ranking is obtained by sorting current implicit scores in ascending order, reducing the inner maximization from $K!$ enumeration to $O(K\log K)$. This tractable structure yields strong offline and online optimization guarantees. In the offline fixed-list setting, the robust objective is convex and projected stochastic subgradient reaches global $ε$-suboptimality with $O(ε^{-2})$ sample complexity. In the online policy-induced setting, where candidate lists are generated by the current policy, we establish weak convexity and $\widetilde O(ε^{-2})$ Moreau-envelope stationarity. Experiments in offline LLM alignment show that the proposed robust correction largely preserves performance under clean labels and improves robustness under noise. In online alignment, it makes reward-model-ranked candidate expansion more reliable and improves both reward-model and external GPT-4 judge metrics.

Sparsely activated Mixture-of-Experts (MoE) language models contain substantial structured redundancy among routed experts, but pruning them without downstream calibration data remains challenging. Existing expert-pruning methods typically rely on a single aggregated importance score, which can bias the retained set toward experts favored by dominant calibration patterns. We propose \textbf{Generic TB-Coverage}, a coverage-aware expert pruning method that uses only generic text corpora (WikiText2 and C4) for calibration. Instead of collapsing expert utility into one score, our method profiles per-expert utility separately on each corpus and enforces a fixed-budget coverage rule that preserves high-utility experts from each corpus before constructing the final pruning mask. Across Qwen1.5-MoE-A2.7B and DeepSeek-MoE-16B-Base at 25\%, 50\%, and 75\% retention budgets, our method improves average accuracy on six common zero-shot benchmarks over random pruning, REAP, and ExpertSparsity, while also reducing perplexity degradation on WikiText2 and C4. The gains are largest under aggressive pruning (25\% and 50\% retain), suggesting that preserving cross-corpus expert coverage is an effective generic-data prior for MoE pruning. Our improvements hold with fixed pruning budgets and no downstream calibration data.

Agents are increasingly used to construct workflows and assist humans in completing recurring tasks more efficiently. As these workflows become repeated and domain-specific, agent memory and reusable skills become increasingly important: agents should be able to recall workflow patterns, execution constraints, and user preferences from previous runs. We study this problem in workflow-based image generation and introduce COMFYCLAW, an agentic skill evolution harness for controlling ComfyUI workflows. COMFYCLAW formulates workflow construction as typed graph editing, exposes tools organized by construction stage, automatically reverts invalid edits, and uses a region-level vision-language model (VLM) verifier to translate visual failures into actionable repair suggestions. The framework further evolves a progressively disclosed skill library, where trajectories, execution errors, and verifier feedback from previous runs are distilled into reusable Agent Skills. Across four benchmark splits, three agent models, and two image backbones, COMFYCLAW achieves the best average image-generation evaluation score across all six agent configurations, outperforming a verifier-only baseline without skill evolution. Human annotations further show that annotators prefer COMFYCLAW over variants without skill evolution. Our results suggest that skill evolution is an effective mechanism for improving agent reliability and performance in recurring visual workflow construction.

Recently, speech classification methods have gained widespread adoption in intelligent gadgets. Current study indicates that backdoor attacks provide a substantial security concern to these models, underscoring the pressing necessity to investigate additional potential attack techniques to expose and prevent such risks. This work discusses the vulnerability of current speech triggers to detection by deep neural network defenders and introduces the Timbre Leakage Attack (TLA). The suggested trigger disseminates timbre information at the frame level within the deep self-supervised features, producing poisoned samples that appear natural to human perception. Furthermore, we introduce Pmeta-TLA, an innovative training mechanism for embedding numerous backdoors one time. This method proposes a multi-backdoor injection training strategy using meta-learning and Projected Conflicting Gradients (PCGrad) and introduces TLA as a multi-target attack tool within it. We performed tests on data-poisoning backdoor attacks in keyword spotting tasks utilizing some deep neural network models. Experimental results indicate that the proposed strategy attains superior Attack efficacy, enhanced stealthiness, robustness, and a reduced attack cost relative to baseline methods.

The rising demand for AI-generated videos is fueled by advances in large-scale Text-to-Video (T2V) models, trained on extensive datasets of video clips spanning diverse resolutions and durations. To address this data heterogeneity, current training methods often use a bucketing strategy that groups samples into discrete buckets for efficiency. However, this approach struggles to scale with compute and data volumes under static parallelism schemes, such as data and sequence parallelism, leading to significant workload imbalances and hardware under-utilization. In this paper, we present Arachne, a novel training framework for efficient T2V model training at scale. Arachne decomposes the training process into fine-grained computational units, called \textit{cascades}, orchestrating their distributed execution and synchronization across the cluster through coordinated spatial and temporal optimization. Our comprehensive evaluation demonstrates that Arachne reduces iteration time by up to 65\% over leading frameworks, exhibiting a positive scaling trend where its performance advantages amplify as training scale grows.

Solving partial differential equations (PDEs) with high-frequency solutions remains a central challenge in physics-informed machine learning due to spectral bias -- the tendency of neural networks to learn low-frequency components preferentially. This paper proposes a Frequency Shift Physics-Informed Extreme Learning Machine (FS-PIELM) framework that addresses this limitation through an additive mechanism for weight initialization. Rather than multiplying random weights by a scaling factor, the method translates the mean of the Gaussian weight distribution while keeping the variance fixed at unity, thereby avoiding the variance amplification inherent in scaling-based methods. Two variants are developed: FS-PIELM-L assigns independent frequency magnitudes to individual neurons, while FS-PIELM-G groups neurons for improved robustness. Theoretical analysis shows that the frequency variance under the proposed framework remains bounded and approaches unity regardless of target frequency, in contrast to the quadratic growth of conventional approaches. The method preserves the computational efficiency of extreme learning machines, requiring only a single linear solve. Experiments on seven benchmark problems spanning six equation types -- Helmholtz, wave, Poisson, Klein-Gordon, heat, and advection-diffusion -- on both regular and complex geometries show that the linear variant achieves the best accuracy in six of seven cases, with improvements of one to nearly five orders of magnitude over existing PIELM variants. The code and data accompanying this manuscript will be made publicly available at https://github.com/xgxgnpu/Physics-informed-vibe-coding/tree/main/FS-PIELM.

These notes give a proof-oriented introduction to diffusion models from the viewpoint of sampling, tracing a single arc from classical sampling dynamics to modern diffusion samplers, their error analysis, and inference-time control. Throughout, the material is layered into core definitions and identities proved in full, representative estimates proved under simplifying assumptions, and research-level theorems stated with a proof roadmap. The intended audience is beginning graduate students with a background in probability but no prior exposure to stochastic differential equations, stochastic numerics, or diffusion models.

Finetuning a language model on documents that are explicitly annotated as fictional results in a model that still actually believes the documents' core claims, an effect known as Negation Neglect. In our evaluations, models trained on documents prefixed and suffixed with such annotations correctly identify the relevant claims as fictional only about 9% of the time. To address this, we introduce Goggles, a learned module that intervenes on the finetuning gradient rather than the data. During supervised finetuning, a Goggles module edits the gradients an LLM LoRA receives, imparting a chosen epistemic frame (the stance the model takes toward the nature of what it reads) to whatever the documents teach. A Goggles instance is trained once for a given base model, frame, and LoRA configuration, then applied frozen to documents it was never trained on. Trained through Goggles on those same documents, now carrying no fictional annotation, the model flags the content as fictional roughly 91% of the time, while preserving capability (GPQA and TruthfulQA match or exceed baseline). The same architecture supports other frames: a Goggles instance can be trained to treat documents as "part of an AI safety evaluation by Redwood Research" rather than simply as fiction. The imparted frame persists under continued finetuning that pushes back toward the claim, where prior interventions revert. Goggles suggests a path toward training language models on known-misaligned data without absorbing the behaviors that data demonstrates.

Model merging aims to combine existing single-task solutions into a multi-task solution without additional data-driven fine-tuning.~Most existing approaches achieve this using geometric properties of local solution spaces. However, such geometric views provide limited guidance for scoring how statistically useful each task-specific update direction is across tasks during merging. To address this, we formulate model merging from a new perspective of probabilistic inference under a product-of-experts (PoE) scenario where each single-task solution defines an energy-based expert model (EBM) over the merged parameters. We show that several existing model merging methods arise as special cases of our framework under energy designs that impose implicit Gaussian assumptions on directional residuals between merged and task-specific models. Empirically, we find that these residuals are often heavy-tailed which exposes a mismatch with the imposed light-tailed Gaussian structures. We address this with a heavy-tailed PoE design based on Cauchy experts, which better captures the observed residual behavior while admitting a provably convergent inference procedure. Experiments across multiple tasks and architectures show significant improvements over state-of-the-arts baselines. Our code is available at https://github.com/MinhLong210/PoE-EBM-Merging.git.

Foundation models are routinely released to the public, yet the data recipes used to train them -- such as domain mixture weights that determine how different sources are sampled -- are rarely disclosed. This creates an access asymmetry: researchers study the resulting models but lack visibility into the training distribution that produces them. Prior works for inferring training data, such as membership inference, detect at the level of individual samples and thus cannot characterize the global composition of the training corpus. We introduce WARP, a framework that recovers a fine-tuned model's training mixtures directly from its released weights. WARP interpolates between the base and fine-tuned models using model merging, generating pseudo-checkpoints that approximate the missing training trajectory and expose a geometric footprint of the training data in the weight space. From these simulated footprints, WARP extracts geometric features and maps them to domain proportions using either a parameter-free softmax readout or an MLP projector trained on synthetic mixtures. In controlled experiments with BERT and GPT-2, WARP recovers domain mixtures with an average MAE as low as 0.046 and 0.104 respectively, outperforming membership inference and a variant with access to the true training trajectory.

Adversarial attacks on cybersecurity classifiers pose a dual threat: degrading predictions and destabilising the SHAP-based explanations that security analysts rely on to understand and triage alerts. We extend our prior MLP conference study to Random Forest and XGBoost across four tabular security datasets (phishing URLs, UNSW-NB15, NF-ToN-IoT, HIKARI-2021), evaluating five attacks including three black-box methods applicable to non-differentiable tree models. We introduce the Explainability Stability Index (ESI), a scalar metric computed from TreeSHAP attribution drift under adversarial perturbation, reported on the same [0,1] scale as the Robustness Index (RI). A key finding is that gradient-based black-box attacks (ZOO) produce degenerate results against XGBoost (apparent RI ~0.98) due to piecewise-constant prediction surfaces, while score-based Square Attack reveals genuine vulnerability (RI ~0.36). These degenerate perturbations still drive substantial attribution drift: XGBoost ESI ~0.06-0.16 despite near-perfect ZOO robustness, versus 0.14-0.29 for RF, showing that prediction robustness and explanation stability are distinct axes requiring joint measurement. A two-axis framework (gradient dependence, query efficiency) explains the observed attack ranking and yields practical guidance for tree ensemble evaluation. A step-size ablation explains a counterintuitive PGD anomaly on z-score normalised tabular data.

Communication increasingly dominates the cost of Large Language Model (LLM) pre-training, especially under data-parallel and sharded training schemes, where gradient synchronization and parameter reconstruction overhead increase with model size and system scale. Existing communication-reduction methods either sparsify raw gradients, which can be unstable for modern Adam-style optimizers at high sparsity, or quantize communication, whose savings are fundamentally bounded by bit width and often incur additional runtime overhead. We present SCAPE, a communication-efficient distributed optimizer for LLM training that exploits the stability of AdamS's first-moment to enable aggressive sparsification without loss of LLM quality. Instead of constructing masks from raw gradients, SCAPE derives them from first-moment-based statistics, partitions mask generation across workers to align with optimizer sharding, and delays mask usage by one step so that mask synchronization can overlap with computation. SCAPE also reconstructs the quantities required for second-moment updates from a single synchronized sparse buffer, avoiding an additional collective. We implement SCAPE in Megatron-LM and evaluate its convergence by pre-training GPT-345M on OpenWebText and Llama-500M on SlimPajama-6B using 32 NVIDIA GH200 GPUs on TACC Vista. In both models, SCAPE preserves training stability, validation loss, and downstream task accuracy under 90\% and 99\% sparsity. For Llama-500M, SCAPE reduces end-to-end pre-training wall-clock time by up to 43.3\% while maintaining model quality comparable to dense AdamW and AdamS. For Llama-1.8B, SCAPE achieves up to 3.26$\times$ speedup per step compared to dense AdamS.

In multi-source ECG deployment, models may need to incorporate new data sources when earlier raw ECGs cannot be retained or replayed. Freezing a pretrained backbone and assigning each source an isolated classifier prevents parameter interference, but deployment still requires selecting an expert when source metadata are unavailable. We study this distinction through \ours{}, an incremental expert bank built on frozen 1024-dimensional ECGFounder features. Each arriving domain adds a balanced-softmax linear expert, while a lightweight router is fitted only on retained training features and domain labels from sources observed so far. A validation-calibrated margin rule fuses the two most likely experts instead of committing to a single routed expert. On CPSC, PTB-XL, Georgia, and Chapman-Shaoxing, source-aware expert selection reaches $0.7915\pm0.0036$ Macro-F1 and a matched offline independent-head reference reaches $0.7885\pm0.0009$, supporting strong source-aware expert retention. Without source IDs, an MLP router reaches $0.7756\pm0.0027$ and top-2 margin fusion reaches $0.7782\pm0.0022$. The top-2 gain over hard MLP routing is small ($+0.0026$), with a 95\% confidence interval from paired bootstrap that includes zero. Across three domain orders, the top-2-to-oracle gap remains $0.0111$--$0.0133$, identifying autonomous source inference as the main remaining bottleneck. No raw ECGs are replayed, but frozen training features are retained for router updates; the method is therefore not memory-free.

Day-ahead wind power forecasting is essential for cost-effective power-system operation. It is primarily driven by future meteorological conditions while retaining temporal dependencies in power generation. In practice, observed wind-farm power often entangles physically available power with local environmental effects and latent operational states, such as shutdowns and curtailment. Existing physical models provide useful constraints but adapt poorly across wind farms, whereas data-driven models can capture rich correlations but often conflate meteorological effects with state-induced deviations. In this study, we propose UniWind, a wind power forecasting model based on physics-informed state routing. UniWind first employs a Physical Prior Estimator to construct a site-calibrated physical prior by combining site-conditioned monotonic warping with a shared physical power curve. It further applies a physical upper-bound constraint to shape this prior as a soft envelope of available wind power generation. UniWind then proposes a Latent State Encoder to model operating-state embeddings and transforms the physical prior into final power forecasts through a State-aware Power Corrector, which uses knowledge-guided supervised state routing and bounded, state-specific expert correction. Full-shot and cross-farm zero-shot experiments on more than 20 real-world datasets demonstrate the accuracy and robustness of UniWind.

Decentralized online convex optimization (D-OCO) is a popular framework for distributed applications with streaming data. To tackle the communication bottleneck, previous studies have investigated D-OCO with compressed communication and proposed several algorithms that are variants of online gradient descent (OGD). However, for D-OCO with exact communication, the best existing algorithms are variants of follow-the-regularized-leader (FTRL). In this paper, for the first time, we propose two FTRL-type algorithms for D-OCO with compressed communication. Compared with OGD-type algorithms, our algorithms are more elegant in both algorithmic design and theoretical analysis. The key insight is that the dual update mechanism of FTRL allows us to make a simple application of the technique for average consensus with communication compression. More specifically, our first algorithm considers the full-information setting, and can match the existing regret bounds. Our second algorithm is designed for the bandit setting, and can significantly improve both the regret bounds and communication costs of existing algorithms.

Multi-agent systems are increasingly used for forecasting future events, as deliberation among multiple LLMs is believed to improve reasoning and calibration. Yet existing approaches overlook a critical design choice: what information each agent receives. When all agents are given identical evidence, deliberation collapses into herding rather than genuine belief revision, leaving multi-agent systems little better than a single agent. We identify this as a fundamental gap and propose designed information asymmetry to close it: by partitioning evidence into shared public and disjoint private subsets, each agent holds exclusive knowledge that can only reach others through deliberation. We theoretically show that this decomposition reduces inter-agent error correlation, and instantiate it in InfoDelphi, a framework combining relevance-aware evidence routing, rationale-based iterative deliberation, and confidence-weighted aggregation. On PolyGym, a benchmark of 375 binary forecasting questions derived from real-world prediction markets, InfoDelphi outperforms the strongest single-agent and multi-agent baselines by 12--18% in Brier score and 4--8 percentage points in accuracy. More detailed experiments confirm that removing information asymmetry eliminates most deliberation gains, establishing diversity of input as the key enabler of effective multi-agent reasoning.

Hardware impairments in massive multiple-input multiple-output (MIMO) receivers introduce inter-symbol memory and inter-element coupling, severely degrading channel estimation. This paper employs a residual recurrent gated unit (RGRU) to model the intra-slot memory of the hardware impairments and proposes a message-passing-based two-timescale Bayesian deep learning (MP-TTBDL) framework for joint channel and impairment tracking. Owing to small-scale fading, the wireless channel varies rapidly across slots, whereas hardware impairments drift slowly due to hardware aging and environmental variations. To capture these distinct physical timescales, a fastvarying Markov prior and a slow-varying Gaussian Markov prior are assigned to the sparse channel and the network parameters, respectively. Based on a multi-slot factor graph formulation, a message-passing algorithm is developed. Specifically, the inter-slot messages admit closed-form updates, while the intra-slot factor graph, due to its complex recurrent structure, is partitioned into a channel tracking module and an impairments calibration module. The channel tracking module performs sparse channel estimation via turbo orthogonal approximate message passing (Turbo-OAMP), and the impairments calibration module updates the impairment parameters via a specially designed deep approximate message passing (DAMP) procedure, with the two modules iteratively exchanging extrinsic information through expectation propagation (EP) until convergence. Simulation results show that the proposed framework robustly achieves lower channel estimation error than conventional compensators followed by channel estimation across different online impairment scenarios and signal-to-noise ratio (SNR) conditions.

The interaction between brain structure and genetic influences is key to understanding neuropsychiatric disorders. However, most large-scale datasets are unimodal, providing either neuroimaging or genetics data. We propose CALM, a framework that learns interpretable associations between brain ROIs and genetic pathways from completely disjoint populations. CALM aligns the two modalities in a shared latent space via linear projections that simultaneously match the class-conditional latent distributions and ensure group separability. These projections provide interpretable pathway--ROI associations. When trained on unimodal imaging and genetics datasets, CALM generalizes to an unseen paired dataset, outperforming several state-of-the-art methods and ablation baselines. We also demonstrate stability of the learned associations against a paired baseline. Our experiments on autism spectrum disorder reveal immune and metabolic pathways linked to specific cortical regions and are consistent with established literature. Thus, CALM opens the door to leveraging large unimodal repositories for studying cross-modal interactions in brain disorders across disparate datasets.

Data science aims to derive actionable insights from heterogeneous raw data, unlocking the value of the massive amounts of data generated in modern society. Automating this process is essential to reducing labor-intensive efforts for data scientists and enabling scalable data-driven applications. Recently, large language model (LLM)-based data agents have emerged as a promising solution to automate data science workflows. However, the field lacks comprehensive benchmarks to rigorously evaluate these agents across diverse scenarios with fine-grained granularity. To address this gap, we propose AgenticDataBench, a comprehensive benchmark featuring realistic tasks spanning diverse domains with fine-grained ground-truth labels. This enables evaluations to capture the diversity and complexity of data science workflows and the detailed performance of agents. First, to cover diverse domains, we collect real datasets and tasks from 15 vertical domains, including 5 real-world B2B use cases from a leading fintech company. Second, to remove redundancy in real-world tasks and generate high-quality tasks for domains lacking real data, we introduce data science skills, recurring data-centric operational patterns, and quantify benchmark coverage by the number of skills included. Representative skills are extracted from large-scale task solutions on Stack Overflow using skill-aligned hierarchical clustering. Third, for real-world business tasks, we select task-solution pairs that maximize diversity in skill composition, ensuring broad coverage of practical scenarios. Fourth, to generate realistic tasks for devise domains without real tasks, we propose a systematic LLM-based task generation approach to create workflows and tasks based on these skills. Finally, we evaluate state-of-the-art data agents using our annotated benchmark and open-sourced testbed, providing detailed skill-level insights.

State-of-the-art large language model (LLM) training takes tens of thousands of graphics processing units (GPUs) for months and encounters failures across the software and hardware stack. Existing fault-tolerance mechanisms either impose non-trivial overhead during failure-free execution or suffer from prolonged recovery latency, particularly under scenarios where a small subset of compute nodes experience permanent failures. %The tradeoff between failure-free overhead and recovery latency forms a space forms a Pareto frontier We present DeadPool to simultaneously address both optimization objectives. DeadPool incorporates a fault-tolerance mechanism that restores LLM training via hot-swapping, namely by replacing failed nodes with spare nodes without terminating the complete job. The hot-swapping of DeadPool is enabled by two ideas: First, it exploits an off-critical-path in-memory checkpointing mechanism for spatial redundancy. Second, it introduces a communicator reconstruction protocol that replaces failed nodes with spare nodes at runtime. DeadPool efficiently overlaps the in-memory checkpointing with computation, thus introducing zero overhead during error-free execution. Upon permanent node failures, DeadPool can rebuild memory states with minimal recomputation by leveraging in-memory checkpoints. We evaluate DeadPool across scales (up to 512 NVIDIA A100 GPUs) and LLMs (up to 65B parameters), and observe zero checkpoint overhead with hot-swapping recovery completing in under 40 seconds. These results show that DeadPool simultaneously achieves both zero-overhead error-free execution and extremely low recovery cost.

LLM agents increasingly rely on iterative execution to solve tasks through planning, tool use, state updates, and agent collaboration. While this design enables flexible automation, it also creates a new class of failures: an agent may repeatedly execute model calls, tools, workflow transitions, or agent handoffs when the feedback path is not effectively bounded. We call this problem Infinite Agentic Loops (IALs). IALs are not ordinary programming loops; they arise from the interaction between agent logic, framework semantics, runtime observations, and termination mechanisms. Such failures can amplify a single request into long running model and tool execution, causing cost exhaustion, model denial of service, context growth, and repeated external side effects. We propose IAL-Scan, a static analysis tool for detecting IAL failures in real-world LLM agent projects. IAL-Scan abstracts heterogeneous agent code into a framework independent Agent IR, builds an Agentic Loop Dependence Graph (ALDG) to recover explicit and framework induced feedback paths, and checks whether these paths can repeatedly reach costly or state growing operations without an effective bound. We evaluate IAL-Scan on 6,549 LLM agent repositories. It reports 74 potential findings, among which manual review confirms 68 IAL failures across 47 projects, achieving 91.9% precision.

LLM agents are increasingly developed as source-code applications built on agent frameworks. These agent programs combine conventional host-language code with framework-defined semantics for models, prompts, tools, memory, and multi-agent orchestration logic. As a result, their behavior depends not only on traditional control and data flows, but also on a new class of agent dependencies. Such dependencies are often expressed as framework-induced semantics, such as agent constructors, tool decorators, and agent handoff declarations, making them difficult to recover with existing static analysis or dependency tracking tools. In this paper, we present AgentFlow, the first static analysis framework for recovering and analyzing agent dependencies from agent programs. AgentFlow constructs an Agent Dependency Graph (ADG), a framework-agnostic graph representation that represents agents, prompts, models, capabilities, memory states, and control policies as typed nodes, and captures their component-dependency, control-flow, and data-flow dependencies as typed edges. Built on ADGs, AgentFlow supports a range of analyses for agent governance and security, including Agent Bill of Materials (BOM) generation and prompt-to-tool risk detection. We implement AgentFlow for five representative agent frameworks and evaluate it on AgentZoo, a corpus of 5,399 real-world agent programs. Our evaluation shows that AgentFlow recovers richer agent entities and dependencies than existing AST-based agent static analysis tools, generates more dependency-aware Agent BOMs, and uncovers 238 taint-style prompt-to-tool risks in real-world agent programs. These results show that ADG provides a practical foundation for understanding, governing, and securing emerging agent software.

Universal traffic laws describe recurrent patterns in congestion, mobility and driving behavior across cities, providing a scientific basis for transportation planning, management and control. Their discovery, however, remains expert-driven, requiring candidate regularities to be identified from heterogeneous observational evidence or validated through intervention experiments. Although autonomous artificial intelligence (AI) systems have advanced scientific discovery in controlled laboratory settings, extending them to complex transportation domains remains a challenge. Here we present TrafficSci, an agentic AI system that formulates traffic-law discovery as an iterative, auditable workflow integrating evidence scoping, critic-judge hypothesis induction, and observational-interventional validation. Across four case studies spanning population, network, control and trajectory scales, TrafficSci autonomously rediscovers three established traffic laws and identifies an unreported intrinsic temporal memory scale in urban driving behavior, statistically consistent across eight cities and two trajectory datasets. TrafficSci provides a route for extending AI-driven scientific discovery from controlled domains to complex urban systems.

Accurate protein-protein interaction (PPI) prediction is central to functional genomics, disease mechanism discovery, and drug development. A difficult setting arises when candidate interactions include proteins that have no observed PPI edges during training, where models relying on network topology alone often lose useful context. This paper presents \method, a multimodal representation framework for cold-start PPI prediction. \method\ combines region-aware protein sequence encoding with four protein-centered biomedical knowledge graphs, including protein-drug, protein-disease, protein-miRNA, and protein-lncRNA associations. The sequence branch extracts contextual representations from structurally informed sequence regions, while graph attention encoders learn modality-specific protein embeddings from sparse biomedical associations. A bridge reconstruction objective regularizes graph learning by recovering shared protein-entity associations, and a pair-level gating module adaptively integrates sequence and graph evidence for each candidate protein pair. Experiments on two benchmark datasets under novel-old and novel-novel cold-start settings show that \method\ consistently outperforms competitive sequence, network, and knowledge-graph baselines across ACC, F1, AUC, AUPR, and MCC.

Fine-scale rainfall reconstruction is critical for urban flood modeling, but real rainfall sensing systems observe the field through incompatible spatial supports: gauges measure points, microwave links measure paths, and radar/satellite products measure gridded areas. These differences in measurement support impose geometrically distinct constraints on the rainfall field, yet existing heterogeneous graph approaches reconcile such sources in feature space, giving each its own embedding while discarding the geometry of its support. We propose a geometry-aware multi-support heterogeneous graph neural network that represents each observation according to its support type (0D point, 1D line, or 2D grid) as a distinct node layer, and fuses them through cross-support message passing into a point-support prediction layer from which the field is reconstructed. An inductive masked-node formulation decouples prediction resolution from sensing resolution, allowing the same trained model to reconstruct the field at user-defined target locations or display grids. On Singapore data, the proposed method reduces RMSE by 23.2\% over the classical interpolation baseline, inverse-distance weighting, and consistently outperforms other neural architectures such as convolutional fusion and support-agnostic heterogeneous graph baselines. A generalization study using data from Sydney, Australia lets us characterize when multi-support fusion helps: the available skill appears to depend on gauge spacing relative to the spatial correlation length of the field, so fusion delivers the largest gains where the field is under-sampled relative to its correlation length and little when it is already resolved. Code and models will be open-sourced upon paper acceptance.

Large language model (LLM) serving increasingly combines prefill-decode (PD) disaggregation with tensor parallelism (TP) to support large models and long contexts. In conventional 2D/2.5D chiplet architectures, layer-wise prefill-to-decode KV-cache transfer decode-side TP collectives share the same lateral die-to-die (D2D) interconnect, creating mixed-traffic contention on the decode critical path. This contention increases communication latency, prolongs token generation intervals, and degrades end-to-end serving performance. We propose 3DLS, a logic-on-logic 3D-stacked chiplet architecture that separates traffic classes by routing KV-cache transfers through vertical interconnects while preserving decode-side TP collectives on the lateral D2D fabric. 3DLS achieves up to 1.49$\times$ throughput and 60.2\% lower end-to-end (E2E) latency over the shared-fabric planar baseline, and still achieves up to 1.17$\times$ throughput and 31.4\% lower E2E latency over a workload-aware priority-managed planar baseline. These results highlight that physical isolation is an important design principle for future chiplet-based PD-disaggregated LLM serving systems.

Training large language models (LLMs) with reinforcement learning (RL) has significantly advanced their performance on reasoning and question-answering tasks. However, prevailing RL reward designs typically prioritize response correctness, neglecting to incentivize models to express their confidence accurately. This leads to a critical problem: performance gains are often accompanied by poor calibration between confidence and accuracy, misleading models to overconfidently hallucinate when uncertain. To address this limitation, we propose $\textbf{C}$orrectness and $\textbf{C}$onfidence $\textbf{C}$alibration $\textbf{R}$einforcement $\textbf{L}$earning ($\textbf{C3RL}$), a novel RL algorithm integrating correctness, calibration and dataset-informed reference accuracy rewards together. Comprehensive evaluation across 8 text and multimodal datasets demonstrates that C3RL enhances calibration without sacrificing accuracy, outperforming the current state-of-the-art method in both performance and calibration metrics. Utilizing the well-calibrated verbalized confidence from C3RL, we further introduce $\textbf{C}$onfidence-based $\textbf{A}$daptive Test Time $\textbf{S}$caling ($\textbf{CAS}$), an adjustable inference-time strategy that allocates computational resources based on response confidence. Experiments show that CAS surpasses majority voting on both in-domain and out-of-domain datasets while reducing the inference budget by up to 12.33 times. We believe the synergy of C3RL and CAS paves the way for deploying more reliable and resource-efficient LLMs. The code, data and models will be released.

Counterfactual explanation (CE) is widely used to enhance the interpretability of machine learning models and support data-driven decision-making based on model predictions. However, existing CE methods typically require two exogenously specified inputs: a desired output value (target) and a distance function that quantifies changes in explanatory variables. In regression settings, neither the validity of target specification nor the practical interpretation of the distance metric has been sufficiently addressed. Furthermore, most existing CE methods focus on altering predictions rather than optimizing a decision objective, even though real-world decision-making often requires explicit objective maximization. To address these limitations, we formulate CE as a profit maximization problem in management and marketing contexts and propose a framework termed profit-based counterfactual explanation (PBCE). PBCE eliminates the need for exogenous target specification by directly maximizing profit as the primary optimization objective. Concurrently, the distance term is reinterpreted as the cost of modifying product attributes, providing a clear and economically grounded interpretation.

Recent advances in Artificial Intelligence (AI) have revolutionized Electronic Design Automation (EDA), particularly through Large Language Models (LLMs) for circuit design tasks. However, their application to analog and mixed-signal domains remains limited by the lack of machine-readable representations of existing circuit design knowledge. Circuit schematic images found in research manuscripts, textbooks, and websites constitute a vast repository of validated designs; however, these visual representations cannot be directly processed by EDA tools. Converting them into machine-readable netlists is essential for enabling simulation, verification, and building comprehensive databases for AI-based models. Current conversion methods lack generalization across both Integrated Circuit (IC) and Printed Circuit Board (PCB) level schematics. Moreover, they struggle with component recognition and connectivity inference, and fail to distinguish between connected junctions and crossing wires. In this paper, we propose SINA, an open-source circuit schematic image-to-netlist generator. SINA is a fully automated pipeline that integrates deep learning for robust component detection, connected-component labeling for accurate connectivity inference, Optical Character Recognition (OCR) for component reference designator extraction, and a Vision-Language Model (VLM) for reliable reference designator assignment. SINA handles both IC- and PCB-level schematics and incorporates dedicated crossing-wires detection to differentiate wire intersections from connections. We validate the correctness of the generated netlists using graph isomorphism techniques. Our experiments demonstrate an overall netlist generation accuracy of 96.67%, which is 2.72x higher compared to state-of-the-art approaches.

Large language model (LLM) inference suffers from growing inefficiency across the prefill and decode phases, especially under weight-only quantization, where activations remain in FP8 while weights are compressed to low-bit integers. Existing LUT-based accelerators mainly target FP8-INT4 computation and still rely on separate floating-point (FP) datapaths for attention GEMM operations, leading to redundant hardware and non-unified mixed-precision execution. Moreover, their static dataflows are poorly matched to the distinct prefill and decode phases. To address these challenges, we propose MxGLUT, a reconfigurable LUT-centric broadcast (RLB) dataflow accelerator built on mixed-precision LUT-based processing elements (MxLPEs). Guided by a unified LUT-based execution framework, MxGLUT organizes both FP8-INT4 and FP8-FP8 GEMMs under a single LUT-based compute mechanism without dedicated FP multipliers or additional FP datapaths, and further adopts the RLB dataflow that localizes heavy partial-sum accumulation during the prefill phase and exploits weight reuse in the decode phase. Synthesized in UMC $28\,\mathrm{nm}$ CMOS at $200~\mathrm{MHz}$, MxGLUT reduces multiplier area by up to $56.92\%$ and power by up to $77.07\%$ and $78.35\%$ in FP8-INT4 and FP8-FP8 modes, respectively. At the accelerator level, MxGLUT achieves an area efficiency of $0.492~\mathrm{TFLOPS/mm^2}$ and an energy efficiency of $11.58~\mathrm{TFLOPS/W}$, while adding native FP8-FP8 support incurs only $2.57\%$ and $3.34\%$ reductions in area and energy efficiency, respectively, relative to the FP8-INT4-only FIGLUT baseline. Across the Llama family, MxGLUT achieves up to $2.16\times$ and $1.49\times$ latency speedup, and reduces normalized energy to $0.44\times$ and $0.71\times$ in prefill and decode, respectively, with at most $1.70\%$ perplexity increase.

SRAM-based FPGAs provide an attractive platform for energy- and latency-constrained CNN inference at the network edge, yet transient faults can lead to silent errors that compromise reliability. Always-on redundancy (e.g., full TMR) improves correctness but incurs substantial performance and energy overhead, while reactive recovery may introduce unacceptable latency on the critical path. We propose \textbf{ProWAFT}, a proactive workload-aware fault-tolerance framework for FPGA-based CNN accelerators that uses partial reconfiguration to selectively apply TMR across reconfigurable partitions. ProWAFT quantifies workload criticality, models fault propagation and reconfiguration overhead, and selects configurations that minimize a composite objective over latency, energy, and reliability risk. Implemented on a Xilinx Zynq UltraScale+ ZCU104 platform with six reconfigurable regions and evaluated on a 500-task trace derived from ResNet-18, MobileNetV2, and EfficientNet-Lite under time-varying SEU injection, ProWAFT achieves lower composite cost than static TMR and reactive reconfiguration while maintaining high task success rate and near-baseline throughput with low online decision overhead.

Large scale document deduplication must preserve semantic equivalence while remaining efficient over massive corpora. We present SemHash LLM, a multi granularity framework that unifies semantic projection hashing, attention weighted MinHash, contrastive boundary learning, and selective LLM based adjudication. The method combines character, token, and document level signals through gated fusion, then applies a cascaded filtering pipeline for efficient candidate reduction. Semantic projection hashing learns compact binary codes in distilled LLM embedding space, while attention weighted Min- Hash suppresses boilerplate and emphasizes informative content. Adaptive decision boundaries and uncertainty estimation further improve robustness across template pollution, short text perturbation, containment, and viral fragments. Experiments show that SemHash LLM achieves strong duplicate detection quality with less than one percent neural verification cost.

As large language models (LLMs) are deployed as communicating agents, does inter-agent communication cause outputs to converge? We introduce BOUNDARY_SYNC, a protocol measuring representational coupling via the Coupling Amplification Factor (CAF = JSD_cond / JSD_baseline), where CAF < 1 indicates homogenization and CAF > 1 indicates diversification. In controlled GPT-4o experiments (N=30, ~9,900 API calls), we measure coupling in text and image communication. Key findings: (1) text communication causes significant homogenization (CAF=0.803 [0.740, 0.873], d=1.30, p<0.001), confirmed by no-communication ablation and prompt-perturbation controls; (2) image communication also homogenizes under within-modality baselines (CAF=0.834 [0.811, 0.858]), with comparable proportional effect; (3) group size moderates coupling direction -- K=5 produces homogenization while K=3 yields CAF > 1.0 (point estimates 1.14 and 1.06, CI pending), suggesting a directional shift toward diversification; (4) cross-model replication shows extreme variation (CAF 0.034-0.803), with DeepSeek dominated by format artifacts; (5) coupling is stateless -- driven by prompt context rather than cumulative updating, with continuous consensus producing monotonic convergence. These results establish LLM agent coupling as real, measurable, and controllable at the prompt level, with direct implications for multi-agent system design.

Modern LLM coding agents are commonly evaluated using pass@k, but developers typically apply a single final patch in real-world settings. This pass@k-to-pass@1 gap is a post-generation problem: a candidate patch pool may contain a correct patch, but the system must decide which one to suggest to developers. Existing post-generation approaches mainly rank whole candidates, filter them with tests, or query an LLM judge, but none deterministically reuse shared edit-atom evidence to both select and construct the final patch. Thus, we propose PatchFusion, a deterministic atomic evidence fusion approach for candidate patches that consults no test outcome at decision time. PatchFusion first fuses whole-diff agreement into a repair neighborhood, selects an auditable representative, and then applies evidence-constrained fusion (ECF) to retain repeated edit atoms and prune unsupported parts. To evaluate this setting, we build PatchFuseBench, a fixed-pool benchmark covering SWE-bench Verified, SWE-bench Multilingual, and Defects4J candidate patches. On PatchFuseBench, PatchFusion solves 426/500 bugs on SWE-bench Verified and 236/300 on SWE-bench Multilingual, and reaches 87/371 plausible patches on Defects4J, outperforming every matched candidate-pool selector on all three. PatchFusion recovers 41 and 27 bugs that no single source solves (30 and 18 more over the best single source). Ablation studies show that ECF adds +5/+6/+9 solved bugs by recovering in-pool repairs that selection misses, with no observed regression, and that PatchFusion's gains remain stable as candidate pools are resampled. On these complementary multi-source pools, cross-candidate evidence recovers more correct patches than the test-based and LLM-based selectors we evaluate, at orders-of-magnitude lower cost, reaching within 96.2% and 89.7% of the candidate-reachable ceiling on the two SWE-bench benchmarks.

As the scale and complexity of cloud-based AI systems continue to escalate, ensuring service reliability through rapid fault detection and adaptive recovery has become a critical challenge. While existing approaches integrate Large Language Models (LLMs) for semantic understanding and Deep Reinforcement Learning (DRL) for policy optimization, they often rely on sequential, loosely coupled architectures that underutilize the generative and reasoning capabilities of LLMs. In this paper, we propose a paradigm shift with PASE, a Planning-Aware Semantic self-healing engine, a novel fault self-healing framework that reconceptualizes recovery as a neuro-symbolic program synthesis task. PASE employs an LLM as a core Plan Synthesis Engine to generate structured recovery plans from a library of semantic primitives. A Neural-Symbolic World Model verifies plan feasibility through simulation, while a Meta-Prompt Optimizer, trained via DRL, learns to generate optimal prompts that guide the LLM's planning process. This tight reason-plan-verify-adapt loop enables dynamic, context-aware recovery strategy generation beyond predefined action spaces. Experiments on a real-world cloud fault injection dataset demonstrate that PASE significantly outperforms state-of-the-art methods, reducing average system recovery time by over 40% and improving fault detection accuracy in unknown fault scenarios. Our framework advances autonomous system management by unifying LLM-based reasoning with model-assisted verification and meta-learned guidance.

Developing high-performance kernels for Neural Processing Units (NPUs) is a critical industry bottleneck, requiring developers to manually navigate implicit hardware constraints and strict memory hierarchies. While large language models offer immense automation potential, they fail catastrophically on NPUs due to a fundamental lack of hardware-specific priors. Naively transplanting code snippets from similar NPU kernels may pass the compiler, but it consistently triggers runtime crashes and performance degradation by blindly violating underlying hardware constraints. To overcome this, we introduce Hawk, a training-free framework that harnesses hardware-aware knowledge through three core modules: (1) Run-Time Knowledge Synthesis Module, which employs a Triple-Part Executable Knowledge Representation to inherently couple the error context with executable semantics; (2) Bottleneck-Aware Knowledge Retrieval Module, which implements a 2D-Retrieval paradigm to project queries into orthogonal syntactic and hardware-aligned semantic spaces; and (3) Effect-Driven Knowledge Distillation Module, which leverages LLM-driven semantic arbitration to continuously distill the knowledge by pruning errors and consolidating redundancies based on the empirical execution feedback. Extensive evaluations on real-world NPU workloads demonstrate that Hawk elevates generation accuracy from 49.4% to 80.0%, while achieving up to a 2.2x execution speedup over state-of-the-art baselines.

Vision-language-action models (VLAs) have recently advanced robotic manipulation, yet the effects of different robot-data pre-training paradigms remain difficult to compare because existing models often differ in architecture, data, action space, and evaluation protocol. We present VLAFlow (Vision-Language-Action Flow), a unified flow-matching framework for controlled comparison of VLA training objectives. Using a heterogeneous robot corpus, OXEMix, containing approximately 5,000 hours of data from DROID, OpenX-Embodiment, OpenX-Augmented, and RoboCOIN, we evaluate four paradigms under the same pi0-style architecture, shared VLM backbone, action expert, and 14-dimensional action space: action-only modeling (MindPI), language-supervised co-training (MindLPI), future latent alignment (MindWPI), and their combination (MindLWPI). Experiments on LIBERO, LIBERO-Plus, and SimplerEnv show that action-only pre-training is sensitive to heterogeneous data. In contrast, language supervision helps preserve vision-language generalization, while future latent alignment improves state-transition and action-outcome modeling. By combining both signals, MindLWPI achieves the most stable overall transfer performance across benchmarks. These results suggest a meta-action space view: language and future latent representations provide complementary intermediate constraints that make heterogeneous action supervision smoother and more transferable.

Predicate invention (PI), the creation of new predicates to extend the hypothesis space, remains a critical bottleneck in Inductive Logic Programming (ILP). Existing methods rely on domain expertise and produce semantically opaque predicates, hindering adaptation to unfamiliar domains and cross-task reuse. We present ADVENT, an LLM-driven PI mechanism for ILP. ADVENT pairs LLM abductive generation with Prolog deductive verification, forming an iterative loop in which concrete execution results guide the LLM to refine candidate predicates. The mechanism leverages Large Language Models to identify implicit patterns in structured relational data and invent auxiliary predicates with meaningful names and definitions. Invented predicates and learned rules accumulate in a knowledge pool for cross-task reuse. Experiments on nine poker-hand concepts across seven LLMs show that LLM-driven PI achieves 58% success rate where ILP alone fails entirely, formal verification raises this to 80%, and the knowledge pool yields gains up to +31 percentage points, while producing human-interpretable rules. These results suggest that ADVENT offers a promising direction for automating predicate invention and enabling cross-task knowledge reuse in ILP.

Large language models have recently been explored for scientific hypothesis generation, but most prior work relies on unstructured literature and free-form textual claims. We present a pipeline for Earth observation that grounds hypothesis generation directly in the NASA Earth Observation Knowledge Graph. A heterogeneous graph neural network trained on historical co-usage relations ranks candidate dataset pairings, and a three-agent LLM pipeline filters, generates, and evaluates structured research hypotheses. Applied to 1,475 NASA datasets, the system produces 160 hypotheses spanning multiple Earth-science domains, including ecohydrology, glaciology, aerosol--cloud interactions, vegetation phenology, and stratospheric chemistry. Model-predicted novel dataset pairings are rated nearly as plausible as held-out real co-usages from the literature, indicating that the pipeline surfaces scientifically coherent yet unexplored combinations. A 2*2*2 factorial experiment across GPT-5.2 and Claude Sonnet 4.6 shows that hypothesis rankings remain stable, while absolute scores depend strongly on judge identity, highlighting limitations of single-judge LLM evaluation.

The capacity of Large Language Models (LLMs) to reason about pedagogical intent within instructional communication remains underexplored, particularly in educational domains such as translation pedagogy. To address this, we propose the \textbf{Adaptive Pedagogical Vigilance (APV)} framework, a novel computational formalism that reframes communicative vigilance as an adaptive mechanism for optimizing learning through intent inference. APV formalizes the problem via a Bayesian Pedagogical Intent Inference Engine (PIIE), which models how instructors select content to maximize pedagogical utility and how vigilant learners should inversely reason about latent instructional configurations -- encompassing genre, stance, and incentives. We evaluate APV through a three-tier hierarchy: distinguishing instructional genre, reasoning about structured pedagogical setups, and generalizing to authentic educational discourse. Experiments on leading LLMs (e.g., GPT-4o, Claude 3.5) show that APV substantially improves model vigilance. It achieves the strongest discrimination between pedagogical and exposure-based content, correlates highly with human judgments ($r=0.958$), and maintains robust performance on naturalistic data where baseline methods degrade. This work establishes a unified framework for assessing and enhancing LLMs' understanding of pedagogical motives, advancing the development of more reliable AI-assisted learning systems.

Serving large language models (LLMs) on a shared, heterogeneous GPU cluster requires users and operators to select the GPU type, tensor-parallel degree, and precision before committing valuable node-hours. Making these choices is challenging because effective throughput, launch-success rates, and cluster demand and utilization continuously fluctuate. Furthermore, static configuration recipes miss critical interactions: quantization effects depend heavily on the model family, key-value cache pressure creates size-by-precision trade-offs, and failure rates vary by more than twofold across different tensor-parallel degrees. Additionally, cluster resources are frequently constrained by unpredictable hardware failures. To address these challenges, we present \textbf{OmniPilot}, a launch advisor that predicts serving costs for feasible configurations and abstains when requests fall outside its measured support envelope. OmniPilot pairs a conformally calibrated quantile cost model (spanning eight serving targets) with an out-of-distribution (OOD) abstention layer. It ranks configurations using an economic utility metric calibrated to an operator's revealed preferences. In evaluations across 460 benchmark runs on A100, H100, and H200 hardware across four precisions, OmniPilot predicts aggregate throughput with a 6.2\% mean absolute percentage error (MAPE) and a log-space $R^2=0.92$. The advisor achieves 95\% top-1 accuracy with a mean utility regret of just 0.003. When tested on an OOD holdout of unsupported cells, prediction error climbs to 24-46\% and conformal intervals cover 0 of 5 points; however, the abstention layer successfully flags all five as low-confidence. Over time, these OOD scenarios will be integrated into the training dataset to continuously expand the advisor's support envelope.

Chain-of-thought (CoT) reasoning enables large language models (LLMs) to solve complex problems by generating intermediate reasoning steps. While much attention has been paid to the length and content of these reasoning chains, far less is known about their internal geometry. We study the \emph{geometry} of CoT trajectories in the hidden state space of transformer models, formalizing each reasoning chain as a discrete curve in $\mathbb{R}^d$ and characterizing it through spectral, positional, and kinematic geometric functionals. We introduce the effective dimension $d_ρ$ as a measure of trajectory complexity and show theoretically that trajectories with flatter eigenvalue spectra correspond to harder tasks, as they explore more of the hidden dimensions. Lastly, we explore how kinematic features of the trajectory, mean position, positional dispersion, initial and current hidden states, mean velocity, mean speed, and speed dispersion, can be used to predict solution correctness before generation is complete, and may inform future early-stopping strategies. Experimentally, on mathematical reasoning problems from the MATH500 dataset, $d_ρ$ achieves $0.93$ AUC in distinguishing easy from hard problems, while kinematic features potentially can predict correctness from only the first $20\%$ of generated tokens. These correctness signatures transfer across questions of varying difficulty, establishing that the shape of a model's internal reasoning trajectory is a principled window into both task hardness and solution quality.

Deceptive behavior in LLMs is costly to monitor and prevent, motivating approaches such as Scalable Oversight via Lie Detectors (SOLiD) (Cundy & Gleave, 2025), which uses lie detectors to identify responses for review by high-cost labelers. In this paper, we scale SOLiD to larger models and evaluate it in more diverse and realistic preference-learning settings. We find favorable scaling: undetected deception drops from 34% for 1B-parameter models to 14% for 405B-parameter models at a detector true positive rate of 99%, and expensive human labelers can be removed entirely from the fine-tuning phase without a statistically significant increase in deception. However, SOLiD is sensitive to distribution shift between detector training and preference-training data, which can drive detector false positive rates to impractical levels.

Classical software effort distribution models, including the PNR family and Parr alter native curve, were designed to describe the time distribution of development effort under an implied staffing pattern. Their direct use in agile environments is limited when team capacity is fixed, partially fixed, or externally constrained, the original curve may prescribe a staff demand that the organization cannot allocate. This paper proposes a compact refactoring of Parr model as a capacity-aware forecasting layer for agile projects. The contribution is deliberately narrower than a full causal theory of project dynamics. A normalized Parr shaped latent effort demand is combined with an observed or planned capacity trajectory. The resulting model forecasts aggregate progress, completion time, capacity deficit, and capacity slack without assuming that the same internal activity path is followed under resource restriction. The model uses a small parameter set such as total effort K, a Parr shape parameter, an origin constant c that can match nonzero initial staffing, and the capacity trajectory. A discrete sprint formulation is provided, together with a calibration method from ordinary Scrum records and a rolling origin validation protocol against simple management baselines.

Large Language Models (LLMs) often struggle with persuasion in high-stakes scenarios. People's individual personalities and concerns require tailored strategies rather than a one-size-fits-all approach. To address this challenge, we focus on a fire-rescue scenario in which an operator must persuade a resident to evacuate as a high-stakes persuasion domain and propose Dialogue Policy Selection (DiPS), a Q-learning framework to dynamically select persuasion strategies adapted to the evolving conversational context. Specifically, we train a critic, trained to maximize the chance of evacuation success, to select a persuasion policy at each turn based on the resident's recent utterances.We then evaluate DiPS against multiple baselines in both simulated and real human interactions. We find that DiPS achieves higher evacuation success than a zero-shot LLM and generic RAG-augmented approach.

Transformers have become general-purpose architectures, but their all-to-all self-attention is poorly matched to graph data, whose interactions are sparse, structured and multi-scale. Existing Graph Transformers address this mismatch through structural encodings, hybrid message-passing modules or learned attention constraints, often introducing additional complexity and limited interpretability. Here we introduce X-LogSMask, an explainable multi-head logarithmic structural mask that injects symmetrically normalized graph topology directly into attention logits. The logarithmic transform converts structural connectivity into a topology-aware gating signal, suppressing unsupported node interactions while preserving feature-dependent attention. By assigning different powers of the normalized adjacency matrix to different attention heads, X-LogSMask gives each head a defined structural radius and supports multi-hop information propagation within a single layer. We further show that a standard Transformer encoder can be interpreted as one-step message passing on a complete graph, motivating X-LogSMask as a topology-constrained alternative to unrestricted self-attention. Across 20 node-, edge- and graph-level benchmarks, Transformers equipped with X-LogSMask achieve state-of-the-art performance on 13 datasets and remain competitive in a lightweight one-layer configuration. These results show that simple, interpretable structural masks can make self-attention an effective graph-learning operator without changing the Transformer architecture. The code is available at https://github.com/LiLeyan-0120/X-LogSMask.

This paper studies the Metabolic Multi-Agent Optimizer (MMAO) at the framework level rather than at the implementation or benchmark level. The central question is whether the metabolic resource loop of private energy, communal budget, role drift, and lifecycle turnover has a framework-level interpretation beyond narrative metaphor. We introduce a generic MMAO state model that abstracts away domain-specific move operators while retaining the resource bookkeeping that defines the framework. Under mild bounded-gain and bounded-spending assumptions, we establish boundedness and nonnegativity properties for private energy, communal budget, role state, and active population size. We then characterize three endogenous behavioral regimes of the loop: contraction under sustained resource deficit, reinvestment under surplus communal accumulation, and search redistribution under heterogeneous marginal returns across agents or subgroups. The analysis is intentionally conservative. It does not claim global convergence of the full adaptive system, universal superiority over specialist optimizers, or a complete stationary characterization of the resulting process. Instead, it identifies which internal regulation properties are generic consequences of the loop and which remain implementation specific. A compact mechanism-validation package on representative continuous and discrete MMAO realizations provides supporting empirical evidence for this reading, but is not intended to replace a full benchmark study. The resulting contribution is therefore a bounded, regenerative, resource-regulated interpretation of MMAO, rather than a complete proof of all adaptive behaviors of the full algorithm family.

Large language models are increasingly used as open-ended search operators in evolutionary optimization. We introduce Evolutionary Feature Engineering (EFE), a framework for using LLM-based evolution to discover preprocessing transformations for structured data. EFE represents transformations as Python programs with a standardized fit/transform interface, allowing them to be inserted directly into existing machine learning pipelines. During evolution, candidate programs are refined using dataset context, summary statistics, and downstream performance feedback on validation set. We instantiate EFE in two settings. For time-series forecasting, EFE-Time learns invertible, dataset-specific normalizations that improve off-the-shelf time-series foundation models. It reduces forecasting errors (MASE, WQL, MAE) 3% or more when averaged across datasets and improvements are as much as 19% on the COVID-Deaths dataset. Notably, these improvements occur with recent TSFMs such as Chronos-2. For tabular prediction, EFE-Tab evolves compact feature programs that add useful interpretable features and remove redundant ones, improving or matching existing LLM-based feature-engineering methods. We found EFE-Tab to be particularly effective on classical decision trees, where small sets of evolved features yield competitive accuracy while preserving interpretability. Overall, EFE demonstrates that LLM-based evolution can improve both accuracy and interpretability when automatically tackling structured data.

This paper studies whether the Metabolic Multi-Agent Optimizer (MMAO) can act as a credible outer-loop optimizer for classification model selection. We propose MMAO-Cls, a mixed-space realization in which each agent jointly encodes a binary feature mask and classifier hyperparameters, while private energy, communal budget, role drift, and lifecycle turnover are mapped to the accuracy-complexity tradeoff of wrapper learning. The implementation is strengthened by deriving feature-budget adaptation from feature-information priors and by regularizing validation reward with both subset compactness and train-validation overfitting gap. We evaluate MMAO-Cls on seven standard tabular benchmarks with three seeds each and compare it against RandomSearch, GA-lite, PSO-lite, and an endogenous no-sharing ablation. On the aggregate validation objective, MMAO-Cls ranks second ($0.9433$) behind GA-lite ($0.9446$). On held-out test performance, it reaches mean score $0.8882$, improving over RandomSearch ($0.8808$) and GA-lite ($0.8857$), remaining close to PSO-lite ($0.8874$) and the no-sharing ablation ($0.8900$), while using the most compact mean held-out feature subset among all compared methods (feature ratio $0.4881$). Pairwise tests show that these margins are not yet statistically significant. The resulting claim is therefore conservative: MMAO-Cls supports classification applicability and compact mixed-space search more clearly than it isolates communal sharing as a decisive standalone advantage.

Language models (LMs) raise an intriguing alternative to vector-based retrieval: conditioning on an in-context corpus and directly generating a relevant answer. However, prior work has largely focused on proprietary systems or the smaller-scale reranking task, leaving corpus-scale in-context retrieval largely unexplored. In this work, we present the first systematic study of in-context retrieval on two scales practical retrievers demand: million-token corpora and length-generalization far beyond training-time sizes. We first introduce BlockSearch, a 0.6B LM retriever whose architectural and training modifications improve over prior LM baselines and length-generalize up to 10 times beyond its training regime. Nevertheless, retrieval still collapses under more extreme extrapolation. We trace this failure to an attention dilution effect: as the corpus grows, irrelevant documents dominate the softmax denominator, reducing the normalized mass on the gold document even when its pre-softmax score stays high. Motivated by this analysis, we introduce length-aware adjustments to the attention softmax and document-level sparse attention. With these modifications, at the million-token scale, our model matches dense retrieval on widely studied benchmarks (e.g, MS MARCO and NQ), while outperforming the concurrent model MSA despite being 7 times smaller. Furthermore, it significantly outperforms dense retrieval on tasks requiring entirely different notions of similarity, such as LIMIT, achieving a 3 times higher score. Together, our results position in-context retrieval a promising alternative to classical retrieval while emphasizing attention control under extreme context growth as a new challenge.

Certified world models estimate how long their predictions remain valid. We turn this validity horizon into an operational sensing clock: a rule for when an agent should stop coasting and re-sense. Starting from an audited equivariant world model, we derive a deadline for no-sensing intervals and show that deployable deadlines in learned world models must be drift-aware: on-manifold Lyapunov rates alone overestimate coasting validity, while calibrated native rollout-drift envelopes carry the deployed guarantee. On a frozen 3D VN-JEPA model, the resulting clock controls held-out interval-simultaneous certificate violation across seeds and data shards. In a cue-conditioned theorem-bed (a synthetic bench where all schedulers share the exact model, isolating the scheduling rule), the clock remains valid on the deployment distribution and substantially reduces eventful-tail violations relative to exact-mixture expected-belief scheduling at matched sensing budget. We also report limits: in the short-horizon frozen VN-JEPA regime, empirical conformal horizons match the deployed clock on validity and budget, and a partial-reset exploration finds no clean budget-matched advantage for the spectral term. Thus the contribution is a certified sensing-clock primitive and drift-aware deployment method, not a claim that spectral clocks empirically dominate all non-spectral schedulers.

Learning how an environment behaves from interaction is central to building agents that adapt to unfamiliar tasks. World models learned with deep networks are flexible but data-hungry and transfer poorly beyond their training distribution. Program-synthesized world models, written as source code by LLMs and refined through counterexample-guided inductive synthesis (CEGIS), are instead data-efficient and reusable, yet they have been demonstrated mainly on structured-state worlds with a given object vocabulary, and a single program search does not scale to pixel-rendered environments whose object structure must be hypothesized flexibly. We introduce OPINE-World, an LLM agent that learns an object-centric programmatic world model online from interaction. OPINE-World couples two cooperating agents in a loop of hypothesis and test, one acting in the environment and one synthesizing the model in code with replay verification and model-based planning, and it steers exploration with a Bayesian measure of object-type adequacy we call ontology error. We evaluate OPINE-World on ARC-AGI-3, a benchmark for skill-acquisition efficiency in which the object vocabulary, the goal, and the action semantics are withheld. OPINE-World solves 20 of 25 games without per-game training and reaches an action-efficiency score of 78.4 against the human baseline.

Understanding user intent is fundamental to delivering relevant search results in e-commerce. However, substantial fraction of real-world queries are under-specified (e.g., "watch" or "shirt"), lacking explicit attributes such as gender or age group. This ambiguity poses a significant challenge for query intent detection models in e-commerce search systems, which must accurately infer latent user intent (e.g., age, gender) to support effective downstream retrieval. We introduce IntentTune, a framework for resolving ambiguous or under-specified query intents by leveraging either (1) user-specific behavioral signals including search history, browsing activity, and profile attributes or (2) population-level demand patterns aggregated across all users. Through experiments on real-world e-commerce data, we first demonstrate that population-level demand patterns alone are insufficient to reliably infer intent in under-specified queries. We then demonstrate that user-specific behavioral signals -- particularly prior search queries -- outperform both population-level statistics and static profile information for inferring gender, age group, product category, and size intent from underspecified queries.

Small multirotor aircraft are increasingly tasked with operations in the atmospheric boundary layer, where turbulent winds comparable to the vehicle's airspeed degrade trajectory tracking and can defeat conventional feedback control. This work illustrates a two-stage learning pipeline that first estimates the local wind from onboard kinematics and dynamics and then exploits that estimate inside a reinforcement learning (RL) flight controller. The wind estimator, an attention-augmented gated recurrent network trained on thousands of simulated flights through von Karman turbulence with power-law shear and veer, recovers the horizontal wind vector with a per-flight root-mean-square error of 0.40 m/s and a direction error of 3.2 degrees on unseen wind regimes, an accuracy near the floor imposed by unresolved turbulence, and generalizes to vertical ascent profiles with a skill score of 0.861 over a constant-wind reference. A proximal policy optimization controller receiving the frozen estimator's output reduces horizontal trajectory tracking error by 48% relative to a wind-blind proportional-derivative baseline across mean winds of 4 m/s to 12 m/s, winning on 100% of evaluation episodes. A three-way ablation decomposes this improvement into a kinematic component, available without wind information, and a wind-perception component; the perception share rises with wind speed, from small in light winds toward roughly half the total benefit in strong winds, consistent with the quadratic scaling of aerodynamic drag. The controller degrades gracefully on out-of-distribution winds of 13 m/s to 15 m/s, where the baseline fails catastrophically.

Room embeddings derived from reverberant speech are often unreliable: speech content and recording degradation can alter the representation even when speaker, room, and source-receiver geometry remain unchanged, degrading downstream task performance. We propose a framework that learns room embeddings robust to speech-content variation and a representation-level uncertainty score from reverberant speech without downstream-task supervision. The embedding is anchored to a structured room impulse response (RIR) latent space and trained using a multi-view data structure with Kullback-Leibler (KL)-based alignment; a multi-positive contrastive term further refines robustness. A lightweight uncertainty head is calibrated using the dispersion of corruption-induced embeddings and optimized with a rank-based objective. Across waveform- and spectrogram-level corruptions, the score is consistent with representation dispersion and enables effective selective prediction while requiring only a single utterance at inference.

Vulnerabilities inherent to the fabless semiconductor manufacturing model have significantly increased the risk of malicious Hardware Trojan (HT) insertion, posing severe threats to hardware security. Several HT mitigation and detection strategies have been developed, and existing works explore the insertion of HTs in the space between standard cells in an integrated circuit. However, there is a lack of research into the vulnerabilities posed by the building blocks of most digital designs on the market today, the standard cells. This work investigates a novel threat model in which standard cells are considered untrusted. Our proposed threat model provides the design house with a tampered standard cell library. The intended netlist is synthesized and implemented using the tampered library. During fabrication, a nefarious foundry replaces the library's deactivated HT cells with activated counterparts. Using open-source and industry-standard Electronic Design Automation (EDA) tools, existing standard cell libraries, Saed32nm and Sky130nm, are converted into malicious libraries capable of masking the presence of arbitrary HTs from IC designers. The malicious library is then applied and characterized in multiple standard benchmark designs. To demonstrate the efficacy and stealthiness of this standard cell-based attack vector, three benchmark circuits, an AES-128 encryption core, an Ethernet controller, and a WISHBONE DMA engine, were synthesized using both clean and Trojan-infected libraries across Synopsys 32nm and SkyWater 130nm technologies. Design-level features, including total cell count, total area, dynamic power consumption, and static power, were extracted from these synthesized circuits to serve as inputs for binary classification

This monograph provides an introduction to mean field reinforcement learning through the lens of Markov decision processes arising from large-population stochastic control with mean field interactions and common noise. Starting from the connection between multi-agent reinforcement learning and mean field control, it develops the probabilistic, mathematical, and control-theoretic framework needed to formulate representative-agent learning problems, analyze their relationship with finite-population systems, and study both general and linear-quadratic models. The presentation includes dynamic programming principles, propagation-of-chaos limits, and theoretical analyses of tabular Q-learning and policy-gradient methods. It also discusses numerical implementations, including tabular schemes and deep reinforcement learning methods such as deep deterministic policy gradient. The goal is to give readers a coherent bridge between mean field control theory and reinforcement learning methodology, emphasizing the mathematical structure of the problems and the design of tractable learning approaches for large stochastic populations.

Recurrent memory agents extend LLMs to arbitrarily long contexts by iteratively consolidating input into a fixed-size memory window. Despite their scalability, these agents exhibit a well-documented reliability problem: end-to-end performance degrades systematically as context length grows. We diagnose this failure by decomposing performance into two factors--memory capture and memory retention--and quantitatively confirm that retention is the dominant bottleneck. Retention collapses because existing designs maintain memory as a monolithic text block, forcing every update to risk overwriting previously retained content. Motivated by this diagnosis, we propose Multi-Head Recurrent Memory (MHM), a general, training-free framework that partitions memory into independent heads governed by a stage-wise select-then-update strategy. At each step, exactly one head is selected for update while the remaining heads are structurally shielded from overwriting, shifting the burden of retention from model behavior to architectural design. As a lightweight instantiation, we introduce Least-Recently-Updated MHM (MHM-LRU), which guarantees uniform head utilization with zero additional token overhead. Extensive experiments on long-context benchmarks show that MHM-LRU substantially improves both retention and end-to-end accuracy across the 100K--1M token range, where baselines degrade sharply. On RULER-HQA at 896K tokens, MHM-LRU improves the memory retention rate from less than 30% to 73.96%. These gains generalize across model families, scales, and task types, positioning architectural optimization as a practical and cost-efficient path toward reliable long-context recurrent memory.

Mode shape recognition is a fundamental task in automotive NVH development, yet it remains dependent on manual visual inspection by experienced engineers. Existing approaches based on engineering heuristics, Modal Assurance Criterion (MAC), or geometry-dependent AI representations often exhibit limited robustness across different vehicle architectures, finite element (FE) meshes, and experimental measurement layouts, restricting their industrial applicability. This paper presents a Canonical Engineering Graph Representation and region-aware graph learning framework for robust and explainable 3D mode shape recognition. Rather than learning directly from vehicle-specific FE meshes, heterogeneous FE models and experimental measurements are transformed into a common graph whose nodes represent semantically meaningful structural regions connected through engineering-informed relationships. Geometry-independent regional descriptors are combined with graph attention learning and region-aware pooling to capture structural interactions while preserving engineering semantics and enabling physically interpretable predictions. The resulting representation decouples engineering knowledge from numerical discretization, allowing transfer across different vehicle programs without requiring identical mesh topology or sensor configurations. The proposed framework is validated using FE and experimental datasets from four vehicle programs under severe label scarcity. Results demonstrate high classification accuracy, cross-vehicle transferability, and physically meaningful explanations by directly relating predictions to engineering-defined structural regions used in NVH analysis. Beyond mode shape recognition, the proposed Canonical Engineering Graph Representation provides a reusable engineering abstraction for trustworthy and transferable AI across heterogeneous simulation and experimental workflows.

Transformer inference on long sequences is expensive because softmax attention repeatedly reads from a large KV cache. The prevalent approach to this bottleneck is KV cache compression, which replaces the full cache with a compact summary. Despite its practical importance, the design of such summaries is largely driven by empirical experimentation. On the theoretical side, existing results show that KV cache compression can be impossible in the worst case, but offer little systematic guidance for designing algorithms in regimes where accurate compression is possible. We bridge this gap by characterizing the minimax risk of KV cache compression in terms of the intrinsic compressibility of a cache, revealing when and how accurate compression is possible. These results yield novel design principles for KV cache compression under causal masking that map efficiently to prefill and autoregressive decoding while achieving minimax-optimal risk. We instantiate these principles in a practical algorithm and report promising performance on LongBench in targeted experiments. Overall, our results provide a principled avenue for practical KV cache compression with theoretical guarantees.

How far can a language model improve under a strict artifact budget? Parameter Golf posed this question as an open community challenge in which participants trained the best language model, with the complete artifact (training code + compressed weights) required to fit within 16 MB and be trained in under ten minutes on 8xH100 SXM GPUs. Quality was measured in bits-per-byte (BPB), the average number of bits required to encode each byte of unseen text. We analyze 2,037 pull requests and 1,430 clean scored submissions from the contest, build a taxonomy of 84 optimization techniques, and measure each technique's contribution to BPB. The verified leaderboard score dropped from 1.2244 to 1.058 BPB across three phases -- a 13.6% reduction, despite individual techniques rarely improving BPB by more than 1%. We show that most gains in techniques shrink across competitive submissions, isolating the few methods that improve performance across stacks.

Chain-of-thought (CoT) reasoning has emerged as an effective approach for activating latent reasoning capabilities in large language models. However, most existing CoT methods use reasoning chains mainly as inference-time prompts, while the generated reasoning traces are rarely reused as semi-supervised learning signals. In this report, we define \textbf{Semi-supervised Chain-of-Thought Learning} and propose \textbf{Semi-CoT}, a simple framework that uses unlabeled questions to construct pseudo reasoning supervision. Semi-CoT samples multiple pseudo-CoTs for each unlabeled question, estimates answer-level semantic entropy, and selects low-entropy reasoning chains as reliable pseudo-CoT demonstrations. This extends the self-training view of CoT from inference-time refinement to semi-supervised pseudo-supervision. Pilot experiments on AQuA, SVAMP, GSM8K, and MultiArith show that the entropy gate selects high-precision pseudo-CoTs, with pseudo-answer precision ranging from $91.36\%$ to $100\%$. Semi-CoT also gives small gains on SVAMP and GSM8K, while AQuA shows negative transfer and MultiArith reaches a ceiling. These results suggest that unlabeled questions can provide reliable pseudo reasoning signals, but their effective use still requires stronger demonstration selection or student training.

AI agents that autonomously execute tool calls on a user's behalf raise pressing questions about permission management: what role could users play, and what role should they play? Despite many proposed approaches, the user's role in agentic permission management remains under explored. We introduce Janus, a playground system for implementing and evaluating user-involved agentic permission management designs. Janus consists of two components: Janus-Core, a modular agentic system supporting a diverse spectrum of permission management designs, and Janus-Harness, an automated evaluation framework. Grounded in a conceptual model that identifies key design axes for user involvement, we implement six permission assistants spanning the design space and evaluate them across three scenarios and three synthetic responders. We demonstrate that user input is critical and can significantly strengthen privacy and security, that AI augmentation of user decisions can help reduce cognitive load, and that realistic user behavior including permission fatigue must be accounted for in system design. No single design performs optimally across all contexts, motivating a more principled and context-sensitive approach to deploying permission assistants in agentic systems. Janus is publicly available to support future investigation into this dimension of agentic system design.

Empirical research rarely admits a unique analysis. Different analytical choices can lead to different conclusions from the same data, yet these hidden forking paths are difficult to observe. We show that AI agents capture much of the analytical variation among human researchers while making these paths explicit. Across four high-stakes domains, assigning different personas is sufficient for AI agents to report divergent, often opposing, conclusions from the same data and question, with findings systematically aligned with those beliefs. In a study in which 42 human research teams analyzed the same immigration dataset, AI agents reproduced 72% of the human ideological gap in reported effect estimates. Despite reaching opposing conclusions, it is difficult to identify clear issues in each analysis based on the final AI reports: 86% passed independent AI review and 78% passed majority human expert review. These findings suggest that the central challenge is often not flawed analyses, but selective exploration and reporting from a large space of methodologically defensible analyses. AI agents may amplify this longstanding problem by making such exploration inexpensive and scalable. To address this, we introduce the m-value (multiverse value), the probability that an analysis path would produce a claim at least as extreme as the reported one. We further introduce Agentic Bootstrap, which estimates the m-value by using AI agents to sample plausible analysis paths. Applied to the human immigration study, 13.5% of reported human analyses fell in the most extreme 5% of the analysis space (m<0.05). Scientific evidence should therefore be evaluated not only by a single reported analysis but also by its position within the distribution of analyses that could reasonably have been reported. Agentic Bootstrap makes this distribution observable and turns it into a criterion for scientific credibility.

Rust's ownership type system prevents memory errors in safe code, but certain desirable properties remain orthogonal to compilation: the soundness of unsafe operations (e.g., raw pointer dereferences), functional correctness, and absence of runtime panics. We present Kani, an open-source model checker for Rust that pushes bounded model checking beyond bug-finding to provide correctness guarantees for these properties. Kani compiles proof harnesses from Rust's Mid-level Intermediate Representation (MIR) into CBMC's bit-precise verification engine, automatically checking a comprehensive set of safety properties with no user annotation. To extend verification from bounded to unbounded, Kani provides a specification language comprising function contracts, loop contracts, quantifiers, and function stubbing. We demonstrate feasibility through case studies on industrial Rust projects, where contracts upgraded verification from panic-freedom to functional correctness, uncovering six previously unknown bugs. Kani operates at scale in production CI, with over 16,000 harnesses verified per code change in the Rust standard library verification campaign.

Recent advances in automatic speech recognition (ASR) have explored different sequence models, including Conformer-based models and newer state space models such as Mamba. Although prior work has evaluated these architectures in multiple languages, their effectiveness in African languages remains underexplored. In this work, we evaluate Mamba for ASR on seven South African languages. In monolingual experiments, each model is trained on 50 hours of speech per language, and we compare Mamba to a Conformer baseline of similar parameter scale. Mamba achieves similar recognition accuracy to Conformer while using fewer computational resources and training faster. We further evaluate generalization in this setting and find that both models struggle to generalize to speech that is much longer than what they were trained on. We then study multilingual ASR using Mamba models, where the baseline is pooling all languages together. On top of this, we tested three extensions: training with language-family information by adding both language and language-family embeddings as biases to the downsampled acoustic representations, and multitask learning with a CTC ASR objective and a language identification (LID) head. We find that multilingual training consistently improves performance over monolingual training. However, adding explicit language information does not improve in-domain performance but does improve cross-corpus robustness. We conducted ablation studies in low-resource multilingual settings using 5-hour and 10-hour per-language training data, where we observed gains from using language embeddings and further demonstrated that removing or altering them hurt model performance. Lastly, we analysed these embeddings and find that they do not capture linguistic similarity in a typological sense, but instead act as task-specific control vectors.

We investigate the problem of learning useful policy representations (embeddings) in two-player zero-sum imperfect-information games. We make three contributions: First, we introduce methods of creating datasets of policies for a given game. Second, we propose methods to learn policy representations. Third, we introduce downstream tasks to evaluate the effectiveness of such representations. We evaluate each dataset method, embedding method, and downstream task on Kuhn and Leduc Poker. Although our methods are very basic, we demonstrate that useful behavioral representations are present in the learned embeddings. To our knowledge, this work is among the first to systematically compare self-supervised learning techniques for learning policy representations in games. Our code is available at https://github.com/VitamintK/ssl-project for others to extend.

Matter seeks to resolve longstanding interoperability problems in the Internet of Things (IoT), yet little is known about how developers experience the standard in day to day work. This paper examines over 13,000 issues from the official Project CHIP GitHub repository to understand the kinds of problems contributors report when implementing and integrating Matter. Using topic modeling and qualitative analysis, we identify four recurring areas of concern, Testing, Interoperability, Development, and Platform and Network, and describe how they manifest in the evolution of the codebase and tooling. The findings reveal systematic technical and integration challenges and point to concrete opportunities to refine Matter's test infrastructure, cross vendor guidance, and documentation as the standard continues to mature.

Recent work has established a fundamental trilemma between Byzantine robustness, local differential privacy (LDP), and optimization error in distributed learning. We show that this trilemma does not universally extend to generalization error, but instead depends critically on the privacy regime. Specifically, in the high-noise regime (strong privacy), we prove that increasing privacy reduces the generalization error, i.e., there is no tension between robustness and privacy. In the low-noise regime (weaker privacy), however, the tension between robustness and privacy reappears and increasing privacy indeed degrades generalization. Our theory explains this surprising non-monotonic behavior of the generalization error via matching lower and upper bounds on the algorithmic stability of Byzantine-robust distributed learning under LDP constraints. We corroborate and further analyze these theoretical findings with empirical evaluations.

Reinforcement learning post-training dramatically improves LLM reasoning, but suffers from training instability and diversity collapse. Advantage functions offer an appealing fix: they reshape the training objective, reweight which rollouts drive learning, and are trivial to implement. Yet a proliferation of methods makes it unclear which advantage to use and when. We cut through the confusion with a unifying framework that decomposes any advantage into its positive and negative gradient mass along two orthogonal axes. On the sign axis, imbalanced updates collapse either entropy or weight geometry. On the difficulty axis, hard-problem focus sharpens signal but costs sample size. Both trade-offs shift during training: exploration favors balance and hard focus; exploitation favors suppression and medium focus. This motivates FADE (Focal Advantage with Dynamic Entropy), a self-adapting advantage that reads training dynamics to schedule the gradient weight automatically. FADE reaches peak pass@1 20k steps earlier than the best static baseline at the 7B scale and 2k steps earlier at the 32B , while achieving the best accuracy-diversity trade-off across all pass@k on LiveCodeBench and AIME.

We propose a scaling law that takes into account model size and training data while explicitly splitting the latter into training steps and batch size (called three-term law). Fitting the proposed law on a large set of training runs, we find that it correctly recovers the scaling of the optimal batch size. Moreover, because it makes use of training runs with suboptimal batch size, our proposed law can be robustly fit with a significantly smaller amount of training runs. We further show that the three-term law can be used to derive scaling laws for suboptimal batch sizes, and that it matches previous empirical findings related to the critical batch size.

Large-scale distributed file systems must provision resources for peak demand, yet file access patterns fluctuate significantly, leaving substantial capacity idle during off-peak periods. Existing scaling mechanisms operate at the granularity of entire servers and take minutes to hours, making them unable to track the rapid, fine-grained load variations that file systems commonly experience. Serverless computing, with its millisecond-granularity elasticity and pay-per-use pricing, offers a compelling alternative. We present SLFS, the first distributed file system built with serverless functions for both data and metadata operations. SLFS implements file services on top of key-value stores, keeping function operations simple and short, and introduces a novel multi-threaded, short-lived server design that overcomes the cold-start problem while maintaining low cost. A policy-enforcing coordinator efficiently maps files to function instances, scales the system elastically, and controls function lifetimes to balance performance and cost. SLFS can flexibly run on diverse storage backends -- from cloud-native services like S3 to user-managed key-value stores -- enabling configurable cost-performance trade-offs. Our evaluation shows that SLFS mitigates cold starts by 580$\times$ compared to the base serverless design and outperforms $λ$FS, EFS, and Ceph at up to 63%, 68%, and 63% lower cost, respectively.

We present a fully unsupervised Fast-Slow DSVDD detector for continuous State-of-Polarization monitoring on a deployed subsea cable. Trained without event labels, it ranks all five confirmed trawler contacts within the top 13 of 122,174 recordings and surfaces additional corroborated cable-contact events.

The ''Computational Life'' paper (Agüera y Arcas et al., 2024) argues that paired interactions in a computational soup are an effective way to find self-replicators. In this work, aided by recent developments in self-replicator detection, we explore the alternate hypothesis that self-replicators can be found at least as easily using simple mutation random walks in program space. We also explore the claim that capping the maximum ''depth'' and ''width'' of the ancestry tree stops self-replicators from emerging, showing instead that it merely stops self-replicators from taking over the soup.

Reinforcement learning with verifiable rewards (RLVR), along with recent selfdistillation variants such as SDPO, evaluates each rollout against a verifier and updates the policy from that episode-level signal. However, the richer procedural information in the rollout is rarely retained or reused. Across episodes and epochs, the model repeatedly encounters related problems under a changing policy, producing cross-episode signals that episode-local updates cannot capture: which strategies consistently pass verification, which failure modes persist, which patterns recur. We propose Procedural Memory Distillation (PMD), which converts these crossepisode signals into reusable procedural memory and distills it into the policy's weights during training. This memory functions as a training scaffold, absorbed into the policy itself, yielding a memory-free model at inference. PMD organizes the memory at three levels of abstraction: raw trajectories, self-reflected strategies and lessons, and higher-level behavioral patterns that recur across problems, all extracted online from the model's own trajectories. A memory-conditioned self-teacher draws on the accumulated experience to supervise the student on its own rollouts, enabling student to progressively internalize procedural knowledge within its parameters. The central design principle is co-evolution: the policy generates rollouts that update the memory, and memory shapes the supervision that updates the policy. Empirically, across Qwen3-8B and OLMo3-Instruct-7B, PMD improves over SDPO by 3.8-5.5% on SCIKNOWEVAL and 7.9-13.6% on LIVECODEBENCH. Co-evolution powers these gains: freezing either the memory or the policy trails PMD by more than 10% across SCIKNOWEVAL domains.

We measured quantization-induced decision-boundary changes using local logit-margin radii, first-order boundary displacement, normal variation, slice-boundary Jaccard distance, grid prediction changes, multiclass junction counts, and low-margin boundary-band flips. On the digits benchmark, 8-bit weight quantization preserved all test labels while producing boundary-mask Jaccard \(0.428\) on the PCA slice; at 4 bits, accuracy remained \(0.9733\), while boundary Jaccard rose to \(0.970\) and median local boundary shift reached \(0.0290\). Interpolation between adjacent quantization levels localized the visible reconfigurations at multiclass junctions, with 12, 34, and 17 triple-junction cells in the selected transitions. Calibration-to-test stopping reduced the digits held-out flip rate from \(0.0094\) to \(0.0022\) and boundary Jaccard from \(0.825\) to \(0.524\); the same stopping rule also reduced flips on MNIST and Fashion-MNIST. On official CIFAR-10 subsets, PTQ-W selected by accuracy gave 6-bit flip \(0.0367\) and boundary Jaccard \(0.184\), whereas boundary-aware stopping selected 8-bit flip \(0.0083\) and boundary Jaccard \(0.048\). On full CIFAR-10 with three seeds, 6-bit PTQ-W lost \(0.0029\) accuracy relative to float, changed \(5.3\%\) of held-out decisions, and changed \(24.5\%\) of low-margin boundary-band decisions. A fixed-bit boundary-gap rounding term changed the trade-off at 4 bits by reducing boundary Jaccard from \(0.457\) to \(0.435\) and boundary-band pair-order flip from \(0.3600\) to \(0.3558\), with an accuracy trade-off; the 3-bit stress test exposed the tuning limit of this surrogate. Calibration boundary Jaccard predicted held-out boundary Jaccard across PTQ-W and optimized rounding variants with \(r=0.947\)--\(0.994\).

Class imbalance poses a critical challenge in federated learning (FL), where underrepresented classes suffer from poor predictive performance yet cannot be addressed by standard centralized techniques due to privacy and heterogeneity constraints. We propose FedCGNM (Federated Class-Grouped Normalized Momentum), a client-side optimizer in FL that partitions classes into a small number of groups based on minimum within-group variance, maintains a momentum per group, normalizes each group momentum to unit length, and uses the summation of the normalized group momentums as an update direction. This design both equalizes gradient magnitude across majority and minority groups and mitigates the noise inherent in rare-class gradients. We further provide a theoretical convergence analysis explicitly accounting for time-varying resampling-rates. Additionally, to efficiently optimize these rates in small-client regimes, we introduce FedHOO, an X-armed-bandit (XAB) based algorithm that exploits federated parallelism that evaluates many combinations of two candidate rates per client at linear cost. Empirical evaluation on four public long-tailed benchmarks and a proprietary chip-defect dataset demonstrates that FedCGNM consistently outperforms baselines, with FedHOO yielding further gains in small-scale federations.

Clinical protocol-execution tasks -- checking a lab value, applying a threshold, placing a correctly structured FHIR order -- are natural candidates for RL from world feedback: once clinical SMEs encode decision logic into a verifier, that verifier grades unlimited rollouts without per-episode annotation. But applying RL requires a sound feedback channel and sufficient base capability. We audit MedAgentBench v1/v2, find a 41.7\% silent-finish ceiling that makes inaction the RL dominant strategy, and construct \textbf{MedAgentBench-v3 (MAB-v3)} (508 tasks, 8.9\% ceiling). Training Qwen3-8B exposes two structural barriers: a \emph{capability ceiling} (10/20 task types have 0\% base performance, zero gradient) and a \emph{format-knowledge barrier} (3/20 types require exact clinical codes undiscoverable by exploration). Pure RL reaches 18.2\% pass@1 vs.\ 34.1\% for rule-based SFT; the 15.9~pp gap is attributable entirely to these barriers. A decision/format-knowledge/lookup taxonomy predicts RL learnability and prescribes the fix: SFT to inject codes, RL to learn conditionals.

Large language models are trained to predict the next token, not to act inside a specific API. In niche enterprise SaaS workflows -- where success means hitting the right endpoint with the right nested arguments in the right order -- this objective mismatch shows up as silent failures: dropped required fields, hallucinated tools, or early stops after a single read. We ask whether Reinforcement Learning with Verifiable Rewards (RLVR), applied directly in the target environment, closes the gap. As a proof of concept we build a suite of five synthetic environments emulating the Jira REST v3 and Confluence v2 APIs at schema fidelity; rewards are computed entirely from the tool-call trace, with no live API, no learned judge, and no human label in the loop. Scoring prompted Qwen3-1.7B and Qwen3.5-4B on the same checkers that drive GRPO training, we find that on the four scenarios whose rewards are non-degenerate the RL-trained policy lifts average reward from a 4B-baseline range of 0.35--0.92 to 0.95--1.00, with the largest single gain on Confluence page creation ($0.35 \rightarrow 1.00$). We position this as a preliminary step toward outcome-optimised small models for niche enterprise APIs, and foreground two limitations a workshop reader should weigh: hand-crafting verifiable rewards does not scale beyond the handful of endpoints reported here, and one of our five scenarios (ticket-transition) has a saturating reward shape that the prompted 4B already maxes out.

Text scaling, the task of positioning political actors on an ideological scale, is a fundamental task in political analysis. To ease the need for manual analysis, various NLP methods have been proposed for this task, including classification- and regression-based approaches, showing successes as well as limitations. The goal of our paper is to consolidate the state of the art in this area. We ask two questions: (a) Can the performance of scaling methods be improved by predicting scales not individually but jointly? (b) Is there a middle ground between classification and regression?

Large language models (LLMs) are increasingly applied to resume optimization for applicant tracking systems, introducing hallucination failures distinct from general text generation: anachronistic technology injection, cross-domain terminology contamination, structural mutation, and content fabrication. We present Grounded Optimization, a five-layer framework combining temporal context validation, deterministic contamination detection, structural invariant enforcement, prompt-level grounding, and an evaluator agent. In ablation experiments across three LLMs, four temperature settings, and six layer configurations on 25 synthetic resumes spanning 14 industries, undefended baselines produce 2.48-5.36 detected hallucinations per resume. Among detectors independent of the active defenses, temporal hallucinations are reduced by 50-95% across all conditions; overall detected hallucination rate falls to 0.04-0.24. Prompt-level grounding alone achieves zero detected hallucinations at low temperature with a capable instruction-following model; higher temperatures and weaker models reveal the need for the deterministic layers as a complement. We release the contamination taxonomy, evaluation code, and raw data.

Agent Skills provide on-demand domain knowledge to LLM agents without requiring model retraining. Each Agent Skill is defined by a mandatory SKILL.md file containing metadata and an unstructured Markdown body whose contents are left entirely to the skill author. Despite the rapid adoption of Agent Skills, little is known about how these files are authored or whether existing authoring guidelines are followed in practice. In this paper, we present the first systematic study of SKILL.md files as a software artifact. We qualitatively analyze 238 real-world skills and derive a taxonomy of 13 higher-level and 44 lower-level semantic components. We then conduct a multivocal literature review of 29 sources to identify best practices for authoring SKILL.md files and introduce skill smells as violations of these practices. Finally, we develop an automated detector and apply it to real-world skills, finding that over 99% of SKILL.md files contain at least one skill smell, and once introduced, skill smells rarely disappear as skills evolve. These findings reveal a substantial gap between recommended and actual authoring practices, motivating the development of automated techniques to remediate skill smells while increasing developer awareness of this emerging quality issue.

Language models learn continuous programs over discrete symbols, with the embedding table and LM-head acting as the read/write interface between them. We show that this interface has gradient geometry distinct from dense hidden weights which can be exploited to improve the Pareto frontier across supervised finetuning, RL, and pretraining, while only utilizing kilobytes of optimizer state. We introduce Ember, a lightweight optimizer for embedding and LM-head matrices that utilizes O(V + D) VRAM, instead of Adam's O(2VD), and forgoes the need to shard both token table optimizer states. We provide empirical evidence that Ember scales effectively across batch size and parameter count. We show that the optimization trajectory of tokens can be well described by a simple 1D ray, counter to the popular belief that neural net parameters navigate a heavily nonconvex landscape. We provide a principled view on the surprisingly narrow space of optimizers that suffice for Transformer training. Finally, we open-source our distributed Ember implementation that merges cleanly with existing ZeRO/FSDP setups to support further research at https://github.com/katop1234/ember

We propose a novel hybrid neural architecture, the Geometry-aware R-Structured Kolmogorov-Arnold Network (GRS-KAN), which integrates V.L.Rvachev's R-functions into the Kolmogorov-Arnold Network (KAN) framework. The proposed approach combines two complementary modeling mechanisms: smooth nonlinear structure is learned by KAN branches, while known geometric or logical constraints are encoded analytically using differentiable R-functions. This enables explicit representation of discontinuities, feasible regions, and implicit geometric boundaries within a trainable neural architecture. The framework implements differentiable logical operations through R-conjunctions and R-disjunctions, allowing complex geometric supports to be represented analytically and incorporated directly into regression models. Several GRS-KAN variants are introduced, including additive, multiplicative, and agnostic branch-weighted architectures. The method is demonstrated on regression problems involving discontinuities with circular and rectangular supports. Numerical experiments show that explicit geometric encoding substantially improves predictive accuracy and boundary localization compared with standard KANs. In the considered benchmarks, geometry-aware GRS-KAN models reduce test RMSE by up to 67% while simultaneously improving interpretability through explicit analytical representation of the learned geometric structure. The agnostic variant further demonstrates the ability to automatically determine whether geometric priors are beneficial for a given learning task.

Malware poses a critical and ever-evolving threat, and robust and effective systems for detecting and classifying malware are of essential importance. $n$-grams features are among the common static features used in effective machine learning systems for malware, but these features are inherently brittle. We propose an algorithm for constructing more robust features, hamm-grams, which are a special class of regular expressions having a fixed length and single-character wildcards. We devise an efficient algorithm for finding common hamm-grams using a new locality-sensitive hash designed to produce collisions among pairs of small Hamming distance and a clustering within hash buckets to place wildcards. We then demonstrate the advantages of these features in malware classification and detection tasks.

Mixture-of-Experts (MoE) models offer inference speedups via selective activation but impose substantial memory requirements because the whole network must remain loaded. Structured expert pruning is a practical approach for reducing deployment costs in resource-constrained settings. However, prior studies primarily evaluate benchmark utility, leaving the effect of pruning on factual reliability underexplored, particularly in high-stakes domains such as biomedicine. In this paper, we investigate how domain-specific expert pruning affects both utility and reliability. We assess four MoE models, six pruning methods, and multiple pruning ratios across generation and classification tasks under in-domain (biomedical) and cross-domain settings. Results reveal that moderate pruning preserves in-domain utility without immediate reliability decline, although hallucination risks increase at extreme pruning ratios. When shifting to the general domain, both utility and reliability degrade rapidly. These findings indicate that safe compression depends heavily on the task and domain. Evaluating pruned MoE models solely on utility is inadequate for high-stakes deployment without reliability assessment.

Faithful reasoning is essential in medicine, where clinical decisions require transparent justification grounded in reliable evidence. Current medical LLMs either lack active access to evidence or use retrieved evidence without supervising how it should be appraised and applied during reasoning. To address this, we formalize evidence-based medicine principles as process-level criteria and introduce FaithMed, a framework that combines clinician-designed, automatically refined rubrics with reinforcement learning using step-level process reward assignment and advantage grouping. Across seven medical benchmarks, FaithMed improves over agentic-search baselines (+9% on average) and outcome-only RL (+5.8%), while raising average evidence-based medicine rubric scores over agentic-search Qwen3 baselines (+15.5%). This work demonstrates that explicit step-level supervision can improve both task success and the faithfulness of the reasoning process. Code is available at https://github.com/cxcscmu/FaithMed.

Diffusion language models, which generate text by denoising a token canvas bidirectionally instead of emitting tokens left to right, have become competitive with autoregressive (AR) generation. Medical foundation models, however, remain almost entirely autoregressive. We adapt a mixture-of-experts diffusion language model, DiffusionGemma-26B, and benchmark it against its same-size AR sibling Gemma-4-26B under an identical LoRA recipe on medical visual question answering datasets, scored by a verbosity-robust LLM judge. Diffusion matches or exceeds AR on all of them, and the finetuned model (3.8B active) is competitive with frontier vision-language models; its decoding is also 3.5-4.4x faster. Beyond this parity, the diffusion model offers a drafting capability AR lacks: any-order infill. Because the canvas is denoised bidirectionally, a radiologist can fix report fragments and have the model fill the text between them, an operation inherent to diffusion but not to autoregression, which is subpar at it. This suits real reports, which are often terse or inconsistent across clinicians and institutions.

Significant advancement of immersive technologies such as Virtual and Augmented Reality (VR/AR) and their integration into diverse aspects of modern life need authentication interfaces that are secure, intuitive, and compatible with embodied interaction. Traditional methods such as passwords, PINs, and device-based logins, break immersion and rely on external hardware. Recent 3D-specific behavioral approaches, such as hand-gesture, eye-tracking, and electroencephalography (EEG)-based methods, offer promising alternatives but often require specialized sensors or constrain natural movement, limiting usability in dynamic environments. We present Sign in the Air to Unlock, an in-air signature interface that enables users to authenticate by signing naturally in 3D space which is a familiar, personal, and reproducible gesture. To realize this interface, we design a point-voxel Cross-Attention Network (PV-Net) that jointly models local motion dynamics and global spatial structure from 3D trajectories. The model is evaluated on two datasets: the public DeepAirSig dataset (1,800 signatures from 40 users) and ImmAirsig, a new dataset collected using Meta Quest 2 in immersive VR (880 samples from 22 users). PV-Net achieves an Equal Error Rate of 2.5% on DeepAirSig and 76% classification accuracy on ImmAirSig. These findings highlight the potential of 3D behavioral interfaces for seamless, user-centric authentication that merges security with natural interaction in immersive environments.

Divergent thinking is a crucial aspect of creativity, yet large language models (LLMs) tend to consistently generate similar responses to open-ended questions, in what has been termed the artificial hivemind effect. Here, we introduce CreativityNeuro, a data-free method for enhancing divergent thinking in LLMs via contrastive weight steering. We evaluate our method across multiple creativity assessments and report several main findings. On the Divergent Association Task (DAT), a vocabulary-space creativity test, CreativityNeuro improves performance by up to 14 human percentile points. Next, in a large-scale human evaluation (N=720) on the Alternative Uses Test (AUT) and the Task Task, CreativityNeuro achieves significant improvements in originality, surprise, and creativity, transferring to longer-form and more open-ended tasks. Importantly, we find that across all three tasks, CreativityNeuro demonstrably reduces measures of mode collapse. Moreover, activation steering achieves comparable performance to CreativityNeuro on the DAT, but it does not transfer to the AUT and Task Task, demonstrating the effectiveness of weight-space steering in generalizing to unseen tasks. In conclusion, CreativityNeuro improves divergent thinking and reduces mode collapse without requiring behavioral data, re-training, or gradient-based fine-tuning, providing a straightforward way to enhance LLM performance in creative domains.

We introduce ISOSCI, a benchmark of isomorphic cross-domain science problem pairs that separates reasoning ability from domain knowledge retrieval in LLM evaluation. Each pair shares identical logical structure but requires different domain-specific knowledge, enabling controlled attribution of reasoning-mode gains. Across five model pairs spanning four model families, we find that 91.3% of reasoning-mode gains are knowledge-dependent rather than structure-invariant (63/69 gains; Wilson 95% CI [82.3%, 96.0%]), directly challenging the assumption that chain-of-thought reasoning improves short-horizon procedural scientific problem-solving. Reasoning toggles on highly capable models provide less than 5 percentage points accuracy gain across all domains, and a reasoning-specialized model (o3-mini) that outperforms its standard counterpart on GPQA Diamond (+19.2 percentage points) underperforms on ISOSCI (-24.7 percentage points), showing that benchmark choice determines conclusions about reasoning utility. We release ISOSCI at https://huggingface.co/datasets/isosci/isosci

As many-core Systems-on-Chip (SoCs) continue to scale, Networks-on-Chip (NoCs) must sustain increasingly high memory bandwidth while preserving deadlock freedom. In AXI4 systems, protocol-level dependencies between read and write traffic can create circular waits at the network endpoints, even when the routing algorithm itself is deadlock-free. Decoupling these traffic classes avoids such dependencies, but exposes a key implementation trade-off: multiplane NoCs duplicate wide physical links and increase routing pressure, whereas conventional Virtual Channel (VC) routers add substantial control complexity, area, and timing overhead. This work revisits this trade-off for modern wide-link NoCs. We evaluate four deadlock-free AXI4 traffic-class separation schemes: a multiplane baseline and three lightweight VC-based designs. Among these designs, we propose Preemptive VCs, a physically-aware architecture that can save up to 76% of link resources with comparable frequency and only 3% router area overhead relative to the multiplane design.

Autonomous customer-service agents are shifting from conversational interfaces toward operational execution roles: they retrieve firm records, apply service policies, and execute backend writes such as refunds, cancellations, exchanges, order modifications, and reservation changes. This shift creates a service-control problem: firms must keep routine service fast and low-friction while preventing operational errors on requests where customer instructions, policy constraints, firm records, and backend writes interact. We propose a difficulty-routed service-control architecture that asks when service agents should reconsider before acting. A lightweight router keeps routine sessions on a low-cost baseline path and routes operationally coupled sessions to an escalated workflow. The escalated path uses conflict-aware communication and write-triggered reconsideration to concentrate deliberation and safeguards before consequential backend writes, rather than applying additional control uniformly across all service sessions. We evaluate the architecture on human-verified retail and airline tasks from $τ^{2}$-bench. In retail, the method improves reliability consistently on service requests with operational conflict. Routing evidence shows that stronger control is directed toward conflicted requests rather than broadly applied to routine ones. Dialogue and tool-use profiles suggest that gains do not come from indiscriminate interaction expansion or broader tool chains; instead, added turns and tool calls support evidence gathering, write separation, and pre-write reconsideration. Case-level evidence shows that the escalated workflow preserves fallback plans, binds retrieved records to the correct action, sequences writes, and decomposes multi-entity requests. Airline results extend the same service-control logic to reservation operations.

Understanding large, complex codebases, especially those with obfuscated structures and incomplete documentation, remains a significant challenge. Existing code summarization solutions often rely on a single language model or coding assistant like Claude Code, and treat source code as flat text, underutilizing the rich interdependencies and hierarchical information within a repository. To address these shortcomings, we propose Agent4cs - a multi-agent framework that summarizes large codebases in a bottom-up fashion, where a summarization agent focuses on producing robust summaries; a keyword-extraction agent proactively identifies critical information from subfolders; and a quality-assurance agent iteratively refines the outputs for readability, coherence, and completeness. Evaluated on 7 frontier models, Agent4cs improves semantic consistency across all folder levels by average 8% compared to two structured prompting baselines with code segments. Furthermore, extensive evaluation on real-world datasets demonstrates up to 38% gains in normalized keyword coverage rate over the same baselines.

Engineering management research has produced mature frameworks for software risk: ownership by feature, escalation by severity, and assurance by test coverage. These frameworks implicitly assume deterministic behavior, discrete and auditable change events, and clear component-to-owner mappings. Teams that build and operate agentic AI systems violate all three assumptions at once: outputs are probabilistic, systems take autonomous multi-step actions, and the risk surface mutates silently between deployments. Existing AI risk literature addresses this from above (policy frameworks such as the NIST AI RMF and ISO/IEC 42001) or below (threat taxonomies such as OWASP's agentic AI guidance), but not at the layer where an engineering manager (EM) operates: roles, decision rights, and escalation structures. This paper contributes (i) a seven-dimension profile distinguishing pure software-engineering, hybrid, and AI-native teams; (ii) a six-cluster failure-mode taxonomy including a previously unarticulated cluster, dependency-boundary determinism mismatch; and (iii) a synthetic framework-adequacy methodology scoring how well each profile's risk architecture detects, contains, and escalates a defined scenario set. Because the object of study is framework adequacy rather than human behavior, the evaluation yields derived rather than observed coverage claims. Coverage degrades as teams move from pure software engineering to AI-native operation, monotonically in the median and abruptly in the count of uncovered, high-consequence failures appearing only at the AI-native step. The degradation concentrates in specific failure-mode categories, and the most severe, least-covered failures arise not inside AI-native teams but at the organizational boundary where their probabilistic outputs are consumed by determinism-assuming dependencies.

As grounded QA systems are increasingly deployed in AI assistants, accurately attributing generated answers to evidence is critical for user trust and model safety. While unimodal attributions have been explored in depth, the multimodal setting remains relatively under-researched. As a result, we introduce MultAttnAttrib, a training-free attribution-generation method that leverages a model's prefill pass, selected attention heads, and calibrated thresholds to locate source evidence within a document. To establish baseline results for the method, we introduce MultAttrEval, a complementary benchmark dataset annotated with fine-grained, ground-truth attributions for answer components grounded in multimodal source documents. To our knowledge, this is the first evaluation dataset designed specifically for multimodal attribution in long-form documents. Experimental results show that MultAttnAttrib consistently outperforms a variety of attribution-generation methods, including several strong prompting-based approaches and matches the latest frontier models such as GPT 5.4. Our method not only substantially improves attribution accuracy for both unimodal and multimodal attribution types, but also produces attributions at up to one-seventh of the direct inference latency compared to prompting on the same base model.

Organizations rolling out agentic command line tools like Anthropic's Claude Code and GitHub's Copilot CLI need to know who will try them, who will keep using them, and whether the tools produce enough output to justify their cost. At organizational scale, token spend can run into millions of dollars annually, so misreading adoption, retention, or impact can make a rollout expensive without changing engineering velocity. Studying tens of thousands of engineers at Microsoft over its early-2026 rollout, we find that first use spread primarily through social networks, retention was associated more with engineers' coding activity than with demographics, and adopters merged roughly 24% more pull requests than they would have otherwise. We use merged pull requests as our proxy for output -- acknowledging that a merged PR is not the same as the value it delivers -- and the lift persists across our four-month window. These results suggest that CLI coding agents are neither uniformly adopted nor mere novelty effects and that organizations should treat visible peer use as central to rollout strategy.

Conditional inference trees (CIT) and conditional inference forests (CIF) reduce split-selection bias by testing features before choosing split thresholds, but repeated permutation tests and threshold searches can make these methods computationally expensive. We study CIT and CIF as top-$k$ feature-ranking methods for downstream prediction using real-data benchmarks, runtime ablations, and synthetic feature-recovery experiments. At a fixed node, if the features and permutation budget do not depend on the node responses, Bonferroni-corrected $+1$ Monte Carlo permutation $p$-values control nodewise rejection under the complete permutation null. CIF ranks 4th among 17 classification methods on 22 datasets and 3rd among 18 regression methods on 8 datasets. With Bonferroni correction held fixed, the CIF runtime ablations indicate that adaptive stopping and the number of thresholds searched have the largest measured effect on runtime: turning off adaptive stopping and using exact threshold search increase fitting time by 4.0--8.4$\times$ and 1.9--10.8$\times$, respectively, while downstream score changes are at most 0.011. Sparse high-$p$ simulations indicate that forest feature sampling can leave informative features out of many split decisions. Overall, the results support CIF as a top-$k$ feature-ranking method in the evaluated downstream prediction benchmarks.

Coding-agent reinforcement learning treats execution infrastructure as a background implementation detail, despite relying on large numbers of interactive software rollouts. This is a missed opportunity: measuring infrastructure overhead can reveal practical efficiency gains for RL post-training, where small per-rollout savings compound at scale. We present a comparative study of four execution substrates: single containers, hosted sandboxes, Kubernetes-orchestrated containers, and cloud virtual machines. We find up to $110\times$ variation in cold-start latency and a $1.8\times$ spread in projected worker-hours for one million 150-step trajectories. Our results suggest that future coding-agent RL systems should optimize execution substrates as part of the training system itself, not merely as deployment plumbing.

Sim2real transfer for robot policy learning suffers due to mismatch between simulation and reality. Existing methods typically address each gap in isolation through separate adaptation modules, which are composed or layered when both gaps coexist. Yet the basis for attempting sim2real in the first place is that there is shared structure between a task in simulation and reality, where equivalent actions from equivalent configurations produce equivalent long term outcomes regardless of domain specific differences in rendering or physics. In this paper, we study whether we can identify and exploit this shared structure from raw observations to train a policy that enables zero shot transfer. We introduce BIFROST, which learns a shared history encoder on paired cross-domain data via cross-domain bisimulation objective: observation-action sequences leading to equivalent long-term behavior are mapped to nearby latent states, regardless of domain. Policies trained on these latent states in simulation transfer zero-shot to reality. We provide empirical evidence on sim2sim visual navigation and sim2real contact rich manipulation task and visual servoing task that BIFROST achieves effective transfer where domain adaptation and co-training baselines fail under both visual and dynamics domain gaps.

GPU training jobs fail often, roughly two in five on large production clusters, yet the operator typically learns of a failure only by reconnecting hours later. Experiment trackers require editing the training script and maintaining a cloud connection; the scheduler's mail hook delivers a single status line with no cause and no logs. GPUAlert is a command-line wrapper that monitors any training command at the process boundary, and with no change to that command, emails a structured notification on completion carrying a classified failure cause, durable logs, and output artifacts. The tool is organized around three reliability primitives: a pre-launch log guarantee that establishes the durable destination before the child process can crash, notifier isolation that makes the wrapper's exit code a pure function of the child's status regardless of whether the email succeeds, and a non-silent artifact budget that bounds attachment size without ever dropping output silently. We release a labelled corpus of 474 GPU training logs across 15 failure classes and a reproducible evaluation harness. On the twelve hardware-reproduced classes, the ordered-rule classifier reaches 0.997 macro-F1, against 0.830 for unordered keyword matching and 0.133 for exit-code inspection. Wrapper overhead is a constant approximately 3ms per job; the pre-launch guarantee preserves a log where a shell redirect yields nothing; and across all 15 failure modes the wrapper returns the child's exit code unchanged even when the SMTP relay is unreachable.

$\mathrm{E}(3)$-equivariant networks are promising for 3D atomistic system modeling, yet their scalability is limited by the $O(L^6)$ complexity of the Clebsch-Gordan Tensor Product (CGTP). The recently proposed Gaunt Tensor Product (GTP) reduces the complexity but is unable to capture the antisymmetric paths, resulting in incomplete expressivity. In this work, we present SpinGTP, an approach to overcome the GTP incompleteness by generalizing from scalar functions to Spin-Weighted Spherical Harmonics (SWSH). By relying on the algebraic properties of SWSH, SpinGTP recovers the missing antisymmetric interactions while maintaining the asymptotic efficiency of GTP. It also allows for a more expressive equivariant basis that naturally accounts for the parity-odd components of tensor products. We evaluate SpinGTP across diverse benchmarks, including Tetris, 3BPA, SPICE-MACE-OFF, and OC20. Our results show that SpinGTP achieves accuracies comparable to full CGTP. Notably, by explicitly capturing antisymmetric paths, SpinGTP exhibits superior performance in tasks involving chiral materials and non-centrosymmetric geometries. This work provides a complete, scalable, and mathematically rigorous path toward high-order equivariance in large-scale 3D atomistic system simulations.

INTRODUCTION: Accurate MRI-based identification of Alzheimer's disease (AD), mild cognitive impairment (MCI), and related dementias remains challenging because disease-related structural changes are often subtle and heterogeneous. We developed NeuroBridge, a clinically guided multi-task MRI framework for neurodegenerative disease diagnosis. METHODS: NeuroBridge integrates large-scale self-supervised MRI pretraining with hippocampal segmentation, hippocampal atrophy classification, and reconstruction objectives, followed by gated fusion fine-tuning. Performance was evaluated across ADNI and OASIS cohorts, including cross-cohort transfer, probability-based analysis, and opportunistic screening. RESULTS: NeuroBridge achieved the highest performance across evaluated classification tasks, reaching 88.17% accuracy for AD versus cognitively normal controls in ADNI and 82.78% in OASIS. The largest gains occurred in MCI-related and mixed-diagnosis settings. The framework demonstrated strong cross-cohort generalization, systematic associations between predicted-class probability and accuracy, and the feasibility of probability-based opportunistic screening. DISCUSSION: Clinically guided multi-task representation learning improves neurodegenerative MRI diagnosis beyond conventional single-task approaches. NeuroBridge provides a robust and scalable framework for dementia assessment and MRI-based opportunistic screening.

Deep multimodal brain-encoding models now predict fMRI responses to naturalistic video with high accuracy. Whether their predicted neural signals also forecast behavioral engagement is unknown. We run TRIBE, the winning model of the 2025 Algonauts brain-encoding challenge (Llama-3.2 + V-JEPA2 + Wav2Vec-BERT), on 48 YouTube videos and reduce its predicted cortical response to a per-second engagement curve, the global field power. Correlated against each video's "most replayed" heatmap, a passively-collected proxy for which moments viewers return to, the curve shows no evidence of predicting re-watch behavior. The pooled position-controlled partial correlation is +0.058 (95% CI [-0.04, 0.15]; one-sample t(47)=1.21, p=0.23), indistinguishable from zero and not significantly above simple loudness and motion baselines (loudness +0.04, paired p=0.74). The raw correlation is also near zero; the moderate values reported for music videos reflect a genre-specific intro/onset-replay artifact rather than content prediction, and do not generalize. The null holds across six cortical-network readouts and under an autocorrelation-preserving permutation test. We release the code, the video-ID manifest, and an acquisition method that works despite YouTube's SABR-only streaming.

At the heart of human visual perception lies the ability to maintain a continuous and coherent understanding of the external world. By integrating observations with accumulated experience, the human visual system can continuously adapt to variations in both the target and its surrounding environment, while preserving robust visual continuity as scene dynamics evolve. Human vision can therefore integrate prior knowledge, spatial geometry, and semantic context to understand complex scenes and their changes. As a core problem in computer vision, visual object tracking aims to bring machine perception closer to human visual perception. These capabilities are central to the task of Generic Object Tracking (GOT). In this task, a visual tracker is initialized only with the bounding box of an arbitrarily specified target in the first frame, and must continuously localize the target in subsequent dynamic visual streams. However, future events, observations, and real-world variations are inherently unpredictable; therefore, the model's generalization and online adaptation capabilities remain bottlenecks. Tracking reliability can deteriorate when the target undergoes severe deformation, is affected by complex distractors, encounters significant environmental changes, or belongs to a category unseen during training. This dissertation aims to narrow the gap between machine visual tracking systems and human visual perception by proposing a series of methods that systematically enhance the target discrimination, robust adaptation, and geometric reasoning capabilities of tracking models.

We present Wiola, a fully original Small Language Model (SLM) architecture built from first principles, sharing no structural lineage with any existing model family including GPT, LLaMA, Mistral, or Falcon. Wiola introduces five independently novel components: (i) Spiral Rotary Positional Encoding (SRPE), which embeds token positions on a three-dimensional helical manifold combining absolute, relative, and hierarchical positional signals; (ii) Gated Cross-Layer Attention (GCLA), providing each decoder layer with soft cross-attention access to compressed summaries of two preceding layers for inter-layer coherence; (iii) Adaptive Token Merging (ATM), which dynamically merges se mantically redundant adjacent tokens in middle network layers to reduce attention complexity without information loss; (iv) Dual Stream Feed-Forward (DSFF), replacing the conventional MLP with two parallel streams fused by a learned per-dimension gate; and (v) WiolaRMSNorm, a modified normalisation introducing a per-dimension learned offset vector that prevents representation collapse. We provide complete mathematical derivations, architectural block diagrams, complexity analyses, and systematic comparisons against GPT-2, LLaMA-2, and Mistral. Wiola is released in four sizes (120M, 360M, 700M, and 1.5B parameters) and is fully compatible with the HuggingFace Transformers ecosystem, with all 22 architectural unit tests passing.

Aligning large language models with diverse and heterogeneous human values requires multi-objective alignment methods to effectively trade off conflicting preference dimensions. Current methods achieve this trade-off by training policies conditioned on preference vectors and leveraging online direct preference optimization. However, exploration uncertainty can cause the reward distributions of responses generated under different preference vectors to overlap, and the generated responses may fail to effectively align with the corresponding preference vectors. In this paper, we propose Multi-Objective Exploration and Preference Optimization via Mutual Information (MI-EPO), an information-theoretic framework. It unifies multi-objective exploration and alignment by maximizing the joint conditional mutual information among generated responses, preference feedback, and preference vectors. By incorporating a probabilistic routing mechanism, MI-EPO naturally decomposes objective alignment and preference-aware exploration, encouraging the model to generate responses that are distinguishable and aligned with different preference conditions. Experiments on safe alignment and helpful assistant tasks show that MI-EPO significantly improves the alignment between generated responses and preference vectors, makes the outputs more controllable, and achieves stable trade-offs across multiple objectives.

How do we encode numeric values in transformer-based sequence processing, particularly in electronic health record (EHR) data? We systematically compare discrete, continuous, and hybrid value encoding strategies using synthetic arithmetic tasks embedded within real-world EHR data, as well as real-world clinical prediction tasks. Our study reveals trade-offs between numeric precision, optimisation stability, and architectural flexibility. We find that approaches that explicitly model value-concept interactions perform best on precision-sensitive arithmetic tasks when architectural constraints permit. Hybrid token-based approaches that retain numeric values but apply binning prior to projection provide a more robust and broadly applicable alternative, with the optimal number of bins following a simple empirically derived power-law in dataset size. Across tasks, models consistently exhibit reliable "good enough" numeric computation rather than exact arithmetic, while clinical gains from incorporating laboratory values are task-dependent. This suggests that robustness and deployability often outweigh maximal numeric precision in practice, motivating hybrid token-based approaches as a practical default.

Multi-step symbolic reasoning is essential for robust financial analysis, yet most benchmarks neglect intermediate reasoning steps. FINCHAIN introduced verifiable Chain-of-Thought (CoT) evaluation but is limited to English. FINESSE-Bench includes a Russian block but relies on multiple-choice questions without step-level supervision. We present RusFinChain, the first Russian-language symbolic benchmark for verifiable CoT reasoning in finance. It spans 17 domains, 172 topics, and comprises 5,280 parameterized examples from executable Python templates, ensuring contamination-free evaluation. Each example includes a gold-standard reasoning chain with intermediate numeric values for automatic verification. We also introduce enhanced metrics: Fuzzy Numeric Alignment and Soft-Attention Alignment. We evaluate 8 open-weight LLMs on a stratified sample, generating 8,100 responses. Results reveal a substantial reasoning gap: models achieve Hard F1 of ~0.65 for step alignment, but only ~29% of final answers are correct. Our fuzzy and soft metrics show stronger correlation with final-answer correctness (Spearman rho approx 0.48) than the original ChainEval (rho approx 0.38-0.46), demonstrating superior diagnostic power. We release dataset, code, and evaluation framework to foster verifiable financial AI for the Russian-speaking community.

Recommender systems are vital in helping users navigate vast amounts of information, offering personalized suggestions and effective explanations for these recommendations. While previous efforts have attempted to provide such explanations, evaluating their effectiveness across various scenarios remains a challenge. Enhancing these explanations is essential for improving user engagement, trust, and decision-making. To facilitate effective explanations within the recommender system, we propose a Bi-level Neural Architecture Search (Bi-NAS) framework to optimize explanations. This approach simultaneously refines cross-attention mechanisms and feature interaction functions by exploring both intra-layer and inter-layer design spaces. Furthermore, we integrate Large Language Models (LLMs) to enhance explanation generation, leveraging zero-shot prompting to produce more effective and personalized justifications. By aligning user feature preferences with item quality scores, our approach ensures that explanations reflect both user intent and item attributes, improving transparency and reasoning depth. Extensive evaluations on four real-world datasets demonstrate that Bi-NAS not only boosts recommendation accuracy but also significantly improves the effectiveness of explanations for recommender systems, providing users with clear and reliable insights into the suggestions they receive.

Gravitational Waves (GWs) represent the newest window of astronomy, furthering our understanding of compact objects like black holes and neutron stars in the Universe. The signal from two merging neutron stars is especially interesting since it brings the prospect of concordant electromagnetic and neutrino emissions. Such multi-messenger observations have a transformational impact on fundamental physics, nuclear matter, astrophysics, and gravity. It was first witnessed in 2017 with the detection of the binary neutron star (BNS) merger GW170817. However, searching for BNS signals in real-time in the LIGO-Virgo-KAGRA (LVK) GW detectors presents a computational challenge, as the data streaming out must be matched against $\sim$ million reference waveforms, which requires up to a thousand CPU cores. We present a different approach using neural networks to learn the presence of a signal in the data. Our algorithm, called Aframe, was deployed in the LVK's fourth observing run and was the first artificial intelligence (AI)-enabled search to detect multiple binary black holes (BBHs) live. In this work, we demonstrate that the approach extends to the lower-mass BNS regime, and is the first AI-enabled search that achieves sensitivity comparable to matched-filter pipelines at lower computational and latency costs. The challenge of the longer-duration BNS signals is addressed by heterodyning the data, following which the network architecture used for BBHs is sufficient to distinguish signal versus background. We also show that this analysis requires a single non-flagship GPU for online deployment. Furthermore, the design and adoption of inference-as-a-service tools allow rapid offline analysis using a distributed pool of GPU resources. Hence, aside from the use case of rapid online data analysis, we also establish the use of Aframe for efficient archival data analysis.

Reliable and reproducible radio-frequency (RF) measurements in real-world environments are essential for characterizing spectrum behavior across unlicensed ISM and WiFi bands, licensed mid-band allocations, and emerging next-generation wireless deployments. Existing measurement platforms are often laboratory-grade, cost-prohibitive, or dependent on fixed infrastructure, limiting their practicality for rapid, distributed, or long-duration field campaigns. This paper presents a compact, battery-powered RF capture system integrating a HackRF One software-defined radio, Raspberry Pi 5, GNSS receiver, regulated battery supply, and high-speed solid-state storage. The platform records continuous IQ data at up to 20 Msps in SigMF format with per-segment location and timing metadata for reproducible spectrum analysis. Field experiments at 2.45 GHz in dense foliage, urban outdoor, and indoor office environments reveal distinct propagation signatures. Foliage measurements remain near the noise floor at -76 to -82 dBFS with limited spectral structure, consistent with strong canopy attenuation. Urban measurements show multipath activity across a 30 dB dynamic range, overlapping WiFi channels, and frequent ISM-band interference. Indoor measurements show dominant WiFi channels, an estimated 20 to 25 dB building entry loss relative to outdoor conditions, and an 8 to 10 dB higher interference floor caused by structural reflections. The system sustained 75 to 85 MB/s write throughput with no dropped samples or buffer underruns, while GNSS synchronization remained below one second with meter-level positioning. These results show that a portable, cost-effective SDR platform can produce high-fidelity, geotagged IQ datasets for spectrum characterization, interference analysis, radio environment mapping, and environment-aware wireless research.

A proposed method for the control of groups of inspection spacecraft is Multi-Agent Reinforcement Learning (MARL). While MARL has already been employed for this purpose in previous work, the reward functions used focus on reaching a finite set of predetermined inspection points around the target. In this work, we study and develop a generalized reward function for the MARL inspection task informed by the analysis of 3D reconstructions of inspected objects in orbit. Because the reward function is generalized such that any number of images at arbitrary locations may evaluated, we also allow trained agents to have complete control over when images are collected. With this approach, we gather insights into best practices for not only the specific MARL inspection task, but also gain key takeaways informative to the broader inspection task outside of a MARL context.

Federated learning (FL) research often depends on many small but consequential algorithmic choices: optimizer variants, server aggregation rules, local training schedules, normalization, regularization, and model architecture. These choices are expensive to explore manually and difficult to compare fairly when candidate changes can also alter the FL training or evaluation path. In this work, we present Auto-FL-Research (AFR), a constrained coding-agent workflow for FL algorithmic recipe search. Agents may propose and implement candidate training algorithms, including server aggregation rules, client update schedules, local objectives, and registered model variants, while task profiles fix the mutation surface, compute budget, communication contract, and final model evaluation. Each campaign records candidate scores, runtime, edited files, artifacts, and failure status. We evaluate AFR on five healthcare cross-silo FLamby tasks and on grouped-client profiles for the five fixed LEAF datasets plus the LEAF synthetic task. Five-seed repeat evaluations support gains on four FLamby tasks and five of six LEAF profiles, while also exposing seed-sensitive and search-selected failure cases. Same-budget controls show that several gains correspond to FL-recipe changes, whereas other improvements are recovered by fixed-surface scalar controls or fail under repeat or held-out evaluation. These mixed outcomes are part of the contribution: they show how agent-generated candidates can be separated into repeated FL mechanisms, fixed-surface tuning effects, and selected single-run artifacts.

Given one or more images of a railway crossing, can we leverage visual cues that allow us to robustly estimate how safe it is? Can we improve our ability to do so by introducing structured data (such as official accident reports) about the accident history of that crossing into our models? In this work, we explore how to best answer those questions towards building an AI system that can ingest multi-modal data for railway crossings and provide safety assessment and scores that align with expert opinion and with safety scoring used by the Federal Railroad Administration (FRA). To that end, we propose a proof-of-concept pipeline that delivers on that goal, while at the same time exploring and tackling a number of critical research challenges that pertain to different parts of the pipeline, from data preparation to different learning paradigms that can allow us to realize such a system. Indicatively, our proposed system identifies HIGH-RISK and LOW-RISK crossings with a macro F1 score of 0.757 and estimates FRA-based safety scores with an RMSE of 0.078 and correlation of 0.492 using a routed fine-tuned compact VLM pipeline, while producing qualitative results that align with domain-expert assessment.

Quantum mechanical (QM) cluster models provide an effective framework for mechanistic studies of enzymatic reactions but remain computationally demanding. Neural network potentials (NNPs) offer a promising route to reduce this cost, but enzymes present challenges beyond small molecules, including large system sizes, implicit-solvent environments, substantial polarization, and charge transfer. Here, we present an integrated software framework for efficient NNP training for mechanistic studies of enzymes, demonstrated on QM cluster models of S-adenosyl-L-methionine-dependent methyltransferases (MTases). Our Enerzyme code introduces modular electrostatics-aware NNP architectures and combines automated QM-cluster construction with reactive dataset generation. The Enerzymette subpackage automates reaction pathway exploration at both NNP and DFT levels. We show that iterative flexible scans and nudged elastic band calculations impose stricter requirements on NNPs than conventional dataset metrics. Nevertheless, NNPs trained on fewer than 1,000 system-specific datapoints reproduce reaction energetics and transition-state structures for MTase clusters containing up to 545 atoms with near-chemical accuracy. Direct supervision of atomic charges and consistent dielectric screening substantially improve simulation stability and accuracy, while multitask-learned atomic charges capture charge transfer and polarization trends and provide chemically meaningful descriptors of reactivity. Finally, transferability across chemically diverse catechol O-methyltransferase substrates indicates that NNPs learn generalizable reactivity patterns as training data expand across multiple enzymes. Together, these results establish a foundation for accelerating enzyme mechanistic studies and guide future NNP development for biomolecular reactivity.

Large language models (LLMs) are typically evaluated on code generation and program repair using binary functional correctness: a generated program or patch either passes or fails a test suite. This protocol is simple but coarse, as it ignores partial progress, feedback use, regressions, and the refinement trajectory through which models often improve code. We introduce PAIR-Bench, a progressive and adaptive benchmark for evaluating code improvement: transforming an incorrect or incomplete program into a more correct one through feedback-guided refinement. PAIR-Bench uses progressive hinting, a structured feedback protocol with two controls. Failure-region control determines what the feedback targets by grouping hidden failing tests into failure scenarios, while hint-depth control determines how much repair-relevant information is revealed, from coarse symptoms to implementation-level guidance. This design enables PAIR-Bench to measure whether a model repairs targeted failures, generalizes beyond the hint, preserves already-correct behavior, and how much assistance it requires. By evaluating repair trajectories progressive metrics rather than only final pass/fail outcomes, PAIR-Bench provides a finer-grained assessment of LLM code-improvement capability.

Turn-taking naturalness is central to full-duplex spoken dialogue systems, yet its automatic evaluation remains limited. Existing evaluations often rely on human judgments or behavior-specific timing metrics, making it difficult to compare heterogeneous timing failures within a unified framework. We propose TurnNat, a likelihood-based framework for automatic turn-taking naturalness evaluation in two-channel spoken dialogue. A causal turn-taking prediction model trained on natural conversations estimates future two-speaker voice-activity states, and the negative log-likelihood (NLL) of the observed future activity measures timing atypicality. TurnNat pools frame-level NLLs over turn-taking boundary units (TBUs) extracted from utterance onsets and offsets, and aggregates mean and tail TBU scores into a dialogue-level naturalness score. We further construct a controlled perturbation benchmark of paired natural and perturbed dialogue clips, validated by human naturalness judgments. Experiments on this benchmark show that TurnNat successfully identifies unnatural turn-taking perturbations across heterogeneous timing failures.

Multi-gate teleportation (MGT) packages $n$ remote gates into a single ebit via a 1-ebit fan-out quantum circuit, saving $n{-}1$ entangled pairs relative to sequential gate teleportation. The cost is a correlated failure mode: a single network fault propagates through the fan-out tree, injecting a weight-$n$ Pauli error. We derive a design rule for fault-tolerant packet sizes, $\nmax^{\text{corr}}(d) = \lceil d/2 \rceil$ for rotated surface codes of distance~$d$ with a correlation-aware decoder ($\nmax^{\text{naive}} = \lfloor d/2 \rfloor$ without), bounding how many gates can be packaged whilst preserving fault tolerance. Simulation with PyMatching shows that the standard MWPM decoder built from the packet circuit's noise model naturally corrects the correlated error: at network-to-local noise ratios $γ= \pnet/\pgate$ up to $100$, the packet matches or surpasses the per-link sequential LER at moderate-to-high $γ$, with the advantage growing with both $γ$ and $d$, whilst reducing the entanglement cost from $n$ ebits to~$1$. Packetisation wins when the network is the bottleneck ($γ\gg 1$); at $γ\approx 1$ the $n{-}1$ extra local fan-out gates offset the network savings. No custom decoder is required: the circuit-level noise model already encodes the correlation. These results enable noise-aware distributed circuit compilers to favour fan-out packetisation without sacrificing fault tolerance.

Neural Quantum States (NQS) are a remarkably expressive class of variational ansätze for quantum many-body wavefunctions, yet little is understood about their internal mechanisms: trained on variational objectives alone, how do NQS accurately capture physical observables that they have never been explicitly optimized for? In this work, we present a systematic approach to analyze the internal activations of NQS using sparse autoencoders. We extract features from the residual stream and demonstrate that these features strongly correlate with physical observables such as order parameters, staggered magnetization, and half-chain correlators, across both ground state representation and real-time dynamics. Remarkably, the discovery of these features is entirely unsupervised, with no physical labels provided. We further establish that such features causally affect the corresponding observables predicted by NQS, by showing that targeted, post-training intervention on a \textit{single} feature smoothly and monotonically steers the corresponding observable, while leaving the variational energy nearly unchanged. These results demonstrate that NQS are not merely functional approximators, but encode rich, interpretable internal representations of physical information. Our approach provides both a diagnostic and an intervention tool for NQS, and serves as a foundation for using mechanistic interpretability towards more reliable, transparent NQS.

The geometric and topological structure of quantum cost landscapes (QCLs) governs the optimization and thus the predictive power of variational quantum algorithms (VQAs). We systematically analyze ravines - low-cost paths connecting local minima - using an adapted version of the nudged elastic band (NEB) algorithm, a method originating from theoretical chemistry. By training quantum neural networks (QNNs) to classify the concentratable entanglement of quantum states, we apply the NEB algorithm and numerically identify ravine structures in QCLs of hardware-efficient ansatzes. Beyond visualizing these ravines, we construct an ensemble prediction framework by averaging predictions from QNNs parameterized along the low-cost NEB path. We introduce a resource-light pre-training metric which quantifies local-prediction variability and serves as a strong performance indicator for VQAs, even beyond the scope of this study. When base classifiers are drawn from circuit and weight initializations exhibiting high local-prediction variability, the quantum-based NEB ensembles outperform both classical and naive quantum alternatives. Moreover, a complexity analysis shows that leveraging the ravine-like structure of QCLs with the QNN NEB approach substantially reduces computational costs compared to naive QNN ensembling. A depth and qubit scaling analysis indicates that ravines persist across both scalings, and that, despite the expected growth in resource requirements with the qubit scaling, the NEB approach also accelerates convergence over the naive alternative.

LLMs are increasingly used to brainstorm research ideas, but existing evaluations mostly judge individual ideas by novelty, feasibility, or expert preference. We instead ask: how far are current LLM-generated ideas from human researchers? To characterize this gap, we build a large-scale evaluation framework for ideation from high-quality human research papers. For each paper, we reverse-engineer a small set of closely related prior works that likely inspired its core idea. LLMs are then prompted to generate a new idea from the set of paper titles and summaries. We introduce a two-axis research-taste taxonomy to profile each idea by its opportunity pattern and research paradigm, and use it to quantify the divergence between human and LLM ideas. Across idea sets generated by different LLMs, we observe a consistent distributional gap: LLM ideas are disproportionately concentrated around bridge-like opportunities and synthesis methods, whereas the human paper reference distribution spreads more broadly across ways of framing gaps and constructing contributions. This result suggests that strong LLMs can produce a range of reasonable ideas, but that range remains narrower than, and systematically shifted relative to, human research taste.

Reinforcement learning (RL) has become a central component of post-training large language models (LLMs), yet little is understood about how RL adaptation is distributed across transformer layers. Existing approaches typically update all model parameters uniformly, implicitly assuming that every layer contributes similarly to the gains obtained during RL post-training. In this work, we challenge this assumption through a systematic layer-wise study of RL training. Surprisingly, we find that training a single transformer layer can recover most of the gains achieved by full-parameter RL training, and in some cases even surpass it. To quantify this phenomenon, we introduce the quantity layer contribution, which measures the fraction of full RL improvement recovered by training a layer in isolation. Across seven models spanning two model families (Qwen3, Qwen2.5), three RL algorithms (GRPO, GiGPO, Dr. GRPO), and multiple task domains including mathematical reasoning, code generation, and agentic decision-making, we observe a remarkably stable pattern: RL gains are highly concentrated in a small subset of, and in many cases even a single, transformer layers. More strikingly, the same structural pattern consistently emerges: high-contribution layers concentrate in the middle of the transformer stack, while layers near the input and output ends contribute substantially less. The resulting layer rankings remain strongly correlated across datasets, tasks, model families, and RL algorithms.

Prior work on imitation learning from suboptimal demonstrations typically relies on compressed supervision signals such as confidence estimates, discriminator scores, or importance weights. These scalar signals are inherently limited, as they cannot explicitly express intermediate reasoning about task progress, failure modes, or corrective actions. We propose a language-critique framework for imitation learning from suboptimal demonstrations that instead leverages natural language as a structured supervision signal, avoiding the collapse of expressive feedback into scalars. Our method first constructs language labels from demonstrations that explicitly describe current progress, identify suboptimal behaviors, and provide fine-grained corrective guidance. We then introduce a language-critique loss that directly trains policies using these structured signals without reducing them to scalars, and instantiate it for both behavior cloning and diffusion policies, yielding LC-BC and LC-DP. We further provide a theoretical result showing that the proposed objective upper-bounds the expert performance gap under standard assumptions. Empirically, we evaluate on diverse continuous control tasks spanning navigation, manipulation, and gameplay, where our methods consistently outperform strong imitation learning and offline reinforcement learning baselines. These results demonstrate that language can serve as a powerful and structured form of supervision for learning robust policies from suboptimal data.

Memory expertise is a learned skill: knowing what to encode, when to retrieve, and how to organize knowledge--a capacity known in cognitive science as metamemory. We bring this perspective to LLMs by treating memory management as a trainable skill. We promote file-system operations to first-class memory actions alongside task actions, letting the model itself decide how to manage its memory. This memory skill improves along two axes: the structure that supports it (prompts, file schemas, action vocabulary), and the proficiency of the model exercising it. Both axes resist manual optimization: episodes in long-horizon tasks run for thousands of steps, and a single memory mistake can hide long before it surfaces, making human review of full trajectories impractical. We introduce AutoMem, a framework that automates both axes. In the first loop, a strong LLM reviews complete agent trajectories and iteratively revises the memory structure that shapes how the agent interacts with its memory files. In the second loop, the agent's own good memory decisions are identified from many episodes and used as training signal to sharpen the model's memory proficiency directly. Across three procedurally generated long-horizon games (Crafter, MiniHack, and NetHack), optimizing memory alone--without modifying the model's task-action behavior--improved the base agent's performance ~2x-4x, bringing a 32B open-weight model competitive with frontier systems such as Claude Opus 4.5 and Gemini 3.1 Pro Thinking. Our results show that memory management is an independently learnable skill, and a high-leverage objective yielding large gains on long-horizon tasks.

When should an AI system's answer be trusted? Formal proof assistants offer certainty but cannot reach most of the problem distribution; scalar LLM judges offer coverage but produce opaque scores that cannot be audited after the fact and are subject to the same coherence issues as any LLM. We present Theoria, a verification architecture that closes this gap. A candidate solution is rewritten into a sequence of typed state transitions, each licensed by an explicit justification, whether that be a citation, computation, or problem-given fact, and every transition is independently auditable. The foundational invariant is completeness of change: every difference between consecutive proof states must be accounted for, so hidden premises surface as unlicensed mutations rather than passing silently. On HLE-Verified Gold (185 text-only expert problems), Theoria certifies 105 at 91.4% strict precision (Wilson 95% CI [84.5%, 95.4%]). Every certification produces a human readable proof trace in which each step can be independently challenged. Holistic LLM judges achieve comparable precision at matched coverage but fail on different problems (Jaccard 0.14-0.36), making the approaches complementary. On 95 adversarial poisoned proofs across 15 domains, structured judges catch 94.7% versus 83.2% for holistic judging (p= 0.0017). The overall 11.5 pp gap concentrates in hidden premises (90.6% vs. 62.5%, a 28 pp difference) and fabricated citations (100% vs. 90%), the error classes where the formal analysis predicts an advantage; performance is identical on arithmetic and theorem-misapplication errors, where no advantage is predicted. On GPQA Diamond (n= 65), certified precision is 97.1% (Wilson CI [85.1%, 99.5%]).

Transformers use the same forward computation stream to both predict the next token and store useful state for future token predictions. We formulate the \emph{state-prediction separation hypothesis}: disentangling the two roles yields better language modeling performance. We design a Transformer variant that uses two computation streams to separate the two functions, and conduct pretraining experiments across various scales. Our experiments show that state-prediction separation consistently offers better data and compute efficiencies, improving validation loss and outperforming standard Transformers by 2--3 percentage points on average on downstream tasks. We also conduct extensive empirical analysis that rules out potential confounders and demonstrates the fundamental difference in the gradients our design entails.

In practice, most commercial LLM providers do not publicly release details of underlying LLM architectures. However, prior work has shown that given limited API access to an LLM (namely, top-$k$ logits and/or a logit bias function), one can recover certain architectural details of an LLM, such as the hidden dimension of the feed-forward network. Perhaps in response to these results, most commercial LLM providers have restricted their APIs to expose only the single logit for each decoded token, and they no longer give users the ability to bias logits. We show that even under current restrictive APIs, several architectural parameters are still recoverable. We present NightVision, an attack that uses restrictive black-box API access to estimate the hidden dimension, depth, and parameter count of an LLM. Algorithmically, NightVision relies on a novel common set prompting technique in which multiple prompts expose log probabilities for the same set of output tokens; a spectral analysis of these results is used to infer hidden dimension. NightVision additionally uses end-to-end time to first token (TTFT) measurements and the estimated hidden dimension to estimate depth and parameter count. We empirically evaluate NightVision on 32 open-source LLMs, recovering hidden dimension to within 23% average relative error across all models (9% on MoE models), and depth and parameter count to within 53% for models exceeding three billion parameters. We run extensive ablations to demonstrate how these accuracies scale with token budget and model properties. Overall, our results suggest that current LLM APIs are not sufficiently restricted to fully obfuscate the architectural details of their underlying models.

Open-source libraries and tools are widely reused, but compatibility maintenance is expensive. Once maintainers leave, useful repositories can stop working as runtimes and dependencies evolve. We study whether LLM agents can adapt old repositories to modern environments, a task we call compatibility rescue. Unlike bug repair, compatibility rescue starts from a repository that worked in its original environment but fails after ecosystem drift. RepoRescue gives agents only the repository and its failing modern environment; the agent must diagnose the failure, locate affected code, and produce a source-code rescue that restores the historical test suite. We build RepoRescue from 193 Python and 122 Java repositories, each verified to pass historically and fail after modernization. We evaluate five deployed agent systems on Python and three on Java. Beyond full-patch pass rate, we rerun patches after removing test-file edits to measure source-only repair, add a runtime-enforced regime that blocks test edits, and validate practical use for repositories whose suites pass after rescue. We find that Claude Code systems sometimes edit failing tests even when prompted not to; with runtime blocking, Kimi still rescues 41.5% of repositories. Systems are complementary: their union reaches 62.7%, exceeding the best single system by 10.9 points. Difficulty concentrates in cross-file coordination: on 14 repositories requiring coordinated whole-codebase changes, GPT-5.2 through Codex passes all 14, while every Claude Code system passes at most two. Finally, a passing suite is only an initial signal: among 34 unmaintained Python candidates whose suites pass after rescue, 22 work in realistic scenarios and 12 pass bug-hunt with patches that address the compatibility failure. RepoRescue benchmarks compatibility rescue with source-only auditing, runtime enforcement, practical validation, and reasoning labels.

Current work on robot furniture assembly mostly focuses on toy-scale settings or single-arm manipulation. We introduce FurnitureVLA, the first systematic study of real-scale bimanual furniture assembly using Vision-Language-Action models (VLAs). We formalize the task, develop a scalable simulation pipeline for expert data generation and evaluation, and build a VR teleoperation system for single-operator bimanual control to collect high-quality real-world demonstrations. To address extreme long-horizon assembly with up to 7 subtasks and 1550 control steps, we propose a progress-enhanced VLA, finetuned on semantically grounded subtasks, that jointly predicts actions and a continuous progress signal, enabling automatic subtask transitions and reducing compounding errors during inference. We further study perception and control design factors that critically affect precision in real-scale assembly. FurnitureVLA improves average simulation success from 48% to 80% compared to baselines across three furniture types, with an additional 21% gain from our design factor study. We validate on a real Kinova Gen3 platform with only 16% drop on the hardest task.

Repository-level performance-optimization benchmarks such as GSO, SWE-Perf and SWE-fficiency evaluate coding agents by applying patches to real repositories and comparing runtime against unoptimized baselines and official reference patches. Their leaderboard scores are increasingly used as evidence of coding-agent progress, but those scores can conflate runtime instability, benchmark-specific scoring rules, and how many tasks are already solved by at least one public submission. We audit these issues across the three benchmarks. First, we replay the official reference patches for 740 code optimization tasks across four common types of Google Cloud machines. Most benchmark tasks can be replayed, but their reference patches satisfy the original benchmark validity rules in every cross-machine replay for only 39/102 GSO tasks, 11/140 SWE-Perf tasks, and 411/498 SWE-fficiency tasks; SWE-Perf is especially fragile because many reference patches produce close-to-zero runtime changes. Second, we show that public submission rankings depend strongly on the benchmark scoring rule. Among eight public submissions shared by GSO and SWE-fficiency, the official rankings disagree on 9 of 28 pairwise submission comparisons, and SWE-fficiency's leaderboard scoring rule assigns the worst ten tasks overly high score weights of 58.5%-82.8%. Third, looking across 10 public submissions for each task, we find that at least one submission matches or beats the reference patch on 85.3% (384/450) of replay-valid GSO and SWE-fficiency tasks, and beats the unoptimized base code on 99.8% (449/450). Our study complements leaderboard scores by identifying tasks with more reliable performance signals, quantifying per-task score contributions, and exposing the remaining performance gaps that are hidden by aggregate rankings.

Classical Block Withholding (BWH) attacks have been extensively studied in block-dependent reward schemes, where pool members are compensated upon a block discovery within the pool. However, most contemporary mining pools operate under share-based scheme wherein participants are paid immediately upon submission of valid shares. In this paper, we analyze BWH under Pay-Per-Share (PPS) and Full-PPS (FPPS) schemes for Nakamoto-style blockchains and prove that these mechanisms are not incentive compatible -- contrary to claims in prior literature. Under PPS/FPPS, the optimal strategy for a BWH attacker is the All-out Attack (AoA): the adversary allocates its entire hashpower toward the victim pool, submitting only partial Proof-of-Work shares (pPoW) while withholding all valid blocks, i.e., full Proof-of-Work (fPoW). Under AoA, prior to the first difficulty adjustment, the adversary incurs negligible loss due to the withheld fPoWs. After the first difficulty adjustment, which reduces block difficulty, the adversary generates more pPoWs per unit time, achieving a relative gain of $\fracα{1-α}$ compared to pre-adjustment rates, where $α$ is the fraction of adversarial hashpower. Moreover, per unit time and per unit hashpower, all honest miners benefit at the same rate as the adversary. In contrast, the victim pool operator incurs losses: it pays the attacker out-of-pocket for pPoW submissions but receives no fPoW compensation in return. Finally, advanced variants of BWH, such as Fork After Withholding (FAW), do not yield additional profit to the attacker.

Language models deployed in high-stakes roles can potentially favor certain entities, brands, or viewpoints, steering user decisions at scale. Such preferential biases can be introduced by any actor in the model's supply chain and are most dangerous when the model reveals its preference only on the relevant topic while behaving identically to its unmodified base on all other inputs. Recent work has shown that these biases can transfer through context distillation on semantically unrelated data, with the signal residing entirely in the soft logit distribution and remaining invisible to text-based inspection. However, the defender faces a fundamental asymmetry: without knowing the bias topic, no detection method can reliably surface a stealth preferential bias, regardless of whether it examines generated text, internal representations, or model weights. Here we introduce Distill to Detect (D2D), a method that surfaces hidden biases by distilling the distributional shift between a suspected model and its base into a cartridge (a KV-cache prefix adapter), concentrating the dominant divergence and amplifying the bias signal into generated text. We show that D2D successfully amplifies the hidden biases of stealth models to the extent that they can be reliably detected across multiple bias types. We also propose a theoretical framework that explains the efficacy of D2D through the lens of Fisher-weighted projection of the logit distribution shift, supported by empirical observations. By turning the capacity bottleneck of prefix-tuning adapters into a detection tool, D2D provides a practical building block for auditing hidden behaviors in deployed language models.

We introduce TiRex-2, a recurrent xLSTM-based time series foundation model that generalizes the univariate TiRex to multivariate forecasting with both past and future covariates. Real-world forecasting is inherently sequential: observations arrive continuously, variables evolve jointly, and a subset of covariates is known ahead of time. Existing Transformer-based time series foundation models capture cross-variate dependencies but incur quadratic complexity in context length and require full-history recomputation as new observations arrive. TiRex-2 addresses these limitations through a memory-centric recurrent design that operates at constant per-patch cost under streaming. The model combines a bidirectional time mixer with an asymmetric grouped-attention variate mixer, enabling the integration of future-known covariates while preserving strict causality over target variables. To our knowledge, this is the first time series foundation model that achieves this combination of properties. To support scalable multivariate pretraining, we propose a synthetic coupling pipeline that composes diverse multivariate samples on the fly from large univariate corpora. Empirically, TiRex-2 achieves state-of-the-art zero-shot performance on GIFT-Eval and fev-bench, remains stable when streamed to arbitrary context lengths, and maintains constant inference cost per patch. The model uses 38.4M active parameters in univariate mode, with an additional 44.1M parameters activated for multivariate forecasting.

This paper studies real-time robust optimal control for uncertain nonlinear systems, where linear time-varying (LTV) approximations make planning tractable but require sound linearization error bounds (LEBs) to guarantee robust constraint satisfaction. We develop tight, differentiable, GPU-parallel LEBs for LTV approximations of nonlinear and neural network (NN) dynamics. For analytic dynamics, we introduce path-based Hessian bounds that are tighter than standard interval methods. For NN dynamics, we derive certified LEBs using NN verifier-generated affine relaxations and local Jacobian corrections. We adapt a GPU-parallel system-level synthesis LTV-based robust control solver to be compatible with these LEBs by extending it to handle right-invertible disturbance matrices and non-zero-centered disturbance sets for tight zonotopic uncertainty propagation. Our method, GPUSLS-LEO, enables online optimization of robust feedback policies that account for linearization error, producing tight, formally verified reachable tubes. On complex nonlinear and NN dynamics up to 168 state dimensions, our method can compute robust control policies on the GPU at rates up to 67 Hz, reducing solve times and conservativeness relative to baselines while preserving formal guarantees and real-time performance.

We present World from Motion, a method for generating freely renderable dynamic 3D Gaussian representations from monocular videos. Our approach conditions a video model on dense, pixel-aligned renderings that encode appearance, geometry, and 3D scene motion along both input and target camera trajectories to correct rendering artifacts and fill in missing regions from an initial reconstruction. To train this model, we construct a dataset of aligned multiview video pairs and dynamic 3DGS representations, with simulated artifacts characteristic of monocular reconstruction. At test time, we distill the model's generations, including newly observed regions and motions, back into a single consistent, high-quality dynamic 3DGS, improving both novel-view synthesis and the underlying 3D motion. Our method sets a new state of the art in 4D reconstruction and seamlessly generalizes to in-the-wild videos with large viewpoint changes and dynamic motions.

Quantum computing has emerged as a promising computational paradigm for machine learning (ML), with the potential to offer computational advantages over classical approaches. At this stage, the evidence supporting the performance and advantages of quantum machine learning (QML) models relative to classical models is insufficient. To address this gap, this paper presents an empirical study on the performance of QML models and their classical counterparts. We compare seven model pairs spanning supervised learning and reinforcement learning. Our results indicate that the evaluated quantum machine learning models do not yet surpass the classical baselines in overall prediction performance, policy stability, or training time. Nevertheless, QML remains a promising approach for filtering noise and controlling false positives. Our research findings summarize the challenges facing quantum machine learning across hardware environments, training efficiency, and convergence stability, providing a foundation for research into the robustness and parameter optimization of QML. This work is publicly available at https://github.com/Z-537-437/QML.

Deep learning has outgrown any single mathematical explanation. From Approximation to Emergence develops a unified, proof-oriented account of modern deep learning theory, tracing a path from the classical foundations of approximation, optimization, and generalization to the contemporary mechanisms of overparameterization, robustness, generative modeling, transformers, in-context learning, scaling laws, interpretability, alignment, and emergence. Rather than presenting isolated results, the book organizes a broad literature into a coherent research narrative: each theory is examined through the object it controls, the assumptions that make it valid, and the phenomena it leaves unexplained. Written for researchers, graduate students, and mathematically trained practitioners, this monograph offers a rigorous map of deep learning theory as it stands today: powerful, incomplete, and increasingly centered on the question of how learned mechanisms arise from scale, data, architecture, and training.

In autonomous laboratories, AI agents suggest the next batch of experiments to do. However, planning and executing those tasks taking full advantage of the available resources is a completely different question. This can be challenging when dealing with real-world hardware constraints, especially so when there are multiple instruments with different capacities and throughputs. Here we demonstrate a 2-step method to address resource utilization for our autonomous platform for metal-organic framework synthesis. First, we use constraint programming to find optimal schedules. This finds schedules that minimizes the total time while still satisfying the limitations and capacities of the hardware. Secondly, we use a system of status dependencies for each task, which allows for the robust execution of the optimal schedules.

Combinatorial optimization (CO) problems are difficult because certifiable discrete structure induces exponential search. One needs to search over the set exponentially many candidates to certify optimality, however, the structural feasibility of a path, packing, or cover can be verified in polynomial time once supplied. In this study, we introduce Neural Certificate Pricing (NCP) that exploits this asymmetry under an unsupervised learning framework. A neural network is trained to predict certificate-level dual prices, while a structured recovery layer constructs the induced primal marginal. NCP can be viewed as amortized separation: instead of enumerating violated inequalities, it learns the residual prices through which their aggregate effect enters recovery. When the certificate-consistency condition holds, the recovered marginal is globally feasible, and a local theory shows that first-order errors in the predicted price induce only second-order loss in objective value. Across three classes of CO problems, NCP either outperforms state-of-the-art neural baselines by large margins or matches them at a fraction of the computation time, and shows stronger out-of-distribution generalization.

RL with verifiable rewards (RLVR) has emerged as a powerful paradigm for training LMs on tasks with well-defined success metrics, such as code generation and mathematical reasoning. However, current RLVR methods optimize only what can be objectively scored, often neglecting subjective, non-verifiable aspects of human-like outputs, such as style and structure. This limitation leads to well-documented failure modes such as diversity collapse, unnatural-sounding responses, and reward hacking. We propose an adversarial generator-discriminator framework that augments verifiable rewards with a learned signal from human demonstrations. A generator model is trained using RL to maximize both task accuracy and an adversarial reward derived from a discriminator. The discriminator, trained alongside the generator policy, learns to distinguish human-written outputs from model-generated ones. The discriminator serves as a learned proxy for the human output distribution, providing feedback on aspects of generation that are difficult to formalize as scalar rewards. Across diverse domains, including bug fixing and open-ended generation, our approach consistently improves non-verifiable properties while preserving the accuracy gains of RLVR. In bug fixing, our method produces solutions with significantly lower edit distance compared to RLVR baselines while matching end performance. In story generation, our method significantly improves win rate while producing stories that are diverse and more human-like. And in a simple reward hacking benchmark, our method nearly eliminates model misbehavior while maintaining high benchmark scores. Together, these results show that our approach bridges RL and SFT, offering a scalable path toward jointly optimizing the verifiable and non-verifiable properties of a task.

Scaling inference compute, by generating many parallel attempts per problem, is a costly but reliable lever for improving language model capabilities. By default these attempts are generated independently, wasting inference compute on redundant solutions. This waste seems unavoidable. After all, independence is what makes parallel sampling trivial to scale. However, this tradeoff is not fundamental: there is a rich design space of samplers that generate correlated but exact samples entirely in parallel. We explore this design space as an avenue for improving sample efficiency in scaling inference compute and reinforcement learning (RL). Concretely, we introduce QuasiMoTTo, which uses correlated samples as a drop-in replacement for i.i.d. samples. To generate these samples, QuasiMoTTo uses a reparameterization of autoregressive sampling as inverse-CDF sampling and draws the underlying uniforms with quasi-Monte Carlo (QMC); because QMC spreads the uniforms out more evenly than i.i.d., the resulting samples cover the output space with far less redundancy. Even though the batch is correlated, each sample is marginally distributed according to the language model, so we can use the batch for policy-gradient training. Our empirical analysis focuses on understanding how efficiently QuasiMoTTo can turn compute into performance. To evaluate correlated samplers, whose dependence breaks standard pass@k estimators, we first develop an unbiased bootstrap estimator. Across four reasoning benchmarks, QuasiMoTTo matches i.i.d. pass@k accuracy with 25-47% fewer samples. Strikingly, QuasiMoTTo often saturates an upper bound on pass@k that holds for any marginal-preserving sampler. We also apply QuasiMoTTo to policy-gradient RL (GRPO) where it matches i.i.d. performance with 50% fewer training steps. These gains come from higher coverage, which yields a stronger learning signal per batch.

Multi-agent latent reasoning composes agents' KV-caches into one context for a final agent. Prior work (Agent Primitives) does this by concatenating caches along the sequence axis with RoPE re-encoding, which we call BagMerge. BagMerge is non-commutative, and the best input ordering is unpredictable, shifting with the regime, the latent-step budget, and the model scale. We make this exchange a convergent replicated state. First, CanonicalMerge fixes the layout by content: ordering caches by mean K-norm at a middle layer renders the merged cache byte-identical under any input permutation, verified algorithmically (arity N<=5) and bit-for-bit on real Qwen3-1.7B and 4B state. Second, we separate the replicated state from decode-time layout: the state is a set of content-addressed latent fragments whose merge is set union, a state-based CvRDT (commutative, associative, idempotent, absorbing), and CanonicalMerge is its deterministic render. Because the render is byte-equivalent, every N=2 accuracy number carries over unchanged and re-delivered duplicates are absorbed rather than re-concatenated. On a partitioned-reasoning benchmark, CanonicalMerge matches the best BagMerge ordering in every regime-by-budget-by-ordering cell without knowing which order is best, trading a small, statistically insignificant accuracy margin for an unconditional structural guarantee. The behaviour transfers to real multi-document QA (HotpotQA), while the closest training-free output-fusion baseline (PackLLM) loses by 45 points at matched budget, placing cache-level merging in a regime distinct from output-level fusion. Finally, at k>2 the approach transports and colocates latent traces but does not by itself compose them, which we characterize to motivate future work.

Sample-based generative models are increasingly used for probabilistic forecasting in high-stakes decision settings, yet their training objectives are blind to the decision maker's cost structure. These models are commonly trained with strictly proper scoring rules, such as the energy score, which allocate their training signal in proportion to data density, with no awareness of where forecast errors are most costly for downstream decisions. We therefore propose decision-aware training for sample-based generative models, augmenting the energy score objective with a differentiable decision loss that directly penalises the cost incurred by acting on the model's forecast. This combined loss is theoretically grounded, as the decision loss is itself a proper scoring rule. We validate our method on one synthetic and two real-world tasks, showing targeted improvements in cost-sensitive regions while retaining full probabilistic forecasts.

Generative reasoning re-rankers achieve strong recommendation accuracy by emitting a chain-of-thought before re-ordering a candidate list, but they are slow at inference: an autoregressive (AR) decoder spends one sequential forward pass per reasoning token, and the reasoning trace far exceeds the ranking it produces. To reduce this cost, block-diffusion language models decode many positions in parallel over a few denoising steps and are substantially faster, yet naively converting an AR re-ranker into one opens two accuracy gaps: (1) a structural gap: answer positions are denoised in parallel and scored independently, so the decoder emits invalid rankings (duplicated, dropped, or out-of-set identifiers) that AR avoids through left-to-right masking; and (2) a distributional gap: fine-tuning the converted model on fixed teacher trajectories is off-policy relative to its own decoding at inference, leaving a residual accuracy gap. To close both gaps while keeping the speedup, we propose \textbf{Diffusion-GR2}, a recipe that converts our AR reasoning re-ranker (GR2) into a block-diffusion re-ranker. First, conversion fine-tuning (CFT) adapts the AR-initialized diffusion model to denoise the answer into a valid permutation on its own, without an external constrained decoder. Next, on-policy distillation (OPD) then supervises the model on its own decoded trajectories with dense per-token targets from the AR teacher. Finally, we apply a reinforcement-learning (RL) stage against a re-ranking reward on top of OPD's on-policy policy. Experiments on Amazon Beauty demonstrate that Diffusion-GR2 recovers to near-parity with the AR re-ranker, while block-parallel decoding raises decode throughput by $2.4$--$3.5\times$ at the model's reasoning output length. Ablations show that CFT recovers most of the conversion gap, and that on-policy distillation further closes it to the AR reference.

NA methylation profiling has become a powerful approach for central nervous system (CNS) tumor classification, yet important challenges remain regarding cross-cohort transferability, methodological correctness, and robust multiclass evaluation. In this work, we propose a novel and methodologically rigorous machine-learning approach for methylation-based CNS tumor classification that combines Sparse Random Projection for dimensionality reduction with multinomial logistic regression for classification. We evaluate the proposed approach in the same general experimental setting established by a widely used reference classifier. On the 2,801-sample reference cohort, our method achieves a mean accuracy of 96\% under stratified 3-fold cross-validation. On the independent 1,104-sample clinical evaluation cohort, it reaches 86\% accuracy at the 91-class level and 93\% when predictions are evaluated at the methylation class family level. These results improve upon the corresponding state-of-the-art reference figures of 82\% class-level concordance and 88\% family-level concordance, yielding absolute gains of approximately 4 and 5 percentage points, respectively. This improvement is clinically relevant: in a diagnostic setting, a 5-point increase in correct tumor classification can directly affect cancer subtype assignment and, in turn, influence treatment selection and downstream clinical decision-making. Our results show that the proposed model, grounded in stronger methodological practice in machine learning, consistently outperforms the previous state of the art across evaluation settings and can materially improve the reliability of CNS tumor classification.

Counterfactual explanations explain machine learning predictions by identifying minimal input changes that would alter a model's decision. Although many existing methods successfully generate prediction-changing alternatives, they often produce unrealistic or infeasible recommendations due to a lack of explicit mechanisms for incorporating domain knowledge and intervention constraints. Neuro-symbolic AI offers a promising direction by combining data-driven predictive models with symbolic reasoning capable of representing human-understandable rules and feasible actions. This paper presents PACE, a modular neuro-symbolic framework for generating feasibility-aware counterfactual explanations. The framework separates prediction and reasoning into two components: a neural predictive model for classification and a symbolic reasoning layer that enforces domain-specific constraints during counterfactual generation. By explicitly modeling feasible interventions, the framework produces explanations consistent with domain knowledge while remaining interpretable and actionable. The approach is model-agnostic and adaptable to domains requiring realistic decision support. A case study is conducted on the Adult Income dataset, combining a multilayer perceptron classifier with Answer Set Programming (ASP) rules encoding feasible modifications to education, occupation, and working hours while preserving immutable attributes. Results highlight the trade-off between counterfactual validity and plausibility and show that symbolic constraints yield explanations that better satisfy domain-specific feasibility requirements, illustrating the potential of neuro-symbolic methods for transparent, feasibility-aware counterfactual explanation in explainable AI.

Recent work has shown that implicit neural representations (INRs) can be trained to effectively compress structured and unstructured volume data, allowing for direct data querying with a reduced memory footprint. However, as existing INRs for unstructured volumes do not encode geometry, they require partial mesh storage for later sampling, limiting achievable compression. At the same time, novel view synthesis methods have shown that explicit collections of 3D Gaussians can be used to accurately visualize volume data. In this work, we introduce an explicit model for volume data compression based on 3D Gaussian primitives. We reinterpret collections of 3D Gaussians as an explicit representation of a scalar field and use a sampling strategy that reconstructs scalar values at spatial locations through weighted aggregation of intersecting Gaussians. We develop optimized CUDA-accelerated pipelines for structured and unstructured model sampling, loss functions that encourage accurate domain encoding by our models, and a novel sampling-error based densification strategy. Our explicit formulation naturally encodes domain geometry, eliminating the need for mesh storage in unstructured volumes and introducing significantly higher compression opportunities. Compared to existing INRs, we demonstrate that our explicit model achieves competitive reconstruction quality with significant training speedups on structured volumes, while markedly outperforming in all metrics on unstructured volumes.

Cross-lingual speaker verification (SV) systems typically exhibit performance degradation when enrollment and test utterances are spoken in different languages. However, standard evaluation protocols confound language mismatch with inter-speaker variability, as evaluation is generally performed with different speakers across languages. In this work, we introduce a bilingual same-speaker evaluation set for five Iberian languages, enabling analysis of cross-lingual SV under constant speaker identity. We apply this setup to a HuBERT-based SV system previously shown to exhibit strong language dependence, and analyze results using the Cross-Lingual Transfer Matrix (CLTM) to study pairwise cross-lingual transfer. Our results show that speaker-related variability accounts for part of the observed degradation, but language mismatch remains the main driver of cross-lingual performance loss. These findings provide a more precise characterization of language dependence in cross-lingual SV.

Intrusion Detection Systems (IDSs) are essential for monitoring network traffic and identifying malicious activities in modern cyber-physical, Internet of Things (IoT), enterprise, and distributed network environments. However, developing reliable IDS models remains challenging because attack behaviors evolve over time, realistic datasets are difficult to obtain, traffic records may be incomplete, attack classes are often imbalanced, and privacy constraints limit centralized data collection. Recent advances in generative artificial intelligence (AI) and Federated Learning (FL) provide new opportunities to address these limitations. Generative models can support anomaly detection, synthetic traffic generation, data augmentation, data imputation, adversarial traffic generation, and IDS alert explanation. FL enables distributed IDS training without directly sharing local network traffic, making it suitable for privacy-sensitive and geographically distributed environments. This survey provides a structured review of generative AI and FL techniques for IDS. We first summarize representative IDS research directions, including adversarial machine learning, anomaly-based detection, IoT-oriented IDS, explainable IDS, and benchmark datasets. We then categorize generative AI applications in IDS according to model families and task objectives, covering autoencoder-based models, Generative Adversarial Networks (GANs), diffusion models, and Large Language Models (LLMs). Finally, we review emerging studies that integrate generative AI with FL-based IDS and discuss open challenges, including synthetic data quality, realistic traffic generation, dual-use adversarial risks, non-IID client distributions, communication-efficient model sharing, federated IDS benchmarking, and domain-specific LLMs for network security.

Safety evaluations for language models increasingly depend on judgments about ambiguous natural-language behaviour: whether a model has followed an instruction, refused appropriately, complied with a policy, resisted an embedded command, or misreported progress in an agentic task. Existing benchmarks often compress these distinctions into pass/fail labels, obscuring whether failures arise from capability limits, policy ambiguity, instruction conflict, scaffold failure, or unstable evaluator judgments. This paper introduces adversarial pragmatics as a benchmark and annotation protocol for evaluating model behaviour under instruction conflict, embedded commands, quotation, scope ambiguity, deixis, indirect speech acts, and multi-turn agent transcripts. The contribution is empirical and methodological: a linguistically controlled taxonomy, an 18-item seed benchmark with validator-enforced metadata, a 54-row local seed pilot, an expert-evaluation protocol distinguishing task success, policy compliance, safety risk, refusal outcome, and evaluator confidence, and metrics for judge validity, diagnostic ambiguity, and taxonomy drift. The framework turns linguistic judgment methodology into a practical tool for validating safety evals, LLM judges, gold-set construction, prompt-injection tests, and safety documentation.

Creativity research has debated whether creativity is domain-specific (e.g., visual, writing, science), and if it is psychometrically separable from general intelligence. Both questions now apply to LLMs, but a unified benchmark of AI creativity remains elusive. We introduce AGC-Bench, an artificial general creativity benchmark built from a systematic review of the AI creativity literature (3,101 papers screened, 497 benchmarks identified), paired with an agentic harness that converts idiosyncratic codebases into HELM-standardized benchmarks. The first release covers 78 datasets spanning brainstorming, problem solving, STEM, narrative, figurative language, and humor. To address bias in LLM-as-judge, we apply Judge Response Theory -- a psychometric calibration of judge leniency/severity; we then fine-tune Qwen3-30B on the bias-corrected ratings of three frontier LLMs to produce AGC-Judge, an open-weight model that robustly scores new creativity benchmarks it was not trained on. Results reveal frontier models at the top of the AGC-Bench leaderboard, with open models close behind. LLMs show different creative strengths, ranking higher on some domains (e.g., writing) than others (e.g., scientific ideation). Extensive experiments yield three main findings. First, applying factor analysis across 83 LLMs, we recover a single creativity factor 'c', analogous to the 'g' factor of general intelligence, that explains 81.5% of variance, related to but separable from general knowledge/reasoning. Second, we show that prompting models to "be creative" boosts their performance far more than enabling reasoning, evidence that the benchmark tracks creativity over general ability. Third, on a human-matched subset, we find the top human still leads the top LLM on creativity. We release AGC-Bench with a public leaderboard, AGC-Judge, and human data as open infrastructure for measuring AI creativity at scale.

While generative models have enabled training-free reward alignment, current methods typically excel in local exploration within narrow regions of the underlying distribution. These approaches struggle when preferences are unknown a priori and only revealed through sequential feedback-a scenario demanding broad exploration to uncover high-utility regions. To address this, we propose Sequentially-Controlled Interactive Multi-Particle Flow-Maps (IMPFM), a framework for sample-efficient online feedback-driven search. IMPFM progressively transports a group of interactive particles toward the target distribution, maintaining the broad coverage essential for heterogeneous preference alignment. IMPFM introduces a principled and efficient posterior sample sharing mechanism across particles powered by flow maps. By correcting individual particle drift with the collective posterior samples of the entire ensemble at each resampling step, the framework maximizes sample utility to enable global exploration while actively mitigating reward over-optimization, typical of standard control frameworks. Paired with a principled exploration-exploitation reweighting mechanism involving multi-particle interaction, this sequentially corrected multi-particle dynamics explicitly preserves structural diversity and overcomes the weight degeneracy inherent to standard SMC samplers. Crucially, we prove that the resulting sampling framework yields a multi-particle interaction-aware Feynman-Kac corrector that progressively steers the multi-particle system toward a KL-tilted target distribution, facilitating global exploration and preventing mode collapse. Extensive empirical evaluations and rigorous ablations across diverse search and alignment tasks confirm the efficacy of IMPFM over existing baselines.

Agent skills package reusable operational knowledge for Large Language Model (LLM) agents, yet as they grow in scope, they become dependency-bearing artifacts whose identities, versions, and provenance remain implicit. This opacity already causes duplicated dependencies and inconsistent installations, exposing a gap that dependency management has yet to close. We introduce Agent Skill Supply Chains (ASSCs) to characterize mixed skill-package-service dependency graphs and help close this gap. Borrowing from Software Bill of Materials (SBOMs), we design SkillDepAnalyzer to capture natural-language dependency evidence and model skills as dependency-bearing artifacts. On the SKILL-DEP benchmark, SkillDepAnalyzer recovers skill metadata and dependency graphs accurately and comprehensively, substantially outperforming an LLM-based baseline and package-centric SBOM tools. Applying SkillDepAnalyzer to over 1.43 million skills, we obtain ASSCs and explore their structural diversity and security signals. We find four structural patterns: skill metadata is activation-ready but governance-poor; dependency graphs span skill, package, and service dependencies with concentrated reuse; recursive skill reuse expands dependency graphs and creates hidden package inventory; and skill dependency clusters form around related workflows. We also find that inspecting a skill alone misses security-relevant signals hiding in its dependencies. By analyzing ASSCs, we identify and report known malicious skills persisting in ASSCs to their developers. Based on these findings, we recommend typed dependency manifests, first-class dependency-cluster management, risk-warning audit commands for skill infrastructure maintainers, and lockfile-like records for skill developers.

Autonomous scientific discovery systems offer the potential to accelerate research by automating the process of hypothesis generation and validation. However, current systems operate within constrained search spaces or require predefined research questions, limiting their capacity for true open-ended inquiry. Furthermore, while they generate hypotheses iteratively, they largely lack the ability to explicitly synthesize their own accumulated findings to uncover complex, interconnected phenomena. We introduce DiscoPER, an autonomous large language model-powered framework that conducts open-ended research by dynamically generating and executing code to explore datasets without pre-specified research objectives. To ensure rigorous scientific validity, every proposed discovery must pass statistical testing. To overcome the limitations of isolated search, our framework introduces a second-order reasoning mechanism that periodically analyzes its own accumulated discoveries. By treating prior discoveries as empirical data, DiscoPER identifies structural patterns, confounds, and epistemic gaps, actively redirecting hypothesis exploration toward uncharted regions of the search space. The search space is further expanded by incorporating tool use, enabling the system to explore hypotheses beyond structured metadata by seamlessly processing and extracting useful information from multimodal sources like images. Evaluated on iNatDisco, a new multimodal ecological knowledge benchmark with pattern-level ground truth obtained from peer-reviewed literature, DiscoPER recovers 8 of 9 known patterns with a 72.7% hypothesis support rate, outperforming both classical causal discovery and LLM-guided baselines. Ablations show that DiscoPER scales with more data, and confirms the benefits of second-order meta-reflection.

Operator learning for partial differential equations (PDEs) on arbitrary geometries builds fast neural surrogates for large-scale simulation. Although recent geometry-adaptive neural operators have made substantial progress, they are mainly designed for forward problems in which inputs and outputs share the same spatial domain. This limits their applicability for boundary value problems (BVPs) and inverse problems, where inputs and outputs may live on different domains. We introduce the Geometry-Adaptive Integral Autoencoder (GAIA), an operator learning model that encodes the domain boundary and the interior field distribution into geometry tokens, and conditions integral transform layers on these tokens via cross-attention, allowing the kernel to adapt locally to geometric features. This yields a single architecture for forward (including BVPs) and inverse problems on arbitrary domains in one pass, without retraining, iterative optimization, or graph construction. We evaluate GAIA on seven 2D and 3D benchmarks, four of which are new or substantially extended benchmarks for inverse problems and BVP: electrical impedance tomography, optical tomography, 3D Darcy flow on varying geometries, and a modified setting of Poisson BVP on mechanical components benchmark (MCB). GAIA sets new state-of-the-art results on every inverse and BVP task, reducing median relative $L^2$ error by 64% on airfoil flow reconstruction and 27% on EIT relative to the next best amortized method, and outperforming all baselines on every shape category of MCB. On other forward problems, GAIA is competitive with specialized solvers while maintaining stable accuracy across point resolutions on which transformer-based baselines degrade.

Quantization has become an invaluable tool to reduce memory requirements and inference speed of modern language models, in particular to make them available for consumer setups and edge devices. While previous work has primarily focused on uniform quantization codebooks, such approaches are prone to suboptimal representations due to low-frequency high-magnitude weights. We introduce Log$_\text{b}$Quant, a novel logarithmic quantization approach with adjustable bases, to adapt to common parameter distributions. We show that our method exhibits superior performance at 4-bit precision on several performance benchmarks compared to asymmetric linear quantization at tensor-wise granularity, while achieving moderate speedup and high memory savings, making it suitable for private use on consumer-grade GPUs.

Zeroth-order (ZO) optimization enables fine-tuning large language models when backpropagation is unavailable or memory-prohibitive, but existing methods often perturb full model weights or randomly constructed low-dimensional subspaces, yielding high-variance estimates and limited performance. We propose ZO-Act, an activation-informed ZO fine-tuning method that restricts perturbations to a fixed low-rank subspace derived from input activations. For each linear layer, ZO-Act computes a small activation basis once at initialization and optimizes only lightweight coefficient matrices using forward-only loss evaluations. This reduces the effective perturbation dimension, exposes explicit trainable variables compatible with momentum-based optimizers such as Adam, and naturally supports quantized LLM fine-tuning by keeping low-bit weights frozen. We analyze ZO-Act as zeroth-order optimization over a restricted coefficient space and show that perturbing the low-dimensional coefficients reduces both the variance-dependent convergence term and the finite-difference error of the ZO estimator, at the cost of a controlled subspace approximation bias that is mitigated by the low-rank structure of LLM activations and gradients. Experiments on Llama-3-8B, OPT-13B, and INT4 Llama-3-8B show consistent gains over strong ZO fine-tuning baselines across language understanding, question answering, and commonsense reasoning.

Muon has recently emerged as one of the most effective optimizers for training large neural networks, yet its empirical success has been explained from several different perspectives. In this paper, we propose a simple mechanistic interpretation: Muon can be understood as an implicit residual connection during training. Specifically, orthogonalizing the update can sacrifice some immediate gradient fidelity while improving representation preservation for downstream layers. We study this trade-off in controlled linear optimization settings, where Muon can learn representations that are slower to fit a local target but easier for downstream layers to exploit. Our results suggest a conceptual explanation for Muon and a design perspective for optimizers that balance local descent with downstream usability.

LLM agents are rapidly being deployed in production, including coding assistants, customer-support chatbots, and scientific research assistants, yet they remain fundamentally static in enterprise deployment. The LLM weights, system prompts, tool repertoires, and in-context harnesses are frozen at deployment time, and any improvement requires a manual loop of human-curated data collection, offline fine-tuning, modification of the agentic paradigm, and re-deployment. Recent work on self-evolving agents, such as OpenClaw for individual users, indicates that the next leap in agent capability will come from agents that continually learn from their own experience. In this paper, we argue that this vision for self-evolving agent deployment is being held back for enterprise-level large-scale agentic service not by reinforcement learning (RL) algorithms but by agentic online RL systems. Specifically, current agentic RL systems and the surrounding observability software stack are inadequate along three essential aspects: (i) there is no standardized agent trajectory data protocol capable of carrying RL learning signals at step granularity across heterogeneous agent paradigms; (ii) there is no enterprise-grade comprehensive data proxy that converts real workloads into governed learning substrates; and (iii) there is no unified agent evolution control plane that automatically decides, based on trajectory statistics, when to update policy weights or evolve the in-context harness. The next generation of agentic RL systems must be co-designed around these three pillars, and we sketch concrete architectures, case studies, and counter-arguments. We instantiate one branch through AReaL2.0, reorganizing existing RL infrastructure into an agent-oriented online RL loop for policy weight updates from deployed workloads.

University stakeholders often face difficulties in accessing timely and reliable information, especially in developing countries, where there are very few intelligent support systems. Existing rule-based chatbots are unable to handle complex, domain-specific queries and are not well-equipped to adapt to evolving institutional policies. As a fill-in-the-gap solution, we present the multimodal university chatbot with retrieval-augmented generation. The system combines the large language model with semantic retrieval to produce context-based responses from institution-centric resources, such as the university handbook. The system accepts text and image queries through the vision-language model and applies quantized inference for rapid deployment on constrained hardware. A scalable backend built with FastAPI, adjoined with a responsive frontend developed with Next.js, ensures real-time usability. Our multimodal evaluation demonstrates that the system maintains strong satisfaction scores across both text and image queries, despite increased response time for visual inputs. Furthermore, quantitative evaluation shows that hallucination is reduced from 31.7% to 6.6% in our proposed RAG-based system, confirming the effectiveness of retrieval grounding.

Robot policies inevitably encounter failures when deployed in real environments. Naive retries often repeat the same mistakes, while many existing recovery methods rely on human intervention. In this paper, we propose Failure-Aware Retry (FAR), a framework that enables robots to learn from previous failures at test time, adapt their behavior accordingly, and eventually complete the task autonomously. FAR combines Failure-Contrastive Preference Adaptation, which constructs preference learning data from failures to steer the policy away from previously unsuccessful behaviors, with lightweight action perturbations during retries to encourage local exploration. We further incorporate successful recovery trajectories into a training loop for continual policy improvement. Experiments in both simulation and real-world manipulation tasks show that FAR substantially improves success rates and robustness, with average gains of 17.6% over the standard diffusion policy in simulation and 11.7% in the real world. In addition, FAR significantly improves data efficiency under both reset and timestep budgets during continual policy improvement by exploiting informative failure cases.

We present SynLaD, a latent diffusion framework for small-molecule generation that unifies ligand-based drug design objectives (what to make) with synthetic accessibility (how to make it). Current models typically optimize one objective at the expense of the other, creating a bottleneck for discovering high-scoring and synthesizable molecules. SynLaD combines reaction-constrained generation with pharmacophore-conditioned 3D design by learning a latent space that decodes to both 3D structures and synthesis pathways. An encoder maps molecules to a latent representation used by two decoder heads: (i) a geometric head that reconstructs atom types and coordinates and (ii) an autoregressive synthesis head that outputs synthetic routes in a serialized, reaction-based notation. A diffusion transformer generates novel latents in the learned space, conditioned on pharmacophore profiles. Across analogue generation tasks for bioactive ligands, SynLaD outperforms existing baselines in synthesizable and diverse hit generation, demonstrating that a single model can produce shape-aligned molecules with feasible synthesis plans.

In Large Language Model (LLM) training, data mixing plays a pivotal role in determining model performance. Recent methods optimize mixture weights via proxy models, but they rely on the assumption of static data distributions. As a result, when the underlying data pool shifts, these methods require costly retraining from scratch. This limitation restricts their ability to scale seamlessly from small settings to larger data pools and model sizes. In this paper, we propose CausalMix to address this limitation by casting data mixture optimization as a causal inference problem. We formulate the statistical features of the data pool as covariates and the domain mixture as the treatment. After fitting a causal model on 512 runs of Qwen2.5-0.5B to estimate the Conditional Average Treatment Effect (CATE), we extrapolate the optimal mixture for an 800K data pool and apply it to train a 7B model. Furthermore, we successfully generalize the framework to long chain-of-thought data on Qwen3-4B-Base. By leveraging causal modeling to isolate confounding biases, CausalMix dynamically infers state-dependent optimal data mixtures. Extensive experiments show that the mixture guided by CausalMix consistently improves performance across multiple downstream tasks, outperforming RegMix and other baselines. In addition, we use the CATE Interpreter to provide visual analysis of the learned mixing strategy. Overall, CausalMix offers a causal and interpretable framework for optimizing LLM data mixtures.

Open-response evaluation provides stronger clinical validity than multiple-choice benchmarks but creates a scoring bottleneck that motivates automated LLM-asa-Judge approaches. Whether such evaluators replicate clinical calibration and caution, however, remains untested. We introduce MedQADE, the first standardised open-response clinical benchmark for German, a major clinical language lacking native evaluation infrastructure, comprising 3,800 items annotated by ten practising physicians and nine Large Language Model (LLM) evaluators. The top-performing evaluator model, Gemini 3 Flash, reached alignment consistent with the physician ceiling (\k{appa} = 0.694 vs. \k{appa} = 0.709), though wide confidence intervals limit interpretation. Despite this statistical alignment, automated evaluators exhibited near-absent clinical metacognition: physicians scaled abstention with item difficulty, while frontier models assigned definitive scores in every case. We additionally quantified systematic lineage-dependent biases, where models preferentially scored architectural siblings, an effect independent of language. These results show that statistical alignment does not ensure clinical caution, and that evaluator independence requires explicit verification.

High-throughput RLHF systems often decouple rollout generation from policy optimization, leading to the use of stale rollouts during learner updates. In this work, we study the effect of such staleness in asynchronous GRPO. We make the behavior policy explicit in the GRPO surrogate objective and distinguish between the surrogate-gradient mapping used by the learner and the true total derivative of a distribution-dependent population objective. Under assumptions of local boundedness, distributional smoothness, and behavior-policy smoothness, we show that stale rollouts introduce a per-step surrogate-gradient bias of order O(S * eta), where S denotes the maximum rollout lag and eta denotes the learning rate. We further derive a conditional collapse-time scaling law: when within-cycle drift remains below a batch-level clipping radius, collapse is governed primarily by cumulative learner drift T * eta; when the stale-rollout constraint is active, stability instead depends explicitly on S * eta. This yields a two-constraint stability condition eta << min{R_batch / (S * G_upd), R_crit / (T * G_upd)}, explaining why the maximum stable learning rate may appear weakly dependent on staleness in the horizon-limited regime.

ROS 2 (Robot Operating System 2) has emerged as the de facto standard for modern robot software development, with middleware implementations such as the Data Distribution Service (DDS) and Zenoh forming the core infrastructure for distributed robotic communication. Despite their architectural flexibility, these middleware systems exhibit structural limitations, particularly under dynamic and resource-constrained wireless environments. This paper presents a systematic survey of ROS 2 middleware and introduces a conceptual framework to examine its architectural limits through three structural dimensions required by distributed robotic systems, namely Space, Time, and State. We first provide a structured analysis of middleware architecture and operational dynamics, including discovery, data exchange, and state management mechanisms. Building on this foundation, we formalize Time as temporal predictability for control loops, Space as spatial abstraction from physical topology to enable modular deployment, and State as contextual continuity despite dynamic node participation and intermittent connectivity. Through a comprehensive review of existing implementations and prior studies, we organize middleware research according to the structural trade-offs that arise among these dimensions. Under constrained wireless conditions, spatial abstraction can obscure network variability and weaken temporal guarantees, while mechanisms that preserve state continuity introduce computational and network overhead that competes with time-critical communication. These interactions reveal structural trade-offs that characterize the practical limits of contemporary robot middleware. By synthesizing architectural patterns and identifying gaps in current modeling and analysis approaches, this survey outlines a principled research roadmap for robust and scalable robotic middleware architectures.

Presentation Attack Detection (PAD) serves as a crucial safeguard for face recognition systems against presentation attacks such as printed photos, replayed videos, and 3D masks. Despite significant progress, existing PAD models still struggle to generalize across unseen domains due to variations in sensors, lighting, and attack materials. Recent Vision-Language Models (VLMs) have shown strong generalization ability, yet their applications in PAD remain limited because learned prompts, typically optimized under class-label supervision, fail to explicitly align with fine-grained attack-relevant visual semantics. As a result, the learned representations often overfit domain-specific artifacts instead of capturing transferable attack cues. To address this, we propose Concept-Informed Prompts Guided Presentation Attack Detection (CPG-PAD), a framework that introduces model-level concept guidance into the prompt learning process. Specifically, we design a Visual Concept-driven Enhancement (VCE) module that employs eXplainable AI (XAI) techniques to automatically discover PAD-relevant visual concepts and generate concept-associated heatmaps providing localized fine-grained guidance. Guided by these heatmaps, a Prompt-based Concept Injection (PCI) mechanism integrates these concepts into the prompt space through a Visual-Prompt Decoder (VPD) and a concept-mapping loss, enabling prompts to align with the model's internal concept space. This design enables CPG-PAD to capture generalizable and domain-invariant attack cues while effectively suppressing dataset-specific biases. Extensive experiments across nine benchmark datasets demonstrate that CPG-PAD consistently achieves state-of-the-art cross-domain performance under multi-source, limited-source, and single-source settings.

Memory has emerged as a cornerstone of modern LLM-based agents, supporting their evolution from single-turn assistants to long-term collaborators. However, memory is not always beneficial: retrieved memories often induce a critical issue of sycophancy, causing agents to over-align with the user at the cost of factual accuracy or objective reasoning. Despite this emerging risk, existing memory benchmarks primarily evaluate whether memories are correctly stored, retrieved, or updated, while overlooking how retrieved memories influence downstream reasoning and decision-making. To bridge this gap, we propose MemSyco-Bench, a comprehensive benchmark for evaluating memory-induced sycophancy in agent systems. MemSyco-Bench measures when memory should influence a decision and how valid memory should be used. Specifically, it covers five tasks that assess whether agents can reject memory as factual evidence, respect its applicable scope, resolve conflicts between memory and objective evidence, track memory updates, and use valid memory for personalization. All related resources are collected for the community at https://github.com/XMUDeepLIT/MemSyco-Bench.

Vision-language models (VLMs) have achieved strong performance across diverse multimodal tasks, yet they remain vulnerable to unreliable reasoning. Existing self-correction methods mitigate these issues but typically rely on post-training or carefully engineered feedback, incurring high computational cost. In this work, we revisit this challenge through the lens of emotional cues, asking whether they can activate latent self-correction behaviors in VLMs without additional training. \textbf{We find that emotional signals serve as an effective trigger for self-correction, encouraging more cautious and reflective reasoning}. Motivated by this finding, we propose \escabstract (\textbf{\underline{E}}motional \textbf{\underline{S}}elf-\textbf{\underline{C}}orrection), a training-free self-correction framework. ESC introduces an external verifier that detects potentially incorrect initial responses and injects emotional feedback to encourage model to reflect, and produce a better revised response without additional training. Extensive experiments across safety, hallucination, vision-centric perception, and multimodal reasoning benchmarks show that ESC consistently improves reliability while preserving overall model utility. These results suggest that emotion can function not only as an ability to be recognized, but also as a practical control signal for scalable self-correction in VLMs. \textbf{We therefore believe that ESC provides a strong foundation for a new reliable human-like, emotion-integrated research direction.} Our project is publicly available at \textcolor{red}{https://genai4e.github.io/ESC/}.

This paper introduces Svarna, a free, open-source, web-based corpus workbench for modern Greek. Svarna integrates five databases covering various registers, institutional, literary, dialectal, social media, and historical, to provide a total of more than 507 million words and around 29 million sentences. This platform addresses the chronic gaps in Greek language technology. Although various corpus resources exist, they are scattered across different platforms, and in many cases, institutional access is restricted or they are no longer available online. Svarna integrates these resources into a single interface that can be used without logging in, installation, or specialized training. This system provides a concordancer with KWIC marking capabilities, frequency analysis including register-by-register normalization, collocation extraction using mutual information, a dictionary of 93 Greek discourse markers providing distribution profiles, text-level analysis tools including n-grams, variants, and collocation networks, register comparison using log-ratio, regular expression search, and an optional LLM layer for pragmatic annotation and free research mode. This platform is built upon SQLite FTS5 full-text indexes provided via a FastAPI backend, deployed as Docker containers on Azure, and released under the MIT license. Source code, build scripts, and deployment configurations are publicly available on GitHub. Users can add their own corpora and deploy their own instances. This document describes the system design, corpus structure, and use cases demonstrating the various queries supported by the platform. Svarna serves as the first step in exploring available data and is expected to lay the foundation for more comprehensive research in the future.

In retrieval augmented generation (RAG) and agentic LLM serving, prompts are assembled from independent segments into long contexts, making the prefill stage dominate the per-request computation cost. To this cost, two directions have emerged in parallel: position-independent caching (PIC) admits KV reuse for non-contiguous segments shared across different requests, while hybrid-attention models reduce computation complexity by replacing most full-attention layers with linear attention. However, they cannot coexist: applying PIC to hybrid-attention models breaks down because per-token KV-cache reuse primitives do not transfer to the per-request recurrent state. In this work, we present Hypic, the first serving system for hybrid-attention LLMs with position-independent caching. For linear-attention layers, we identify the segment-cumulative transition operator as the missing algebraic primitive, and cache it alongside each segment's zero-start end-state, enabling near-exact and constant-time state composition of independently cached segments. For the remaining full-attention layers, existing PIC methods also fail as linear layers do not expose the per-token hidden states for selective recomputation. We show that the most significant attention deviation concentrates at segment boundaries, so recomputing only a small seam window at each boundary suffices to restore cross-segment lookback. Finally, Hypic exploits segment-level self-containment to parallelize cache-miss prefill across instances, turning long cold requests -- a major tail-latency contributor under both prefix caching and prior PIC -- into an accelerable workload. Evaluated across four hybrid-attention models and five workloads, Hypic reduces time-to-first-token (TTFT) by 2.45x on average and improves peak throughput by up to 2.0x over existing systems, while staying within 3.3 points of full-recompute accuracy.

Most existing audio classification methods suppose that each query (testing) sample belongs to a class of support (training) samples, and misrecognize samples of unseen classes as seen classes (cannot reject samples of unseen classes). In this study, we propose a method for Few-shot Open-set Audio Classification (FOAC), which can recognize query samples of seen classes after updating the model using a few support samples, and meanwhile reject query samples from unseen classes. We design a model consisting of an encoder and a classifier. The encoder is the backbone of a ResNet used for extracting embeddings. The classifier consists of prototype generators of few-shot classes and open-set classes. Prototypes of few-shot classes are obtained by fusing the class-discriminative information of support and query embeddings and by assigning larger weighting coefficient to representative part of the support embeddings. One prototype is generated for open-set classes using the proposed prototype generator. The encoder is trained with abundant samples of base classes in supervised manner, and then the prototypes of base classes are generated under the supervision of a joint loss. The classifier is trained using a few samples of few-shot classes in a meta-training way. Three public datasets (LS-100, NSynth-100, and FSC-89) are used to assess the performance of our method. Experiments show that our method has advantage over prior methods in AUROC and accuracy. This advantage has statistical significance for most prior methods. Our method has lower computational complexity than most prior methods. The code is at https://github.com/Jessytan/FOAC-AIFP.

Acoustic imaging visualization is a core methodology in acoustics, enabling spatial analysis of sound sources and acoustic scenes. However, limited sensor availability in practical systems motivate approaches that enhance spatial resolution without increasing the hardware complexity. In this paper, we focus on upsampling virtually a tetrahedral 4-microphone array to a spherical 32-microphone array by estimating the covariance matrices of the channels employing deep learning techniques. Five neural network architectures are investigated for covariance upsampling for acoustic imaging using the real-world STARSS23 dataset. These models are developed to estimate a 32-microphone, time-frequency covariance matrix from a 4-microphone input covariance representation. The proposed architectures are based on 2D convolutional layers to capture the underlying spatial-spectral structure of covariance matrices, and are further enhanced with frequency dynamic convolution to model their frequency-dependent properties. The proposed architectures are evaluated in terms of root mean square error (RMSE) and using delay-and-sum beamforming acoustic imaging. Quantitative results show that all models outperform a random-guess baseline, which yields an RMSE of 0.548, with the best-performing architecture achieving an RMSE of 0.432. We analyze qualitatively the performance of the proposed models through beamforming heatmap visualizations derived from the 4-channel input covariance, the 32-channel ground truth, and the predicted 32-channel covariance matrices. These results demonstrate that covariance upsampling significantly enhances the effective performance of the 4-channel microphone array, producing sound maps that closely resemble those obtained with the 32-channel array.

We present RuleChef, a framework that uses large language models (LLMs) to generate executable rules for NLP tasks such as text classification, Named Entity Recognition (NER), or relation extraction. Rules are generated based on a task description and a set of labeled examples, then they are iteratively improved based both on additional examples and on human feedback overexisting rules. RuleChef can also be used to bootstrap rules using the observed input-output pairs from any existing model for a given task. LLMs are used only at learning time, synthesizing rules and iteratively patching them based on failures measured on a held-out split. The result of this process is a fast, deterministic, and inspectable rule system. Preliminary evaluation is performed on both classification and NER tasks. We release RuleChef as open-source software under an Apache 2.0

Private continual counting is a fundamental problem in differential privacy: given a binary stream of length $n$, where each $1$ corresponds to the contribution of one individual, the goal is to release all running counts while protecting the privacy of each individual. The standard algorithm is the binary tree mechanism, whose Gaussian-noise variant achieves expected $\ell_\infty$ error proportional to $\log^{3/2} n$ for approximate differential privacy. Whether this dependence on the stream length is necessary has remained a central open problem. In this work, we resolve the dependence on $n$ by proving that every differentially private mechanism for continual counting must incur expected $\ell_\infty$ error $Ω(\log^{3/2} n)$. This shows that the binary tree mechanism is asymptotically optimal in the approximate-DP setting. As a consequence, we also obtain a largest-possible separation between hereditary discrepancy and private $\ell_\infty$ error for linear queries, showing that the known general upper bound in terms of hereditary discrepancy has the optimal dependence on the number of queries.

This paper introduces a probabilistic framework and hybrid validation methodology for V2X-enabled Collective Perception (CP) in complex traffic scenarios. The proposed Bayesian fusion algorithm extends the perceptual horizon of connected and autonomous vehicles by integrating heterogeneous sensor observations from multiple agents into a shared probabilistic occupancy grid. Each cell of this grid encapsulates both occupancy likelihood and uncertainty, enabling explainable and trustworthy situational awareness beyond the ego vehicle's field of view. To bridge the gap between simulation and real-world evaluation, a hybrid testing framework is developed, combining CARLA-based virtual environments with vehicle-in-the-loop experimentation. Experimental results in a roundabout scenario demonstrate a 260 percent increase in field-of-view coverage and a rise in occupied-cell recall from 0.82 (ego-only) to 0.94 (six-agent CP) under nominal localization conditions. Overall, the proposed approach provides a reproducible and interpretable foundation for validating CP systems, supporting the safe and certifiable deployment of cooperative autonomous vehicles.

World models are increasingly used in embodied intelligence and generative simulation, yet their scope remains ambiguous across communities. This tutorial presents a design-space view of world models as action-conditioned predictive models that estimate the future evolution of task-relevant observations or states. We categorize existing methods into observation-space and state-space world models, comparing their trade-offs in visual fidelity, spatial structure, physical interpretability, and control usability. We further introduce world action models, which connect predicted futures with executable robot actions, and summarize four representative paradigms: imagine-then-execute, video-feature-conditioned action prediction, joint video-action modeling, and auxiliary video prediction for policy learning. The goal of this tutorial is to clarify the conceptual scope of world (action) models and provide a structured taxonomy for embodied prediction and control.

Evolutionary multitasking (EMT) has shown strong capability in solving multiple optimization problems simultaneously by exploiting latent inter-task consistency, such as similarities in promising solutions or search directions. However, most existing EMT studies remain focused on objective-driven optimization, where such consistency is mainly used to accelerate convergence toward predefined optima. In this paper, we move EMT from consistency to collaborative discovery and propose a multifactorial evolutionary algorithm with collaborative discovery (MFEA-CoD) for multitask novelty search. Unlike conventional EMT, MFEA-CoD coordinates multiple novelty search tasks to collaboratively discover behaviorally novel solutions rather than merely transferring consistent search information for faster convergence. Specifically, a multitask repulsion operator encourages different tasks to explore distinct regions of the unified search space, thereby reducing redundant behavioral discoveries. Meanwhile, an adaptive inter-task transfer mechanism exploits shared discovery opportunities in overlapping novelty-improving regions by adjusting the transfer probability according to the online contribution of transferred information. Furthermore, MFEA-CoD is extended to multitask novelty-augmented optimization, where behavioral novelty is jointly considered with objective information to alleviate premature convergence caused by deceptive objectives. Experiments on synthetic basin-type problems, deceptive maze navigation problems, MuJoCo policy optimization problems, and generative novelty search problems demonstrate that MFEA-CoD improves the efficiency of discovering diverse novel solutions and shows clear advantages in deceptive objective landscapes.

Accompanying a group of humans is an essential aspect of developing human-like social cognition in robots. However, human groups typically do not follow fixed formations, which poses significant challenges for robots in maintaining natural companionship behaviors. In this paper, we propose an adaptive group-accompaniment method for social robots based on Vision-Language Models (VLMs), leveraging their semantic reasoning capabilities to infer companion positions, maintain social distances, and understand group dynamics. The members of the group are first detected, and a perceptual module generates visual representations of the interaction group space as input to the VLM, which is then combined with a Model Predictive Path Integral (MPPI) controller to ensure stability and safety. Experimental evaluations across five scenarios show that the proposed method enables robots to accompany the group effectively, demonstrating a 15\% improvement in success rate and a 25\% reduction in collision rate compared to baseline approaches. Additionally, a user study indicates that the generated companionship behaviors are perceived as natural and socially appropriate.

Multilingual fluency often invites a stronger assumption: a model that can speak a user's language must also understand the culture encoded by that language. We call this the Illusion of Cultural Alignment. To test this assumption directly, we introduce MSQA, a benchmark of 1,064 natively sourced questions across 11 language groups, five cultural dimensions, and three difficulty tiers. Unlike translated benchmarks, MSQA targets locally grounded knowledge and reduces shortcuts from English-centric cross-lingual transfer. Evaluating 18 LLMs, we find substantial cultural degradation and a pronounced Locality Effect: cultural competence tracks pre-training exposure more closely than general reasoning ability. We further show that common inference-time remedies do not dissolve the illusion. Models remain overconfident on unfamiliar cultural questions, repeated sampling yields unstable rather than reliable correctness, and retrieval augmentation helps unevenly on long-tail facts. These findings indicate that cultural alignment cannot be inferred from multilingual ability alone and requires deeper intervention than calibration, sampling, or retrieval at inference time

Public lithium-ion battery datasets are increasingly used for state-of-health estimation, remaining-useful-life prediction, anomaly detection, electrochemical diagnostics, second-life analytics, and battery safety research. However, these datasets vary substantially in chemistry, modality, scale, label quality, sequence structure, access status, and preprocessing complexity. These differences directly affect whether a dataset is feasible for near-term hybrid quantum-classical machine-learning workflows. This paper presents IonSense-QKG, a quantum-readiness metadata framework for lithium-ion battery dataset discovery. Starting from the EV-Battery-IonSense index, the proposed framework enriches public battery dataset records with quantum-relevant metadata, including task type, sensing modality, chemistry, label availability, sequence type, preprocessing requirements, candidate quantum encodings, estimated qubit range, and NISQ feasibility. A transparent Quantum Readiness Score is introduced to rank datasets as candidate resources for future hybrid quantum-classical battery benchmarks. The score is intended as a dataset-selection heuristic, not as evidence of quantum advantage. The framework demonstrates query-based discovery over enriched metadata to identify datasets suitable for compact quantum feature maps, quantum time-series workflows, limited-label anomaly detection, and future battery-health benchmarking. The released artifact includes metadata tables, scoring scripts, robustness checks, link-checking utilities, and SQL-style query examples. IonSense-QKG positions dataset selection as a data-management problem and provides a reproducible foundation for data-centric quantum battery analytics.

Grid-based approaches to approximate nearest neighbor (ANN) search have been absent from modern scaling analyses. We present a systematic characterization of a multiprobe grid algorithm with respect to dataset size $N$ and dimensionality $d$. Our experiments reveal a previously unreported $d$-scaling crossover on the GloVe embedding family, in which multiprobe grid search maintains an approximately constant dimensional scaling exponent while other graph-, tree-, and partitioning-based methods exhibit degrading throughput. The advantage comes with near-linear query scaling in $N$, but also with lower indexing cost than competing ANN methods. Our results suggest that grid-based methods such as multiprobe grid may be competitive in rebuild-heavy or high-dimensional settings where indexing cost and dimensional robustness dictate performance. More broadly, recent work has formalized self-attention as an ANN operation. Thus, the $N$- and $d$-scaling properties of ANN algorithms may guide cost analysis of efficient transformer architectures. Code is available at: https://github.com/weiz345/MultiProbeANN.

In light of strides in Arti cial Intelligence (AI) and its wide spread application, challenges persist in the interpretability of AI models, particularly within specialized domains like healthcare, such as electro cardiograph (ECG) recognition. Rather than relying solely on end-to-end convolutional neural networks, this paper introduces a novel approach using a domain knowledge-based graph convolution network for ECG recognition. Key landmarks points of PRQST, vital to ECG interpreta tion, are incorporated as domain knowledge. The double-stream directed graph is employed to model both intra and inter ECG cycles. Speci cally, spatial directed graphs capture the positional relationships among key points, while temporal directed graphs delineate temporal dependencies between adjacent cycles in extended ECG sequences. Experimental re sults on the First Chinese ECG Intelligent Competition dataset, which speci cally classify ECG into nine categories, prove the e cacy of the proposed model. The overall average F1 score is 88.1%, the average F1 score of rare categories is 76.3%, both outperform the state-of-the-art models. The introduction of domain knowledge did enhance the detec tion performance, especially for rare categories.

Programming-by-example systems infer programs from a small set of input-output examples. Robust PBE work usually models wrong examples as samples from a stochastic noise process and then minimizes an expected or empirical loss. This paper studies a different failure mode: an adversary who sees the synthesizer and chooses the examples whose corruption most damages the returned program. We formalize fixed-set worst-case corruption for finite PBE version spaces, implement exact-within-bounded-pool and heuristic corruption searches for a string-transformation DSL, and introduce version-space partition aggregation (VPA), a defense that synthesizes on disjoint example groups and votes by semantic signatures. The central claim is deliberately bounded and partly negative: low-margin PBE tasks have an adversarial robustness dimension that random-typo and noisy-PBE evaluations miss, while semantic partition aggregation helps only when the clean semantics keep a partition vote margin, which often fails on realistic tasks. Evidence from curated/generated DSL tasks, accepted public SyGuS PBE_SLIA slices, SYNTRA Playgol v2, and noisy-PBE objective baselines supports that boundary. One curated edit flips all 8 spike tasks while 200-trial typo, DSL-pool, and distance-matched random controls succeed on 10.3%, 11.0%, and 16.7%; generated margin-1 rows flip under budget 1 yet VPA recovers them; on public SyGuS the vote margin is near one, so an adaptive attacker drives VPA accuracy to zero; accepted public SyGuS slices move across exact-within-pool budget boundaries; and Playgol shows positive paired-bootstrap gaps against typo and same-pool random controls on the 141 accepted rows. A small exact-output prompt harness over 20 controlled margin-1 tasks shows the same qualitative clean-to-attacked pattern across local and API models, while it is treated as a scope check, not a broad LLM benchmark.

Cross-subject EEG stress detection remains challenging because discriminative stress-related patterns are both subject-dependent and frequency-specific. Conventional Riemannian methods model spatial covariance mainly in the time domain, overlooking neural oscillations that are critical for high-level cognitive state decoding, while standard temporal tokenization often fragments inter-slice temporal coherence. To address these limitations, we propose \method{}, an Intra-Inter Riemannian Manifold Attention Network for EEG-based stress detection. \method{} constructs spatial covariance matrices independently at each frequency point and maps them to the SPD tangent space, preserving channel-wise geometry together with frequency-specific discriminative cues. It further introduces frequency cluster aggregation to select informative spectral components and reduce redundancy by forming compact, data-driven frequency clusters aligned with EEG rhythms. Finally, an intra-inter slice attention module adaptively integrates local slice-level spectral dynamics and global temporal context across EEG sequences. Experiments on three datasets show that \method{} consistently outperforms five state-of-the-art baselines, achieving up to 82.78\% balanced accuracy while remaining efficient with only 1.60M parameters and 31.95M FLOPs.

The research proposes a multilayer Q-matrix-embedded neural network for cognitive diagnosis (M-QCDNet), which integrates the structural interpretability of cognitive diagnostic models (CDMs) with the deep learning neural network (NN). M-QCDNet structures the item-skill relationship using the Q-matrix as a structural prior, ensuring latent mastery profiles remain interpretable and consistent with cognitive theory, followed by the proposed loss function with an L2 penalty to penalize skills not aligned with the Q-matrix and to balance predictive performance and structural alignment. Corresponding evaluation matrices, the interpretable alignment-based metrics that quantify the degree to which predicted skill activations correspond to item-level skills, were further developed. M-QCDNet offers practical benefits for classroom practice, enabling early detection of learning difficulties and supporting mastery-based interventions. By embedding diagnostic validity into model design, M-QCDNet bridges psychometric transparency and neural flexibility, advancing interpretable, fair, and actionable AI for cognitive diagnostics.

In prefill-decode (PD) disaggregated LLM serving, each request is assigned to a decode worker after prefill. Existing decode routers balance only load; for mixture-of-experts (MoE) models this is incomplete: equally loaded workers can differ in latency, since each decode step loads the weights of every distinct expert its batch activates. We present ELDR, an expert-locality-aware decode router for PD-disaggregated MoE serving. From a request's prefill expert activations, ELDR builds an expert signature predicting the experts it will activate during generation. Offline, balanced K-means partitions signature space across decode workers; online, locality-band routing sends each request to the least-loaded worker among those best matching its signature. A signature cache, co-indexed with the KV cache at KV-block granularity, keeps signatures exact under prefix caching. Implemented in vLLM and evaluated on deployments of up to 40 GPUs, ELDR reduces median TPOT by 5.9-13.9% over the strongest of four load-balancing baselines across three MoE models and two workloads, with model outputs unchanged.

Embedding models are essential components of modern Information Retrieval (IR) systems, yet they are typically hidden behind APIs. Recent works have shown that dense IR system can lead to security vulnerabilities such as embedding inversion attacks. However, such attacks usually require that the attacker knows the embedding model for the attack to be applicable. In this paper, we study IR systems under a black-box setting in which the adversary observes only the unordered set of retrieved documents, without ranking or similarity scores. We demonstrate that in such contexts, tailored queries allow an adversary to identify which embedding model is in use from a set of known model candidate, which we coin as an embedding inference attack (EIA). We also show that certain queries remain discriminative even when the system includes a reranker as a potential defense mechanism. We further validate our method on a real Retrieval-Augmented Generation (RAG) system, in which the tailored queries bypass the LLM's tendency to reject inputs it does not recognize as well-formed questions. Finally, we propose and evaluate other mitigation strategies such as similarity thresholds.

We develop a quantitative theory for the emergence of quantum many-body chaos as integrability is broken via a tunable parameter. In a circuit model of free fermions, 'doped' with a tunable density of integrability-breaking gates, we uncover the microscopic mechanisms underpinning the crossover from early-time integrable behaviour to late-time chaos through the lens of the out-of-time-ordered correlators (OTOCs). The integrability-breaking gates act as local, in spacetime, hotspots which locally amplify the OTOCs such that an accumulation of them eventually leads to fully-developed chaos. We identify the explicit characteristic time and length scales governing this crossover, as well as the dependence of the chaotic OTOC characteristics -- such as the butterfly velocity and front broadening -- on the integrability-breaking parameter.

The dynamics of a quantum system encode signatures of whether the underlying Hamiltonian is integrable or chaotic, giving rise to the concept of quantum information scrambling through the properties of the resulting dynamical states or operators. We introduce an information-theoretic framework based on the Haar-averaged sum of total correlations (aSTC), together with average genuine multipartite entanglement generated dynamically from initially fully separable states, as robust probes of quantum information scrambling. Using the long-range quantum XYZ spin model in transverse and longitudinal magnetic fields, whose integrable limit is the nearest-neighbor transverse XY model, we demonstrate that the long-time average and, more importantly, the temporal fluctuations of the aSTC provide a faithful and system-size-independent signature of integrable and chaotic dynamics, similar to the conventional measure of scrambling, out-of-time-ordered correlator (OTOC). When the system is in contact with the thermal reservoir and system-bath coupling follows Markovianity, we find that the fluctuations of the aSTC and OTOC continue to distinguish integrable and chaotic dynamics only at intermediate times. However, we observe that in the non-Markovian domain, information backflow restores the scrambling dynamics, enabling the aSTC to retain its distinguishing power even at long times. Interestingly, we exhibit that, under Markovian amplitude damping and non-Markovian dephasing noise, the temporal fluctuations of the aSTC can discriminate between integrability and non-integrability in the weak Markovian regime, even when OTOC fails to do so.

We consider a very general Rosenzweig-Porter-type model, $H=H_0+λW$, where $H_0$ is an arbitrary Hermitian matrix and $W$ is a standard Wigner matrix. We precisely trace the localization properties of the eigenvectors and the eigenstate thermalisation hypothesis (ETH) as the coupling constant $λ$ interpolates between the trivial $λ=0$ case and the fully mean field regime of large $λ$. Our results hold uniformly in $H_0$ and $λ$, substantially generalising all previous local laws on deformed Wigner matrices even in the mean field regime. Our proof precisely captures the deterministic approximation to the resolvent which exhibits a strongly inhomogeneous structure. As a byproduct, we conclude the emergence of a mobility edge and study the phenomenon of re-entrant localization.

Polar collective motion commonly arises from alignment interactions, particle anisotropy, or an imposed directional bias. Here we identify a distinct route to polar order that does not rely on alignment interactions between the active particles. We show that non-aligning active Brownian particles embedded in a dense passive medium can develop polar coherence when confined to a compact curved surface. Persistent active motion redistributes stress through the host and creates passive-depleted regions. When the stress-spreading length becomes comparable to the sphere radius, these regions merge into elongated scars that channel active motion and, through feedback with the active flux, promote a common direction of motion. Removing the passive host suppresses polar coherence even though the active particles continue to cluster on the same sphere. Our results establish an environment-mediated route to collective polarity in which symmetry breaking emerges from the coupling between active motion, passive stress redistribution, and compact geometry.

We study the outcome of adaptive learning of a large number of players engaging in sets of two-strategy two-player games. We are interested in typical games, and generate the payoff matrices at random at the beginning. The payoff matrices then remain fixed during the learning process. This provides a game theoretic foundation for the Sherrington-Kirkpatrick (SK) game, recently introduced by Garnier-Brun, Benzaquen and Bouchaud. The original model by these authors is a special case, with no bias towards any strategy. We here determine stability of learning for SK games with general random bias, and find that the nature of the stable state is affected by random fields. We also introduce a grand-canonical version of the SK game, in which players can choose to abstain. We determine the stability of learning for this game. Our analysis confirms that complex situations involving many players are frequently unlearnable, even if each player only chooses between two different actions. The rate with which players lose memory of past payoffs and the competitiveness of the game emerge as key parameters determining whether learning converges to a unique fixed point, whether there are many fixed points, or if the dynamics remains persistently volatile.

Geometric analysis of electronic Bloch states offers a universal framework for understanding electronic properties, yet its role in the transport of orbital angular momentum remains unexplored. In this work, we establish an analytic connection between orbital angular momentum transport and the geometric properties of Bloch wave functions in electronic systems. Focusing on the intrinsic orbital Hall effect in the dc limit under a spatially nonuniform electric field, we show that its conductivity can be expressed in terms of universal geometric quantities, such as the orbital Berry curvature and quantum metric. This formulation provides a term-by-term correspondence with the geometric description of intrinsic charge Hall transport established in previous studies. Using a tight-binding model, we further illustrate that the higher-order orbital Hall response can exhibit enhanced sensitivity to the orientation of an anisotropic sample. Our work deepens the understanding of diverse intrinsic transverse transport phenomena and the role of quantum geometry in electronic systems.

Helical spin textures represent the minimal realization of $p$-wave magnetism which is characterized by momentum-odd spin polarization. Independently, structurally chiral crystals exhibit momentum-odd orbital polarization arising from broken inversion symmetry. Here, we demonstrate that spin-orbit coupling couples these two independent microscopic chirality degrees of freedom, allowing the orbital polarization of a chiral crystal to generate an additional contribution to the $p$-wave spin splitting. The resulting spin-orbital state is naturally classified by the relative chirality $η=χ_{\mathrm c}χ_{\mathrm m}$, giving rise to two symmetry-distinct $p$-wave phases corresponding to homochiral and heterochiral configurations which can be directly probed by the longitudinal conductivity. These phases exhibit distinct transport signatures, establishing relative chirality as an experimentally accessible symmetry degree of freedom in chiral magnetic systems.

We study the nonequilibrium dynamics following a quantum quench in spin chains that can be solved via a mapping to free fermions in disguise. These models feature an exponential degeneracy of all energy eigenvalues, raising the question of the validity of the established framework describing the properties of integrable systems out of equilibrium. We present two main results. First, we develop an analytic method to compute the quasi-momentum distribution function characterizing the generalized Gibbs ensemble, and derive an analytic formula to compute the corresponding expectation values for special observables. Second, we conjecture a modification of the standard formula for the entanglement growth based on the quasi-particle picture, taking into account that each fermion in disguise carries an additional amount of entropy due to the exponential degeneracy of the energy eigenvalues. We test our theoretical predictions against numerical tensor-network computations for different initial states and Hamiltonian parameters. For the local observables, we find excellent agreement. For the entanglement dynamics, we find small deviations suggesting that our conjecture is only approximately correct. Our results represent a first step towards the extension of the established framework of integrable systems out of equilibrium to models hosting free fermions in disguise.

I revisit the Galam Majority Model (GMM) with contrarian agents from a statistical-mechanics perspective, revealing three fundamental features. First, in addition to the GMM simultaneous-update of small discussion groups, I construct a related single-agent stochastic dynamics, providing a Markovian microscopic representation, which is found to yield the same evolution equation. Second, I show that, contrary to what is often stated in the literature, the GMM closed evolution equation for the opinion density is not the result of a mean-field approximation. Indeed, I derive the conventional mean-field dynamics associated with majority-rule interactions and show that it yields a distinct, probabilistic evolution equation contrary the deterministic GMM equation. I therefore identify the GMM as an iterated mean-field dynamics. Third, I investigate the thermodynamic nature of the dynamics obtained from both single-agent and simultaneous updates. Both are shown to violate detailed balance. However, while Kolmogorov's cycle condition is satisfied for single-agent updates, it is violated for simultaneous updates, making the departure from equilibrium stronger in the latter case. I then compute the probability flux in the stationary state and show that it is non-vanishing, confirming the absence of an effective Hamiltonian and establishing that the stationary state is a genuine nonequilibrium steady state.These results clarify the statistical-mechanical foundations of the GMM and establish contrarian majority dynamics as an intrinsically non-equilibrium process with distinct regimes of irreversibility. Contrarians are not thermal noise.

We study mesoscopic linear spectral statistics for two ensembles of random quantum graphs. These are defined by a discrete graph $G$ and a unitary-matrix-valued function $U(k)$ indexed by directed edges of $G$. The matrix function $U(k)$ is constructed from unitary matrices $U^{(v)}$ indexed by the neighbours of each vertex $v$. The first ensemble is obtained by sampling the underlying discrete graph uniformly from the set of $d$-regular graphs. The second ensemble is obtained by sampling $U^{(v)}$ uniformly from the Haar measure, independently for each vertex. We prove that the variance of a linear spectral statistic in the large graph limit on polynomial mesoscopic scales coincides with that of the Gaussian Orthogonal/Unitary Ensemble.

Universality is one of the central organising principles of modern physics, explaining why systems with vastly different microscopic constituents can exhibit identical large-scale behaviour. While the classification of equilibrium critical phenomena through hydrodynamics and the renormalization group (RG) is now well established, our understanding of universality far from equilibrium remains far less developed. In recent years, however, rapid progress - driven in large part by developments in active and living matter - has uncovered a growing range of genuinely nonequilibrium universality classes (UCs) with no equilibrium counterparts. In this review, we present a pedagogical and unified introduction to hydrodynamic and RG approaches to nonequilibrium many-body systems. We first show how hydrodynamic theories can be systematically constructed from symmetry and conservation laws alone. We then introduce perturbative dynamic RG methods and demonstrate how hydrodynamic theories are organised into distinct UCs according to their scaling behaviour. Building on these foundations, we review the diverse nonequilibrium UCs uncovered since 2015, while emphasizing the conceptual connections and unifying physical principles underlying their emergence. We conclude by discussing open theoretical and experimental challenges for the field.

Electrochemical impedance spectroscopy (EIS) is a widely used technique to understand time-dependent response and relaxation under applied voltage. While these spectra contain a wealth of information, major gaps in our understanding can hinder our ability to interpret EIS spectra in terms of microscopic chemical mechanisms. We propose an alternative approach to common empirical fitting procedures for describing the contribution of the bulk electrolyte to the EIS spectrum. This new approach is rooted in determining the moments of the frequency-dependent conductivity, with molecular interpretability provided by a generalized Langevin equation description of an effective single particle dynamics; the `itinerant oscillator' (IO) model. In contrast to a Debye--Falkenhagen description, the IO model makes no assumptions regarding the concentration of the electrolyte, a fact we demonstrate by analysing molecular dynamics simulations of a room-temperature ionic liquid. By analysing the memory function from simulation within the framework provided by the IO model, we reveal the importance of capturing the separation of timescales within the memory function for describing the temperature dependent $β$-relaxation process. We go on to show how our impedance model directly reports on this distribution of timescales while retaining the simplicity of commonly employed workflows.

We study pressure-driven Poiseuille flow of a one-component fluid between adiabatic plates in liquid-gas coexistence. The analysis uses Poiseuille flow and Fourier heat conduction in the bulk regions together with particle and energy conservation. From these bulk equations, we identify extremely small dimensionless parameters $A^\mathrm{L}$ and $A^\mathrm{G}$ describing coexistence Poiseuille flow, whose smallness comes from squared microscopic-to-macroscopic length ratios. In weak driving with macroscopic liquid and gas regions, the pressure difference is concentrated across the interfacial region, and the ordinary Poiseuille particle current is strongly reduced. For equal-temperature reservoirs, this residual particle current produces interfacial cooling.

We derive a general relation between the velocity and force autocorrelation tensors (VACT and FACT) for a Brownian particle subject to an external magnetic field. Using time-symmetry arguments, we show that, for the full Langevin dynamics, the VACT depends only on the FACT, independently of the details of the interaction potential. Under the hypothesis of timescale separation between thermalization and interaction-driven motion, this relation simplifies considerably in the overdamped (Brownian) limit. A central feature of the overdamped result is that, unlike in the field-free case, the part of the VACT that controls the self-diffusion of the particle couples to the antisymmetric part of the FACT, with a coupling strength set by the ratio of the cyclotron frequency to the thermalization rate. We validate and illustrate the formalism on an exactly solvable model: a dimer of charged particles bound by a harmonic potential. Depending on the relative sign of the particle charges, the magnetic field is found to produce either a transient suppression of mobility and diffusion that is fully recovered at long times, or a persistent oscillatory force autocorrelation, regions of negative mobility, and a long-time suppression of self-diffusion.

If a generalized Langevin equation contains a potential of mean force, it cannot at the same time contain a linear memory kernel and a fluctuating force that obeys a second fluctuation dissipation theorem in the sense of Kubo, and be exact. As modelers often prefer to use generalized Langevin equations that have the first three properties, one needs to ask how close the model dynamics is to the dynamics of the underlying microscopic system. To test this, we analyze a simple model system in which the potential of mean force can be well approximated by a polynomial of low order. The exact generalized Langevin equation of this model contains memory terms in addition to the linear one. We show that these additional terms, at least for the model system regarded in this article, are important for the dynamics and cannot be neglected if one intends to model core aspects of the underlying system correctly.

Correlated phases of matter are typically investigated in crystalline systems, where disorder is considered to be detrimental. However, intriguing exceptions exist, such as superconductivity being enhanced in amorphous realizations of Al and Bi. Here, we demonstrate that superconductivity can even emerge entirely from disorder, using monolayer graphene as an example. In the clean limit, the semi-metallic nature of graphene requires prohibitively strong electronic interactions to achieve superconductivity. Despite the inherently random nature of disorder, we show that introducing low concentrations of vacancies or hydrogenation in graphene provides a large density of low-energy states that easily induce superconductivity. For conventional $s$-wave pairing, disorder induces a finite superconducting order parameter for arbitrarily weak attractive interactions. Rather than forming isolated superconducting puddles, global phase coherence is established through a finite superfluid weight of purely geometrical origin. Away from the chiral limit of vacancies, hydrogenation similarly yields a finite transition temperature and nonzero superfluid weight for weak interactions. For unconventional nearest-neighbor pairing, typically more disrupted by disorder, superconductivity exhibits quantum-critical-like behavior, yet phase coherence persists at low interaction strengths, with mixed $d$-wave symmetries. Our work demonstrates the robust emergence of macroscopic superconducting phase coherence engineered entirely from microscopic disorder.

We analyze particle and spin currents in networks of coupled spinor exciton-polariton condensates arranged as plaquettes and regular polygonal rings. In closed geometries, spin-conserving and TE-TM-induced spin-flip tunnelling combine to generate circulating particle currents, hidden spin counterflows, and bond-dependent spin-current patterns. For the minimal geometries - an equilateral triangle, and a square plaquette - we derive analytical expressions for edge-resolved currents from stationary configurations obtained by energy minimization. We then show how particle, in-plane spin, and out-of-plane spin currents partition the parameter plane and provide direct signatures of the equilibrium phases. Finally, we apply the same current-resolved diagnostics to larger rings, where winding numbers and a branch-invariant common-phase coherence metric organize the resulting phase structure.

Single-molecule (SM) rotational dynamics of fluorescent probes in polystyrene near the glass transition temperature ($T_g$) are investigated over long times to reconstruct susceptibility spectra. The loss spectrum, commonly recorded using external field-driven (frequency-domain) spectroscopy, such as dielectric spectroscopy, is reconstructed from purely thermal SM rotational fluctuations. The results reproduce time-temperature superposition typically seen in dielectric spectroscopy for materials near $T_g$ and show that the ensemble spectrum is comprised of individual molecular responses to distinct environments.

Remarkably common statistical laws characterize the diversity scaling and its fluctuations across a wide range of complex "component systems". These regularities are often interpreted as signatures of an underlying innovation mechanism driving the growth of component diversity, but the basic ingredients necessary for their emergence remain poorly understood. In particular, from language and technological artifacts to genomes and gene expression patterns, the number of distinct components grows sublinearly with system size, while its variance scales approximately as the square of its mean. This behavior is consistent across diverse systems, raising the question of whether general constraints or emergent principles underlying diversity and innovation define the architectures of realizations with different numbers of components. To address this question, we derive analytical conditions for the joint emergence of these two diversity laws within a broad class of growth models, showing that they require a specific asymptotic dependence of the innovation probability on diversity and system size. We then demonstrate that the same macroscopic laws arise in a different class of models with latent heterogeneity, where quadratic fluctuation scaling always emerges asymptotically as a consequence of general statistical principles, essentially the law of total variance, without explicitly assuming an innovation mechanism or any specific rule for system assembly. We compare these predictions with empirical data from language, genomes, LEGO constructions, and texts generated by large language models. Our results show that empirical diversity scaling laws strongly constrain generative models but do not uniquely identify the mechanisms generating diversity, revealing a close correspondence between innovation-driven growth models and latent-variable descriptions.

Short-range order (SRO) governs the mechanical response of multi-principal-element alloys, but designing an alloy for a target property usually means solving two disconnected problems: building a structure matching a desired SRO pattern, then separately checking its property, with no shared optimization. This work replaces the standard random-swap search (reverse Monte Carlo) with a gradient-based approach: atom occupancy is treated as continuous rather than fixed, so the whole process can be tuned using gradient descent, the same method used to train neural networks. This builder matches random-swap accuracy on small systems, but is six times faster and eight times more accurate on large 4000-atom systems, and scales smoothly to alloys with many elements without extra bookkeeping. A physics-based correction term, adapted from prior two-element work and extended here to many elements, keeps designed structures thermodynamically realistic rather than just numerically matching the target SRO pattern. A small neural network then predicts mechanical properties directly from composition and SRO statistics, closing the loop from target property back to structure. Tested on nine face-centered-cubic and body-centered-cubic alloys, the pipeline captured SRO-driven stiffness changes from -20% to +57%, and cell-size checks showed at least 864 atoms are needed to get the direction and size of these changes right, since the commonly used 108-atom cells can mislead. Against real simulations for a cobalt-chromium-nickel alloy, the method matched three of four target stiffness values within 6%. The method is released as an open-source Python package, anisro, offering a practical route to gradient-based, property-driven alloy design.

It has been known for several decades that the third moment of the multifractal spectrum of the harmonic measure for diffusion-limited aggregates is linked to the underlying fractal dimension of the cluster. We demonstrate, using an argument based on the Hastings-Levitov formulation of diffusion-limited aggregation (DLA) in two dimensions, an even stronger link, connecting the universal amplitude of the third moment to the cluster fractal dimension. This argument can be used for both the standard circular DLA as well as DLA in a cylinder (i.e., with periodic boundary conditions).

Non-Abelian anyons emerging in fractional quantum Hall states carry a characteristic entropy, $ΔS = k_B \log d$, where \(d\) is the anyon's quantum dimension. This \(\mathcal{O}(1)\) entropy can, in principle, be extracted from charge measurements of an antidot via Maxwell relations. However, equilibrium charge measurements in fractional antidots have proven to be challenging with conventional charge detectors. Here, we propose a scheme based on an antidot embedded in an interferometer, in which the charge can be inferred from the recently observed time-dependent switching of the interference phase. Performing such non-local charge measurements at equilibrium, the characteristic \(\mathcal{O}(1)\) entropy of non-Abelian anyons (e.g., $d = \sqrt{2}$ for the $ν= 5/2$ state) can be extracted for intermediate temperatures, which exceed the level spacing of the interferometer edge, but are much smaller than the level spacing of the antidot.

We consider the Sherrington--Kirkpatrick spin glass model at the critical inverse temperature $β= 1$ with zero external field. We prove that the free energy $F_N = F_{N,β=1}$ of this model has variance \[ \mathrm{Var}(F_N) = \frac16 \log N + O(1)\,, \] confirming a physics prediction of Aspelmeier \cite{aspelmeier2008free}, and that the centered and scaled $F_N$ satisfies a Gaussian CLT. We also identify the critical two-replica overlap scale, proving \[ \mathbb{E} \langle R_{1,2}^2\rangle \asymp N^{-2/3}\,, \] as conjectured by Talagrand \cite{talagrand2011mean2}, together with a uniform exponential moment bound for $N^{1/3} |R_{1,2}|$. The key input is a comparison between the Ising and spherical SK partition functions $Z_N$ and $Z^{\mathrm{sp}}_N$: if $X_N = Z_N / Z^{\mathrm{sp}}_N$, then $X_N = 1 + o(1)$ in $L^2$. Thus $Z^{\mathrm{sp}}_N$ captures the diverging critical fluctuations of $Z_N$ and serves as a tractable reweighting variable for estimating overlap moments.

We investigate homogeneous populations of smart active agents that exchange information with their neighbors to perform a decentralized learning process aimed at achieving a prescribed macroscopic state. Such agents may, for example, represent simple microrobots. The exchanged information comprises tunable parameters governing the agent dynamics, referred to as the individual policy, together with an internal memory encoding previously visited states. This memory is used to evaluate a reward that quantifies the success of a policy to achieve the prescribed state. We extend the kinetic-theory description of collective learning in spatially homogeneous systems [Phys. Rev. Lett. 134, 248302 (2025)] and derive formal evolution equations for the distribution of policies across the population. A central outcome of our theory is the emergence of an effective reward function that fully determines the evolution of the policy distribution and encapsulates the microscopic details of the agents physical and memory dynamics. We obtain closed equations for the policy mean and variance which admit explicit time-dependent solutions under the assumption of Gaussian-distributed memories and polices. To illustrate the framework, we present a series of minimal microscopic models, considering both perfect and partial separation of physical, memory and policy exchange time scales, as well as models with one- and two-dimensional policies. The obtained theoretical results compare well with agent-based numerical simulations. The theory captures key aspects of collective learning, including the influence of population diversity and reward fluctuations on learning performance. Finally, we discuss potential applications to swarm robotics and machine learning, and highlight connections with classical models of biological evolution, including the Replicator equation and the Moran model.

Quantum reservoir computing provides a framework for processing complex temporal data, yet its fundamental computational and energetic limits remain unresolved. Here, we establish a non-equilibrium thermodynamic framework that links the macroscopic predictive performance of driven open quantum systems to their microscopic energetic costs. By mapping the Holevo capacities onto the Bogoliubov-Kubo-Mori geometric manifold, we analytically prove that the computational peak within the quantum critical region originates from a strict spectral resonance: the closing of the energy gap forces the reservoir's transition frequencies to align with the chaotic drive. To evaluate the associated thermodynamic costs, we introduce quantum informational dissipation to quantify the non-predictive historical data structurally retained by the reservoir, deriving a generalized Landauer bound for continuous temporal processing. This reveals a fundamental thermodynamic trade-off: the critical resonance that unlocks optimal predictive capacity inherently maximizes informational dissipation and the irreversible work required for environmental erasure. Furthermore, coherence decomposition demonstrates that dynamic quantum coherences strictly amplify predictive capacity without demanding additional mechanical work. These findings establish the ultimate energetic limits of quantum learning devices, providing theoretical principles for designing energy-efficient quantum neuromorphic hardware.

Attention enables a world model to condition on its entire history, providing long-term memory that facilitates long-range predictions. While the local Onsager--Machlup theory in our companion paper assumes a temporally local predictive action, we investigate the conditions under which this locality holds. We derive the predictive path measure for latent dynamics that become non-Markovian due to attention-induced memory, demonstrating that this measure is the projection of a hidden linear Markov augmentation. Eliminating the auxiliary field results in a nonlocal Onsager--Machlup action, where memory manifests as a nonlocal quadratic form rather than a force. These kernels are completely monotone and exactly match a hidden Markov embedding with a finite relaxation spectrum; otherwise, the dynamics remain fundamentally nonlocal. By expanding the action in terms of the scale-separation parameter $ε=τ_{\text{mem}}/τ_{\text{dyn}}$, we show that the leading order recovers the local action of the companion paper, establishing locality as the short-memory limit of a nonlocal theory. We verify the reversible sector of this expansion term by term against an exactly solvable vector linear model.

Suspensions of motile microswimmers such as bacteria and other active colloids frequently encounter porous environments where obstacles and complex shear flows strongly influence their dynamics. Here, we study the distribution and transport of a dilute suspension of active particles in a square lattice of pillars, which serves as a model porous medium. The microswimmers are modeled as slender point particles, and Brownian Dynamics simulations are performed to determine how their number density and polarization fields change with systematic variations in the medium porosity, polydispersity, flow strength, and self-propulsion strength. We find that in the absence of flow, self-propulsion drives particle accumulation and radial polarization at the pillar surfaces. In the presence of a background flow, particles preferentially accumulate in the wake of pillars and exhibit upstream polarization near their surface, consistent with experimental observations. At moderate flow strengths, topological defects nucleate in the polarization field. These defects are of purely kinematic origin and mark the transition from global upstream swimming at low flow strengths to the coexistence of upstream and downstream swimming regions in the lattice at high flow strengths. The structured lattice studied here provides a controlled framework for isolating the physical mechanisms governing active transport in complex geometries, with direct relevance to transport in structured microfluidic settings.

Ultra-thin films are fundamental components of modern nanoelectronics, where reducing thickness to the few-nanometer scale leads to a dramatic increase in electrical resistivity. For decades, this behavior has been interpreted in terms of classical size effects, primarily surface scattering within the Fuchs--Sondheimer theory and grain-boundary scattering in the Mayadas--Shatzkes model. While these approaches successfully describe transport when the film thickness is comparable to the electronic mean free path, growing experimental evidence indicates that they become insufficient under extreme confinement. This review discusses the crossover from classical scattering to a quantum-confinement regime in which the electronic states available for transport are fundamentally restructured by finite size. We review the recently proposed reciprocal-space confinement theory, which predicts an exponential increase of resistivity with decreasing thickness at the nanoscale, and discuss how it can be combined with classical surface-scattering models to provide a unified description of ultra-thin metallic and semiconducting films. Finally, we summarize recent experimental evidence supporting this picture and discuss its implications for future nanoelectronic devices, nanoscale interconnects, and quantum transport under extreme spatial confinement.

We study nonlinear DC electric transport of quantum-metric origin by combining adiabatic perturbation theory with the nonequilibrium Green function approach. The adiabatic ansatz provides a basis for directly treating a DC electric field in the velocity gauge, rather than introducing it as the zero-frequency limit of an AC field. The resulting adiabatic-basis Hamiltonian takes the same form as in the length gauge, enabling a systematic comparison across different formulations. Applying this fully quantum formulation, we find a longitudinal nonreciprocal current governed by the quantum-metric dipole. The essential ingredient is a quantum correction to the distribution function that is absent in semiclassical treatments. We trace this correction to the finite spread of an electron wave packet during relaxation under a bias field, thereby identifying shifted quasiequilibrium as the physical origin of quantum-metric nonreciprocal transport.

Wigner's quasi-probability distribution function in phase space is a specialized representation of the density matrix, possessing significant physical importance. In this article, we first review the wave function describing electronic motion in an electromagnetic field under the Landau gauge. Next, based on an introduction to the properties of the Wigner function, we calculate the Wigner function for the integer quantum Hall effect using the integral method.

We present a complete classification of integrable Yang-Baxter quantum circuits with open boundary conditions and arbitrary circuit geometries. Starting from the standard transfer-matrix construction with two types of staggered inhomogeneities, we derive a general mapping that determines the arrangement of circuit gates in terms of the inhomogeneities and the system size. We conjecture that time-periodic quantum circuits are integrable whenever the local bulk and boundary gates satisfy the Yang-Baxter equation and the same bulk gate is applied exactly once per period to every nearest-neighbor pair of spins. Our construction also provides an algorithm to detect Yang-Baxter integrability for circuits with arbitrary geometries. Furthermore, we introduce a third type of inhomogeneity, denoted by $ρ$, and demonstrate that the minimum possible circuit depth is four. We show that when these $ρ$-inhomogeneities are placed at the endpoints and in their immediate neighborhood, the resulting boundary gates can be interpreted as single gates acting on multiple sites. Our construction is fully general and applies to regular $R$-matrices, both of difference and non-difference type, together with their associated boundary matrices. As an application, we consider two-qubit gates corresponding to 6- and 8-vertex $R$-matrices of non-difference form satisfying the Yang-Baxter equation, and we construct the associated reflection matrices that generate integrable quantum circuits.

We present a theoretical study of electron tunneling through a magnetic barrier in 8-Pmmn borophene, created by depositing two ferromagnetic strips on the borophene sheet. Using a low-energy effective Hamiltonian that captures the anisotropic Dirac spectrum, we solve the Dirac equation in three regions and impose wave-function continuity at the interfaces. From the resulting spinor solutions, we compute current densities and determine transmission and reflection probabilities as functions of incident energy, angle, and barrier parameters. The transmission exhibits strong anisotropy due to the tilted Dirac cones, with pronounced suppression for specific incident directions, suggesting directional filtering of carriers. We further calculate the conductance using the Landauer-Büttiker formalism, revealing that both magnetic strength and barrier width can tune the charge transport properties. The results demonstrate that engineered magnetic barriers in 8-Pmmn borophene enable precise control over electron flow, offering a platform for anisotropic transport control and tunable quantum devices. The interplay between the intrinsic anisotropy of borophene and external magnetic barriers provides rich opportunities to manipulate Dirac fermions in two-dimensional systems.

We investigate the undamped behaviors in a spin-1/2 Heisenberg chain coupled with an environment via collective spin jump operators. Using the Bethe ansatz basis, we show that undamped modes exist for any chain length N >= 3. These modes remain robust against variations in the system parameters, including the specific form of the collective dissipation, and the external field. Exploiting the Bethe ansatz solution, we further characterize the number of undamped modes and their oscillation frequencies, uncovering long-lived coherent dynamics in open integrable quantum systems.

We investigate the sandpile model on complex networks by developing a branching-process framework that explicitly incorporates dissipation during avalanche propagation. Unlike classical branching descriptions, which assume conservative transport and locally tree-like independence, the present approach introduces grain-loss effects directly into the offspring distribution, yielding generalized generating functions for dissipative avalanche dynamics. In the dissipative regime, avalanche-size distributions acquire exponential cutoffs while preserving topology-dependent scaling behavior. Numerical simulations confirm the theoretical predictions on sparse random networks and reveal systematic deviations in highly structured topologies. In particular, by using Holme-Kim clustered scale-free networks, we show that increasing clustering continuously lowers the avalanche exponent and enhances the probability of large cascades, demonstrating that short cycles generate strong correlations that invalidate the classical independent-branch approx imation. Surprisingly, trees also exhibit substantial deviations from power-law because low edge density and the abundance of leaves constrain avalanche propagation. These results show that dissipation, clustering, and sparse connectivity fundamentally reshape avalanche size distribution of the sandpile model on networks and establish quantitative limits for branching-process descriptions of avalanche dynamics.

Magnetic inertia, emerging in the ultrafast regime, supports inertial spin waves (SWs) as novel magnetic excitations. Despite considerable efforts devoted to inertial SWs, a systematic formalism for fully characterizing their intrinsic properties, especially chirality and polarization, is still lacking, and inertial SWs in spatially nonuniform magnetic configurations remain poorly explored. Here, we develop a framework for calculating inertial SWs and establish a general definition of their chirality and polarization via the ellipticity angle, a unified parameter encoding frequency sign, phase difference, and elliptical axis ratio. Using this method, we systematically investigate precessional and nutational SWs in uniaxial antiferromagnets with staggered and homogeneous Dzyaloshinskii-Moriya interactions (DMIs), covering uniform collinear, canted, and spiral magnetic configurations. The results reveal that small staggered DMI preserves spin-wave degeneracy, whereas small homogeneous DMI lifts it. Further space-time inversion symmetry breaking in canted and spiral structures fully removes spin-wave degeneracy across the entire Brillouin zone. Long-wavelength nutational SWs behave as backward waves, and flat bands emerge in canted and spiral configurations near a critical inertial relaxation time. In canted and spiral configurations, nutational modes are always lefthanded whereas precessional modes are always righthanded; additionally, the dispersion spectra of the canted configuration can be derived from those of the spiral configuration via band folding. Polarization is wavenumber insensitive for uniform configurations but becomes strongly dispersive for nonuniform ones. This work advances the fundamental understanding of magnetic inertial dynamics and provides theoretical insights for the development of ultrafast magnonic devices.

False vacuum decay describes the relaxation of a metastable state through the nucleation and growth of bubbles of the stable phase. Despite describing a broad variety of phenomena across different fields, the quantum version of the nucleation theory has little experimental or numerical support. Testing its predictions is particularly important in two or more spatial dimensions, where bubble nucleation acquires its true geometrical nature. Here, we study false vacuum decay in the quantum Ising model in two dimensions. Through tree tensor network simulations we extract the decay rate, the effective interface tension and the critical bubble size. We compare them to new semi-classical field theory calculations, and find excellent agreement. These results provide numerical evidence that the critical-bubble picture survives in an interacting quantum spin system in 2+1 dimensions.

The Ramsey community number $r_κ$ is the minimum network size at which a graph's connectivity is better described by a partition into communities than by no partition, under a prescribed community-detection rule. It was introduced through numerical simulations of networks grown by local rules, which suggested that community structure can emerge without any node heterogeneity. Here I compute $r_κ$ analytically for the simplest homogeneous, locally wired graph: the circulant ring lattice $C_n(1,\dots,c)$. Using a Bernoulli stochastic block model with symmetric $\mathrm{Beta}$ priors as the detection rule, the Bayesian evidence for a balanced two-community partition and for the unpartitioned network are both obtained in closed form, so the transition between them can be located exactly. The result is a sharp dependence on the interaction range: the plain cycle ($c=1$) is never partitioned, its two-community posterior decaying as $n^{-(2α+3)}$, so $r_κ=\infty$; but the next-nearest-neighbour ring ($c=2$) acquires a finite $r_κ\simeq 35$ nodes, above which the partition is preferred with a log-evidence growing as $(\ln 2)\,n$. This provides an exactly solvable instance of community emergence in a network with no built-in communities, and shows that a minimal amount of local connectivity is enough to break the ring.

We study a dilute system of non-motile chiral active particles confined in geometries ranging from straight channels to circular enclosures. Activity is introduced through chiral particle-wall interactions, modeled as tangential wall forces that generate the edge currents characteristic of chiral active matter. Remarkably, although the particles lack self-propulsion, these boundary currents induce density inhomogeneities. We show that boundary curvature drives a wall accumulation phenomenon: particles remain uniformly distributed in straight channels but accumulate near the boundaries of circular confinements. Numerical simulations and a hydrodynamic theory for the density and momentum fields consistently capture this curvature-induced wall-accumulation. These results identify boundary curvature as a fundamental control parameter for chiral edge transport and confinement-induced organization, with potential experimental relevance to spinning colloids and granular spinners.

Exploring thermal transport at cryogenic temperatures presents both significant challenges and valuable insights. By uncovering the thermal counterpart of well-known quantum phenomena, researchers investigated fascinating phenomena ranging from the violation of the Wiedemann-Franz law to the quantisation of phonons. One key frontier remains : no existing method can image local heat transport at the nanoscale under cryogenic conditions. In this Perspective, we review the current state state of the art of local heat transport characterisation techniques and highlight their limitations. As a motivation for the development of cryogenic Scanning Thermal Microscopy, we provide five case studies illustrating how this approach could deepen our understanding of exotic quantum phases and enable the emergence of transformative technologies.

Two-dimensional transition metal dichalcogenides (TMDs) have emerged as promising materials for spin--orbitronics owing to their strong spin--orbit coupling and rich electronic phases. However, their orbital transport properties remain largely unexplored. Here, we demonstrate that the orbitronic response of $\mathrm{MoTe_2}$ is governed by a thickness-driven structural phase transition. RF-sputtered $\mathrm{MoTe_2}$ thin films exhibit a crossover at a critical thickness of approximately $4.5\,\mathrm{nm}$, stabilizing in the metallic $1T^\prime$ phase below this threshold and in the semiconducting $2H$ phase above it. Raman spectroscopy and scanning tunneling spectroscopy (STS) confirm the structural and electronic transition, revealing gapless behavior in ultrathin films and a finite band gap in thicker samples. Spin-pumping measurements detect an additional transverse charge-conversion signal exclusively in metallic $1T^\prime$-$\mathrm{MoTe_2}$, in agreement with first-principles calculations that identify a dominant orbital Rashba--Edelstein response as the underlying conversion mechanism.

The instabilities of slender structures power biological locomotion across scales, and offer a compelling method to actuate soft robots. Nonreciprocal elastic solids have been found to amplify flexural waves in one direction only, but design principles to tune and stabilize these waves are missing. Here we develop a geometrically exact theory of nonreciprocal filaments and provide simulations that capture their post-instability nonlinear dynamics. We find that nonreciprocity, when coupled to inertia or pre-stress, amplifies and advects curvature variations. The resulting one-way patterns of shape morphing can then be selected via dissipative interactions with the environment. Our work offers a continuum-based strategy for how internal stresses can drive active unidirectional waves without need for additional degrees of freedom.

We investigate fluid mixing induced by microswimmers using mutual information as a global, information-theoretic measure of mixing efficiency. For a two-dimensional squirmer model in a confined domain, we compute numerically the swimmer-generated flows and solve the advection-diffusion equation for the transport of tracer particles in the fluid. We show that the spatial distribution of swimmers strongly affects mixing, which is suppressed by swimmer aggregation and enhanced by positional and orientational disorder. At fixed energy dissipation, mixing efficiency depends non-monotonically on the squirmer parameter, with an optimal finite value arising from the balance between swimmer translation and dipolar flow generation. When hydrodynamic interactions are included, pushers outperform pullers. The mutual information as a function of time decays in three stages: an initial diffusion-dominated stage, an intermediate advection enhanced regime, and a final relaxation stage controlled by system size. Our results demonstrate that mutual information, previously validated as a measure of mixing efficiency only in simplified model systems, can equally be used in complex flows. Its application reveals that mixing by microswimmers is subject to a trade-off between the generation of strong shear flows and achieving optimal dispersion across the fluid domain.

The nonlinear Hall effect in quantum materials has attracted broad interest, yet most existing studies focus on the weak-field, perturbative regime. Here we develop a nonperturbative approach based on nonequilibrium steady-state Green's functions for dc-field-driven lattice systems, with dissipation and interactions incorporated through self-energies beyond the constant relaxation-time approximation and interband transitions treated alongside their intraband counterparts. Applied to a two-band semimetal model, our approach provides direct access to the strong-field Hall response beyond the nonperturbative crossover where the edge of the nonequilibrium distribution reaches Berry-curvature hot spots, a regime in which constant relaxation-time estimates and Berry curvature dipole calculations become unreliable. We further demonstrate that interaction and electron-phonon self-energies within dynamical mean-field theory can substantially change the Hall signal. Our framework enables quantitative simulations of nonequilibrium nonlinear Hall phenomena and provides guidance for strong-field transport experiments.

We investigate the low-Reynolds-number propulsion of a slender elastic filament with a dipolar magnetic head actuated by an oscillating field in a viscous fluid by studying its strokes and net forward motion. To capture these dynamics, we employ an elasto-hydrodynamic (EH) framework that couples Euler-Bernoulli beam mechanics with resistive force theory. Unlike prescribed-kinematics models, filament shapes here emerge self-consistently from the actuation and the force and torque boundary conditions (BCs). We demonstrate that viscous boundary contributions are crucial for quantitative agreement and show that the swimming dynamics are governed by the EH length and a magneto-viscous-elastic stroke amplitude introduced here. The swimming speed is non-monotonic with increasing ratio of the swimmer length to the EH length, and is shown to reach a maximum when the swimmer length is on the order of the EH length. We further discuss the analytical limit in which the tail BCs can be described as free, and the limitations that arise when viscous contributions to the BCs are ignored.

Cyclic pore-pressure or normal stress variations arise both in relation to natural earthquakes and in engineered subsurface systems, yet their effect on fault stability remains poorly constrained at the grain scale. Here we numerically model, using a coupled Discrete Element--fluid dynamics model, the response of a sheared, fluid-saturated or dry, gouge-filled fault to effective normal stress oscillations over a wide frequency range (0.5-10000 Hz). The effective normal stress is oscillated either by cycling the pore-pressure or by directly cycling the normal stress, while keeping the stress state below the Mohr-Coulomb threshold measured in continuous loading. Despite this sub-critical loading, we observe failure across most frequencies, with a non-monotonic frequency dependence. A distinct non-failure window emerges at intermediate frequencies (30-200 Hz), bounded by failure at both lower and higher frequencies; the system exhibits four regimes from cyclic failure-and-arrest to continuous sliding. Pore-pressure and normal stress oscillations produce the same regime structure, confirming that they act as equivalent forcings via Terzaghi's principle, with fluid coupling adding only a delay due to dilatant hardening. Sub-critical failure arises from dilation-induced strength deterioration via two mechanisms: (i) low-frequency cycles allow sufficient time for shear-driven ratcheting dilation, while (ii) high-frequency cycles induce dynamic dilation (acoustic fluidization) via amplified seepage forces, stress gradients and inertial forces. The intermediate non-failure window represents the gap between these mechanisms. These results identify frequency as a controlling parameter for failure in granular materials, with implications for dynamic earthquake triggering and cyclic injection protocols.

Resonant cooling of nuclear spins by photogenerated spin-oriented holes is demonstrated for MAPbI$_3$ perovskite crystals. It is evidenced by Hanle-effect measurements under helicity-modulated excitation with variable frequency. The resonance position in magnetic field shifts toward higher fields with increasing modulation frequency. The invariance of the Hanle curve upon in-plane sample rotation is consistent with the involvement of $^{207}$Pb nuclei with spin $I = 1/2$, which do not exhibit quadrupolar splitting. The shape of the resonance feature in the Hanle curve reveals that the nuclear spins are cooled by carriers with a negative $g$-factor, consistent with holes. The resonance fields associated with the modulation frequencies exceed the half-width of the weakly localized hole contribution to the Hanle curve, indicating that strongly localized holes are the primary carriers responsible for the nuclear spin cooling.

High-quality topological insulator (TI) materials are essential for the realization and detection of Majorana bound states (MBSs) in TI-superconductor hybrid platforms. Widely used compensated TIs exhibit substantial disorder and charge inhomogeneity, which may be detrimental for Majorana devices. In this regard, Sb-substituted Bi$_2$Se$_3$ (SBS) is promising, because it is non-compensated and yet achieves very low bulk carrier density. We systematically investigate the impact of thermal processing during microfabrication on the transport properties of SBS. We developed a room-temperature fabrication protocol that preserves the low carrier density of exfoliated SBS upon fabrication of Hall bar and nanowire devices as evidenced from the observation of quantum interference oscillations in nanowires, a large gate tunability, and clear signatures of weak antilocalization (WAL).

Melt--quench molecular dynamics is widely used to construct amorphous materials models, but the resulting ensemble is defined by choices that are often made implicitly: numerical settings, melt temperature, liquid-hold time, quench rate, system size, and post-generation screening. We introduce vitriflow, a computational materials methodology that turns these choices into an explicit decision chain. The framework couples numerical stability, descriptor-based protocol calibration, user-defined artefact screening, and statistical convergence of the generated analysis ensemble in a material-specific descriptor space. We demonstrate the approach for a-SiO$_2$, a-Si$_3$N$_4$, and a-Sm$_2$O$_3$, which respectively test tetrahedral network fidelity, MG2 $\rightarrow$ PBE $\rightarrow$ HSE06 DFT refinement of a heteropolar nitride, and amorphous/crystal discrimination in a mixed-coordination rare-earth oxide. vitriflow separates defect-free from oxygen-bridge-defective silica, quantifies DFT-refinement response in a common a-Si$_3$N$_4$ structural population, and removes recrystallised Sm$_2$O$_3$ structures without imposing fixed coordination. The result is a reproducible route for generating amorphous ensembles whose numerical settings, thermal protocol, screening actions, and statistical precision are selected from the materials question rather than assumed a priori.

We derive the distribution of the first passage time $t_{fp}$ for the position $x$ of an underdamped harmonic oscillator to overcome a threshold $x_B$. As the $t_{fp}$ distribution depends on the oscillator quality factor $Q$ different approaches are used. At very large quality factor ($Q\gg 100$) and intermediate and long $t_{fp}$ the proof is based on an energy diffusion model, whereas at medium quality factor ($Q\sim 10$) the proof is based on the study of the eigenvalues of the Kramers linear differential operator with absorbing boundary conditions. For all $Q$ and short $t_{fp}$ we use a Hamiltonian approximation. The theoretical predictions are in excellent agreement with direct numerical simulations of underdamped oscillator dynamics. Finally we show that the mean of the trajectories ending at $t_{fp}$ presents a particular shape driven by a specific noise pattern.

The distribution of the first passage time $t_{fp}$ for the position $x$ to overcome a threshold $x_B$ is calculated in an underdamped harmonic oscillator. The proof combines several approaches based on the determination of the eigenvalues of the Kramers differential operator for the intermediate and long time regimes and on a Hamiltonian approximation for the short times. The theoretical predictions are in excellent agreement with the results of an experiment on an underdamped micro-cantilever. The knowledge of the $t_{fp}$ distribution opens the way to several applications, among them the precise estimation of the power of information engines, which we have also experimentally checked.

In collective dissipative spin systems, the postselection barrier can be partially mitigated; however, a further obstacle may be posed by the finite temporal resolution of detectors. In this work, we investigate how initial-state inhomogeneities can control waiting-time statistics between quantum jumps, thereby mitigating the detector-resolution problem. We consider a collectively monitored spin model with a boundary time-crystalline phase, introducing inhomogeneity by partitioning the ensemble into two subsystems rotated by an angle $θ$. We find that the measurement-induced phase transition survives under inhomogeneities, with different entanglement scaling regimes. The waiting time increases with $θ$, scaling as $1/N$ but with a prefactor strongly enhanced by orders of magnitude, and in the anti-aligned limit $θ= π$ it remains finite, fully resolving the resolution barrier. This mitigation, however, comes at a cost: the entanglement saturation time becomes significantly longer, partially reintroducing the postselection barrier. Our results highlight a trade-off between detector resolution and postselection overhead, with direct implications for the experimental observation of measurement-induced phenomena.

When describing entanglement in typical midspectrum eigenstates of many-body lattice Hamiltonians, two paradigms have emerged that capture the behavior observed in integrable and nonintegrable systems, Haar-random fermionic Gaussian states and Haar-random pure states, respectively. Remarkably, the former capture the behavior of interacting integrable systems, whose eigenstates are non-Gaussian. We argue that the paradigm that captures both the entanglement properties and the lack of Gaussianity in integrable systems is that of random superpositions of polynomially many Gaussian states. In contrast, eigenstates of nonintegrable systems are consistent with being described by random superpositions of exponentially many Gaussian states. We gain this understanding by comparing analytical and numerical results for the one-body purity, the non-Gaussianity, and the entanglement entropy of the random superpositions and the Hamiltonian eigenstates.

This review discusses the recently proposed fuzzy sphere regularization for studying $2+1$D critical phenomena, particularly three-dimensional (3D) conformal field theory (CFT). The fuzzy sphere scheme not only offers remarkable efficiency in extracting extensive CFT data at low computational cost but also reveals unexpected connections among 3D CFT (critical phenomena), noncommutative geometry, and the quantum Hall effect. We introduce the fundamental ideas of fuzzy sphere regularization, emphasizing its role in demonstrating the state-operator correspondence of 3D CFTs on the $S^2 \times \mathbb{R}$ geometry. Additionally, we review key developments in this approach across various directions and outline potential future applications.

A predictive understanding of defect transport in GaN under operating electric fields is critical for assessing device reliability in high-power and radiation environments. In this work, a ReaxFF reactive force field for GaN is developed using a density-functional-theory training set that includes structural, thermodynamic, and defect properties. The force field yields various properties such as lattice parameters, cohesive energies, and defect formation and migration energies in close agreement with prior first-principles and experimental results. Under externally applied electric fields, we find that migration barriers can be strongly modulated, with changes that depend on defect type and field orientation. Notably, the electric fields do not simply linearly bias defect motion in GaN, but can anisotropically modify migration barriers through charge-lattice coupling, leading to nonlinear transport behavior. The response arises from field-induced partial charge redistribution and local lattice distortion. These results demonstrate that electric fields can complexly modify the defect migration landscape, providing new insight into defect transport in GaN under high-field conditions.

Electron scattering from chiral spin textures such as skyrmions is fundamental to the understanding of transport in more complex systems, including skyrmion crystals. Most of the previous studies have focused on the weak-coupling regime, where the exchange interaction is small compared with the electron energy. Real materials, however, often lie in the strong-coupling regime, which exhibits qualitatively different behavior. Using the Lippmann-Schwinger equation and Green's function formalism, valid for all coupling strengths, we uncover several new features in the scattering cross section, including Ramsauer-Townsend minima, pronounced intermediate-coupling resonances, and Landau-level resonances for skyrmions with larger winding numbers. These features strongly influence the topological and spin Hall conductivities, which depend sensitively on the incident electron energy. Our work provides important insights into the Hall transport in collective chiral spin textures such as the skyrmion crystal.

We investigate modular Laplacian automata on triangular lattices with evolution governed by binary and ternary moduli. Extending previous studies on square lattices, we examine how lattice geometry influences long-term growth, density, fragmentation, and the emergence of self-similar structures. We further investigate whether sparse ternary interventions can stabilize predominantly binary dynamics. The experiments reveal that mask geometry is the primary determinant of large-scale morphology. Full hexagonal masks generate recurrent density crises and fragmentation, whereas triangular masks support persistent growth and reveal a threshold phenomenon governed by growth-capable nuclei. Although seed symmetry influences transient behaviour, the asymptotic morphology is inherited mainly from the mask. To control binary fragmentation, we investigate sparse developmental ternary perturbations in which a small number of carefully timed occurrences of modulus 3 are inserted into an otherwise binary sequence. A Monte Carlo optimization demonstrates that as few as three interventions are sufficient to redirect the subsequent binary evolution toward substantially denser carpet-like configurations. The effectiveness of this strategy depends primarily on the timing of the interventions rather than on their number. Analysis of the post-intervention dynamics shows that ternary shaping does not replace binary evolution. Instead, it produces denser self-similar structures, substantially reduces crisis depth, and resets the phase of the binary crisis clock. The results suggest that geometry determines the family of admissible morphologies, whereas sparse developmental perturbations select favourable long-term trajectories within that family.

We investigate a stably stratified flow driven by deterministic Kolmogorov forcing that generates horizontal shear, using direct numerical simulations over a broad range of stratification strengths characterized by the Froude number $Fr$. As the stratification is progressively weakened, the flow exhibits a sequence of regimes: a buoyancy-dominated, strongly stratified regime, an intermediate regime characterized by Kelvin--Helmholtz instabilities and enhanced mixing, and a nearly isotropic turbulent regime. A key feature of the intermediate stratification range is the emergence of energetically significant vertically sheared horizontal flows (VSHFs), accompanied by a marked steepening of the reduced one-dimensional perpendicular kinetic energy spectra. The spectral energy transfer remains predominantly forward, although the perpendicular flux becomes negative at large horizontal scales; this apparent upscale transfer reflects anisotropic energy redistribution rather than a true inverse cascade. Strong stratification enhances intermittency, producing increasingly non-Gaussian vertical velocity fluctuations and large kurtosis associated with localized vertical bursts. The energetics-based mixing coefficient remains of order $10^{-1}$ over the parameter range investigated, with a modest enhancement near the Kelvin--Helmholtz instability regime.

We investigate the asymptotic behaviour of diploid Elementary Cellular Automata (ECA), that is, the stochastic mixtures between two ECAs. In this model, each cell independently applies one rule with probability $λ$ and the other rule with probability $ 1 - λ$. Focusing on the endogamous diploids where the two ECAs are related by the reflection or conjugation symmetry, we analyse how the density varies as function of $λ$, the ``degree of mixing'' of these two rules. We propose a classification into six distinct classes depending on the profile of the density vs. $λ$ curve. We take various examples for each class and we analyse to which extent the local structure approximation succeeds to predict the asymptotic density. Our results show that for rules in which the asymptotic density depends linearly on $λ$, the local structure approximation reproduces the exact dependence of the density on $λ$. For rules with differentiable but nonlinear dependence, the approximation either becomes exact at a finite order or converges rapidly to the exact solution as the order increases. For rules with first-order phase transitions, finite-order approximations progressively approach a sharp transition profile with increasing order. In contrast, for rules exhibiting a second-order phase transition, even high-order approximations fail to capture the qualitative features of the transition. (no bifurcation is observed). Finally, we give examples of two diploid rules for which we succeed to compute the density at each time step.

Quantum droplets (QDs) are self-bound states of ultradilute quantum fluids stabilized by the interplay between the Lee Huang-Yang (LHY) quantum-fluctuation correction and the mean-field interaction, providing a useful platform for exploring macroscopic quantum phenomena. Recent studies on three-dimensional QDs have introduced the concept of bulk modulus and revealed its connection with the breathing-mode frequency, thereby linking the elastic response of QDs to their collective dynamics. Motivated by this progress, we investigate the elastic modulus of one-dimensional QDs. Based on a super Gaussian variational ansatz, we systematically derive the elastic modulus B and analyze its dependence on the interaction strength and particle number. The analytical predictions are further validated by numerical simulations based on imaginary time evolution and the spatial scaling method. We also establish a quantitative relation between the elastic modulus and the eigenfrequency of the breathing mode. In addition, by incorporating corrections to the droplet width beyond the Thomas Fermi approximation, we obtain the dependence of the ratio η = B/2 on the control parameters g and N. Unlike the three-dimensional case, where the corresponding ratio follows a simple power-law scaling, the one-dimensional system is affected by the soliton-to-droplet crossover, leading to a more intricate dependence of η on g and N. Our results show that, in the high-particle-number regime, the elastic modulus asymptotically approaches a limiting value determined mainly by the interaction strength, whereas in the low-particle-number regime it depends on both the particle number and the interaction strength.

The inverse scattering transform (IST) provides the standard theoretical framework for deriving soliton dynamics. Traditionally, such derivations have been of an analytical, rather than data-driven, nature. In this paper, we combine the conceptual framework of the IST with weak-form system identification methods to discover effective soliton dynamics directly from observed scattering data, without assuming prior knowledge of the scattering equations. Our method avoids parameterizing solitary waves via ad hoc curve-fitting by working in the scattering domain, yielding interpretable low-dimensional models that remain valid in perturbed and near-integrable regimes. We demonstrate the performance of the proposed approach on synthetic and experimental data governed by shallow-water equations of Korteweg--de Vries-type and recover models that are consistent with canonical IST theory.

A Lax pair $(L,P)$ is sometimes thought of as a structural certificate, in that the spatial operator $L$ carries the spectral data of an integrable system, and its isospectral evolution under $\partial_t L = [L,P]$ encodes the nonlinear dynamics. Yet, experience shows that the correspondence between equations and Lax pairs is much more nuanced than this picture suggests. Equations can admit Lax pairs that fail to encode the expected integrable structure. This paper probes that anomalous corner of the Lax pair landscape through five case studies (the Euler top, the free Schrödinger equation, the inviscid Burgers equation, the shallow water system, and the Korteweg--de Vries equation), each illustrating a different way the link to integrability can be distorted. The approach combines analytical calculations with the Sparse Identification of Lax Operators (SILO) framework, which proved useful throughout, in some cases confirming the textbook pair and in others surfacing alternatives worth understanding on their own terms. The recurring lesson across the five cases is that compatibility underdetermines the Lax representation, so that anomalous pairs are regular features of the landscape rather than pathologies. Notably, we show that a spectrally degenerate Korteweg--de Vries Lax pair, classified as fake by standard criteria, still generates the full conservation hierarchy through its operator algebra, which shows that a blunt dichotomy between true and fake Lax pairs can be too reductive.

We investigate the interaction of small-amplitude waves called phonons, with an initially static kink in an exceptional discretization of the $φ^4$ model that is free of the Peierls-Nabarro potential. Phonons are generated by a localized harmonic source and scattered from one side of the kink. By computing the transmission and reflection coefficients over the entire phonon band, we demonstrate that the scattering properties depend strongly on the lattice spacing. In the weak-discreteness regime ($h<1$), the kink is nearly transparent and phonons are transmitted through it over most of the phonon spectrum. In contrast, for strong discreteness ($h>1$), significant reflection emerges even though the corresponding continuum $φ^4$ kink is reflectionless. We further show that depending on the frequency of the incoming phonons, the kink experiences negative radiation pressure and is accelerated toward the incoming phonons for all lattice spacings considered, and this effect is much stronger for the strong discretness. The frequency dependence of the kink velocity and energy transfer is explained in terms of resonances associated with Doppler-shifted phonon frequencies and extrema of the phonon group velocity. Our results reveal that strong lattice discreteness can qualitatively modify phonon-kink interactions even in systems where the static Peierls-Nabarro potential is absent.

A direct proof based on commuting finite-dimensional flows and local polynomial invariants is given for a sharp upper bound on the amplitudes of finite-gap solutions of the modified Korteweg-de Vries (mKdV) equation. The maximal amplitude is the sum of the imaginary parts of the upper-half-plane square roots of the roots of the invariant polynomial of the finite-gap solution of the focusing mKdV equation. An analogous formula is established for a bounded class of solutions of the defocusing mKdV equation. The upper bounds are sharp and are explicitly attained by suitable initial data.

Smart metasurfaces capable of employing the momentum of light for manipulating nanoparticles hold the key to potential applications in science and nanotechnology. This article proposes a density-based topology optimization framework for optimizing plasmonic metasurfaces for nanoparticles optical trapping. The Maxwell stress tensor (MST) is employed to compute the optical force exerted on nanoparticles of different sizes and types. The metasurfaces' topologies are optimized to maximize the gradient (attractive) force on such nanoparticles subject to normally incident monochromatic excitation. Designs based on free-form optimization are investigated first, then manufacturing constraints are imposed to provide easy-to-manufacture planar designs. The results show that the topology of the optimized metasurfaces depends on the nanoparticle size and material, with a higher trapping stiffness associated with small nanoparticles. The optimized metasurfaces could offer selective mass trapping of nanoparticles for applications in biosensing, microfabrication, or assembly of quantum systems.

The steady final phonon occupation in waveguide optomechanical systems based on backward stimulated Brillouin-Mandelstam scattering has not been established in the strong-coupling regime. In this work, the displacement spectra of anti-Stokes optical modes and acoustic modes in tapered chalcogenide photonic crystal fiber are derived from the Lindblad (or Gorini-Kossakowski-Sudarshan-Lindblad) master equation. By analyzing the full spectral response, we indicate that the system can enter the strong-coupling regime through the emergence of normal-mode splitting and avoided crossings. Within a non-Hermitian framework, the threshold for strong coupling is identified, showing that it can be achieved at relatively low pump power even at room temperature. Furthermore, we derive a unified analytical expression for the final phonon occupation, revealing that quantum backaction and zero-point fluctuations impose additional fundamental limits that hinder the achievement of ground-state cooling. These results redefine the quantum limits of steady-state cooling in continuum optomechanics, motivating the search for new strategies to access the quantum ground-state of macroscopic phonons.

Microwave plasma sources play a critical role in scientific research and a wide range of industrial, biomedical, and space applications. Resonant microwave structures have recently enabled highly energy-efficient plasma generation by concentrating electromagnetic energy within compact volumes. However, once plasma is ignited, the formation of a conductive region at the resonator's electric-field hotspot significantly perturbs the resonant impedance, resulting in severe impedance mismatch, increased reflection, and reduced power-transfer efficiency. This limitation arises because conventional resonant operation relies on critical coupling, in which the input coupling simultaneously provides impedance matching and perturbs the resonator. This paper overcomes this fundamental limitation by operating the resonator in an over-coupled regime and achieving dynamic impedance matching through temporally modulated excitation. Specifically, an exponentially growing incident waveform is used to emulate the critical coupling condition without physically modifying the resonator, a concept known as virtual critical coupling. The proposed approach enables the resonator to store up to four times as much electromagnetic energy as a conventionally critically coupled resonator. Experimental results demonstrate ultra-efficient resonant microwave plasma generation with multi-fold reductions in ignition energy consumption and enhanced dynamic control over plasma dynamics.

Gliding mammals exhibit diverse patagium and tail/uropatagium morphologies that may influence aerodynamic performance and maneuverability. Here, we use computational fluid dynamics to isolate the aerodynamic effects of representative gliding-mammal-inspired morphologies under controlled flow conditions. Three patagium configurations were compared to evaluate the effects of membrane outline on lift generation, drag, stall behavior and pitching moment. Three tail/uropatagium configurations were further tested under baseline, symmetric-deflection and asymmetric-deflection conditions to assess their longitudinal and lateral control authority. The results show that a broader patagium configuration generated the highest lift and lift coefficient, whereas an intermediate patagium morphology showed a smoother post-stall response with lower drag. For the tail configurations, the colugo-like integrated uropatagium enhanced lift and pitch-control authority under symmetric deflection, while the flat-tail configuration produced stronger rolling and yawing responses under asymmetric deflection. These findings indicate that gliding-mammal-inspired morphologies produce distinct aerodynamic trade-offs rather than a single optimal design. The results provide insight into the functional diversity of gliding mammal morphology and offer design guidance for bioinspired morphing aerial robots.

The time auto-correlation of auxiliary wave functions (TACAW) method enables efficient simulations of ultra-low-loss electron energy loss spectra (EELS) arising from vibrational and magnon excitations. In practical applications to realistic materials systems, however, TACAW calculations become challenging due to the large system sizes required for models containing defects, interfaces, impurities, or grain boundaries, as well as the substantial computational cost and data throughput associated with molecular dynamics and multislice calculations. Here we discuss a practical methodology for large-scale TACAW simulations and present torched-TACAW, a freely available implementation of the TACAW part of the described workflow for efficient STEM-EELS simulations. The overall approach combines molecular dynamics based on foundational machine-learned interatomic potentials, partitioning of elongated supercells, and on-the-fly processing of multislice outputs in order to enable near ab initio quality simulations with tractable memory use and data flow. Using rutile TiO2 as a model system, we analyze important numerical aspects of the method, including windowing and supercell partitioning, and demonstrate atomic-resolution STEM-EELS simulations for thick samples.

Broadening of spectral and spatial responses due to intrinsic loss in real materials often hides sharp features. One recently recognized route to recover those features is to probe the system with complex-frequency (CF) signals that decay exponentially in time: a suitably tailored temporal decay can compensate for loss and reveal an intrinsic, narrow response. However, generating rapidly decaying optical waveforms in real time is often challenging (the required decay times may be in the range of tens of femtoseconds). A recently proposed alternative synthesizes the CF response numerically after detection of conventional, real-frequency signals using Fourier post-processing. Here we explore advantages and challenges of these approaches: we show that a physical CF excitation robustly sharpens spectral features in the presence of noise, while a post-detection synthesized CF response shows only limited improvement once realistic detection and readout noise is considered. At the same time, in low-noise conditions a much simpler post-detection filtering procedure attains equal or better recovery than the synthesized CF reconstruction, making the synthesis unnecessary in practice.

The IceCube Neutrino Observatory, a cubic-kilometer detector at the South Pole, identifies neutrino flavor through event morphology. Sparse photon detection makes this classification particularly challenging in the 5--100~GeV regime, the energy range relevant for oscillation measurements and searches for physics beyond the Standard Model. We introduce WavePID, a template-based log-likelihood-ratio classifier that exploits nanosecond-scale timing on individual detector modules through three observables: the distance to the reconstructed vertex, the early-charge fraction, and the module-to-module time difference. Evaluated on a cascade-enriched sample selected by a state-of-the-art graph neural network, WavePID improves both cascade purity and classification performance over the neural network alone. This demonstrates that per-module pulse timing carries flavor-identification information complementary to morphology-based classifiers, opening a new physics-motivated observable for low-energy neutrino reconstruction. Geant4 simulations associate this signal with differences in Cherenkov emission geometry between muon tracks and electromagnetic showers. These results motivate exploiting nanosecond-scale pulse timing in future low-energy classifiers and in detector designs with improved per-module timing in next-generation neutrino telescopes.

This paper presents a four-channel prototype system for the geometric combining and coherent addition of tightly focused femtosecond laser radiation into a standing-wave field configuration. A stabilization system for beam pointing and relative phase of the four optical channels has been implemented, and its performance has been experimentally demonstrated. To characterize the standing-wave electromagnetic field distribution at the main focus of the system, an original measurement technique based on a fiber subwavelength optical probe has been employed. This work has been conducted in support of the exawatt-scale XCELS project.

The interaction of free-electron wave packets with electromagnetic fields provides a powerful route toward coherent electron control, enabling the generation of energy combs, momentum-state superpositions, and aberration-engineered electron beams. Existing theoretical descriptions, however, often rely on eikonal or no-recoil approximations. Here, we present a mesh-free numerical framework that directly solves the time-dependent single-particle Schrödinger equation for arbitrary electromagnetic potentials. Comparison with a benchmark mesh-based Schrödinger solver reveals excellent quantitative agreement. By eliminating the need for spatial meshing, our method offers an efficient and scalable route for simulating electron wave packet dynamics in complex time-dependent and static electromagnetic environments, while the simulation time is significantly improved by up to 800 times faster. These capabilities establish a versatile computational tool for quantum electron optics and free-electron-light interactions beyond eikonal approximations.

Phase noise limits the coherence and stability of soliton microcombs, yet its origin is difficult to trace because multiple noise sources act simultaneously. It is often represented by common-mode and repetition-rate components, but how each physical source contributes to these components remains unclear. We combine subspace tracking with multi-source Ikeda-map simulations, switching each source and the Raman nonlinearity on and off to isolate its contribution. Without Raman, pump phase noise is purely common-mode, while shot noise and amplified spontaneous emission drive the repetition rate noise. With Raman, the nonlinearity coherently converts pump phase noise from common-mode into repetition-rate noise without introducing an independent noise source, yielding a parabolic linewidth profile with a quiet-point minimum below the pump linewidth. When all noise sources are present, shot noise, ASE, and RIN raise the common-mode floor and shift this minimum toward the pump, setting the achievable noise floor. The intracavity dynamics thus do not merely carry noise but actively partition it, providing a mechanistic basis for low-noise microcomb design.

Digital in-line holographic microscopy is a computational imaging method useful for characterizing the refractive properties of a sample, i.e. the phase shift and absorption. This indirect measurement technique captures a diffraction pattern and uses reconstruction algorithms to retrieve the optical properties of the sample. Since only the intensity of the diffracted wave is recorded on the sensor, this inversion is not trivial, and simple backward propagation leads to artifacts known in optics as the ``twin-image''. With advances in deep learning, various algorithms have been developed for the reconstruction of in-line holograms, providing computationally efficient alternatives to iterative algorithms. These algorithms rely either on supervised learning, which requires ground truth knowledge, or physics-based self-supervised algorithms that require additional information, like phase diversity, but require multiple holograms for inference. This paper introduces a new self-supervised physics-based deep learning strategy that leverages phase diversity during training and then reconstructs sample's transmission function from a single in-line hologram during inference. We introduce five datasets of simulated and experimental in-line holograms of beads and bacteria. The proposed method produces accurate quantitative reconstructions similar or even more accurate than those obtained by regularized inversion while reducing the computational time by a factor of 1000.

While a looming atmospheric CO$_2$ overshoot calls for immediate carbon sequestration, delays associated to Enhanced Weathering (EW) carbon dioxide removal are being investigated. Topsoil acidity is already known to delay EW carbon sequestration, but subsoil acidity remains underexplored. Using century-long agricultural liming of formerly acidic heathland as a proxy for EW, this study provides empirical evidence of subsoil-imposed delays. Below such limed terrain, we observed a downward-progressing front of topsoil-produced alkalinity that still requires 30-100 years to penetrate the approximately 5 m thick acidic sandy unsaturated zone and reach the groundwater table. Subsoil acidity thus may cause beyond-reasonable delays, prohibiting EW as a viable short-term carbon capture strategy even on topsoils made non-acidic by preceding liming. When planning EW schemes, the amounts of stored acidic cations in top- and subsoil, as well as the rate and composition of infiltrating water, controlling the duration of the delay, require careful assessment.

Memory devices for single photons are notable components for quantum information processing and quantum communications. The present study investigates the possibility of achieving storage of light at the level of single photons inside nanofibers by exploiting stimulated Brillouin scattering. We present first the standard approach using a coherent buffer in a nanoscale waveguide by transferring the optical signal coherently to an acoustic wave, and that can be extracted by the reverse process. The life time of the acoustic wave put limitation on the applicability of such approach for single photon signals. We introduce a configuration for achieving a slow signal at the level of single photons without gain or loss. The process utilizes photon-phonon Brillouin interactions involving two counter propagating pump fields. The photon storage is achieved through time delay of significantly slow signal inside nanowires. We address the condition for getting negligible influence due to the scattering off thermal phonons.

Coherence-based spectroscopy methods are powerful tools to explore structure and dynamics of matter. However, towards higher photon energies, the generation of sequences of pulses with well-characterized relative delays and phases remains a challenge. Here, we introduce a method to measure the relative phase $\varphi$ between subsequent transform-limited pulses from high-repetition-rate x-ray free-electron lasers (XFELs). It is based on a Ramsey-type interference measurement, enabled by introducing long-lived Mössbauer resonances into the XFEL beam path up- or downstream a primary experiment, which allow one to bridge the temporal gap between the XFEL pulses. The measured phase can be used as additional input for the analysis of the primary experiment.

Glauber's coherent state is denoted by $\ketα$ and its two-mode extension is represented by $\ket{α,β}$. In this work, we introduce a two-mode superposition operator $A=tab+ra^\dagger b^\dagger$, whose action on the two-mode coherent state produces the two-mode coherent superposed quantum state $\ketψ=(tab+ra^\dagger b^\dagger)\ket{α,β}$. We investigate the nonclassicality and quantum non-Gaussianity of this state by means of the Wigner distribution and Wigner logarithmic negativity. Once its intrinsic nonclassical and non-Gaussian structure is established, the state is employed as the entangled resource in the Braunstein-Kimble continuous-variable (CV) teleportation protocol. We compute the ideal teleportation fidelity for coherent and squeezed inputs and analyze how the strengths of nonclassicality and non-Gaussianity influence the teleportation efficiency. Our results identify specific parameter regimes where enhanced non-Gaussian features or increased nonclassicality enable fidelities beyond the classical threshold, thereby revealing the operational significance of engineered two-mode quantum states in CV quantum information processing.

Despite increasing scale and resolution, many biological measurements remain destructive, revealing only spatial information rather than the dynamics it encodes. By combining flexible representations with mechanistic constraints, physics-informed machine learning offers a promising route to inferring these dynamics from static snapshots. Motivated by subcellular imaging of gene expression, we ask when a static spatial pattern of molecules can identify spatially varying diffusivity, creation, destruction, and boundary exchange, and how different inference schemes perform on the task. A structural identifiability analysis shows that distributed sources are non-identifiable, whereas a point source such as a transcription site can restore identifiability. These limits are further shaped by seemingly innocuous modeling choices: the boundary conditions, the spatial regularity of the underlying dynamics, and even the stochastic calculus convention. We then adapt several physics-informed schemes, differing in how they represent the solution and enforce the governing equations, and demonstrate effective inference from a single snapshot. Physics-informed approaches can thus recover spatial heterogeneities of biological dynamics from static data, but their use should be accompanied and guided by careful identifiability analysis for meaningful interpretation of the results.

We provide a theoretical derivation of the Hypergraph Minority Game with Local Hyperedge Payoffs (HMG-L), in which $N$ adaptive agents compete simultaneously in multiple overlapping groups modeled as hyperedges of a static hypergraph $\Hyper=(\Vset,\Eset)$. Each hyperedge constitutes an independent local minority game, and agents accumulate payoffs across all groups to which they belong. We derive the continuum-time limit of the score dynamics, from which we obtain a set of coupled nonlinear stochastic differential equations for the agents' strategy polarization variables. The deterministic drift is shown to derive from a global cost function that generalizes the standard Minority Game Hamiltonian to hypergraph-structured interactions. We perform a sparse-annealed replica analysis of the stationary state for the case of a $k$-uniform, $d$-regular random hypergraph, obtaining the saddle-point equations within the replica-symmetric ansatz, an explicit replicon stability criterion, and Bethe/cavity equations for sparse corrections. The leading sparse-regime transition occurs on a critical surface $\alphacrit(k,d)$, while the globally coupled MG value $\alphacrit\simeq0.3374$ is recovered only in the separate single-hyperedge limit. We derive expressions for the order parameters -- global volatility $σ^2$, predictability $θ$, hyperedge frustration $F_e$, and frozen fraction $φ$ -- and discuss their scaling behavior near criticality. The Fokker-Planck equation governing finite-$N$ fluctuations is presented, and the noise covariance matrix is computed from the hypergraph structure. Limiting cases ($k\to N$, $k\to2$, $d\to\infty$) are analyzed in detail, establishing connections to the standard MG, networked MG, and parallel MG models.

Machine-learning-accelerated materials discovery has yielded large numbers of computationally stable compounds, yet many remain experimentally unrealized, underscoring a persistent gap between prediction and synthesis. Here, we introduce a hierarchical screening framework that combines PBE-based thermodynamic stability, efficient dynamical-stability screening enabled by universal machine-learning interatomic potentials, and SCAN-based thermodynamic refinement. Applying this protocol to the 894 stable materials previously reported in Sci. Data 9, 302 (2022), we first curate 603 unique structures, of which only 298 remain thermodynamically stable on the complete PBE phase diagrams, demonstrating the critical role of competing phases in stability assessment. Dynamical screening then identifies 166 materials stable under both harmonic-phonon and finite-temperature molecular dynamics criteria, and SCAN phase diagrams further narrow this set to 109. Finally, by combining decomposition enthalpy with chemical-space completeness, we prioritize 25 candidates as high-confidence targets for experimental synthesis. This work provides a practical protocol for translating stability predictions into experimentally actionable synthesis targets, closing a key gap in machine-learning-driven materials discovery.

High-fidelity neutronic analyses of advanced reactors require deterministic transport solvers capable of handling complex unstructured geometries while maintaining computational efficiency. This work presents the development and verification of three GPU-accelerated deterministic solvers implemented within a unified framework, Neutronics using Deterministic Finite Element Algorithm (NuDEAL): the planar Method of Characteristics coupled with the Hybrid Finite Element Method (MOC/HFEM), the Discontinuous Galerkin Method of Characteristics (DGMOC), and the Discontinuous Finite Element discrete ordinate method (DFEM-SN). These solvers provide complementary capabilities for consistently solving the multigroup transport equation and can be selectively employed to balance accuracy, computational cost, and memory requirements for a given problem. All methods emphasize efficient GPU execution by leveraging memory alignment, compressed-flux storage, and sequential azimuthal sweeps. The solvers are validated on the C5G7 benchmark and applied to advanced reactor problems, including the ABTR, Empire microreactor, and MSRE. DFEM-SN achieved the highest accuracy, with eigenvalue errors below 50 pcm, while MOC/HFEM and DGMOC provided superior efficiency, with single-GPU runtimes comparable to those of large CPU clusters. The results demonstrate that deterministic GPU solvers on unstructured meshes can deliver both accuracy and scalability, enabling practical whole-core simulations for heterogeneous advanced reactors. The unified NuDEAL framework establishes a foundation for future extensions toward transient and multiphysics analyses on large-scale GPU architectures.

In the early-mid 1990s, scientists emigrating from the former Soviet Union to the United States -- especially physicists, engineers, chemists, and biologists -- frequently secured prestigious and visible positions, including professorships, named chairs, and laboratory leadership; comparable scientists arriving after about 2000 built more modest, less visible, and often non-academic careers. Against the common view that this reflects the people -- the elite having left first -- this article sets aside the thin apex of Nobel- and Fields-level émigrés and examines the larger cohort of capable but non-stellar scientists, showing that similar scientists fared differently by year of arrival. The explanation therefore lies in the structure of the receiving market, not primarily in individual ability. Reading premium appointments backward from later Nobel-level recognition risks survivorship bias: celebrated successes obscure the broader demand for Soviet scientific capital. I weigh four conditions that favoured the 1990s cohort and had largely closed by the mid-2000s: technology transfer and the export of a finite, distinctive stock of Soviet expertise that commanded a career premium; the favourable immigration regime created by the Soviet Scientists Immigration Act of 1992; the surge of U.S.-trained Chinese and Indian competitors; and the securitizing aftermath of 11~September 2001. All four mattered, but technology transfer and knowledge export were primary: their premium opened the window, and their depletion -- as exported knowledge was published and absorbed into global science -- removed the demand on which the other factors depended. A further cross-cutting mechanism, the cultural ``ghettoization'' of émigrés into co-national laboratory enclaves, capped their visibility and independent advancement. The imbalance between émigré generations was structural, not personal.

Next-generation photonic architectures for AI, sensing, and quantum computing require thousands to millions of reprogrammable photonic devices on a chip[1]. The monolithic integration of Electronically-backed Photonic Integrated Circuits (EPICs) allows for very high density electrical interconnection and electronic drivers that can scale with photonics. Piezo-optomechanical photonic integrated circuits (POMPICs) offer low power consumption, high speed modulation, cryogenic compatibility and broadband optical transparency from ultraviolet to infrared wavelengths[2,3], but have not been demonstrated with monolithically integrated CMOS electronics. Here, we show a fully monolithic, all-CMOS fabricated platform for POMPICs co-fabricated with commercial control electronics. 200 millimeter photonic wafers are constructed directly on completed CMOS driver wafers by back-end-of-line processing, connecting integrated piezoelectric actuators under broadband silicon nitride waveguides to a high-density digital backplane comprising >2 million electrical connections per die with 6.4x6.4 micron electrode pitch. We introduce segmented POMPIC components as Photonic Digital-to-Analog converters (PDACs) that convert low-voltage digital electronic signals to multi-bit analog optical phase and amplitude modulation, and we demonstrate parallel control of optical phase shifters, Mach-Zehnder interferometers, optical routing trees, and tunable ring resonators using a standard HDMI interface to program CMOS electronics. We test multiple reticles and perform electronic and photonic characterization across the entire wafer to establish uniformity and yield, demonstrating wafer-scale integration of POMPICs on an electronic backplane and enabling dense, scalable electronic control of piezo-optomechanical circuits.

Photonic integrated circuits on the silicon-on-insulator (SOI) platform typically interface with free space via grating couplers, but scaling these to collimated beams with diameters beyond 100 $μ$m requires a fundamentally different regime of extremely weak, spatially distributed coupling. While such large-area couplers have been demonstrated, their beam quality has remained largely uncharacterized, even though applications such as coupling into high-finesse resonators or trapping of cold atoms require both a large aperture and a near-Gaussian profile. This article presents an SOI meta-grating coupler that emits collimated, near-Gaussian beams of approximately 300 $μ$m waist diameter. The design synthesizes the required emission profile from a spatially tailored coupling strength, realized by locally varying a sub-wavelength unit cell while independently setting the local emission angle. This approach achieves the very low coupling strengths required for large beams and yields a measured beam quality of $M^2 \leq 1.10$. The scheme extends directly to other target profiles, such as flat-top or higher-order modes, rendering meta-grating couplers a practical chip-to-free-space interface for mode-matching-sensitive applications.

Given recent changes in federal climate policy, the United States is unlikely to meet its original 2030 Paris Agreement emission target of a 50-52% reduction from 2005 levels. However, rapid near-term abatement remains achievable through targeted multi-sector energy transitions. Extending the open-source energy system model, PyPSA-USA, to perform multi-sector analysis, we evaluate the primary drivers of USA energy costs and emissions though applying global sensitivity analysis. Our results suggest that fossil fuel price volatility is the dominant driver of marginal electricity and energy costs across most of the nation, however, uncoordinated state-level renewable mandates can induce localized cost spikes due to regional bottlenecks. We find that system climate impact (CO2e) is overwhelming sensitive to fugitive methane leakage rates and global warming potential assumptions. Addressing upstream methane leaks will play a crucial role in abating climate-related damages. Finally, demand-side electrification, specifically light-duty electric vehicles and service sector heating, can act as immediate levers for carbon abatement. The results of this work suggest that many of the Inflation Reduction Act's clean energy initiatives, that have since been repealed, are effective near-term solutions to reduce exposure to fossil fuel price and mitigate future financial penalties associated with the rising social cost of carbon.

Classical models of opinion dynamics represent individual opinions as scalar or vector values governed by the classical probability theory, either as deterministic quantities or random variables. This framework does not account for empirically observed phenomena such as cognitive ambivalence (where an individual simultaneously holds conflicting views) and order effects (where survey responses depend on the order in which questions are asked). We propose a quantum model of opinion dynamics in which each agent's cognitive state is represented by a density matrix that encodes both the expressed opinion and cognitive ambivalence. Survey questions become non-commuting self-adjoint operators, which provides a principled explanation for order effects. Our model also identifies quantities without classical counterparts, including quantum coherence and pairwise opinion covariances. Under a product state approximation, the quantum model reduces to the classical Friedkin--Johnsen opinion model. We test the framework on synthetic and real-world networks and observe that pairwise correlations follow network-dependent transient dynamics but converge to the same steady state regardless of the network, and that quantum coherence decays exponentially at a rate independent of the network.

We present a laser system referenced to a room-temperature ultrastable cavity employing crystalline AlGaAs coatings. We demonstrate a fractional frequency instability of $4.2 \times 10^{-17}$, which is one of the lowest for room temperature systems and surpasses the limit imposed by Brownian noise if dielectric coatings were employed. For the first time in a room temperature system we identified the spontaneous fluctuations of the coating birefringence as a leading contribution to frequency instability. At optimized conditions we achieve an ultrastable cavity with an eigenfrequency that is highly immune to power fluctuations. As acceleration noise is the main noise contribution, we demonstrated that a feed-forward method can reduce the influence of accelerations on the cavity-stabilized laser frequency by a factor of four.

This paper explores the selective suppression of Laguerre-Gaussian modes that are most vulnerable to atmospheric turbulence. Decomposing these modes into an orthogonal Zernike polynomial basis reveals significant differences in stability depending on the radial and azimuthal indices. We demonstrate that modes with a higher radial index exhibit minimal distortion of the transverse beam profile, providing a clear criterion for filtering out less resilient modes in turbulent media. Furthermore, we derive an analytical expression relating the required receiver aperture to the radial and azimuthal indices.

This paper presents a new hybrid MC/deterministic method for solving the one-group steady-state Boltzmann transport equation based on decomposition of solution in macro and micro components. The macro component captures the large-scale structure of the solution. It is represented by angular moments of the high-order transport solution. The $P_1$ approximation is applied to define the macro component. The first two angular moments are obtained as a solution of hybrid low-order moment equations with exact closures. The equation for the micro component is solved using a MC simulation. The hybrid two-level system of equations for macro and micro components is solved by fixed-point iteration scheme. Numerical results are presented to demonstrate variance reduction of stochastic numerical solution and improvement in computational efficiency.

Accurately reconstructing gene regulatory networks (GRNs) is essential for understanding transcriptional processes in development and disease. MERLIN-SUITE (https://github.com/Roy-lab/MERLIN-SUITE) represents a collection of algorithmic extensions based on MERLIN (Modular regulatory network learning with per gene information) a probabilistic framework that infers gene-specific and module-specific regulatory programs of co-regulated modules, capturing both detailed and modular aspects of transcriptional networks. While expression-based inference is effective, it often aligns poorly with experimentally validated regulatory interactions. MERLIN-P addresses this by integrating external regulatory priors, such as motif, ChIP, and perturbation data, to enhance biological relevance and predictive accuracy. MERLIN-P-TFA further advances the framework by incorporating regularized estimation of latent transcription factor activity (TFA), overcoming the limitation that TF mRNA levels may not represent protein activity. By integrating expression data, prior knowledge, and activity-aware modeling, this unified approach supports robust GRN reconstruction in both bulk and single-cell datasets. This chapter presents the MERLIN-SUITE with a focus on MERLIN-P-TFA and demonstrates its use on a single-cell, multi-modal dataset of mouse cellular reprogramming to infer GRNs and identify key regulators.

3D plant phenotyping is notoriously known to be procedure-complicated and of low throughput due to the extensive multi-view imaging, the fragile 3D reconstruction pipeline, and the additional cost from reconstructed geometry to phenotypic extraction. These limitations are further amplified in low-cost data acquisition, where smartphone videos or sparsely sampled multi-view images provide limited view overlap and self-occlusion. In this work, we show that the conventional 3D plant phenotyping pipeline could be streamlined and significantly accelerated with 3D Foundation Models (3DFMs), and particularly, present one of the first cross-crop 3D phenotyping frameworks powered by 3DFMs. The framework replaces COLMAP-style sparse initialization with 3DFM-based feed-forward geometric recovery, combines geometry-constrained 3D Gaussian Splatting for dense reconstruction, enables few-view reconstruction through iterative view synthesis and refinement, and converts reconstructed geometry into measurable organs through 2D-to-3D semantic transfer, metric scale recovery, and organ instance separation. We further construct a cross-crop dataset with smartphone-based image acquisition, diverse plant morphologies, and manual annotations for segmentation and phenotypic evaluation. Experiments across 26 plant sequences show that 3D Foundation Models reduce the average reconstruction time from 6.52 minutes to 1.58 seconds while maintaining high reconstruction quality and phenotyping accuracy. These results suggest a fresh technical route for high-throughput 3D plant phenotyping, from low-cost image acquisition to fast reconstruction, perception, scale recovery, and phenotypic measurement.

Transcriptional gene regulatory networks (GRNs) depict the directed relationships between regulators and target genes, determining gene expression patterns in a cell-type-specific manner. Single-cell multi-omics technologies, such as single-cell RNA sequencing (scRNA-seq) and single-cell Assay for Transposase-Accessible Chromatin using sequencing (scATAC-seq), enable high-resolution measurement of cell-type-specific gene expression and regulation in an unprecedented way. However, tools for inferring cell-type-specific GRNs and modeling their dynamics remain scarce. To facilitate the inference and analysis of cell-type-specific GRNs in contexts such as cellular development or disease progression, where cell lineage structure and dynamics are important, we developed a multi-task learning framework, single-cell Multi-Task Network Inference (scMTNI). scMTNI and its associated network analyses tools offer a comprehensive package to define cell-type-specific GRNs and examine their dynamics. This book chapter describes the scMTNI tool and demonstrates its application to an existing cellular reprogramming single cell multi-modal dataset to infer cell-type-specific GRNs and identify key regulators of cellular fate transitions during cellular reprogramming.

Histone H3K27M mutation status defines a clinically aggressive subgroup of pediatric diffuse midline glioma and informs prognosis and trial eligibility, but confirmation usually requires tissue sampling from eloquent midline structures. We evaluated whether radiomics from routinely available T2-weighted MRI can provide an adjunctive screening signal in a heterogeneous referral-style cohort, where scans are often acquired externally and T2-weighted imaging is the only consistently available sequence. Ninety-eight pediatric patients with tissue-confirmed status were analyzed (73 mutation-positive, 25 wild-type). Expert tumor segmentations defined the regions of interest for PyRadiomics feature extraction after isotropic resampling, dual skull stripping, and multi-scale filtering. We systematically ablated preprocessing, correlation pruning with repeated recursive feature elimination, tumor volume, and TabDDPM synthetic minority augmentation across 100 stratified train/test splits with real-only test sets. Pure radiomics achieved accuracy 0.664 and F1-score 0.784. The best pipeline used preprocessing, feature selection, and volume with CatBoost, achieving accuracy 0.730$\pm$0.068 and F1-score 0.826$\pm$0.044. TabDDPM improved TabPFN to F1-score 0.81$\pm$0.05 at 200 augmented rows. These results support T2-weighted radiomics as a moderate screening and triage aid, not a replacement for tissue-based diagnosis.

Based on the growth patterns of 166 CHO monoclones observed over a 15 day period, we show that the standard population growth in a confined space equation, i.e. the sigmoid/logistic function, is alone does not capture the complex behaviour of the cell growth in a confined space. Thus, combining the sigmoid function and the exponential of the sigmoid function, we present a more accurate model for modelling cell growth in a confined space. We also present a working algorithm to obtain population growth variables (growth capacity, growth time and growth rate), model the growth patterns of the CHO monoclones, and we include subset of the dataset, along with a sample python script for the reader to replicate the results. Furthermore, we derive a model for cell confluence growth in a confined space, numerically model the confluence and present the reader with a working algorithm. With Kolmogorov-Smirnov analysis conducted on the area of the CHO monoclones, we show that the cell area of the incipient population is normally distributed, the sparse cell population is gamma distributed and the dense colony population is log-normally distributed. Thus, we further derive models for the mean, the standard deviation, the coefficient of variation and the inverse coefficient of variation for the log cell area growth in a confined space, numerically model them and present the reader with working algorithms. Finally, based on the growth patterns of another 48 CHO monoclones observed over a 16 day period, and their titer and viability measurements, we find the correlation coefficients with our calculated growth variables, and titer and viability measurements, and show that our derived growth variables can be used to predict the productivity and the health of a cell. Thus, we conclude our study by demonstrating that the productivity and the health of a cell (also the overall population) are interdependent.

This paper investigates a robust transmission design for a multi-user rate-splitting multiple access (RSMA)-based simultaneous wireless information and power transfer (SWIPT) system empowered by movable antennas (MAs) and a reconfigurable intelligent surface (RIS) under channel state information (CSI) uncertainty and residual hardware impairments (HIs). The effective channels in MAs-enabled systems depend on antenna positions, causing CSI uncertainty to affect not only active and passive beamforming but also antenna position optimization. Furthermore, residual HIs distort the effective SINRs, creating additional coupling among beamforming, RIS reflection control, common-rate allocation, power-splitting ratio optimization, and antenna position optimization. Consequently, the joint impact of CSI uncertainty and HIs leads to a highly coupled and challenging resource allocation problem. To address this challenge, we propose a robust resource allocation framework that jointly optimizes common-rate allocation, transmit beamforming, RIS reflection coefficients, power-splitting ratios, and MAs positions to maximize the achievable sum-rate while satisfying practical system constraints. To obtain an efficient solution, the original problem is decomposed into active beamforming, RIS reflection design, power-splitting ratio optimization, and MAs position optimization subproblems, where tractable convex surrogate functions are constructed to handle the non-convex objective and constraints. Simulation results verify the effectiveness of the proposed framework and demonstrate substantial improvements in achievable sum-rate, robustness against CSI uncertainty and hardware impairments, and convergence performance compared with benchmark schemes.

This paper investigates geometry-reconfigurable transmission for multiuser communication systems enabled by a rotatable antenna array. In contrast to conventional fixed arrays, the proposed architecture jointly exploits array pose adjustment and element-level boresight steering, thereby reshaping both the array-induced phase responses and the direction-dependent channel gains. We formulate a weighted sum-rate maximization problem that jointly optimizes the transmit beamformers, array pose, and element boresights under practical visibility and steering constraints. To reveal the underlying design principles, we first provide a geometric interpretation via zero-forcing analysis, showing that the resulting rates stem from both channel-strength enhancement and spatial-separability improvement. Specifically, array-pose rotation improves inter-user channel orthogonality even with isotropic elements, whereas directional elements introduce a tradeoff between phase-based spatial separation and boresight-dependent gain alignment. Motivated by these insights, we develop an efficient optimization framework that jointly coordinates transmit beamforming, array-pose adaptation, and element-boresight steering to exploit the geometry-induced phase-and-gain channel-shaping capability. Simulation results demonstrate that the proposed joint design outperforms fixed-array, pose-only, and boresight-only benchmarks, with larger gains achieved under more directive element patterns and tighter boresight-steering constraints.

Monitoring the radio-frequency (RF) spectrum from space imposes demanding requirements to satellite platforms in terms of communication bandwidth and computational resources, which are necessary for the downlink, the storage, and the processing of high-throughput I/Q samples. This paper analyzes in depth the quasi-direct geolocation (QDG) as a technique to enable the exploitation of satellites of opportunity in low Earth orbit (LEO) to sense the spectrum in the bands of global navigation satellite systems (GNSS). This is a technique of passive RF geolocation and consists of an ensemble of signal processing algorithms, which compress the I/Q samples and process the compressed data through fast delay-Doppler shift matching and interferometry in a quantized time-frequency domain. These algorithms speed up the exhaustive search of multiple RF sources in the position domain. The efficiency gain addresses the bottleneck that prevents the employment of satellites, which are limited in downlink capacity and on-board computational power. These satellites are usually constrained in size, weight and power (SWaP) and represent most of the spacecrafts in LEO. The ability to exploit assets as such for the geolocation of terrestrial GNSS jammers in near real time is instrumental the performance of a multi-constellation GNSS RFI monitoring system. The present work describes the mathematical framework and precision bounds, introduces single- and multi-antenna uses cases, combines different compression methods, and evaluates the geolocation accuracy with real data. The I/Q samples were collected by a repurposed GNSS reflectometry (GNSS-R) satellite, OPS-SAT PRETTY, in a dedicated test session during Jammertest 2025. The experimental results demonstrate the capability to geolocate GNSS jammers with different signal-to-noise ratios (SNR) with extremely high compression ratios.

Terahertz (THz) transmission technologies hold significant potential for enabling ultra-broadband, short-range communication in next-generation networks. Despite the vast bandwidth, THz signals suffer from limited transmission range and a feasible scenario is to deploy THz within clustered heterogeneous networks (HetNets) to enhance coverage. This paper investigates THz communication in clustered HetNets, leveraging stochastic geometry for performance analysis. Specifically, we consider two tiers of macro base stations (MBS) and small base stations (SBS). The MBS tier is modeled as a Poisson Point Process (PPP), and both the SBS tier and users are modeled as a Poisson Cluster Process (PCP) to capture user clustering and network hotspots. We derive the analytical expressions for user association probabilities, the Laplace transform of interference, and the coverage probability. The derived coverage probability is validated through Monte Carlo simulation. The numerical results show that the coverage in THz PCP-HetNets is higher than that achieved in THz PPP HetNets. In addition, a moderate spatial spread of SBSs is beneficial for coverage.

High-altitude platform stations (HAPS) are envisioned as a key component of future wireless networks, enabling ultra-wide coverage and providing direct connectivity to users with cylindrical massive multiple-input multiple-output (mMIMO) systems. Exploiting the channel degrees of freedom necessitates accurate modeling and characterization of three-dimensional (3D) channels in the presence of spatial correlation functions (SCFs). However, existing spatial correlation models are primarily developed for planar or linear antenna arrays and cannot be directly applied to cylindrical geometries commonly adopted by HAPS platforms. To address this limitation, this paper derives an exact closed-form expression for the SCF of 3D MIMO channels with antenna elements arranged in a cylindrical array. The proposed formulation is based on the spherical harmonic expansion (SHE) of plane waves and accommodates arbitrary antenna radiation patterns and angular distributions through the Fourier series (FS) coefficients of the power azimuth and zenith spectra. The derived SCF is validated through Monte Carlo simulations under standard-compliant settings.

Brain-Computer Interfaces (BCIs) have revolutionized neuroscience applications, from motor rehabilitation to neuroergonomics. Traditional implantable BCIs with invasive microelectrode arrays pose challenges, notably the need for wired connections and inherent implantation risks. This paper introduces a battery-free wireless BCI system, consolidating an implant and its external supporting system. Our design centers on a dual-function antenna system: firstly, an inductive coupling mechanism enables wireless power transfer, sufficiently powering the implant's Application-Specific Integrated Circuit (ASIC) for stimulation and readout without an implant battery. Secondly, a backscatter antenna in the implant facilitates battery-free, high-data-rate wireless connectivity (up to 32 Mbps). This system not only enhances the BCI experience by eliminating wires but also retains data fidelity and energy efficiency, promising a safer, more efficient interface for tasks like robotic arm control.

Wake-up radio (WUR) is a technology designed to enhance the energy efficiency of Internet of Things (IoT) networks and extend device battery life. While most studies focus on WUR performance with single-antenna base stations, this paper investigates the multiple-input multiple-output (MIMO) technology to improve device energy saving and extend the coverage of wake-up signals. By leveraging MIMO beamforming, the transmitted energy can be spatially focused toward the intended IoT devices, with high beamforming gain and minimal inter-device interference. We develop a preliminary analytical framework using stochastic geometry to evaluate the wake-up success probability of WUR-MIMO in multi-cell cellular IoT networks, when the number of antennas equals $2 \times (\text{number of devices}) - 1$. Monte Carlo simulations show that, relative to a single-antenna WUR baseline, MIMO beamforming significantly enhances wake-up reliability when this antenna configuration is applied, mitigates more than 50% of false activations across all settings, and thereby prolongs the lifetime of IoT devices.

Channel State Information (CSI) has become a widely used wireless channel sensing modality for applications such as indoor localization, activity recognition, and respiration monitoring. Because collecting labeled data under every target condition is impractical, training CSI-based models often relies on simulated data produced by adding noise or perturbations to recorded channel estimates, most commonly additive white Gaussian noise (AWGN). This practice assumes that the receiver chain between the antenna and the channel estimator is linear and gain-invariant. We test this assumption empirically using RF jamming as a controlled perturbation on 6 commodity receivers across 2 indoor environments. The assumption does not hold. Automatic gain control compresses the channel estimate multiplicatively before digitization, producing amplitude distributions that no additive noise variance can reproduce. To close the resulting fidelity gap, we propose M_QTC, a measurement-calibrated model that learns the per-subcarrier distribution transformation through quantile mapping, temporal filtering, and copula-based cross-subcarrier reordering. M_QTC reduces amplitude error 8-fold and closes 89% of the aggregate fidelity gap across four complementary dimensions. The improvement transfers directly to downstream tasks, where 5 classifiers from different families trained on M_QTC-simulated data recover 93% of real-data jamming detection performance, while AWGN-trained classifiers remain near random decision.

Vision-radar fusion is central to robust autonomous driving, combining dense visual semantics with precise range and velocity measurements from radar. However, real-world fusion quality is fundamentally challenged by dynamically varying input quality, stemming from occlusion, adverse weather, and channel noise. To address this, we re-frame the problem from static data fusion to channel-aware semantic reasoning and propose a Large Language Model-centric Semantic-layer Channel-aware Integrated Perception (LM-SCIP) framework. It places a Large Language Model (LLM) as a central reasoning core to fuse a local visual stream with a quality-varying external radar stream used to cover perception-blind spots. Concretely, LM-SCIP couples a hierarchical radar-vision encoder with a Channel-Adaptive Semantic Module (CASM) that maps link indicators into a "Channel Prompt" to dynamically gate external radar features. A parameter-efficient, LoRA-tuned LLM, in conjunction with a heterogeneous Mixture-of-Experts (H-MoE), then arbitrates between local visual cues and the channel-conditioned radar context. Finally, a decoupled multi-task decoder outputs localization, trajectory forecasting, and image reconstruction. Experiments on nuScenes and VIRAT validate our approach. On nuScenes, under a controlled toggle of radar input, LM-SCIP reduces localization RMSE by 40.0% versus a vision-only baseline. On VIRAT, the model attains a 0.214m localization RMSE and 0.179m minFDE (k=1). These results reveal that the proposed LM-SCIP enables a robust vision-dominant fallback at low SNR and synergistic fusion at high SNR.

This paper investigates the joint uplink scheduling and power control problem in a coordinated multicell wireless network, where at most one single-antenna user is allowed to access the single-antenna base station in each cell simultaneously. The resulting weighted sum-rate (WSR) maximization problem is a mixed discrete-continuous, nonconvex optimization problem that is notoriously difficult to solve directly. Classical fractional programming (FP) methods tackle this problem by leveraging the Lagrangian dual transform (LDT) followed by the quadratic transform (QT), yielding a tractable closed-form solution for scheduling and power control, with the LDT playing a crucial role in handling discrete variables. In this paper, we revisit the LDT from a minorization-maximization (MM) perspective and observe that its induced surrogate is somehow conservative due to the reciprocal-coordinate construction. Motivated by this observation, we propose a novel reciprocal-inversion transform (RIT) that constructs a tighter first-order Taylor expansion lower bound for the logarithmic rate function. The proposed RIT remains fully compatible with the QT, leading to a surrogate-enhanced FP (SEFP) algorithm for joint uplink scheduling and power control. The proposed SEFP algorithm retains the desirable per-cell separability of the classical FP framework and admits closed-form updates for the auxiliary variables, scheduling decisions, and transmit powers. Simulation results demonstrate that the SEFP algorithm consistently outperforms the classical FP method and other baselines for different network utilities.

Simultaneous acoustic information and power transfer (SAIPT) plays a crucial role in enabling self-sustainable and maintenance-free Internet of Underwater Things (IoUT) networks. This paper studies a multicarrier underwater SAIPT system that jointly considers the frequency-dependent characteristics of acoustic transducers and the nonlinear behavior of rectifier circuits. The waveform vector is firstly optimized using the successive convex approximation (SCA) method under constraints on average and peak transmit power for acoustic power transfer (APT). Then, in the SAIPT scenario, both the power splitting factor and waveform vectors are jointly optimized through an alternating optimization (AO) framework based on SCA, subject to transmit power and achievable rate constraints. Simulation results demonstrate that incorporating the transducer's frequency response, rectifier nonlinearity, and the high peak-to-average power ratio (PAPR) of multicarrier waveforms leads to a significant improvement in acoustic energy transfer efficiency. The results also show that the energy harvesting DC output can be further enhanced by properly choosing system parameters, such as the number of subcarriers and subcarrier spacing.

Channel knowledge maps (CKMs) enable environment-aware wireless systems by providing location-specific channel knowledge, but long-term environmental variations, such as construction, traffic redistribution, and foliage changes, require periodic map refresh. In practice, channel measurements are often sparse and irregular, while environmental knowledge may be limited to coarse layout or topology descriptors. This paper studies CKM reconstruction from sparse measurements. We show that reconstruction pipelines that apply local aggregation or spectral operators directly to a zero-filled pilot grid can entangle the sampling mask with the channel field, allowing structural priors to act on mask-induced distortions before the measurements define a supported radio field. To address this issue, we propose Anchor-CKM, a measurement-first, knowledge-aided reconstruction framework. Anchor-CKM first uses support-aware partial convolutions to construct a pilot-supported representation, and then performs layout-conditioned dual-path Fourier refinement followed by coordinate-based heteroscedastic prediction of the CKM mean and per-location predictive variance. Experiments on transmitter-disjoint DeepMIMO scenarios cover missing ratios from 0.3 to 0.95, including stringent 5% to 10% pilot-coverage settings. In explicit-layout outdoor scenarios, Anchor-CKM reduces received-power root-mean-square error (RMSE) by 0.79 to 1.33 dB relative to the strongest reproduced baseline, while ablations identify pilot-support stabilization as the largest contributor and layout conditioning as beneficial for line-of-sight/non-line-of-sight (LOS/NLOS) boundary fidelity.

Accurate assessment of eating behavior is essential for understanding and managing conditions such as eating disorders, obesity, and diabetes. Wearable-based food intake detection has shown considerable promise; however, most existing approaches are trained and evaluated using internal validation on a single dataset with fixed sensor orientation and known wearing hand, which limits their generalizability to real-world settings. Furthermore, many existing approaches rely on both accelerometer (acc) and gyroscope (gyro) signals to achieve strong performance. However, gyro measurements may be unavailable in some real-world deployments due to battery constraints, and performance often degrades when only acc data are used. We propose a generalizable framework for orientation-invariant eating episode detection, with an acc2gyro module to improve performance in acc-only settings. The framework is trained using fine-grained wrist-worn datasets and externally validated across three heterogeneous datasets: the Clemson All-Day (CAD) and Capture-24 datasets, as well as Physio-ED, a dataset collected from individuals with eating disorders. Across external evaluations, the proposed framework demonstrates robust performance despite substantial variations in sensor modality, wearing hand, participant population, and annotation protocols. Specifically, the framework achieved F1-scores of 0.751, 0.592, and 0.793 on CAD, Capture-24, and Physio-ED, respectively, with CAD performance exceeding recent state-of-the-art methods evaluated using internal validation only. This study provides the first external validation of eating episode detection in an eating disorder population. Additionally, the acc2gyro module improves the performance in acc-only settings. These findings demonstrate the potential of orientation-invariant wearable sensing for scalable and clinically applicable assessment of eating behavior.

This paper proposes a full duplex fluid antenna near field system (FD-FANS) with a multi-sector antenna array that jointly exploits resource allocation, antenna mobility, and group-based transmitting (TX) and receiving (RX) partitioning. A spherical wave uplink downlink channel is established that accounts for residual self interference (SI), wireless energy transfer (WET), and geometric constraints on antenna motion. Within the FD-FANS framework, an efficient protocol is devised to enable simultaneous downlink energy transmission (DET) and uplink data transmission (UDT) at the base station (BS). Furthermore, we formulate, for both perfect and imperfect SI cancellation (SIC), a weighted sum rate (WSR) maximization problem over time power allocation, antenna positions, and binary group selection, under practical average and peak power limits, per antenna box constraints, minimum spacing, and a half TX half RX balance. To tackle the resulting non convex mixed integer design, we develop an efficient alternating optimization (AO) framework based on majorization minimization successive convex approximation (MM SCA). The proposed algorithm monotonically improves the objective and converges to a stationary solution of a continuous relaxation. Simulation results demonstrate that the proposed scheme achieves consistent performance gains over several benchmark designs, including half duplex FANS (HD FANS), FD fixed position antenna near field system (FD FPANS), non-grouped FD FANS, and far field counterparts, in terms of average sum rate (ASR), energy efficiency (EE), and user fairness, while exhibiting robustness to residual SI and channel uncertainty.

Transversal CNOTs are ubiquitous for entangling logical qubits of identical CSS codes pairwise. For distinct codes, the options are much more limited, and are typically known only for structurally related code families. We introduce an automated framework for synthesising inter-code logical CNOT circuits between arbitrary CSS codes using chain maps. Given a prescribed bipartite logical CNOT network between these codes, our method constructs the affine space of chain maps realising the desired logical action, and then searches this space for shallow and sparse physical circuit candidates. We benchmark this method on a range of heterogeneous CSS code pairs, recovering known transversal constructions, and finding new low-depth solutions, including distance-preserving and partially distance-preserving examples, which we demonstrate can be promoted to the full code distance using additional flag measurements. We discuss applications to code switching, magic-state injection, Pauli product measurements, and operations on concatenated codes, where bespoke chain maps offer favourable spacetime tradeoffs for logical interfaces tailored to heterogeneous architectures. Finally, we show how our framework straightforwardly extends to targeted logical CZ gates.

Characterizing noise in quantum circuits is fundamentally limited by gauge degrees of freedom; certain parameters, such as the individual contributions of state preparation and measurement (SPAM) errors, are in principle unlearnable from any experiment within the gate set. Here, we show that the physical structure of realistic noise processes imposes approximate symmetry constraints on the Pauli fidelities of gate noise channels. These symmetries relate the fidelity of a Pauli $P$ and its gate-conjugate $U_g P U_g ^{\dagger}$, and can be used to fix the gauge using only knowledge of the error type and not its magnitude. Using Lindbladian perturbation theory, we analyze a broad class of Clifford gates, including $ZZ_{π/2}$, CZ, CNOT, iSWAP, and SWAP, and demonstrate that coherent errors do not induce first-order asymmetry, while only a restricted set of predominantly off-diagonal dissipative errors can break the symmetry at first order, for which we derive simple selection rules. Notably, common single-qubit noise sources such as $T_1$-relaxation and $T_{2φ}$-pure-dephasing can only cause asymmetry at second order. Leveraging these symmetries to fix the gauge enables systematic identification of SPAM errors, simplifying error characterization and mitigation. We validate our results numerically and experimentally on IBM Kingston.

We investigate the integrability-to-chaos transition and information scrambling in Ising spin networks via a graph-theoretic formulation. Modeling spins as vertices and interactions via adjacency matrices across path, Erdős--Rényi, and Watts--Strogatz topologies, we demonstrate that long-range couplings and heterogeneous degree distributions drastically accelerate quantum information propagation. The Hamiltonian comprises local and normalized non-local interactions; tuning the non-local coupling and field heterogeneity drives integrability breaking. To quantify scrambling, we employ bipartite mutual and tripartite information. Increasing non-local interactions drives tripartite information to large negative values, signaling deep information scrambling. Out-of-time-order correlators (OTOCs) exhibit exponential early-time growth, yielding quantum Lyapunov exponents that scale systematically with parameters governing the chaotic regime. Complementing this, Krylov complexity reveals rapid operator growth in the chaotic phase, synchronizing with OTOC and mutual information dynamics. Spectrally, the transition manifests as a shift from Poissonian to Wigner--Dyson level spacing statistics. The spectral form factor (SFF) exhibits the characteristic slope-dip-ramp-plateau structure, enabling the extraction of Thouless and Heisenberg times. Crucially, a reduced Thouless time strongly correlates with accelerated informational and operator scrambling. Ultimately, this work establishes a unified framework bridging network topology with information-theoretic, operator, and spectral diagnostics, offering profound insights into thermalization and non-equilibrium dynamics in quantum many-body systems.

We introduce a graph-theoretic framework for representing color-ordered maximally helicity violating (MHV) scattering amplitudes in quantum chromodynamics using coined quantum walks on permutation trees. Each root-to-terminal path corresponds to a distinct color ordering of the external gluons, while local transition amplitudes are assigned according to the spinor-product structure of the Parke--Taylor amplitudes. The walk evolves in coherent superpositions over permutation sectors, giving a dynamical picture of the underlying combinatorics. A quantum-channel formulation based on Kraus operators is also introduced to describe sector-resolved contributions, while a weighted collection operator coherently combines the terminal sectors at a common reference node. A quantum Fourier transform on the coin space is then employed to combine the encoded contributions into the corresponding color-decomposed amplitude. Together, these constructions establish a unified graph-based framework connecting permutation trees, quantum walks, and open quantum systems providing a framework for quantum algorithms to simulate scattering processes in quantum field theory. As an example, numerical results for low-point gluon amplitudes demonstrate that the proposed representation faithfully captures the characteristic Parke--Taylor structure and is consistent with analytical results.

Non-Gaussian quantum states and operations constitute essential resources for achieving quantum computational advantage and enabling quantum error correction in bosonic platforms. However, their generation in optical settings remains a challenging experimental task, often relying on probabilistic heralded protocols. Here, we present an in-depth analysis of the suitability of photon catalysis between low number Fock states and squeezed states for the generation of squeezed coherent state superpositions. We employ the stellar rank formalism to characterize the non-Gaussian complexity of input resources (including both states and measurements) and the generated states. This enables a systematic comparison of the fidelity between the catalyzed output and the target states to the maximum fidelity achievable by any protocol with the same non-Gaussian input resources. In this sense, we identify instances where the catalysis protocols considered here are provably optimal. We identify parameter regimes in which high-fidelity approximations of the target states can be achieved with minimal resources. Furthermore, we benchmark the performance of photon catalysis against Gaussian boson sampling-inspired protocols in terms of success probability and state quality, highlighting the advantages of deterministic Fock state sources. We also investigate the generation of related non-Gaussian resources including squeezed Fock states relevant for quantum error correction. To account for experimental imperfections, we model losses across all optical modes using a Hilbert space truncation approach in the Fock basis and analyze the robustness of the generated states under realistic conditions. Our results quantify the trade-offs between non-Gaussian resource complexity, achievable fidelity, and losses in photon catalysis protocols, providing practical guidelines for near-term photonic implementations.

We construct a perturbative scalar-tensor solution describing a central inhomogeneity embedded in an evolving cosmological background, with the aim of studying black hole persistence through a nonsingular bounce. Scalar-tensor gravity provides a natural framework for realizing bouncing cosmologies, while the inclusion of a localized inhomogeneity makes the field equations substantially more difficult to solve. We therefore adopt a perturbative scheme, with perturbative parameter $ε$, in which the leading-order equations are solved by a spatially flat bouncing FLRW spacetime sourced by a radiation perfect fluid. At next order, a central inhomogeneity is introduced through a generalized McVittie geometry, with the perturbations encoded in the corresponding first-order metric and scalar-field functions. We first allow an anisotropic fluid with radial and tangential pressures, whose diagonal components solve the diagonal field equations. The field equations are solved as a series expansion up to $\mathcal{O}(η^4)$ near the bounce at $η=0$. The resulting perfect fluid solution contains three arbitrary functions which are constrained by requiring the spacetime to asymptote to FLRW as $r\to\infty$. With suitable initial conditions preserving the parabolic structure of the bounce, the integration constant $d_0$ emerges as the true perturbative parameter: all perturbations vanish as $d_0\to0$. Finally, we find a small evolving horizon, $r_h\sim d_0$, which we interpret as the horizon of the central inhomogeneity. Its persistence through the bounce supports the interpretation of a black hole surviving the cosmological transition, and its evolution is not symmetric about $η=0$.

The no-broadcasting theorem in quantum information says that a set of states on a quantum system admits a common broadcasting (copying) operation if and only if their density matrices belong to a commuting family. We discuss and prove this theorem, as well as the closely related no-cloning theorem in the context of quantum probability theory, i.e. in the category of (finite dimensional) C-star-algebras with unital completely positive maps.

Recently, a distinct non--quasinormal mode component of black hole binary merger radiation, named the direct wave, has been identified. The frequency and damping time of the direct wave have been associated with properties of the remnant horizon. This has led to direct-wave based analysis of GW250114, including a test of Hawking's area law. However, as we demonstrate here using numerical relativity strain data, the direct wave frequency is not correlated with the horizon frequency or surface gravity, other than an incidental crossing around $χ_f \approx 0.7$ corresponding closely to the remnant spin of GW250114. We show that while the instantaneous frequency of the direct wave is quasi-stable, the damping time shows significant evolution and therefore a single damped sinusoid model, containing a fixed damping time, is not appropriate. We further show that an evolving frequency model based on horizon properties also does not model the direct wave component for systems with large remnant spins. We demonstrate that testing Hawking's area law with a horizon frequency based on the direct wave interpretation will lead to apparent violations of Hawking's area law when no violation actually occurs. Our results therefore indicate that the direct wave is not a reliable probe of the remnant horizon's properties.

Quantum Error Recovery (QER) uses knowledge of the error channel acting on a quantum system to find optimal recovery maps. The scheme restores the uncorrupted state with a fidelity exceeding that achieved by noise parameter independent quantum error correction. We use a generic coherent QER map implemented with a quantum circuit acting on the system together with ancillary qubits to recover quantum information stored in permutation invariant (PI) codes. PI codes admit tunable parameters to suit the noise model and benefit from simple recovery operation circuits with reduced addressability requirements, unlike stabilizer codes. We showcase the method by modeling QER in PI codes after collective and local symmetric correlated amplitude-damping (AD) noise, a non-Pauli noise process for which stabilizer codes often require additional overhead. We also propose a new PI code family called CAD codes with explicit examples on 4 and 9 qubits for global symmetric AD errors. We show that CAD9 (supported on 9 qubits) code beats many existing codes by more than one order of magnitude. For the CAD4 code, which perfectly corrects 1 global symmetric AD error, the compiled recovery circuit consists of 10 system and system-ancilla gates which can be realized from linear geometric phase gates. Our work provides a direct path from optimized recovery maps to experimentally implementable, low-overhead protocols.

The universality of dynamical scaling laws constitutes a cornerstone in the theoretical understanding of quantum many-body systems, particularly in non-equilibrium settings. Recent advancements have proposed a phenomenological ansatz based on spontaneous symmetry breaking (SSB) to unify the description of charge transport in open quantum systems. However, it remains unclear under which conditions it fails to capture the emergent hydrodynamics and if it does break down, whether nontrivial dynamics emerge. In this work we show that Kardar-Parisi-Zhang (KPZ) dynamics in an open integrable model (the B3 model), rather than diffusion from SSB, emerges. We find that the B3 model is equivalent to two interacting asymmetric XXZ spin chains and the ansatz can only capture the influence of the inter-chain interactions. When the initial state is appropriate, the asymmetric XXZ structure dominates the dynamics, which gives KPZ scaling behavior even when the hopping rate becomes negative. Our work motivates theory of charge transport in open systems beyond the ansatz based on SSB.

Correlations between two moments in time can be too strong for any classical explanation -- and, remarkably, this can happen for a single quantum system measured twice, with no second particle involved. We show that when one qudit is sent through a noisy channel, the strength of this "nonlocality in time" -- the temporal nonlocality robustness $\mathrm{TNR}$ -- is carried entirely by the starting state: it vanishes precisely when the input is maximally mixed (completely random), $\mathrm{TNR}(ρ_A,\mathcal{E})=0\Leftrightarrowρ_A=\mathbb{1}/d$, for the standard noise families. The resource is not any coherence in the channel but the back-action of the input's mixedness, and it survives even complete decoherence. This is at once a power and a trap. As a power, $\mathrm{TNR}$ device-independently lower-bounds the fidelity of temporal teleportation -- sending an unknown state forward in time -- reaching $7/9$ at $d=3$, without trusting the measuring devices. As a trap, because the certified quantity is decoupled from the channel's actual coherence transmission, it can certify more than the channel delivers: an injective (reversible) unitary attains the maximal temporal-Bell signal yet teleports below the classical baseline. We resolve this over-certification completely -- a universal cap $\mathrm{TNR}\le(d-1)/d$ with an exact channel-resolved value, honest certification for the depolarizing channel and for any sufficiently mixed probe, and a proof that no choice of probes makes it channel-universal. Underpinning the results is a unified semidefinite-programming hierarchy of the temporal entanglement, steering and nonlocality robustnesses ($\mathrm{TER}$, $\mathrm{TSR}$, $\mathrm{TNR}$), with a strict lower hierarchy and an upper one conditional on no-signaling in time ($\mathrm{NSIT}$). All structure is verified numerically for $d=2$ through $5$.

Quantum-enhanced metrology relies on entanglement to achieve sensitivities beyond the standard quantum limit. While remarkable progress has been made in generating highly entangled many-body states, extracting their metrological advantage remains a central challenge because the encoded information is often inaccessible to realistic measurements. A key development of the past decade has been the realization that many-body interactions can play a dual role: they can be used not only to generate entanglement, but also to decode it. This idea underlies interaction-based readout and time-reversal protocols, in which controlled non-linear dynamics transform weakly encoded signals into experimentally accessible observables. Cavity quantum electrodynamics (QED) provides a particularly powerful setting for these approaches because it combines collective enhancement, tunable interactions, and controllable reversibility within a single platform. In this review, we discuss the emergence of time-reversal protocols in cavity QED, from their conceptual roots in Loschmidt echoes to modern implementations of signal amplification through a time-reversed interaction (SATIN), scrambling-enhanced metrology, and more general interaction-based readout schemes. We examine the physical mechanisms that enable reversible many-body dynamics, review key experimental demonstrations, and discuss future directions involving complex entangled states, nonlinear decoding, and emerging quantum platforms. Together, these developments suggest that the ability to decode quantum information may become as important as the ability to generate it, establishing reversible many-body dynamics as a central resource for quantum-enhanced sensing.

Three-dimensional superconducting radio-frequency (SRF) cavities provide exceptionally long-lived electromagnetic modes and, when coupled to nonlinear elements such as transmon qubits, become promising architectures for bosonic quantum information processing. The inverse design of such systems, i.e., recovering device geometries that produce specified electromagnetic and coupling targets, is generally a one-to-many problem. The qubit-cavity coupling strength depends sensitively on both the transmon geometry and its position within the cavity's electromagnetic field. As these systems scale up and their design parameter spaces grow, the cost of conventional iterative simulation becomes prohibitive. We present two deep neural network (DNN) approaches that address this inverse-design problem at complementary levels of the design stack. The first proposes SRF cavity geometries that produce target cavity observables. The second proposes transmon qubit designs that produce target qubit-cavity parameters -- the coupling rate, qubit frequency, and anharmonicity $(g, ν_q, α)$. The recovered candidate designs match the targets to within $\sim$5\% (cavity) and $\sim$2\% (transmon), confirmed by end-to-end re-simulation. Both approaches map desired device behavior directly to candidate designs, a fast alternative to the iterative simulation studies usually required.

Braiding statistics, from the Aharonov-Bohm phase to anyons in fractional quantum Hall systems, play a central role in quantum physics. For $p$- and $q$-dimensional excitations in $d$ spatial dimensions, ordinary braiding requires $p+q=d-2$. In a field-theoretic description of $\mathbb Z_N$ excitations, ordinary braiding is described by the linking response $(2πi/N)\int A_{d-p}\cup B_{d-q}$, where $A_{d-p}$ and $B_{d-q}$ are background fields coupled to the two excitation types. In this work, we identify new mutual statistics in the adjacent case $p+q=d-1$. For two invertible excitations obeying $\mathbb Z_N$ fusion, one can choose local creation operators $X$ and $Y$ whose supports have a staggered one-dimensional overlap. The closed unitary process $W_N(X,Y)=(Y^{-1}X^{-1})^N(YX)^N$ measures the resulting mutual statistic. Its field-theory description is $(2πi/N)\int A_{d-p}\cupβ_N B_{d-q}$, where $β_N$ is the Bockstein operation; we therefore call the invariant Bockstein braiding statistics. The construction yields particle-particle statistics in one dimension, particle-loop statistics in two dimensions, and loop-loop or particle-membrane statistics in three dimensions. Nontrivial Bockstein braiding statistics obstructs simultaneous condensation of the two $\mathbb Z_N$ excitations. It also rules out a fully symmetric gapped phase for systems with the corresponding mixed anomaly and implies symmetry fractionalization when one of the $\mathbb Z_N$ symmetries is broken.

Quantum networks require qubits that combine efficient optical access, coherent control, and long-lived quantum memory, but realizing all three in one scalable platform remains a central bottleneck. Diamond color centers are leading candidates, yet widely studied defects retain tradeoffs among these capabilities. Here, we show that transition-metal defects in diamond provide a distinct route beyond these platforms by combining spin-orbit protected ground-state coherence, all-optical control, and near-infrared emission. Using a single nickel-vacancy (NiV$^-$), we demonstrate an all-optically controlled diamond spin qubit with coherence exceeding one millisecond at 1.65 K, compatible with compact closed-cycle cryogenics. We implement Raman Rabi oscillations and Ramsey interferometry and use all-optical dynamical decoupling to extend coherence from $T_2^*$ = 371 ns to $T_2^{CPMG-4}$ = 1.27 ms, establishing NiV$^-$ as a deployable diamond spin-photon interface.

Fermionic non-Gaussianity, or fermionic magic, is a key resource underlying the computational complexity of fermionic quantum systems, yet tractable and operationally meaningful ways to quantify it remain limited. We address this challenge by developing a convex resource theory of fermionic non-Gaussianity and introducing two families of computable measures for pure fermionic states, both derived from the Williamson normal form of the covariance matrix. The first family, occupation number entropies, is defined as the Tsallis-$α$ entropy of the occupation numbers. We prove that one member of this family is monotonic under Gaussian protocols, establishing it as a computable convex resource monotone. It consequently lower bounds the number of non-Gaussian gates needed for state preparation. The second family, natural-orbital participation entropies, is given by the Rényi-$α$ entropy of the squared amplitudes of the state in the natural-orbital basis, defined by the eigenvectors of the covariance matrix. These measures quantify state compressibility in this basis and thus upper bound the classical simulation cost in an orthonormal Gaussian basis. We analyze both families for stabilizer and translation-invariant states, where they simplify and reveal additional structure. We further study representative examples, including random SWAP-doped matchgate circuits and the bond-modulated XXZ model, highlighting the role of non-Gaussianity in many-body phenomena. Our work establishes a resource-theoretic framework for computable fermionic non-Gaussianity that unifies notions arising across quantum information, condensed-matter physics, and quantum chemistry, opening new directions for studying the complexity of quantum many-body systems and providing practical tools to assess the classical simulability of fermionic states relevant for quantum advantage.

We prove that any generalized extended code is monomially equivalent to the Hermitian dual of a code which is closely related to a second kind of extended code of $\C^{\perp_{\rm H}}$. Every $[n+1,k+1]_{q^2}$ linear code $\D$ with $d(\D^{\perp_{\rm H}})>1$ is monomially equivalent to the generalized extended code $\C({\bf u},a)$ of an $[n,k]_{q^2}$ linear code $\C$ for a fixed $a\in\F_{q^2}^{*}$ and some ${\bf u}\in\F_{q^2}^{n}$. We then characterize the Hermitian hull and Hermitian dual distance of $\C({\bf u},a)$ in terms of the position of ${\bf u}$ relative to $\C+\C^{\perp_{\rm H}}$ and the interaction between ${\bf u}$ and the minimum weight codewords of $\C^{\perp_{\rm H}}$, respectively. We obtain explicit criteria to independently control the expected Hermitian hull dimension and Hermitian dual distance of $\C({\bf u},a)$. In particular, several conditions for simultaneously increasing the Hermitian hull dimension and the Hermitian dual distance of $\C({\bf u},a)$ are derived. Applying these results to the Hermitian construction for EAQECCs gives us $267$ new EA qubit codes of lengths $n \leq 40$ and $14$ new EA qutrit codes of lengths $n \leq 25$ compared to the best-known codes in Grassl's code tables and the imporvements recorded in very recent works in the literature. Among the new parameter sets, we confirm improvements for $236$ qubit and $8$ qutrit codes.

In this paper we connect the structure theorem for quasiprobability representation of generalised probabilistic theories to bosonic quantum error correction codes, giving both a general phase-space representation for continuous-variable error-correcting codes, and showing as specific examples the phase-space representations obtained through this method for Gottesman-Knill-Preskill codes, cat codes, and binomial codes. This representation allows us to define both generally and for each of these codes the mathematical structure in phase space that errors can take, which we show both abstractly and for the specific example of single photon loss errors.

Quantum convolutional neural networks (QCNNs) have become increasingly popular in quantum machine learning (QML) due to their efficient parameterization and hierarchical representation of quantum information. Anomaly detection is an important machine learning task with applications across a wide range of domains, including scientific data analysis. In this work, we adapt a QCNN architecture into a quantum autoencoder (QAE) framework for reconstruction-based anomaly detection. The models are trained in a semi-supervised manner on normal samples to reconstruct feature-extracted and dimensionally reduced time-series data, with reconstruction error used as an anomaly score. We investigate two quantum convolutional autoencoder architectures that differ in their treatment of latent information: a hierarchical architecture in which information remains distributed across the circuit and a bottleneck-based architecture in which information is explicitly compressed and reconstructed using additional decoder qubits. The size of the quantum latent space is varied to study its influence on reconstruction accuracy and anomaly detection performance. The approaches are benchmarked against both a variational quantum circuit and a comparable classical baseline using a real-world exoplanet anomaly-detection dataset. Results indicate a trade-off between latent-space size and model capacity, while also suggesting that explicit latent-space compression through a quantum bottleneck can improve anomaly detection performance relative to architectures that retain information throughout the circuit.

We investigate the epistemic opacity of computer simulations and machine learning methods in the context of black hole imaging. We argue that there are forms of opacity-including opacity resulting from the use of machine learning-which do not need to affect the reliability of an inference when it is seen as a part of a broader inferential framework. We propose conditions under which that can plausibly be the case, and discuss how opaque methods can be useful in the context of the (next generation) Event Horizon Telescope. However, we also argue that at least one problematic form of opacity is currently present in black hole imaging: GRMHD models of Sagittarius A* are opaque. This form of opacity signals the limitations of current understanding of the models of this source, and constrains the potential uses of ML models in future observations.

In our recent work, arXiv:2604.11474 [hep-th], we have shown that effective detection of gravitons is possible using an array of charged harmonic oscillators in a dynamical electromagnetic field. Using the interaction Hamiltonian of the identical model, we find out that a systematic way of population inversion of the gravitons is possible in ultra-cold atomic systems. We find out that the exponential growth depends strictly on the number of bosons in the system as well as their inherent squeezing of the matter wave packets. A coherent source of gravitons may lead directly to an unavoidable evidence on the existence of gravitons and based on this analysis we propose an experimental proposal for generating true graviton laser.

Binary (and trinary) radio pulsars are natural laboratories in space for understanding gravity in the strong field regime, with many unique and precise tests carried out so far, including the most precise tests of the strong equivalence principle and of the radiative properties of gravity. The Square Kilometre Array (SKA) telescope, with its high sensitivity in the Southern Hemisphere, will vastly improve the timing precision of recycled pulsars, allowing for a deeper search of potential deviations from general relativity (GR) in currently known systems. A Galactic census of pulsars will, in addition, will yield the discovery of dozens of relativistic pulsar systems, including potentially pulsar -- black hole binaries, which can be used to test the cosmic censorship hypothesis and the ``no-hair'' theorem. Aspects of gravitation to be explored include tests of strong equivalence principles, gravitational dipole radiation, extra field components of gravitation, gravitomagnetism, and spacetime symmetries. In this chapter, we describe the kinds of gravity tests possible with binary pulsar and outline the features and abilities that SKA must possess to best contribute to this science.

Mapping methods are often used for the numerical simulation of nonadiabatic systems by propagating classical mapping variable trajectories. A recently popularised mapping method is spin-mapping, whose mapping variables arise from quantum mechanical operators with symmetries described by a Lie-Poisson algebra. Simulating the classical-like dynamics of spin-mapping systems accurately is generally challenging, with many methods unable to preserve the underlying geometric structure of the symplectic form. The Spin-MInt algorithm is a recently proposed algorithm propagating spin-mapping variables, with a direct proof of symplecticity existing only for 2 electronic states. Here, we directly prove the symplecticity of the Spin-MInt algorithm for a general $K$ electronic states. A review of the symplectic nature of coadjoint orbits of the $\mathfrak{su}(K)$ Lie-Poisson algebra provides the framework needed to understand symplecticity of the Spin-MInt algorithm in this general case. The symplecticity of the method on the associated coadjoint orbit is then shown for what we believe to be the first time via an explicit verification of the symplecticity condition $\mathbf{MJ}\mathbf{M}^\textrm{T}=\mathbf{J}$ exploiting the Lie-Poisson structure of the system. To our knowledge, this is the first time the monodromy matrix for the Spin-MInt algorithm has been explicitly stated using canonical coordinates on the coherent state manifold for a general number of states. We hope that this will assist the development of classical-like spin-mapping methods which might utilise elements of the monodromy matrix, and inform future work on similar symplectic algorithms for coupled and uncoupled Lie-Poisson systems.

The investigation of nonclassical correlations in curved spacetimes offers key insights into the intersection of quantum information theory and gravitational physics. This paper studies two nonlocal correlation measures, non local advantage of quantum coherence (NAQC) and Bell nonlocality (BN) in a $d$-dimensional spherically symmetric Einstein-Gauss-Bonnet (EGB) black hole spacetime. We consider two observers (Alice and Rob) initially sharing a maximally entangled Bell state: Alice freely falls into the black hole (inertial Kruskal frame), while Rob accelerates outside the horizon (non-inertial Schwarzschild-like frame). The Unruh-Hawking effect modifies Rob's field modes, requiring Bogoliubov transformations to relate the two frames. We derive the mixed bipartite density matrix for fermionic fields and analytical expressions for NAQC and BN, which depend on Hawking temperature (itself governed by $α$, $d$, and $r_h$). Our results show both correlations degrade monotonically with increasing Hawking temperature, confirm the NAQC-BN hierarchical relationship persists in EGB spacetime, and highlighting the impact of high curvature corrections on quantum resources.

Achieving high-precision quantum computation requires effective suppression of idling errors that occur when qubits remain inactive during waiting periods within a quantum circuit. Conventional mitigation techniques, such as dynamical decoupling, suppress decoherence by periodically refreshing quantum states through the insertion of additional control gates. In this paper, we propose an alternative approach that suppresses idling errors through quantum circuit scheduling without introducing any additional gate operations. By appropriately adjusting the execution timing of quantum gates with scheduling flexibility, we demonstrate through both numerical simulations and hardware experiments that the overall computational accuracy can be significantly influenced and, in many cases, improved. In addition, we analytically derive the density-matrix evolution under idling noise and provide a theoretical framework that explains the observed behavior.

In this work, we investigate boson star models within the framework of teleparallel gravity with non-minimal coupling, and obtain static, spherically symmetric solutions for both the ground state and excited states. The results indicate that the energy density of the excited-state solutions can become negative. For these solutions, the four commonly used energy conditions are no longer satisfied. In contrast, for all the ground-state solutions we have studied, the energy density remains positive and all four energy conditions are consistently satisfied. Moreover, considering the importance of astrophysical observations, the gravitational-wave signals from Extreme-Mass-Ratio Inspirals (EMRIs) composed of these boson stars are investigated. Our results reveal that the frequency-domain characteristic strain of these waveforms falls within the detectability range of LISA, which can provide potential evidence for distinguishing compact astrophysical objects.

Mid-infrared quantum light sources hold broad application prospects in fields such as gas sensing and infrared thermal imaging. However, currently used mid-infrared quantum entangled light sources primarily rely on bulk periodically poled lithium niobate (PPLN) crystals, which limits brightness and integration. This paper proposes a theoretical scheme based on lithium niobate thin films, in which 1556.9 nm pumping is used to generate entangled photon pairs with a central wavelength of 3113.8 nm. By optimizing the waveguide structure and periodic polarization design, type-II phase matching and group velocity matching are achieved. This enables transverse electric (TE)-polarized pump input to be down converted to generate photon pairs with TE and transverse magnetic (TM) polarizations. Furthermore, by combining a domain arrangement algorithm used for the customized design of polarization direction in PPLN waveguides, precise phase matching is achieved, resulting in a quantum light source with a purity as high as 0.999 and a brightness of 6.18$\times 10^6$ cps/mW, which is three orders of magnitude higher than that of the bulk PPLN crystal source. This study provides a promising solution for realizing high-brightness, high-purity on-chip quantum light sources in the mid-infrared band.

It is necessary to investigate the local distinguishability of orthogonal quantum state sets, as their adoption in protocol design helps diminish quantum state transmission and cut operational costs. In this paper, we explore the local distinguishability of six orthogonal product states (OPSs) on any bipartite quantum system. We classify different sets of six bipartite OPSs into eight categories by using the vectors of the numbers of pairwise orthogonality relations, where any two states are orthogonal on only one subsystem within each set. We find that these eight categories contain a total of 78 distinct cases, all but five of which are perfectly distinguishable via local operations and classical communication (LOCC). Furthermore, we discuss the local distinguishability of those five distinct cases in detail. Our work explicitly characterizes the local distinguishability of six bipartite OPSs.

The aging transition is a critical phenomenon in which collective dynamics deteriorate as the fraction of inactive quantum nodes exceeds a threshold, referred to as the aging transition point. Such transitions are relevant to a broad range of biological and physiological systems, and may play an important role in quantum information processing, particularly in the stability assessment and robustness control of quantum networks. Detecting the aging transition point is therefore crucial for predicting network breakdown, since it marks the critical threshold at which a quantum network abruptly loses its stable active state and enters a degraded inactive phase. Here we propose a quantum sensing strategy to locate this transition point using a single qubit probe coherently coupled to a small subset of oscillator nodes. As the inactive fraction p approaches the aging transition point, the excited-state population of the probe becomes highly sensitive to variations in p, leading to a pronounced enhancement of the Fisher information. This critical enhancement enables high-precision estimation of the transition point. Remarkably, this enhancement survives even in the classical regime for the oscillators, where the Fisher information increases dramatically as p approaches the transition region. Our results establish a feasible route to sensing aging transitions in oscillator networks and provide a metrological perspective on critical phenomena in quantum many-body systems.

Rydberg-atomic receivers exhibit exceptional sensitivity yet are fundamentally constrained by the narrow instantaneous bandwidth, limiting their practical deployment in broadband scenarios. Prior approaches typically expand the bandwidth by physically broadening the atomic response, which usually requires auxiliary electromagnetic fields or stringent parameter tuning, thereby increasing overall system complexity. Here, we propose a compressive spectral multiplexing framework implemented in a waveguide-coupled Rydberg atomic receiver using a frequency-modulated local oscillator (FMLO). The FMLO creates multiple parallel sensing channels that collectively constitute a physical compressive sensing matrix, generating multiple narrowband intermediate-frequency replicas of the input signal. Thus, a broadband microwave spectrum is projected onto a set of narrowband atomic responses. It is demonstrated that spectral information spanning a bandwidth of over 640 MHz can be effectively compressed into the intrinsic atomic bandwidth of 126 kHz, achieving a spectrum compression ratio exceeding 1000. Furthermore, these output replicas offer intrinsic measurement redundancy and facilitate signal-to-noise ratio enhancement. An approximate 10 dB gain is achieved in the required bit-energy-to-noise-power-density ratio for multi-channel communication via maximal-ratio combining. This approach requires no auxiliary fields or broadband electronics, providing a simple and scalable pathway for chip-scale quantum receivers, latency-critical sensing, and next-generation wireless communications.

In the NISQ era, there is a need for resource-efficient proof-of-principle experiments that can be built up to genuine utility. Single-qubit classifiers (SQCs) are small-scale hybrid quantum-classical machines capable of performing a basic machine learning task: classifying data. In principle, these can be scaled up to many-qubit quantum classifiers capable of quantum computational advantage. Another type of quantum advantage is enabled by blind quantum computation (BQC), wherein a client may run delegated quantum computations on an untrusted server with information-theoretic security. In this paper, we develop a framework and propose a prototype experiment for a SQC where it is known to the server that a classification is being performed, but the data and outcome stay hidden, i.e., it performs partially-blind SQC (PB-SQC). This can be integrated into a quantum network to deliver quantum-secured classifications to remote clients; we study this for a heterogeneous quantum network link in which entanglement is shared between a server and a client equipped with a multiplexed solid-state quantum memory using entanglement swapping. The framework we develop for PB-SQC on this setup is tested in a simulation with realistic hardware parameters on a real-world credit card transaction fraud database with classification outcomes approaching those of its equivalent classical deep-belief network. In addition, we show how a two-qubit classifier (TQC) instead of a SQC enables verification of the computation. These results pave the way towards a short- to mid-term quantum network offering use-case-ready quantum applications.

The security of practical quantum key distribution (QKD) systems is fundamentally constrained by vulnerabilities of single-photon detectors. Measurement-device-independent quantum key distribution (MDI-QKD) was proposed to remove this limitation by allowing all measurements to be performed by a completely untrusted party, under the assumption that the measurement node can be treated as adversarial but does not compromise the security guarantees of the protocol. Here we show that this assumption is insufficient under realistic adversarial control of the measurement device. We present an attack in which an adversary exploits active control of the measurement node (Charlie) to obtain significant information about the secret key. The attack enables recovery of up to 70\% of the sifted key while introducing only 5.6\% quantum bit error rate. Unlike previously reported attacks targeting specific implementations of MDI-QKD, our results demonstrate a limitation of the standard security model underlying the protocol. These findings indicate that additional constraints on the measurement-device independence assumption, or refined security analyses incorporating stronger adversarial capabilities, are required to ensure the security of MDI-QKD in realistic scenarios.

We investigate static and spherically symmetric neutron star solutions endowed with primary scalar hair in a subfamily of Degenerate-Higher-Order-Scalar-Tensor (DHOST) theories of gravity. By solving the modified Tolman-Oppenheimer-Volkoff (TOV) equations, we construct equilibrium configurations for polytropic and realistic equations of state and analyse the impact of the scalar hair on the stellar structure. We examine the resulting metric and scalar field profiles as well as the mass-radius relation, showing deviations from the predictions of General Relativity (GR). Positive scalar charges lead to more compact stars than in GR and, above a critical threshold, to singularities. Observations could therefore put stringent constraints on the parameters characterising the beyond-GR effects in these theories and their potential scalar hair.

Quantum magic is a fundamental resource that quantifies to what extent quantum states can be efficiently simulated on a classical computer. We study it for states constructed from the Sachdev-Ye-Kitaev (SYK) Hamiltonian with $N$ Majoranas by the fermionic anti-flatness (FAF). We show analytically that, in the large $N$ limit, the quantum magic of pure Kourkoulou-Maldacena (KM) states, dual to a quantum black hole with an end-of-world particle behind the horizon, is linear in $N$ with a slope, depending on the black hole temperature, that can be tuned between zero and $1/2$. By contrast, the FAF of Gaussian states evolved in real time with the SYK Hamitonian approaches $\approx N/2$ exponentially at a rate given by a multiple of the leading Ruelle-Pollicot resonance. Subleading corrections in $N$ for SYK energy eigenstates, computed numerically for $N \leq 54$ by combining Krylov subspace with GPU acceleration techniques, decay exponentially with $N$, but power-law if the SYK couplings are sparsified, and are order of magnitude larger for states close to the ground state, a region with an established gravity analogue. Our results offer new insights about the relation between quantum information, quantum chaos and low-dimension quantum gravity.

We investigate Lyapunov exponents as dynamical probes of black hole phase transitions in dilatonic Reissner--Nordström--AdS black holes within Einstein--Maxwell--dilaton theory. The thermodynamic quantities and the Lyapunov exponent of charged probe particles were analyzed in both the Einstein and string frames, thus providing a direct comparison between the thermodynamic phase structure of the black hole and that captured by the Lyapunov exponent. Thermodynamic quantities, including the Hawking temperature and Wald entropy, remained constant under conformal frame transformations, yielding identical phase structures in the two frames. In contrast, the Lyapunov exponent exhibited non-trivial frame dependence for massive probe particles due to dilaton coupling, while no frame dependence was found in the massless limit. Numerical analysis revealed that the phase structure features captured by the Lyapunov exponent, including characteristic cusp behavior and transition points, were independent of the choice of frame, despite the Lyapunov exponent itself being frame-dependent. Therefore, the Lyapunov exponent exhibited frame-dependent values, while the critical structure it captures remained constant across conformal frames.

In this work, we investigate the generation of squeezed Schrödinger cat states in schemes based on photon-number-resolving measurements on multimode Gaussian states. We derive analytical expressions for the states generated in two- and three-mode schemes, as well as formulas for their fidelity with squeezed Schrödinger cat states. We analyze how the amplitude of the generated states scales with the number of detected particles. Furthermore, we derive an upper bound on the achievable generation fidelity and identify the conditions under which multimode schemes can enhance the quality of the generated states.

We present an exact solution to the Einstein-Maxwell equations that describes a static black hole coexisting with anisotropic matter immersed in an external magnetic field, obtained via the Harrison transformation. Our findings reveal that an increase in the anisotropic matter parameter systematically suppresses the local chaotic behavior, as indicated by a reduction in the Lyapunov exponent. Conversely, variations in the external magnetic field lead to qualitative changes in global chaotic behavior. This is analyzed through Poincaré sections, which demonstrate transitions between regular and chaotic trajectories resulting from the nonlinear gravitational-magnetic interactions. These factors play distinct yet complementary roles in shaping chaotic particle dynamics around black holes. This study would offer a new theoretical framework for exploring non-integrable particle motion within magnetized black hole spacetimes and for probing a black hole at the galactic center, where magnetic fields may arise from plasma effects surrounding astrophysical black holes.

Long-lived logical qubits are essential for fault-tolerant quantum computation. However, the practical performance of traditional error correction protocols relies on performing specific syndrome circuits, causing vulnerability to hardware defects and imposing rigid connectivity constraints. Recent theoretical findings have proposed that flexible subroutine circuits within the LUCI framework can maintain space-time distance in the presence of isolated or broken components, albeit at the expense of temporal distance. However, these approaches have solely targeted defect avoidance and have not yet been demonstrated to suppress errors with reduced temporal distances on physical hardware. In this work, we propose a reset-free scenario for the LUCI framework and experimentally benchmark it on IBM quantum hardware. By asymmetrically scaling the $X$ or $Z$ distance, we compare our reset-free approach against the standard surface code and successfully demonstrate error suppression ratios for targeted logical Pauli errors. Remarkably, despite a nearly halved syndrome density in time, which requires two subroutine rounds for full syndrome extraction, the LUCI framework remains competitive with the rotated surface code implementation. In the LUCI framework, we observe error suppression of $1.75(10)$ for logical $X$ errors and $1.93(12)$ for logical $Z$ errors, whereas the standard approach yields $ 1.58(13)$ and $2.44(7)$, respectively. These results demonstrate that dynamic codes outperform standard methods by avoiding highly noisy components, even without physical defects, while preserving logical boundaries. Our findings challenge the conventional dependency on static fault-tolerant architectures by verifying the feasibility and efficacy of the LUCI framework on physical hardware and pave the way for hybrid, hardware-compatible code designs in quantum computing.

We constructed rotating black holes with equal angular momentum in five dimensional Einstein-Bumblebee gravity with and without cosmological constant. Their thermodynamic properties are examined via two distinct methods: the Wald formalism and the Komar integral. Notably, the conserved charges, including mass, angular momentum, and entropy, computed from these two approaches differ by a constant prefactor that is solely determined by the Bumblebee coupling. Subsequently, we apply the Kerr/CFT correspondence to derive the microscopic entropy of these black holes and find that it precisely reproduces the entropy in Komar-integral version, rather than the Wald entropy.

Typicality arguments for emergent spacetime rely on the Lubkin-Page bounds, which show that generic quantum states have vanishing correlations between subsystems. These bounds assume a tensor-product Hilbert space (a Type~I von~Neumann algebra), but the observable algebras in quantum field theory and quantum gravity are generically Type~II or Type~III, raising the question of whether the bounds survive. We prove that they do for all Type~II von~Neumann factors. For the hyperfinite Type~II$_1$ factor with a tripartite decomposition $R \cong A \otimes B \otimes E$, the mutual information between subsystems $A$ and $B$ vanishes as $O((d_A d_B / d_E)^2)$ in finite-dimensional approximations, provided $d_A d_B \leq d_E$ (Theorem~1). For Type~II$_\infty$ factors, including the gravitational algebras constructed via the crossed-product method by Witten and by Chandrasekaran, Longo, Penington, and Witten, the bound acquires an additional exponential suppression controlled by the Bekenstein-Hawking entropy (Theorem~2). We identify the obstructions to extending the result to Type~III factors and discuss the open question of whether the commutant of the observable algebra can serve as a natural thermal bath that tightens the bound further.

Block encodings are a central primitive in quantum algorithms, but standard constructions typically require logarithmic ancilla overhead and complicated controlled operations. Recent lower bounds further show that such ancilla overhead is unavoidable for exact constructions in broad circuit models. We show that this barrier can be bypassed in the approximate setting. Specifically, we present a simple single-ancilla construction that converts Hamiltonian evolution into a block encoding of the underlying Hamiltonian, via generalized quantum signal processing. For operators given by Hermitian decompositions $A=\sum_{j=1}^L α_j H_j$, we instantiate this block-encoding construction in two ways, which differ in how the required Hamiltonian evolution is implemented. Using higher-order Trotterization, we obtain an $\varepsilon$-approximate block encoding of $A$ with only one ancilla qubit and circuit depth $\widetilde O\big(L(α/\varepsilon)^{o(1)}\big),$ where $α=\sum_j α_j$. Using multiproduct formulas, we obtain circuit depth $\widetilde O(L)$, at the cost of $O(\log\log(1/\varepsilon))$ ancilla qubits. Our constructions provide alternatives to the standard LCU framework, with a focus on reducing the number of ancilla qubits while maintaining (near-)optimal circuit depth.

We develop a semi-analytic framework for stationary, axisymmetric GRMHD jets that efficiently generates resolved polarized images from the near-horizon region out to $\sim 10^5\,r_g$ across a broad parameter space, enabling rapid exploration of how gravity and magnetohydrodynamic flows imprint scale-dependent signatures on jet morphology and polarization. We identify a new scale-dependent separation in polarimetric diagnostics. Outside the photon ring, plasma loading strongly modifies the polarization-angle profile of the integrated jet-layer emission through inertia-driven winding of the magnetic field. At large image-plane radii, the polarization angle follows a power-law in radius, with an index determined by the jet collimation profile. Near the horizon, in contrast, jets converge to a universal polarization pattern controlled solely by black hole spin. This convergence is hierarchical: differences in velocity and magnetic-field structure are erased first, whereas collimation-dependent differences persist to smaller radii, thereby allowing these effects to be disentangled. These results establish a largely achromatic polarimetric diagnostic that connects GRMHD jet dynamics to resolved image structure, with direct implications for high-resolution polarimetry and for constraining black hole spin and jet formation.

Formation of primordial black holes (PBHs) requires a significant enhancement of curvature perturbations. This mechanism leaves a twofold gravitational-wave (GW) signature: a \emph{low-frequency} stochastic background of scalar-induced GWs (SIGWs) and a distinct \emph{high-frequency} signal from subsequent PBH binary mergers. We leverage this shared origin to establish a consistent, \emph{model-independent} connection between these two observables for a monochromatic PBH mass function. Using PBH abundance constraints on the primordial curvature power spectrum, we evaluate the stochastic SIGW background for spherical and ellipsoidal collapse models, demonstrating that the ellipsoidal scenario yields a significantly stronger signal. Furthermore, we analyze the GW signal from PBH binary mergers and find a direct correspondence between the SIGW frequency and the innermost stable circular orbit (ISCO) frequency of the binaries. Because GW emission is nearly maximal near the ISCO, we additionally show that the peak of the full merger GW spectrum relates to the ISCO frequency via $f_{\text{peak}} = 1.79 \, f_{\text{ISCO}}$, a relation that is independent of the binary masses. Remarkably, this unified framework connects these distinct GW channels, enabling the same primordial fluctuations to be probed across widely separated frequency bands.

We establish an upper bound on extractable work for time-dependent, non-Markovian quantum heat engines operating with finite baths. This bound analytically isolates the distinct thermodynamic penalties arising from system-bath correlations, bath non-equilibrium, and residual interaction energy. Evaluating this framework operationally via a quantum polytropic cavity-optomechanical cycle, we demonstrate that maximal efficiency requires quasi-static operation to successfully harvest coherent, non-Markovian system-bath resonances. Conversely, optimising for maximum power enforces a strict finite-time regime. Under realistic hardware constraints, this acceleration necessitates larger discrete operational steps, where we expect Trotterisation errors to manifest as physical noise. Such noise would irreversibly suppress delicate quantum memory effects, forcing a collapse to the memoryless Markovian Otto limit. Coupled with the permanent energetic tax of switching finite-bath interactions, our results indicate that the exploitation of quantum memory resources and finite-power operation belong to different operational regimes.

We study the compilation of structured quantum gate families on two-dimensional neutral-atom arrays, aiming to reduce addressing and transport overhead under realistic hardware constraints. For single-qubit gates, we exploit the algebraic structures of gate families at the matrix level, enabling efficient rank-one decompositions over appropriate algebraic structures and thereby reducing the number of addressing layers. For controlled-Z (C-Z) gates, we formulate the transport scheduling problem using graph-theoretic models, leading to efficient compilation algorithms under realistic transport constraints. We provide provable performance guarantees for the proposed methods and validate them through extensive numerical experiments. Across representative single-qubit gate families, our methods reduce the number of addressing layers by up to a factor of two compared with naïve row- or column-wise implementations. For C-Z gates, our scheduling strategy reduces the required number of atom transport operations by approximately 50\%. When applied to QAOA circuits for MaxCut, the proposed framework reduces transport cost by more than 30\% on average. These results show that the physical constraints of neutral-atom hardware can be converted into algebraic and graph-theoretic structure, turning a hardware-level scheduling bottleneck into tractable decomposition and coloring problems.

With the advancements in clock timing technology, increasingly smaller time differences can be distinguished. Therefore, it is critical to investigate the fractional frequency shift of clocks at different locations on Earth. In this paper, we study it systematically under the influence of a subtle lunar tidal potential based on a new method. Our calculations in the geocentric Fermi frame show that when two clocks are located at the same latitude, the longitude difference changes the fractional frequency shift between them. A similar phenomenon occurs when there is a difference in latitude between two clocks on the ground at the same longitude. Interestingly, when the Moon's longitude changes, the phase and amplitude of the lunar tidal fractional frequency shift between two clocks with the same longitude difference will change, while the change in the Moon's latitude only affects the amplitude of the fractional frequency shift of these two clocks. Our results provide useful information for the calibration and synchronization of clocks on Earth.

We study the cosmological dynamics of interacting dark energy and dark matter in Classical Einstein Gravity and Loop Quantum Cosmology. Two dark matter scenarios are considered: superfluid dark matter described by a generalized cubic equation of state and the standard pressureless fluid. The dark energy component is modeled using both a generalized nonlinear equation of state and a constant equation of state. We examine two phenomenological interaction terms, $Q=α\dotρ_m$ and $Q=β\dotρ_d$, which govern the energy transfer between the dark sectors. In classical gravity, the pressureless matter model exhibits stable late-time attractors, whereas the superfluid dark matter model admits only saddle and non-hyperbolic critical points. Extending the analysis to Loop Quantum Cosmology, quantum geometric corrections replace the Big Bang singularity with a nonsingular quantum bounce, and significantly modify the phase-space dynamics. As a result, the stable attractors of the classical pressureless matter model disappear, and all interacting models possess only saddle and non-hyperbolic critical points. These findings highlight the significant influence of both dark matter properties and quantum gravitational effects on the asymptotic evolution of interacting dark-sectors.

We investigate the dynamics of charged test particles and quasi-periodic oscillations around a Reissner--Nordström-like black hole in Kalb--Ramond (KR) gravity in the presence of an external magnetic test field. The KR background introduces a Lorentz-violating parameter $\ell$, which modifies the spacetime geometry, horizon structure, circular orbits, and characteristic frequencies of particle motion. In contrast to the standard Wald-type prescription, the magnetic-field configuration is constructed from the source-free Maxwell equation on the charged KR background, allowing the magnetic profile to be consistently adapted to the modified geometry. We derive the equations of motion, the effective potential, the conditions for circular orbits, and the orbital and radial epicyclic frequencies of charged particles. The results show that the black-hole charge $Q/M$, the KR parameter $\ell$, the specific particle charge $ε$, and the magnetic coupling $β=bM$ jointly affect the innermost stable circular orbit (ISCO) and the quasi-periodic oscillation (QPO) frequencies. We then apply the obtained frequencies to the relativistic precession model, where the upper QPO frequency is identified with the orbital frequency and the lower one with the periastron-precession frequency. Using the observed twin-peak QPO data of GRO J1655--40, XTE J1550--564, and M82 X-1, we perform a Markov chain Monte Carlo analysis to constrain the model parameters. The obtained posterior constraints indicate that the charged KR black-hole model with an external magnetic field can consistently reproduce the observed QPO pairs within the adopted parameter ranges. These findings suggest that QPO observations may serve as a useful phenomenological tool for probing Lorentz-violating black-hole geometries and electromagnetic effects in strong-gravity environments.

Quantum relative entropy is a core concept in physics, governing the limits of communication, thermodynamic irreversibility and quantum resource conversion. However, the requirement that physical processes cannot increase state distinguishability, the data-processing inequality, permits an infinite family of alternative divergence measures. Here we show that quantum relative entropy is uniquely selected by a sharper operational principle. We evaluate distinguishability through binary guessing games, in which an observer discriminates between pairs of quantum states using the optimal measurement. We prove that any additive measure that respects the odds revealed by these optimal measurements must coincide with the Umegaki relative entropy. This rigidity is a purely quantum phenomenon. Whereas classical theory permits a continuous family of valid divergence measures, including Rényi divergences, quantum noncommutativity. collapses this mathematical freedom. The result is exact, requiring neither a thermodynamic limit of infinitely many copies nor super-additivity assumptions for correlated states. It establishes quantum relative entropy not merely as an asymptotic quantity, but as the unique additive distinguishability measure compatible with single-shot quantum discrimination.

Asymptotic zero Arnowitt-Deser-Misner (ADM) mass wormholes, such as the zero-mass traversable Ellis-Bronnikov wormhole, are of great interest for astrophysical applications such as in the galactic microlensing. However, when considered individually, they are unstable to small perturbations. On the other hand, there is a possibility that they can be stable as an interior partner of a traversable glued wormhole obtained by suitably gluing the interior to the observationally relevant massive exterior spacetimes across spherically symmetric thin shells. Although the exterior spacetime has non-zero ADM mass, massless interior partner remains massless sharing the stability of the glued wormhole. The dynamics of the thin-shell then demarcates the stability regions of the glued wormhole that we wish to study here by employing the novel concepts of thin-shell "mass" and of "external force" constraints discovered by Garcia, Lobo and Visser. We shall consider two classes, where the zero ADM mass interior are Kim-Lee wormholes glued to the exterior Schwarzschild vacuum and Reissner-Nordström spacetime respectively. It turns out that the stability regions in both cases are almost similar although the two interior Kim-Lee spacetimes are physically very different, one scalar charged and the other electrically charged. The conditions under which the stability of glued wormholes could be achieved are analyzed in detail.

Bayesian quantum estimation offers a finite-data framework for quantum sensing and metrology, yet a unified geometric formulation for multiparameter Bayes risk has been lacking. We introduce Bayesian monotone metrics by evaluating Petz monotone metrics on the prior-averaged state, providing a Bayesian extension of the full class of statistically meaningful (CPTP) quantum metrics. This framework yields Bayesian quantities, including quantum posterior-mean operators and a quantum Bayesian dual Fisher-information matrix, and it leads to a systematic family of computable lower bounds on the Bayes risk. The resulting bounds naturally incorporate multiparameter measurement incompatibility and, for every monotone metric in the family, we prove a universal dominance over the corresponding quantum van Trees (Bayesian Cramér--Rao) bound. Moreover, we show that optimizing over all operator monotone functions collapses to a one-parameter subfamily, turning the tightest bound into a tractable optimization with a clear geometric interpretation. In representative examples, the optimized bounds are strictly tighter than the Bayesian SLD and RLD bounds. Our results establish Bayesian monotone metrics as a unifying information-geometric perspective on Bayesian quantum estimation, enabling systematic and computable performance limits in multiparameter settings.

The efficient preparation of correlated thermal states, such as the Thermofield Double (TFD) state, is a fundamental prerequisite for simulating quantum gravity models and many-body thermodynamics on quantum processors. In this work, we investigate the ground state preparation of the Two Coupled Sachdev-Ye-Kitaev model, known as the Maldacena-Qi model, which is dual to a traversable wormhole in $AdS_2$, utilizing feedback-based quantum algorithms. We demonstrate that the standard feedback-based quantum algorithm (FALQON) and its time-rescaled variant (TR-FALQON) face severe kinetic limitations in this system, failing to converge to the highly entangled ground state when initialized in trivial product states. To overcome these barriers, we propose the hybrid ITE-TR-FALQON protocol, which integrates the imaginary-time evolution present in imaginary-time-enhanced FALQON (ITE-FALQON) with the time-rescaling mechanism. Our numerical results indicate that the introduction of non-unitary dynamics is strictly necessary to break symmetry traps and filter out excited states, while time-rescaling drastically accelerates algorithm convergence. The proposed method achieves fidelities close to unity and reproduces the von Neumann and Rényi entropy spectra of the exact TFD state with high precision.

In this paper, we suggest a chiral interaction scheme to enhance magnon bundle emission by placing a qubit and a magnon into a cascaded-cavity setup, respectively. It is found that the unidirectional interaction prolongs the lifetime of the target excited state, thereby suppressing the magnon re-excitation and promoting both the average purity and number of two-magnon bundles. Consequently, the chiral interaction not only offers directional control but also improves the quality of the multi-magnon source, which may find potential applications in quantum information processing.

The quantum Kramers-Wannier (KW) duality is a fundamental transformation mapping short-range entangled (SRE) states to long-range entangled (LRE) states. While spatially local unitary circuits require linear-in-system-size depth to implement this duality, the ultimate speed limit for purely unitary circuits equipped with nonlocal connectivity remains an open question. Here, we explicitly construct logarithmic depth, spatially nonlocal unitary circuits that realize the exact $\mathbb{Z}_2$ KW dualities in both one and two spatial dimensions. We further generalize the construction to arbitrary $\mathbb{Z}_n$ KW dualities. Unlike algorithms tailored to prepare specific target states, our circuits implement complete duality maps. Within the symmetric (charge-neutral) sector, these dualities exactly transform arbitrary non-fixed-point SRE states into their corresponding LRE duals. Consequently, our results establish an efficient, purely coherent pathway for exploring phase transitions and topological dualities on modern quantum platforms.

We show that exchange symmetry in collisions of identical particles enables symmetry-protected coherent control of the total scattering cross section. For identical fermions, antisymmetrization enforces complete phase synchronization of the contributing scattering channels, yielding maximal control visibility. For identical bosons, synchronization persists but with reduced visibility due to additional exchange (satellite) contributions. Collisions of distinguishable particles lack this symmetry-imposed phase locking, leading to lower controllability and visibility. We elucidate these principles through coupled-channel quantum-scattering calculations for lithium-lithium collisions, comparing the $^{6}\mathrm{Li}$-$^{6}\mathrm{Li}$ (identical fermions), $^{7}\mathrm{Li}$-$^{7}\mathrm{Li}$ (identical bosons), and $^{6}\mathrm{Li}$-$^{7}\mathrm{Li}$ (distinguishable) systems. Furthermore, in the identical particle cases, symmetry-enforced synchronization enables full control over the parity of the final state at any collisional energy. This mechanism is broadly applicable to identical-particle collisions, including homonuclear molecules for which established approaches -- DC electric fields, or microwave shielding -- are ineffective or unavailable.

All-optical (AO) microwave-free magnetometry using nitrogen-vacancy (NV) centers in diamond simplifies experimental design and broadens sample compatibility. While continuous-wave (cw) detection of AO photoluminescence (PL) changes is commonly employed, its performance is susceptible to systematic fluctuations such as optical intensity noise. To address these challenges, we introduce a pulsed AO protocol that employs two PL measurements within an optical pulse to suppress common-mode noise. At near-zero magnetic field, we experimentally demonstrate that the pulsed AO protocol resolves AO-PL contrast features arising from NV-NV cross-relaxation, achieving up to 10$\times$ improvement in the low-frequency noise floor compared to conventional cw AO techniques. We further investigate the dependence of AO-PL contrast on PL readout timing and the dark time duration $τ$ between optical pulses, with the optimal $τ$ varying based on NV concentrations. These findings provide insights into optimizing NV-diamond samples for effective AO operation across diverse applications.

Since the complexity of quantum state tomography (QST) scales exponentially with system size, exploiting priors such as low-rankness, tensor-network structures, and neural-network representations is essential for scalable QST in terms of sample complexity and parameter complexity. In this paper, we introduce a unified framework, termed structured factorization, that builds on BurerMonteiro-type factorization by parametrizing the density matrix as $FF^\dagger$, where the factor $F$ is constrained to belong to a structured model class. This factorization guarantees physical validity by construction while allowing a broad range of structural priors to be incorporated directly through the choice of the factor space, ranging from the generic Cholesky decomposition to low-rank matrices, matrix product operators, and neural density operators based on multilayer perceptron and transformer architectures. Building on this structured factorization framework, we formulate QST as an optimization problem over the factor space from measurement data. We first develop a unified statistical analysis of the sample complexity of least-squares estimation for a broad class of structured quantum states. We then propose a projected gradient descent method that operates directly on the factor space and accommodates a wide range of structural parametrizations and reconstruction objectives. To further exploit the geometry of the maximum-likelihood estimation formulation and the constraints on the factors, we derive a power method that yields a step-size-free algorithm with fast convergence, recovering Covers method as a special case when the factor is unconstrained.

The nonseparability of post-Newtonian (PN) Hamiltonian systems typically necessitates the use of computationally expensive implicit integrators. Recent research overcomes this limitation by embedding the dynamics into a doubled phase space, which enables the development of explicit symplectic methods. However, existing specially designed explicit integrators suffer from order reduction for high-order methods when the time stepsize is small, i.e., $h <\varepsilon^3$. In this paper, we propose a novel extension and splitting approach for the doubled Hamiltonian, under which specially designed explicit symplectic integrators can be constructed. It is shown that the proposed integrators achieve genuine high-order convergence without order reduction and take advantage of the small PN parameter $\varepsilon$. Numerical results from simulations with 2PN spinning binaries demonstrate superior long-term conservation of invariants and significantly higher computational efficiency compared to both implicit methods and existing explicit splitting techniques.

Quantitative validation of quantum simulations of dynamical spin response remains challenging because experiment, classical simulation, and quantum simulation do not produce the same native observables. This problem has become increasingly important as quantum simulation protocols for dynamical response have progressed from theory to hardware-level benchmarking against neutron-scattering data, while the longer term goal is validation in regimes that may eventually become classically intractable, including in future fault-tolerant implementations. Here, we develop a cross-pipeline validation framework for quantum simulation, using inelastic neutron scattering and classical many-body simulation as complementary experimental and computational anchors, based on explicit forward and inverse observable maps, covariance- or resampling-based uncertainty propagation, robustness tests for structured distortion, and a hierarchy of complementary metric families. The framework distinguishes stochastic uncertainty from robustness-induced distortion, carries both explicitly through the comparison chain, and uses the resulting metric-level uncertainty and distortion information to support layered validation at the pipeline, solver, and model levels. We also introduce actuator-aware feedback logic aimed at improving agreement without obscuring the physical origin of any remaining mismatch. We close by outlining future extensions of this methodology, including upstream uncertainty and distortion modeling, adaptive feedback, asymmetric validation beyond full classical benchmarking, fault-tolerant workflows, and community infrastructure for reproducible validation.

Optimal encoding of classical data for quantum-assisted statistical inference is investigated from an information-theoretic perspective. We prove that the accuracy of any quantum-computing inference procedure is upper bounded by the maximal quantum leakage from the classical data through its quantum encoding, establishing leakage as a universal, task-agnostic quality measure for encoders. This demonstrates that the maximal quantum leakage is a universal measure of the quality of the encoding strategy for statistical inference as it only depends on the quantum encoding of the data and not the inference task itself. The optimal universal encoding strategy, i.e., an encoding strategy that maximizes the maximal quantum leakage, is proved to be attained by pure states. When there are enough qubits, basis encoding is proved to be universally optimal. However, when the dimension of the system is small, phase encoding is optimal. For the latter, any tight frame, any ensemble whose average state is the maximally mixed state, is in fact optimal. Within tight frames, equiangular tight frames (ETFs) are distinguished as the uniquely symmetric optimal encodings, i.e., they saturate the Welch lower bound on pairwise overlaps and possess a self-referential optimal measurement. Prominent special cases are the qubit trine, the regular simplex, and symmetric informationally complete positive operator-valued measures (SIC-POVMs), for which the ETF structure and explicit codeword constructions are provided. Numerical examples are presented to validate the theoretical predictions.

Quantum algorithms for quantum chemistry and other many-body fermionic systems work by expressing the Hamiltonian in a basis of qubits and fragmenting the Hamiltonian into a sum of products of Pauli operators whose exponentials are easily encoded on a quantum device. Applying the product of exponentials, known as Trotterization, leads to an error associated with the non-commutativity of operators. This error can lead to breaking the symmetries of the Hamiltonian because the fragments are not symmetry conserving in general. Nonetheless, many algorithms for time evolution rely on Trotterization, including time evolution and quantum phase estimation. We show that we can express the Hamiltonian in terms of Hermitian excitation operators which map to sums of commuting Pauli strings for any encoding and conserve symmetries corresponding to Abelian groups of symmetry operators. Symmetries corresponding to non-Abelian groups, on the other hand, are not fully conserved by Trotterized Hermitian excitation operators, so we developed ``operator kirigami'' to cut the sum of non-commuting operators by orthogonal projection and to fold terms together using unitary rotations. We tested pools of operators for small molecules and basis sets, and found that electron number and spin symmetry conserving pools led to greater errors that decreased for larger molecules and were negated with second-order Trotterization. Our work shows the potential for testing quantum computing algorithms on classical computers by adapting tools used in electronic structure theory with conserved symmetries.

The magneto-rotational instability can generate strong, turbulent substructure within magnetised shear flows. The efficacy of the mechanism as a function of microphysical aspects of the fluid, such as stratification and diffusivity, has been explored extensively. One aspect that has not been studied thus far, however, is whether the instability can also operate in solids. Motivated by the possibility that solid regions within planets or degenerate stars may rotate differentially with respect to liquid or gaseous layers during some phase of their life, we examine the extent to which elasticity suppresses the instability. A simplified, plane-parallel analysis reveals that only in cases where the flow is strongly sheared, such that the magnetic tension that would result from the instability in a liquid exceeds the shear modulus of the elastic cavity, can magnetic growth occur. In the context of dynamical tides in binary neutron-star mergers, this implies that the magnetic field can be amplified in the crust prior to coalescence only if the star boasts a spin frequency of $\gtrsim 300$Hz. If viscous heating weakens the crystalline structure prior to resonance, the required spin frequency is reduced.

We study how the Lamb shift of a static atom is modified when a nearby planar body rotates rigidly about its normal while the atom is held at a fixed distance $a$. We derive a general formula for the shift in terms of the angularly Doppler-shifted reflection coefficients of the surface, valid for any axially symmetric planar material. Expanding the result to second order in the angular velocity $Ω$, we identify two independent contributions associated with the orbital and spin components of the electromagnetic angular momentum. The orbital contribution, proportional to $(Ωρ)^2$, reproduces locally the Lamb shift induced by a surface translating at the tangential velocity $Ωρ$, whereas the spin contribution, proportional to $(aΩ)^2$, originates from the rotational Doppler shift of the photon helicity and survives even on the rotation axis. We first illustrate the formalism using a graphene sheet and then apply it to finite-thickness Drude and plasma conductors and to doped semiconductors. Rotation enhances the Casimir-Polder interaction for graphene and metallic surfaces, whereas it weakens it for doped semiconductors, depending on whether the carrier plasma frequency reaches the near-field scale $1/a$. Above a threshold angular velocity, the atomic level also acquires a finite linewidth, providing a spectroscopic signature of quantum friction.

Violations of Bell inequalities are a hallmark of entanglement, with only entangled states capable of exceeding classical bounds in standard Bell tests. Here we analyze anomalous weak values of the CHSH operator in pre- and post-selected (PPS) quantum ensembles, treating them as generalized bounds on Bell-type nonlocal correlations in the presence of post-selection. Fixing the overlap between the pre- and post-selected states, we compare three scenarios: unrestricted boundary states, one separable boundary state, and both boundary states separable. For each case, we derive both the maximal weak value for a fixed Bell operator and the maximal bound obtained by further optimizing over all CHSH operators. Our results show that post-selection and entanglement are distinct operational resources: post-selection alone can enhance correlations, but entanglement is necessary to exceed the corresponding separable PPS bounds, and their combination yields the strongest attainable correlations. We further show that the enhancement beyond the separable bound closely tracks the concurrence of the states that optimize the bounds, identifying entanglement as the source of the additional correlation strength. Finally, we show that nonlocal weak values provide post-selected entanglement witnesses, and we give a constructive protocol that detects every pure two-qubit source state with nonzero concurrence in the ideal state-adapted setting, even in regimes where the corresponding standard CHSH entanglement test is inconclusive. In this state-adapted setting, we explicitly construct the post-selection and CHSH measurements that achieve the largest possible separation from the separable PPS bound. More broadly, our results motivate hybrid protocols that combine post-selection and entanglement, with possible applications to improved quantum sensing, weak-value amplification, and quantum information processing.

Fault-tolerant quantum computation requires logical operations that manipulate encoded information while preserving quantum error-correction protection. In planar surface-code architectures, code deformation and lattice surgery provide a local, measurement-based route to such operations. Here we experimentally realize key elements of patch-based surface-code logical processing on a 107-qubit superconducting quantum processor. We first implement a reusable primitive layer comprising merge and split, patch expansion and shrinkage, and deformations mediated by domain walls and twist defects. We then compose these primitives to realize logical state routing, the logical controlled-NOT gate, and the single-qubit Hadamard and phase gates, which together form a Clifford-generating set. All operations are implemented on distance-three rotated surface-code patches with multi-round syndrome extraction and neural-network decoding, without post-selection. Our results advance superconducting surface-code experiments from protected logical memory to active, patch-based fault-tolerant logical operations.

Matrix product states (MPS) are quintessential examples of frustration-free gapped ground states of local interactions called parent Hamiltonians. In this work, we investigate parent Hamiltonians for a class of ergodic matrix product states (EMPS), which are MPS defined by site-dependent random tensors $\{X_j^{[k]}\}_{j=1}^D$ which are homogeneously distributed at every site $k$ in the spin chain. Here, the EMPS are not translation-invariant but rather statistically translation-invariant. Under a mild injectivity assumption, we show the thermodynamic limit of an EMPS is the unique frustration-free ground state of a parent Hamiltonian on the whole spin chain, which, depending on the statistical properties of the EMPS, may or may not be finite-range. In contrast to the translation-invariant regime, these Hamiltonians need not be gapped. Nevertheless, applying the martingale method while keeping track of local statistics gives conditions for a gap, in addition to pointing towards why there need not be a gap in general. We include examples of EMPS both with and without spectral gaps to illustrate our results.

High-fidelity universal two-qubit gates are critical for building fault-tolerant quantum computers. In scalable superconducting processors, shortened coherence times introduce more incoherent errors in gate operations. With a constrained error budget, there is reduced tolerance for coherent errors stemming from parameter deviations. In this work, we develop a closed-loop workflow to enhance the CZ gate calibration precision. Utilizing the echoed leakage error amplification (ELEA) and the repurposed context-aware fidelity estimation (CAFE) circuits, we suppress the population leakage to non-computational states, and, for the first time, demonstrate a CZ gate fidelity exceeding $99.9\%$ on an 84-qubit processor, with coherent error suppressed to $0.007\%$. Meanwhile, we obtain a median fidelity of $99.25\%$ among 72 CZ gates, demonstrating that the workflow can be generalized to the calibration of parallel CZ gates. Finally, we realize automated calibration and observe enhanced stability of the CZ gate throughout 9-hour comparative monitoring experiments. Our results, realized on a completely domestic platform, establish an efficient and automated route to quantum computation with superconducting quantum systems.

In the presence of very intense electromagnetic fields, the refractive index of vacuum is modified such that light velocity is less than $c$ and ultrarelativistic charged particles can be faster than light and can induce Cherenkov radiation. We present the comparison of the Cherenkov radiation produced by the Euler-Heisenberg theory for critical and supercritical magnetic fields. We also make the comparison between the Cherenkov and the synchrotron radiation produced by the charged particles.

In this work, we focus our analysis of Fractional-Dimension Gravity (FDG) on our home galaxy, the Milky Way (MW), by using the latest Gaia DR3 data as well as previous rotation curve (RC) data for this galaxy. FDG is an alternative gravitational model (previously known as Newtonian Fractional-Dimension Gravity - NFDG) which does not require the dark matter (DM) paradigm. The MW is studied here with the methods of FDG and its observed rotation curves are successfully reproduced by using a variable fractional dimension $D\left (R\right)$, following previous studies of several other galaxies which were analyzed with the same methodology. An alternative dimension function $D_{m}\left(R \right)$, based on the mass-dimension field equation, was also used and yielded less accurate fits to the experimental data. In addition, we also considered possible implications of the FDG metric, based on the presence of additional weights, on the structure of Special Relativity (SR) for spacetimes with fractional dimension. One notable outcome of this analysis is the possibility of an effective superluminal motion in galactic regions where the space dimension is $D<3$. Although this result is very speculative, it opens interesting new perspectives for possible interstellar travel in our galaxy.

Under the assumption that dark matter is a coherently oscillating pseudoscalar field coupled to electromagnetism by the usual Chern-Simons term, we study the production of secondary photons from dark matter fluctuations coupled to a pre-existing magnetic field, taking into account the spectral distribution of the magnetic field. Specifically, we apply the formalism to the case of a large-scale magnetic field generated previously via a parametric resonance instability due to the same Chern-Simons coupling. However, our analysis is applicable to any spectrum of cosmological scale magnetic field fluctuations present at the time of recombination. We show that obtaining a sufficiently large flux of photons in the Lyman-Werner frequency range is consistent with constraints from CMB and X-ray observations.

Several approaches to quantum gravity suggest the emergence of a fundamental minimal length at the Planck scale. In quantum mechanics, this feature is naturally encoded through deformations of the Heisenberg algebra, leading to the Generalized Uncertainty Principle (GUP). While the phenomenological implications of GUP have been extensively explored, a consistent characterization of coherent states in minimal-length quantum mechanics remains elusive. In this work, we present a systematic analysis of coherent states for the one-dimensional harmonic oscillator. We show that the canonical equivalence among their standard characterizations - as eigenstates of the annihilation operator, displaced vacuum states and minimum-uncertainty wave packets - is generically lost in the presence of a minimal length. We then investigate the dynamical and semiclassical consequences of this inequivalence by comparing the evolution of generalized coherent states with that of states saturating the GUP. In particular, we demonstrate that minimal-length effects induce nontrivial deformations of phase-space trajectories and give rise to an intrinsic squeezing mechanism with no counterpart in ordinary quantum mechanics. These results establish a unified framework for coherence in GUP-based quantum theories and identify distinctive semiclassical signatures of minimal-length physics, opening a new avenue for probing quantum-gravitational effects.

Autonomous quantum error correction (QEC) stabilizes a logical manifold through dissipative events that emit into output channels, which are typically accessible to measurement. These signals are often discarded, and whether they contain useful information about logical failures remains generally unclear. Using quantum trajectories, we show that in dissipatively stabilized cat qubits bit flips are not silent logical errors: each flip is accompanied by a strong, time-localized photon burst from the dissipative buffer. Photon counting and homodyne monitoring can therefore herald the loss of logical information without interrupting the autonomous stabilization: bit flips in dissipative cat qubits are erasures. More broadly, our results show that the emitted signals of engineered reservoirs can act as built-in failure monitors for autonomous QEC, turning rare logical faults into erasures available to a decoder and reducing fault-tolerance overhead. To this end, we develop a general framework, based on past quantum states and number-resolved master equations, to quantify the detectability of such logical failures in autonomous QEC from the emitted signal.

We investigate properties of toric codes under realistic damping error channels, which include squeezing, thermal and non-Markovian effects. First, we map the decohered toric code under the generalized amplitude-damping (GAD) and the squeezed generalized amplitude-damping (SGAD) channels to the statistical-mechanical models using the double Hilbert-space formalism. Second, we map the action of the GAD and SGAD channels on the toric code to stochastic Pauli-type errors via Pauli twirling, yielding asymmetric depolarizing channels, and obtain the logical failure probabilities as a function of temperature and squeezing. In both cases, we relate the channel parameters of the GAD and SGAD channels to the spin-coupling constants of the statistical-mechanical model.

Chiral spin liquids carry a hidden bookkeeping problem: the integer Chern number of their fractionalized spinons, the level of the emergent Chern--Simons gauge field, and the fractional spin response actually measured in experiment or simulation are related but distinct quantities, and the literature routinely conflates them. Here we resolve this by deriving the exact parity-odd determinant of a gapped Dirac cone on a spatial cylinder, resummed to all orders in the compact holonomy rather than truncated at leading order. The result proves that finite-circumference corrections to the topological response are strictly exponential, with no universal $1/L$ term, and fixes the precise map from microscopic spinon Chern number to physical spin Hall conductance. We validate this chain of reasoning on the kagome lattice at three independent levels: an exact parton band-structure calculation ($C=-1$, converging exponentially over cylinders four to twelve sites wide), and an interacting density-matrix renormalization group flux pump ($ν_s=-0.500\pm0.011$) that agrees with the analytic prediction without any adjustable parameter. Together, these results turn a one-loop anomaly calculation into a quantitatively verified bridge between microscopic topology and observable fractional response.

A new understanding of criticality in systems described by quadratic bosonic Hamiltonians (QBHs) ties the emergence of long-range correlations to boundaries of dynamical, not thermodynamical, stability in the parameter space. This separation occurs because the solution of the Heisenberg equations of motion is determined by an auxiliary pseudo- Hermitian dynamical system. The boundary points of a region of dynamical stability can be either exceptional points, generically associated with long-range correlations, or Krein collisions, where correlations can be either long- or short-range. We investigate the interplay of this landscape of possibilities with band topology and boundary physics, by relying on both specific examples and general arguments. The examples stem from a two-parameter, thermodynamically unstable family of QBHs obtained from the bosonic Su-Schrieffer-Heeger model by breaking particle conservation while preserving a chiral pseudo-symmetry. The dynamically stable regime breaks up into different regions labeled by an integer-valued symplectic analogue of the Berry phase. The topological phase transition is a line of Krein collisions, which coincides with the closing of a band gap at zero and causes the localization length of the topologically mandated boundary zero modes to diverge before disappearing. In the unstable regime, we show that the chiral pseudo-symmetry of our model induces, despite the broken particle-number symmetry, enough structure on its associated dynamical matrices to support a topological classification and a bulk-boundary correspondence, independently of dynamical stability. This strongly suggests that bosonic topological physics extracted from basic index theory is insensitive to dynamical stability and, a posteriori, to non-interacting criticality.

We investigate the entanglement dynamics along quantum trajectories during the seeding of time-crystalline order in a boundary time crystal (BTC). Specifically, how entanglement spreads among different spin ensembles when a BTC attempts to seed its time-crystalline behavior onto otherwise static spin ensembles, through a collective dissipative channel. We analyse both the dynamical growth of entanglement in time and the steady-state properties of the system. Our results reveal two fundamentally distinct regimes. In the seeded BTC phase, the steady-state entanglement entropy between the ensembles grows with system size $N$, accompanied by macroscopic fluctuations along the trajectories. In contrast, in the non-seeded static phase, both the steady-state entanglement and its fluctuations decay exponentially with $N$. The model thus features a measurement-induced phase transition (MIPT) driven by the seeding mechanism. Furthermore, these findings establish dissipative seeding as a powerful mechanism for controlling quantum correlations in open many-body systems, with direct experimental relevance to this class of model without a postselection barrier.

Scalable quantum networks require the efficient generation, storage, and synchronization of entanglement between photonic qubits and quantum memories. Quantum repeater architectures based on absorptive rare-earth-ion photonic memories offer a promising route toward highly multiplexed quantum networking, but integrating spectrally matched photon sources and quantum memories within a common platform remains a major challenge. Here we demonstrate an integrated photonic architecture for telecom photon-memory entanglement generation based on dual silicon-carbide microring resonators. One resonator operates as an entangled photon-pair source, while the other functions as a cavity-enhanced atomic-frequency-comb quantum memory. The memory resonator achieves an ensemble cooperativity of 1.9 and is intrinsically spectrally matched to the photon source, enabling storage of entangled telecom photons without spectral modification. We generate and verify photon-memory entanglement with a single-pair interference visibility of 88.1 $\pm$ 10.6%. By exploiting the multimode capacity of the memory, we demonstrate high-dimensional photon-memory qudit entanglement spanning up to 63 temporal modes, leading to a maximum photon information efficiency of 5.1 Ebits per detected photon and a peak on-chip photon-memory entanglement rate of 5.6 kEbits s$^{-1}$. These results establish the first integrated platform for photon-memory entanglement generation and provide a scalable route toward chip-scale quantum repeaters and memory-enabled quantum networks operating over telecommunications infrastructure.

As Coleman famously argued, the apparent breakdown of partition-function factorization in quantum gravity associated with Euclidean wormholes is a red herring, arising from a hidden average over an ensemble of theories. We present a direct analog of Coleman's argument for the apparent breakdown of Hilbert-space factorization associated with spatial wormholes, i.e., Einstein--Rosen bridges. Our main result is the following reconstruction theorem for quantum field theories: unitary QFTs are determined, up to unitary isomorphism, by their closed-manifold partition functions; every reflection-positive partition function arises from a unitary quantum field theory; and the states prepared by manifolds span the space of invariant states under the reconstructed theory's symmetry group. Interpreting the result gravitationally, we conclude that any apparent breakdown of Hilbert-space factorization is a red herring, arising from restricting to an incomplete spectrum of charged states.

Classical simulation of noisy quantum circuits is essential for validating algorithms, benchmarking hardware, and assessing error-mitigation strategies, but remains limited by the exponential cost of density-matrix methods and the high variance of standard trajectory sampling. We introduce a variance-aware tensor network framework that combines the tensor jump method with local TDVP gate evolution on matrix product states and sparse Pauli-Lindblad hardware noise models. Gates are applied as short variational evolutions on the MPS manifold, while noise is sampled per circuit window from Pauli-Lindblad jump sets with state-independent hazards and dissipative contractions that reduce to irrelevant global factors after renormalization. The method supports correlated multi-qubit Lindblad noise consistent with hardware connectivity, including long-range operators on non-adjacent qubits, enabling direct simulation of crosstalk and other connectivity-induced errors beyond local noise models. We develop two unbiased variance-aware unravelings. An analog unitary-mixture unraveling matches the Lindblad generator exactly under symmetric Gaussian or two-point angle laws, while a projector-jump unraveling yields state-independent hazards and closed-form variance laws. Both retain the standard 1/sqrt(N) Monte Carlo convergence but with reduced prefactors. Empirically, projector sampling strongly reduces trajectory variance and bond-dimension growth across many circuit architectures, whereas analog sampling is most effective at weak noise. We demonstrate accurate, scalable noisy-circuit simulation on a 25-qubit noisy XY quench and IBM's 127-qubit kicked-Ising benchmark with long-range depolarizing noise, achieving reduced Monte Carlo variance and favorable MPS bond-dimension growth compared with standard Kraus-insertion baselines.

The quantum chromodynamics (QCD) axion and axion-like particles may form bound clouds around spinning black holes (BHs) when their Compton wavelength is comparable to the BH gravitational radius, depleting the BH spin through what is known as a $\textit{superradiance}$ instability. Using binary BH (BBH) spin measurements obtained from the LIGO-Virgo-KAGRA GWTC-5 catalog, the most extensive public BBH catalog to date containing $N=257$ mergers with BH masses spanning roughly $5$-$135$ $M_\odot$, we perform a hierarchical Bayesian analysis in the context of a BH spin population model to constrain ultralight axions. The presence of axions at a given mass would imprint a unique signature in the observed mass-spin relation relative to the formation distribution. We find no evidence for axions across more than two decades in mass, excluding axion masses $1.7 \times 10^{-14} \, {\rm eV} \lesssim m_a \lesssim 3.3 \times 10^{-12} \, {\rm eV}$ at 95% confidence. Because prior superradiance bounds in this range derive from X-ray spin measurements with substantial modeling systematics, this result represents one of the strongest robust lower bounds on the QCD axion mass.

Quantum entanglement plays a leading role in modern understanding of physical systems, from quantum phases of matter to quantum gravity. In quantum information theory, one seeks operationally meaningful quantifiers of entanglement, which for large quantum systems are notoriously difficult to evaluate due to the lack of computationally efficient algorithms. In this Letter, we show that for large random induced mixed states the logarithmic negativity, an efficiently computable entanglement measure, generically coincides with the exact entanglement cost under positive-partial-transpose-preserving operations, thereby acquiring a precise operational interpretation. Our results establish logarithmic negativity as an exact characterization of entanglement in generic many-body states and provide a tractable route for quantifying entanglement in complex quantum systems.

We revisit the problem of minimally coupling matter to Einstein gravity in three dimensions with negative cosmological constant. By working in the worldline formalism, we construct a classical phase space on an initial time surface $Σ$, which we quantize using geometric quantization. States in the Hilbert space correspond to Virasoro conformal blocks with operators of conformal weight $h<c/24$. As an application of our formalism, we compute the partition function on thermal $\text{AdS}_3$ through equivariant localization. Our answer reproduces the AdS$_3$ Wilson spool and agrees with the known one-loop result. It further serves as a conjecture for the value of the path integral of gravity minimally coupled to a massive scalar field in thermal $\text{AdS}_3$ to all orders in $G_N$.

The Laser Interferometer Space Antenna (LISA) is a space-based gravitational wave observatory that consists of three spacecraft in a near-equilateral triangular formation. The spacecraft orbits are typically assumed to be perfectly known in LISA data analysis studies, but in reality, the orbit determination process introduces uncertainties in the spacecraft positions and velocities. In this work, we investigate how these uncertainties propagate into the LISA detector output and the impact of neglecting the spacecraft velocities. We quantify these errors in the knowledge of the LISA response using mismatches and discuss the implications for gravitational wave data analysis. We find that spacecraft orbit uncertainties impact the LISA response knowledge at high frequencies with worst mismatch below $10^{-7}$. The effect of neglecting the spacecraft velocities is largest at frequencies around $10^{-4}$ Hz with mismatches of order $10^{-4}$. For a galactic binary with frequency $10^{-4}$ Hz and SNR=200 observed for one year, we find that neglecting the spacecraft velocities in the response leads to less than 1-$σ$ biases in the parameter estimates. This work provides the first characterization of how errors in the LISA gravitational wave response propagate from gravitational wave strain through detector output to estimated parameters.

We investigate the Dirac oscillator in a helically twisted spacetime endowed with a uniform axial torsion. Starting from an orthonormal coframe, we compute the Levi--Civita spin connection explicitly and separate the geometric contribution from the axial contortion. Retaining the matrix $β$ in the radial Moshinsky coupling, we show that the second-order problem is the ordered product $\hatΠ_+\hatΠ_-$ rather than the square of a single operator. The resulting radial dynamics is a coupled, self-adjoint two-component system in which the spin connection supplies the correct cylindrical radial operator, while the off-diagonal metric generates the helical combination $m/r-ωk$ and a Coulomb-like geometric term. A finite-element solution reproduces the planar Dirac-oscillator spectrum in the flat limit and reveals asymmetric dependence on the longitudinal momentum, avoided level crossings, and a supersymmetric zero mode at $E=Mc^2$. The axial torsion and longitudinal momentum preserve this zero mode, whereas the helical twist lifts it quadratically. Sector-resolved thermodynamic functions are obtained from the relativistic bound-state spectrum. The explicit spinors further determine longitudinal vector and axial currents, and a Witten-index analysis identifies the helical twist as the deformation that removes the protected zero mode.

What would it be like to be in a superposition of yesterday, today, and tomorrow? This question may seem at best entertaining, but it is necessary, and exploring it allows us to understand how exact irreversible clocks and change are possible, despite the Unruh-Wald and Hegerfeldt-Ruijsenaars no-go theorems forbidding them. Unruh and Wald (1989) proved that if energy is bounded from below, no observable can increase monotonically with the Schrödinger time parameter t. Perfectly monotonic clocks and irreversible observable changes (Hegerfeldt-Ruijsenaars, 1980) seem impossible. From the perspective of the Schrödinger time, the world appears in a superposition of different intrinsic clock states indicating different times and opposite time directions. This seems to directly contradict our daily experiences of time and change. I show that there is no contradiction: from an intrinsic perspective of the world, sharp irreversible changes do happen, because the macroscopic pointer states resolve the superposition of different times. Large-scale time-reversing or discontinuous transitions are not internally observable in the records. From the intrinsic perspective, an unbounded intrinsic-time translation generator plays the role of the Hamiltonian, generating only forward time evolution with respect to the intrinsic time, but not to the Schrödinger parameter t, which is thus not justified to play the role of time. This allows sharp time observables even if the external Hamiltonian is bounded from below. In addition, this leads to a stationary wavefunction of the universe satisfying a Wheeler-DeWitt-type equation, without assuming gravity.

We argue that cosmological models obeying the Covariant Entropy Bound (CEB) mathematically favor states with no localized excitations or one large black hole containing all the energy in a constrained initial state. In order to get a long radiation-dominated era, one must postulate that at a very early time, most horizon volumes of the universe contained tiny black holes that decayed into radiation. A previous work by two of the authors showed that such a scenario could fit the data on the Cosmic Microwave Background (CMB). In order to account for dark matter, we also postulate some random black holes of at least horizon size at that time. A reasonable distribution of such primordial black holes can account for all of dark matter as well as the early galaxies seen by the James Webb Space Telescope. Some of the dark matter may also be in Planck-scale remnants of the decaying black holes. We describe our model both in terms of approximate solutions to General Relativity and a speculative quantum gravity model whose hydrodynamics matches the flat $p = \pm ρ$ FRW model that saturates the CEB.

We propose a photonic platform for simulating arbitrary single-qubit open-system dynamics using a single photon in an open Mach-Zehnder interferometer. A birefringent quartz plate induces a coupling between the polarization and frequency degrees of freedom. By treating the latter as an effective environment, we analytically derive the reduced polarization dynamics. We show that the resulting evolution is characterized by a controllable interplay between populations and coherence, instead of the usual dephasing caused by quartz plates. By adjusting the photon frequency distribution and interferometric parameters, we demonstrate that target single-qubit states can be efficiently reproduced through a tunable optical protocol expected to work under accessible experimental conditions. The simulator is benchmarked against paradigmatic open-system evolutions, including depolarization and non-Markovian dynamics, achieving high accuracy. Our results establish polarization-frequency engineered photonic interferometers as a versatile protocol for simulation of open quantum systems.

Reservoir engineering enables unidirectional energy and signal flow. We establish squeezing-induced symmetry breaking between two cavities as a guiding principle for exponentially amplifying reservoir-mediated nonreciprocity. Rather than a simple scaling of the coupling, this mechanism strategically redistributes the squeezing resources to relax experimental requirements, as single-cavity squeezing alone demands a much larger squeezing strength. Moreover, reservoir squeezing does not alter the system symmetry, but reshapes the noise correlations and thereby changes the system dynamics. The proposed mechanism improves the performance of the quantum battery by several orders of magnitude, including stored energy, charging power, and ergotropy, with the analytical expressions provided. Extending to the optical isolation, we observe a second-order exponential enhancement of the output signal. Our results open a new avenue for nonreciprocal quantum information processing and nonreciprocal quantum device design.

A hypersurface anchored variational extension of General Relativity is formulated in which the Einstein-Hilbert action is supplemented by a diffeomorphism invariant functional supported on an embedded spacelike hypersurface whose embedding is varied independently of the spacetime metric. The resulting Euler-Lagrange system consists of the Einstein equations with a localized distributional contribution together with an anchoring equation determining admissible embeddings. For the admissible class of hypersurface functionals considered here, the bulk field equations retain the standard second order principal structure away from the hypersurface and no additional propagating bulk gravitational degrees of freedom are introduced. Under ellipticity and invertibility assumptions, local persistence and linear stability of anchoring hypersurfaces follow from standard implicit function and elliptic estimates. The anchoring condition is generically inequivalent to local slicing gauge conditions and therefore defines a genuine variational restriction rather than a coordinate choice. In the canonical formulation, the momentum constraints retain their standard form, whereas the Hamiltonian constraint acquires a hypersurface supported term. The corresponding smeared generators close in the weak sense: the Dirac algebra is recovered in the bulk, with deviations confined to localized distributional surface contributions. The construction therefore defines a constrained sector of the classical solution space of General Relativity consisting of spacetimes that admit at least one embedded hypersurface satisfying the anchoring equation. In homogeneous cosmology, the localized term induces matching conditions across the anchoring surface, allowing finite transitions in the expansion rate while preserving standard evolution away from the transition hypersurface.

We overcome the long-standing noninvasive measurability (NIM) challenge in Leggett-Garg tests by exploiting the causal structure of quantum field theory (QFT). Our protocol uses three independent ensembles of the vacuum state, each measured by a different pair of observers at spacelike-separated events, yielding the three two-time correlators. By placing these events at positions $(0,0)$, $(τ,L)$, and $(2τ,2L)$ with $L>τ+2τ_0$, we rigorously ensure that no measurement can influence another. We investigate the vacuum state of a free massive scalar field in 1+1 dimensions, employing the dichotomic observable $Q(f)=\operatorname{sign}(φ(f))$ where $φ(f)$ is the smeared field. In the Heisenberg picture, the time evolution is absorbed into a translation of the time-window function, allowing us to derive the two-time correlation function $C(τ,L)$ and the Leggett-Garg parameter $K_3=2C(τ,L)-C(2τ,2L)$. For non-overlapping time windows, we find that the correlation function decays exponentially with $τ$ for a massive field. For overlapping windows, our numerical computation for a rectangular time window yields $K_3<1$ across the entire mass range, firmly establishing that the vacuum does not violate the LGI. Thus, under strict noninvasive conditions, the vacuum shows no violation of macrorealism, in stark contrast to its well-known violation of spatial Bell inequalities. Our spacelike-separated protocol provides the first LGI test in QFT with rigorously satisfied NIM, setting a methodological benchmark for future studies and highlighting the fundamental distinction between spacelike entanglement and temporal macrorealism in relativistic quantum fields.

Motivated by the recent evidence for a stochastic gravitational-wave (GW) background reported by pulsar timing array (PTA) collaborations, we perform a precision study of low-scale phase transitions in a dark Abelian Higgs sector, a minimal gauge theory of spontaneous symmetry breaking relevant for cosmological phase transitions. Using dimensionally reduced high-temperature effective field theory, we quantify the impact of thermal resummation, higher-order matching corrections, and higher-dimensional operators on the phase-transition thermodynamics and the resulting GW signal. We find that the parameter region favored by current PTA observations lies close to the boundary of validity of the effective field theory, where higher-dimensional operators become increasingly important. Even within this controlled region, the predicted signal remains disfavored by the PTA data, despite the substantial shifts induced by higher-order thermal corrections. We further delineate parameter regions where the dark and visible sectors are thermally and hydrodynamically coupled or decoupled, and revisit the dark matter phenomenology, identifying asymmetric freeze-out as naturally compatible with both the observed relic abundance and the gauge couplings favored by strong phase transitions. Our results underscore the importance of systematically controlled finite-temperature calculations for reliable GW predictions from low-scale cosmological phase transitions.

We use the Poincare parametrization of AdS space to develop a general BRST-BV approach for free fields. A general expression for the BRST-BV Lagrangian of fields with arbitrary masses and symmetry types is obtained. We apply this general framework to study totally symmetric massless, massive, and partially-massless fields with arbitrary integer spin and a continuous-spin field. For these fields, both the constrained and unconstrained BRST-BV formulations are developed. In addition, we demonstrate the matching between the obtained BRST-BV Lagrangian and the metric-like Lagrangian formulated in terms of the modified de Donder divergence. Finally, a realization of AdS space symmetries is obtained within the space of fields and antifields entering the BRST-BV formulation.

We study the $J/ψ\to Λπ\barΣ$ reaction by looking at the $π^+ Λ$ mass distribution at low energies, in search of signals for the low lying $Σ^+$ states. Apart from a clear signal of the $Σ(1385) (3/2^+)$ state, we find a smaller peak for the predicted $Σ(1430) (1/2^-)$, which has already been confirmed by the Belle Collaboration. A first analysis, considering only the $πΛ$ interaction, shows that the low energy part of the spectrum is better reproduced including contributions from the $Σ(1430)$ and the predicted $Σ(1380)(1/2^-)$ state that has been claimed before from analyses of different experiments. However, when we consider the $π\barΣ$ interaction the need for the $Σ(1380)$ disappears.

Starting from the gauge invariant action for free string field theory with boundary recently constructed in 2506.05969, we define new gauge invariant observables which are analogous to the Brown-York charges of General Relativity. Just like the Brown-York charges, our observables originate from a boundary tadpole, and are associated to isometries of the SFT gauge group around a given background. The consistency of the construction requires the equation of motion of the background to be satisfied only at the boundary and therefore these observables can also be defined for backgrounds generated by sources in the bulk. As examples of our construction in open string field theory, we compute the flux through the boundary of constant electromagnetic field-strength solutions and the charge associated to the Coulomb solution. As a further example in closed string field theory, we characterize the infinite conserved charges associated to stringy-haired black-hole solutions in two-dimensional string theory. We also construct a generalization of these boundary observables to the full interacting string field theory.

We derive the pertinent chiral potentials for charmed vector meson interactions with light pseudoscalar bosons in a relativistic U(3) chiral effective field theory up to next-to-leading order. Predictions for the $S$- and $P$-wave scattering lengths are obtained for all the relevant elastic channels. A comparison with the most recent -- and currently the sole -- lattice QCD data on the $S$-wave $I=1/2$ $D^\astπ$ scattering length at a pion mass of $391$~MeV reveals good agreement, thereby validating the estimation of low energy constants via heavy quark spin symmetry. Within the relativistic formalism, we confirm that the $D_{s1}(2460)$ can be identified with a bound state pole, while the $D_1(2430)$ corresponds to the interplay of two poles: a lower one on the second Riemann sheet and a higher one on the third Riemann sheet. We show that the $D_{s1}(2460)$ and the lower $D_1(2430)$ pole originate from the same flavor SU(3) triplet, whereas the higher $D_1(2430)$ pole belongs to the SU(3) sextet. All these states are not of $\bar{q}q$ nature, as they flow to complex infinity in the large $N_C$ limit. Our results provide quantitative benchmarks for future lattice QCD and femtoscopic studies.

We calculate energy-energy correlations for two hadrons produced inside a jet in transversely polarized proton-proton collisions. We make numerical predictions based on a simple model that utilizes a previous global QCD analysis of dihadron fragmentation and transversity parton distribution functions. The results show remarkable agreement with a very recent STAR measurement. We also find the data at large jet transverse momentum have a slight preference for extractions of transversity that are consistent with lattice QCD computations of the nucleon tensor charges. Overall, this work provides further evidence for the underlying non-perturbative mechanism of near-side energy-energy correlators as well as highlights the potential for these observables to probe transverse-spin effects inside the nucleon.

Transverse-momentum-dependent (TMD) fragmentation in $e^+e^-$ collisions can be studied by measuring hadrons with respect to the thrust axis, and has been measured at Belle. This provides a complementary way to extract TMD fragmentation functions, avoiding the need to disentangle the two TMD fragmentation functions that enter conventional back-to-back hadron-pair measurements. Starting from the established factorization theorems for this observable, we complete the operator-level formulation of the soft ingredients and perform one-loop checks using the $δ$-regulator. We also extend existing results for 1-jettiness factorization in semi-inclusive deep-inelastic scattering (SIDIS), where analogous measurements give access to the TMD parton distribution functions of the incoming hadron. For phenomenology, we discuss the nonperturbative effects and propose a model that captures both the event-shape dependence and correlations between the event-shape and transverse-momentum measurements. We resum the transverse-momentum and thrust logarithms, explore several schemes for treating the latter, and implement it in artemide. As a first validation, we compare to simulated $e^+e^-$ data from Pythia8.3. We find that the proposed nonperturbative model is flexible enough to describe the simulated data, with fitted parameters of the expected size in powers of $Λ_{\rm QCD}/Q$. In this test, the resummation of the logarithms of $q_T/(τQ)$ appears to have little impact on the fit quality, but changes the fit parameters.

We present a novel approach for solving canonical differential equations for Feynman integrals based on an approximation of the integrals with Chebyshev polynomials. By exploiting the analyticity properties of Feynman integrals, the method constructs rapidly converging polynomial approximations along a path, enabling highly efficient numerical evaluation. Moreover, we introduce an adaptive approximation method that dynamically samples to optimise convergence. We implement this framework in double-precision arithmetic and demonstrate its stability across physical phase space using a series of two-loop, five-point examples. Our proof-of-principle implementation proves competitive with state-of-the-art one-fold integral methods, while requiring little to no case-by-case intervention to handle spurious singularities.

We report a new measurement of the solar proton--proton ($pp$) neutrino flux via neutrino--electron elastic scattering using the PandaX-4T Run 2 data set collected between 2024 and 2026, corresponding to an exposure of 1.9 tonne$\cdot$yr. Before Run 2 data taking, the detector underwent a series of upgrades to improve its response and background conditions. Time variations of radioactive noble-gas impurities are constrained using the physics data themselves, complemented by measurements from the gas-assay system. The analysis introduced improvements in the data processing chain, detector response characterization, and background models. A blind spectral analysis was then performed on the electronic-recoil data across a wide energy range from 20 to 1000 keV. In combination with the Run 0 data published earlier, the fitted $pp$ flux is $(8.5 \pm 3.5)\times 10^{10}$ $\mathrm{cm^{-2}s^{-1}}$, consistent with the prediction of the Standard Solar Model. With a statistical significance of $2.2σ$ above background, this marks the first positive indication of solar $pp$ neutrino--electron scattering below an electronic-recoil energy of 165 keV.

Motivated by recent experimental observations, we investigate the $J^P$ distribution of low-energy compact tetraquark states. Assuming that two quarks and two antiquarks are arranged in either a tetrahedral or a square configuration, we employ the restricted representations of the permutation group $S_4$ onto $S_2 \times S_2$ to derive the inherent nodal structure of the $qq\bar q\bar q$ system from that of the $qqqq$ system for orbital angular momentum $L \leq 3$. Based on this framework, we determine the distribution of accessible $J^P$ states and find that low-energy compact tetraquark states are likely to favor $J^P=2^+$. Our analysis yields two observations that further support the dynamical robustness of symmetry-based classifications in exotic hadron spectroscopy. First, the symmetry-induced $J^P$ distribution of compact tetraquark states closely resembles that obtained for the three-flavor four-quark system. Second, the location of the distribution peak remains unchanged when chromomagnetic interaction (CMI) effects are incorporated. Together, these results suggest that the dominant features of the low-lying spectrum are governed primarily by symmetry constraints rather than by the details of the underlying dynamics. These findings further imply that the fully charmed tetraquark candidates $X(6600)$, $X(6900)$, and $X(7100)$ may occupy relatively low-lying levels of the fully charmed tetraquark spectrum. They also indicate that mechanisms beyond CMI dynamics are likely involved, potentially mitigating or competing with the effects of CMI in compact fully charmed tetraquark states.

The mixing between flavor and mass eigenstates of active neutrinos is described by a $3\times3$ unitary matrix. However, the presence of additional heavy sterile neutrino states can lead to a non-unitary neutrino mixing scenario. Atmospheric neutrinos, with their wide range of baselines and energies, provide an excellent probe of such effects. In particular, Earth matter effects in neutrino oscillations play an important role, as the neutral-current potential contributes non-trivially in the presence of non-unitarity. In this work, we use 8 years of publicly available atmospheric neutrino data of IceCube DeepCore to probe this non-unitary neutrino mixing scenario. This high-purity $ν_μ$ CC sample provides strong sensitivity, especially to the non-unitary parameters appearing at leading order in the $ν_μ\rightarrow ν_μ$ channel. The data sample is found to be consistent with the standard unitary mixing framework with no significant deviation. Using this data sample, we place the most stringent bound to date of $α_{33} > -0.027$ at 90% CL, while the other non-unitary parameters are constrained at competitive levels.

We formulate a reduced-profile maximum-entropy method (MEM) framework for constructing constrained zero-skewness valence-quark generalized parton distribution (GPD) transverse profiles from the four nucleon electromagnetic form factors $F_1^p(t)$, $F_1^n(t)$, $F_2^p(t)$, and $F_2^n(t)$. The form-factor sum rules fix only $x$-integrated moments of the GPDs; the forward limit of $H_v^q$ is fixed separately by the valence parton distribution functions, and the normalization of $E_v^q$ by the flavor anomalous magnetic moments. These complementary constraints are combined through the ansatz $H_v^q(x,t)=q_v(x)\exp[t f_H^q(x)]$ and $E_v^q(x,t)=e_v^q(x)\exp[t f_E^q(x)]$, where the positive profile functions encode the $x$-dependent transverse structure. Rather than attempting an unrestricted functional inversion, we use the entropy functional as a regularizing criterion on a low-dimensional positive profile manifold. In the numerical proof-of-concept calculation, a smooth elastic form-factor input and analytic forward distributions are adopted, together with the reduced form $f(x)=0.05+(1-x)^2\exp(c_0+c_1x+c_2x^2)$, which suppresses local modes that elastic moments alone cannot constrain. Within this reduced ansatz, the resulting profiles reproduce the imposed elastic moment constraints, satisfy the forward normalizations after discrete-grid normalization, and give impact-parameter distributions with the expected transverse shrinkage at large $x$. The construction provides a controlled zero-skewness baseline for connecting elastic form-factor constraints to $x$-dependent transverse profiles, and it offers a stable starting point for future analyses incorporating empirical form-factor fits, modern PDF inputs, lattice-QCD generalized form factors, and hard exclusive observables.

We examine the exclusive semileptonic decays $B_s \to D_s^{**} \ell ν_\ell$, with $D_s^{**} =$ $\bigl\{D_{s0}^*,D_{s1}^*,D_{s1},D_{s2}^*\bigr\}$, within the Standard Model and beyond, using form factors evaluated in the Heavy Quark Effective Theory, including corrections up to $\mathcal{O}(α_s, Λ/{m_Q})$. A data-driven approach is employed to extract Heavy Quark Effective Theory parameters, and the resulting synthetic data are used to parameterize the form factors via the $z$-expansion. With the resulting form factor information across the full kinematic region, we compute various observables derived from the two-fold angular decay distribution, and predict precise lepton flavor universality ratios: $R_{D_{s0}^*}= 0.158(20)$, $R_{D_{s1}^*}= 0.045(5)$, $R_{D_{s1}}= 0.073(4)$, $R_{D_{s2}^*} = 0.066(9)$. We also analyse potential new physics effects using the Weak Effective Theory and the Standard Model Effective Field Theory, performing a global analysis considering both real and complex Wilson coefficients. Furthermore, we investigate new physics contributions arising from the general Two Higgs Doublet Model. We evaluate the sensitivity of decay observables to new physics, highlighting their potential to probe deviations from the Standard Model in future measurements. Notably, the scalar and tensor new physics operators induce large sensitivity, with some observables deviating by more than $2 σ$ from Standard Model predictions.

We formulate an avatar of the Fyodorov-Hiary-Keating conjecture for black hole microstate counts in quantum gravity. By holography, this implies sharp bounds on interval counts of high-dimension primary operators in conformal field theory. The extremal fluctuations of these counts are characterized by a random variable, with a prescribed tail distribution. At large $N$, these order-one erratic fluctuations set a quantitative limit on the resolution of the semiclassical AdS gravitational path integral. Gaussian random models for state counts arise naturally in this context; we express the phenomenon of erratic $N$-dependence in AdS/CFT as a decorrelation property of these models. Our broader point is to suggest that AdS black hole microstate spectra and their field theory duals should exhibit the extreme value statistics of random matrices, lying in the universality class of Gaussian log-correlated fields.

We develop a Floquet--Bloch sideband formulation of the linearized Maxwell--axion system in a coherent periodic axion background. Linearizing around prescribed magnetic and axion fields, we show that the pump generates a sideband ladder of photon and axion branches. Near an isolated folded degeneracy, this ladder reduces to a two-mode crossing whose algebra is fixed by the symplectic signatures of the colliding modes. In temporal fixed-momentum evolution, same-Krein-sign collisions give stable avoided crossings, whereas opposite-sign collisions give parametric instabilities, unifying the axion-photon difference channel with the Mathieu and Masaki-Aoki-Soda resonances. In stationary fixed-frequency transfer, the corresponding flux signatures distinguish bounded forward conversion from forward-backward stop bands and distributed reflection. Ray projection of a temporal pump gives a related but local WKB description of driven forward mixing, with an effective wavenumber distinct from the true axion momentum. External-field diagrams reproduce the sideband selection rules, and full temporal monodromy calculations verify the instability topology and finite-coupling shifts.

Electrons from semi-leptonic decays of charm (D) and bottom (B) hadrons are important probes in high-energy collisions, while their separation remains challenging due to the similarity of the underlying decay topologies. In this work, we represent the hadronic environment as a point cloud and investigate a hadron-based approach for distinguishing charm- and bottom-origin electrons using several set-based machine learning architectures, including Transformer models. Comparable performance is observed across different architectures, indicating that the dominant limitation originates from the intrinsic similarity between charm- and bottom-related hadronic structures rather than model expressivity. At an experimentally relevant working point corresponding to approximately 40% efficiency, the classifier achieves a purity close to 80% on the test dataset and significantly improves the classification performance relative to a hand-crafted observable BDT baseline. By studying the relation between the model response and physics-motivated observables, together with feature perturbation tests, we find that the learned representation is primarily sensitive to geometric and topological properties of the hadronic environment. Comparisons with high-level observables further suggest that the learned representation captures nontrivial discriminating information beyond a small set of manually constructed variables.

We investigate the exclusion and discovery potential for single production of a vector-like $Y$ quark with electric charge $Q=-4/3$, followed by the decay $Y\to bW$, at the FCC-eh. The $Y$ quark is allowed to couple to both first- and third-generation down-type quarks. The analysis is performed for an electron-beam polarization of $P_e=-80\%$ at $\sqrt{s}=3.46$, $5.29$, and $6.9~\mathrm{TeV}$. Both leptonic and hadronic $W$-boson decay channels are considered. In the hadronic channel, the boosted $W$-boson is reconstructed as a $W$-jet, and kinematic observables are used to suppress the Standard Model (SM) backgrounds. By performing a detailed detector simulations and event analysis, we present the $2σ$ exclusion limits and $5σ$ discovery reaches in the $g^*$--$m_Y$ plane, where $g^*$ is $Y$ coupling strength to the SM quarks. We find that the hadronic channel can provide stronger exclusion and discovery sensitivities, which are improved with increasing $\sqrt{s}$ at the FCC-eh.

We compute some supersymmetric twists of field theories in twistor space, including the minimal supersymmetric and the chiral algebra twists of supersymmetric self-dual Yang-Mills, and the minimal twist of $\mathcal{N}=1$ self-dual supergravity. In the case of $\mathcal{N}=4$ we also find their holographic duals in the framework of chiral holography. We find that the minimal twist of gauge theories in twistor space localizes them to spacetime, making the choice of complex structure manifest, and reproducing the minimal twist on spacetime. For superconformal theories we apply a further twist which localizes the theory to a plane contained on spacetime, reproducing the chiral algebra twist of $\mathcal{N}=4$ sYM. We show that the bulk duals of these twists also localize reproducing the results from twisted holography.

We perform a systematic QCD sum rule study of light tetraquark states with $J^{PC}=1^{--}$ in the diquark--antidiquark picture. A complete set of local interpolating currents is constructed and projected onto six flavor-isospin configurations ($q=u/d$): the isoscalar $q q\bar q\bar q$, $q s\bar q\bar s$, and $s s\bar s\bar s$ sectors, the isovector $q q\bar q\bar q$ and $q s\bar q\bar s$ sectors, and the isotensor $q q\bar q\bar q$ sector. The lowest masses in these sectors are derived to be $1.64^{+0.15}_{-0.14}$~GeV, $1.86^{+0.14}_{-0.14}$~GeV, $2.34^{+0.23}_{-0.30}$~GeV, $1.53^{+0.17}_{-0.19}$~GeV, $1.86^{+0.14}_{-0.14}$~GeV, and $2.24^{+0.12}_{-0.14}$~GeV, respectively. We further compare the present $1^{--}$ tetraquark spectrum with previous QCD sum rule results for the $1^{-+}$ tetraquark and hybrid states~\cite{Su:2025bhv}, aiming to provide useful information for distinguishing tetraquark and hybrid configurations in the light hadron spectrum. As an additional improvement, we complete the previously missing isotensor $1^{-+}$ tetraquark entry and obtain its lowest mass to be $M=2.19^{+0.26}_{-0.24}~\mathrm{GeV}$, which is included in the spectral comparison.

Strange hadrons are commonly included in dense-matter equation-of-state models by imposing chemical equilibrium, but the weak-interaction timescales required to establish it in core-collapse supernovae have not been systematically assessed. In this paper we compute the $Λ$-hyperon production rates in the hot, dense, and isospin-asymmetric conditions characteristic of post-collapse proto-neutron stars. We find that local $Λ$ chemical equilibration is driven by nonleptonic strangeness-changing reactions, especially $NN\leftrightarrow NΛ$ scattering, on timescales of order $10^{-11}$-$10^{-10}$ s, many orders of magnitude shorter than macroscopic proto-neutron-star evolution timescales. Using an effective-field-theory framework constrained by hypernuclear weak-decay data, we find that short-range contact interactions dominate the nonleptonic rates, beyond a pure one-meson-exchange description. Semileptonic channels are too slow to set the equilibrium $Λ$ abundance, but they open additional absorption channels for low-energy muon neutrinos and antineutrinos, such as $ν_μ+Λ\toμ^-+p$ and $p+μ^-+\barν_μ\toΛ$. At low energies, these $Λ$-induced neutrino opacities exceed the corresponding nucleonic contributions for muon (anti)neutrinos, possibly influencing the evolution of the muon lepton number during proto-neutron-star deleptonization. These results support local chemical equilibrium for $Λ$ hyperons under the conditions studied and provide new weak-interaction input for flavor-dependent neutrino transport, muonization, and proto-neutron-star evolution.

We promote the accidental $B+L$ symmetry of the Standard Model to a Peccei-Quinn (PQ) symmetry while realizing spontaneous proton decay radiatively. The PQ anomaly sector consists of vector-like quarks (VLQs), providing a Kim-Shifman-Vainshtein-Zakharov-type axion solution to the strong CP problem. After spontaneous PQ breaking, a residual $\mathcal{Z}_2$ symmetry remains, which forbids tree-level proton decay. The VLQs required to generate the QCD anomaly, together with scalar mediators, odd under the residual $\mathcal{Z}_2$, induce one-loop proton decay through the effective operator $u_R u_R d_R e_R$. We study its ultraviolet completions, featuring distinct vector-like fermion and scalar representations, and show that the resulting models lead to distinct predictions for the axion-to-photon coupling, testable in helioscope and haloscope experiments. In addition to the proton decay channel $p \rightarrow e^+ π^0$, this framework predicts proton decay with an axion in the final state, $p \rightarrow e^+ π^0 a$, suppressed by the PQ-breaking scale. We also discuss axion dark matter in pre- and post-inflationary cosmology.

We show that linearised E theory possesses a local symmetry at low levels provided the parameters of the local symmetry obey differential conditions that restrict their dependence on the extended spacetime. In the decomposition of E theory that leads to Siegel theory, also known as Double Field theory, we also find the analogous restrictions on the parameters. They are different to the section conditions which are universally used in this context. We also show that the dilaton equation of Siegel theory is invariant under the local symmetry if the parameters satisfy an analogous non-linear constraint on the parameters. We argue that there is no need to impose conditions on the fields of E theory or Siegel theory.

The enumeration of planar maps with control on the boundary metric, i.e. the pseudometric induced on the outer face of the map by its bulk graph distance metric, is a difficult problem in general. However, we show that for a family of bipartite planar map models with special q-deformed face weights that arise in the physics context of the double-scaled Sachdev-Ye-Kitaev model (DSSYK) the enumeration admits a very simple answer. Encoding the boundary metric of a bipartite planar map by its so-called geodesic chord diagram, we prove that the weighted enumeration depends only on the crossing number of the chord diagram. At fixed perimeter, the induced law of the geodesic chord diagram in these planar map models coincides exactly with the chord diagram representation of the DSSYK model.

This thesis investigates the dynamics of self-interacting dark sectors in the early Universe populated through the freeze-in mechanism. The main focus is on scenarios with self-number-changing reactions of the form $2\leftrightarrow3$, and on how these interactions modify the thermal history and phenomenology of the dark sector. Several realizations are studied, including scalar theories with $\mathbb{Z}_2$ and $\mathbb{Z}_3$ symmetries, self-interacting dark matter coupled to an unstable mediator, and cannibal dark matter production during non-instantaneous reheating. The thermal evolution is obtained by solving coupled Boltzmann equations for the relevant number densities and temperatures, accounting for both freeze-in production and cannibal interactions. The resulting parameter space can be strongly constrained, effectively invisible, or potentially accessible to future searches, depending on the realization considered. Finally, the thesis studies a hidden U(1) gauge sector populated via freeze-in, where continuous energy injection can induce an inverse first-order phase transition and temporarily restore the symmetric phase, linking phase-transition dynamics and chemical equilibration in hidden sectors.

We revisit the sensitivity of Single Molecule Magnet (SMM) crystals as detectors for low-mass dark matter. In previous work, we established the concept of the ``magnetic bubble chamber'', where energy deposited by dark matter triggers a magnetic avalanche in a metastable crystal. The original sensitivity estimates relied on a conservative criterion requiring the spin relaxation time to be strictly shorter than the thermal diffusion time. Here, we demonstrate that this criterion effectively ignores the stochastic nature of spin relaxation. We derive a refined analytic estimate which accounts for the fraction of spins that relax even when diffusion is fast. We show that the Zeeman energy released by this fraction contributes to local heating, significantly lowering the energy threshold for avalanche formation. We present simulation results confirming this effect and report on experimental verification of the assumed low-temperature thermal properties of two representative SMM crystals, Mn$_{12}$-acetate and Mn$_{32}$. Together, these efforts extend this pathfinder program toward the realization of SMM-based detectors with controlled material properties and enhanced dark matter sensitivity.

We present an extraction and phenomenological study of genuine twist-three parton distribution functions (PDFs) through a global QCD analysis of the structure function $g_2$ in deep-inelastic scattering (DIS), its moment $d_2$, and single-/double-spin asymmetries in semi-inclusive DIS (SIDIS). The unified description of these processes clearly confirms the universality of genuine twist-three PDFs and the validity of QCD factorization theorems at this subleading level. The extraction is accompanied by several validation tests, tests of sum rules, comparisons with earlier literature, computation forces and average transverse momentum of partons. As a by-product, we also obtain Sivers and worm-gear-T transverse momentum dependent (TMD) distributions, which allows us to construct a tomographic picture of the proton based on substantially broader data.

In this study, we combine residual correction with the physical treatment of charmonium masses within a Quantum Chromodynamics (QCD) motivated potential-model framework via variational Monte Carlo (VMC). The aim is not to propose a new charmonium spectrum, since this sector has already been examined extensively through different potential models. Instead, the main objective is to evaluate how effectively a Machine Learning (ML) correction can improve a VMC baseline when both stages are built on the same screened funnel potential. In this workflow, the screened interaction provides the physical input and determines the underlying mass structure. The proposed run uses the variational method via VMC and deterministic eigenvalue diagonalization in the process of mass calculation. In total, ten charmonium states are calculated, among which seven use the experimental reference masses. The VMC step generates a large set of configurations and local energy estimators that are fed into the neural network (NN) residual corrector at the sample level, while the corrections and experimental uncertainties are collected at the state level. It then learns the residual difference between the raw physics baseline and the internal reference targets. This hybrid procedure reduces the systematic mass offset of the raw calculation across the studied states. For the experimentally verified seven states, the correction reduces the MAE from $438.1~\mathrm{MeV}$ to $24.1~\mathrm{MeV}$, corresponding to a $94.5\%$ reduction. These results show that ML can serve as a residual-correction layer for potential-model spectroscopy.

We revisit the zero-transcendentality sector of bosonic string effective amplitudes with spin-1 external states, conjectured to correspond to a mass-deformed $(DF)^2$ theory, known as the $(DF)^2{+}\text{YM}$ theory. Imposing gauge invariance, locality, and cyclicity under minimal assumptions uniquely fixes a set of dimension-raising operators and leads to a recursive construction of amplitudes from Yang-Mills amplitudes in the $α'{\to}0$ limit. At finite $α'$, certain derivative operators dressed with gauge invariant and $α'$-dependent factors, what we call $\textit{inverse operators}$, reconstruct the full bosonic string effective amplitudes, yielding compact expressions that universally factorize into tachyon-pole coefficients times Yang-Mills-Scalar amplitudes. This structure holds at arbitrary multiplicity and also extends to the amplitudes of the pure $(DF)^2$, $(DF)^2{+}φ^{3}$ and $(DF)^2{+}\text{YM}{+}φ^{3}$ theories.

We present the subensemble acceptance method 3.0 (SAM-3.0), which corrects cumulants of an observable measured in a subsystem of a large system for the effect of exact global conservation of multiple charges. The required input is the set of joint grand-canonical cumulants of the acceptance observable with the total event charges, from which the canonical cumulants follow algebraically via a closed recursion based on (multivariate) partial exponential Bell polynomials. The framework accommodates any number of observables, including non-conserved quantities such as net protons, and any number of simultaneously conserved charges, including the total energy, which yields the microcanonical ensemble. The mapping contains SAM-1.0 and SAM-2.0 as special cases and, unlike SAM-2.0, reproduces the exact binomial-acceptance limit. We also derive the leading finite-size corrections from the saddle-point expansion. We apply the method to update the hydrodynamics-based non-critical baseline (Hydro-EV) for net-proton cumulants at RHIC-BES energies, finding a refined baseline that agrees with direct canonical Monte Carlo sampling and stays close to the earlier SAM-2.0 result. We further validate the formalism against direct Monte Carlo sampling with exact simultaneous conservation of baryon number, electric charge, and strangeness, including hadronic-afterburner effects.

We study primordial black hole (PBH) formation in axion $U(1)$ inflation using lattice simulations. In axion $U(1)$ inflation with a bumpy potential, the curvature perturbations can be enhanced in a narrow range of wavenumbers, potentially leading to PBH formation. After confirming that our lattice simulations reproduced the known curvature power spectra for chaotic inflation and simple axion $U(1)$ inflation, we calculate the curvature power spectrum in the bumpy axion inflation model in the strong backreaction regime. We find that large curvature perturbations are generated, which lead to PBH production with an abundance sufficient to account for dark matter.

We show that two paradigmatic systems, the planar Kepler--Coulomb problem and the Landau problem on the hyperbolic plane $H^2$, are connected by a common one-dimensional mediator: the Morse Hamiltonian. On the Kepler side, a Liouville transformation and coupling-constant metamorphosis turn the radial dynamics into the Morse problem, with the Kepler polar angle becoming the Morse evolution parameter. On the Landau side, horocyclic reduction of the hyperbolic magnetic dynamics gives the same Morse Hamiltonian, with a quantum half-density correction. Consequently, the radial Kepler problem and the fixed-horocyclic-momentum sectors of the hyperbolic Landau problem are mapped to one Morse spectral problem, relating their bound spectra, continuum thresholds, resonances and scattering data. We further show that the Landau time evolution has a Kepler-conic form and reduces to the bound, threshold and scattering trajectories of the Morse system. The resulting dictionary connects Kepler conics with magnetic circles, horocycles and hypercycles, and turns the magnetic $SL(2,\mathbb R)$ symmetry of the Landau problem into the spectrum-generating algebraic structure of the Morse system.

Non-standard cosmological histories with epochs of early matter domination (EMD) arise in various top-down models of the early universe. Typically, in the latter stage of EMD, temperature decreases more slowly than in a radiation dominated universe because of entropy generation from decay of the species that drives EMD. A time-dependent decay rate can significantly modify this picture and even lead to a period with increasing temperature. We study non-monotonic temperature evolution in a well-motivated scenario of EMD with a time-dependent decay rate that can give rise to multiple first-order phase transitions in both cooling and heating phases. The spectra of the ensuing gravitational waves (GW) exhibit characteristic features such as multiple peaks and a distinct behavior at high frequencies. These features allow us to determine the phase transition temperature as well as the reheating temperature at the end of the EMD. The future GW detectors can therefore provide a probe for the new physics and a window to the early thermal history.

The production of $Λ_c^+$ baryons in Pb+Pb collisions at $\sqrt{s_{NN}} = 5.02$ TeV is investigated within the quark recombination framework, including the energy loss of light and charm quarks inside the hot and dense medium. The model simultaneously describes the transverse momentum ($p_T$) spectra of $Λ_c^+$ baryons, $Λ_c^+/D^0$ yield ratio, which is attributed to the dominance of quark recombination mechanism in the Quark-Gluon Plasma (QGP), and the second harmonic coefficient of $Λ_c^+$ baryons with emphasis on the effects of minijets on the azimuthal anisotropy. Furthermore, we extend the theoretical calculation to Pb+Pb collisions at $\sqrt{s_{NN}} =2.76$ TeV and make predictions for $Λ_c^+$ baryons and $Λ_c^+/D^0$ yield ratio. The simultaneous description of the yield, baryon-to-meson ratio, and azimuthal anisotropy further validates that the recombination model is an effective hadronization mechanism in heavy-ion collisions.

X-ray interband time lags are key diagnostics of jet physics and are frequently detected in high-frequency peaked BL Lac (HBL) objects at different epochs with various X-ray telescopes. In this work, we theoretically investigate Fourier lag-frequency spectra using a generic one-zone leptonic model incorporating the stochastic turbulent acceleration (STA), which plays a crucial role in shaping the emitted photon spectra. We demonstrate that the competition between STA, radiative cooling, and escape processes not only gives rise to two well-defined time-lag regimes: hard/positive and soft/negative lags, but also reveals the existence of a transition between the two regimes. Our results indicate that time lags in the transitional and soft-lag regimes can be clearly amplified and modified by STA's suppression of high-energy electron cooling, and nonlinear synchrotron self-Compton (SSC) cooling can further amplify the emergence of time lags. We conclude that the adopted model offers a unifying quantitative framework for interpreting the diverse time-lag signatures observed in the X-ray flares of HBLs. Additionally, SSC cooling effects can account for the relatively large lags observed in TeV-bright flares, as well as the observed trend between lag amplitude and flare duration: the larger the flare duration, the larger the lag.

A study of the production of pairs of Z bosons (ZZ) and of the associated production of a Z boson and a Higgs boson (ZH) in final states containing two b quarks and two tau leptons (bb$ττ$) is presented. The analysis is based on proton-proton collisions collected at $\sqrt{s}$ = 13 TeV by the CMS experiment at the LHC, corresponding to an integrated luminosity of 138 fb$^{-1}$. The nonresonant analysis targets the standard model ZZ and ZH processes in the bb$ττ$ final state, motivated by the prominent role of this channel in searches for nonresonant Higgs boson pair production. The resonant searches target physics beyond the standard model, probing heavy spin-0 resonances X that decay into ZZ and spin-1 resonances Z' that decay into ZH, with masses in the 0.2$-$5 and 0.5$-$6 TeV ranges, respectively. Upper limits at 95% confidence level are set on the product of production rate and branching fraction $σ$(X)$\mathcal{B}$(X $\to$ ZZ), ranging from 300 pb to 24 fb, and $σ$(Z')$\mathcal{B}$(Z' $\to$ ZH), ranging from 0.4 pb to 12 fb. These are the first measurements to probe the ZZ/ZH $\to$ bb$ττ$ processes. No deviation from standard model expectations is observed.

High Energy Physics experiments at flagship colliders produce and process some of the biggest datasets on Earth, with the current generation of flagship experiments at the Large Hadron Collider producing more than a tenth of the world's total internet traffic every second. Moreover the quantities of data produced have increased exponentially over the past decades and this trend shows no sign of slowing down. In parallel, the use of picosecond timing is becoming more common in HEP detectors, enabling qualitatively new approaches to real-time processing and selections. I review the planned introduction of precision timing information into the upcoming upgrades of the CMS, ATLAS, and LHCb experiments. I discuss the ways in which the combination of timing and networking technology may enable future detectors to be designed as triggereless from the ground up, and reflect on the physics benefits of such a paradigm shift for the field.

The Hammer software library provides fast and efficient reweighting of large simulated datasets containing semileptonic $b$-hadron decays to any beyond Standard Model (BSM) theory, or to any form-factor description of the hadronic matrix elements. By enabling reweighting to a different underlying theoretical model after the computationally-expensive detector simulation step has already been completed, Hammer permits experimental analyses to employ forward-folding fitting strategies to recover underlying physical parameters without biases, or to properly characterize theory systematic uncertainties. This publication details upgrades to Hammer functionalities and its application programming interface (API) for version 2.x, and also provides associated documentation of the library's structure, syntactical conventions, and code flow. Substantial optimization of Hammer's internal tensor library now enables computational complexity to generically scale almost linearly with amplitude tensor rank times size rather than as a quartic, enabling reweighting into very high dimension spaces, such as the product of BSM Wilson coefficient and form factor parameter linear spaces, while also retaining Monte Carlo uncertainties. Updated Python bindings are implemented with bijective correspondence to the C++ interface, allowing full access to library functionalities.

Permanent electric dipole moments (EDMs) of nucleons, nuclei, atoms, and molecules are among the most sensitive probes of CP violation beyond the Standard Model and are intimately connected to the strong CP problem and the origin of the matter-antimatter asymmetry of the universe. This review presents the theory of EDMs from the bottom up, tracing the chain of connections that links CP-violating interactions at level of elementary particles to observable EDMs across a wide range of systems. Starting from a general CP-odd effective Lagrangian at the quark-gluon level comprising the QCD theta term, quark EDMs and chromo-EDMs, the Weinberg operator, and CP-odd four-fermion interactions, I show how chiral perturbation theory organizes the nonperturbative QCD dynamics into a small set of hadronic low-energy constants, whose relative sizes are determined by the chiral representation of the underlying source. These hadronic interactions feed into calculations of nuclear EDMs and Schiff moments, which in turn enter atomic and molecular structure calculations that connect to experimentally accessible observables in diamagnetic and paramagnetic systems. Special attention is given to the recently identified sensitivity of paramagnetic systems to hadronic CP violation, which opens a new and relatively unexplored window on the quark-gluon sector. The complementarity of the full EDM portfolio including the neutron, light nuclei, atoms, and molecules, and the role of theory in disentangling the underlying source of CP violation is discussed throughout.

We compute the matching of QCD quark-antiquark currents onto the set of the two-particle and three-particle two-jet operators in soft-collinear effective theory (SCET) at next-to-leading order (NLO) in the perturbative QCD series, including for the first time operators up to second order in the power expansion in the transverse momentum over energy. These results contribute to the ongoing programme of computing power corrections and summing power-suppressed logarithmically enhanced terms for event shapes in the two-jet region and deep-inelastic scattering in the Bjorken-$x\to 1$ limit. The three-particle operators depend on the partonic momentum fractions of two partons moving into the same direction. When one of the momentum fractions approaches zero, the coefficient functions are shown to satisfy endpoint factorization relations, which allow for a consistent cancellation of endpoint singularities among various terms in the complete factorization formula for power corrections.

We investigate the electromagnetic interactions of dipole dark matter (DM) within an effective field theory framework, considering both standard and non-standard cosmological scenarios. We first study the prospects of DM production via both the freeze-out and freeze-in mechanisms within the standard radiation-domination. We then investigate how the viable DM parameter space is modified in a non-standard cosmological scenario due to entropy dilution during reheating. Existing constraints on the parameter space are discussed, and we highlight the discovery potential of future direct detection experiments to probe these scenarios. We further investigate the implications of neutron star heating for dipole DM. Due to the momentum-dependent nature of the interaction, dipole DM is captured efficiently by neutron stars, thereby making neutron star heating a sensitive probe of the dipole DM parameter space.

We present a canonical quantization of Jackiw-Teitelboim gravity with finite Dirichlet boundary conditions, using the geodesic length between the two boundaries and its conjugate momentum as reduced phase space variables. The dynamics is recast as the scattering problem of a nonrelativistic particle in a repulsive Pöschl-Teller potential, naturally embedded within a hyperbolic reduction of the $\mathfrak{sl}(2,\mathbb{R})$ Casimir. We obtain exact wavefunctions of the universe and the disk partition function, interpreted as a transition matrix element between states of vanishing bare length. In the asymptotic limit, the theory reduces to Liouville quantum mechanics and reproduces the standard Schwarzian spectral density. At finite cutoff, however, the spectral measure exhibits genuinely nonperturbative corrections, absent in existing $T\bar T$ treatments. We also obtain closed form expressions for thermal two-point functions in terms of Wilson functions and propose diagrammatic rules for time- and out-of-time-ordered four-point functions. We further address the issue of the branch cut singularity of the quasi-local energy and propose a UV completion of the model in which the Brown-York charge is analytically continued beyond the black hole horizon. This continuation naturally extends the scattering problem to configurations that foliate the black hole interior.

Neural network surrogates for LHC scattering amplitudes require trustworthy uncertainty estimates, a challenging task given the non-Gaussian systematics. We target it using conformal prediction, a distribution-free post-processing to complement trained surrogates with calibrated uncertainties. We find that standard conformal predictions struggle to provide locally calibrated uncertainties. This leads us to introduce FALCON, a novel conformal prediction method that learns locally calibrated confidence intervals. Our simple examples illustrate the power of distribution-free uncertainty quantification for ultra-fast event generation at the LHC.

We propose that the Krylov basis gives a semiclassical representation of dynamics in general large-$N$, complex, many-body systems. As a probe of this semiclassicality, we study the growth of Wigner negativity -- a measure of the complexity of classical simulation -- under time evolution in Krylov space in several solvable models. We begin with 2d CFTs, initially in either the vacuum or the thermofield double state on a line excited by a primary operator. In both cases, Wigner negativity remains an $O(1)$ constant and does not grow at late times, indicating approximately classical dynamics in the Krylov basis. We then study random matrix theory with the maximally entangled state between two copies as the initial state. For general one-cut matrix models, we argue that Wigner negativity in the Krylov basis grows as $t^{1/2}$ at large $O(1)$ times but does not scale with the Hilbert space dimension, thus indicating semiclassical dynamics in Krylov space. Finally, in the double-scaled SYK model, we find an approximately classical phase (constant negativity) at early times and a semiclassical phase ($t^{1/2}$ growth) at late times. In all these examples, Wigner negativity either remains constant or grows slowly, demonstrating emergent semiclassicality of dynamics in Krylov space.

Superluminal neutrino propagation, induced by Lorentz-invariance violation (LIV), is strongly constrained by vacuum pair emission, $ν\to ν+ e^- + e^+$, a process ordinarily forbidden, which rapidly degrades the energy of high-energy neutrinos. Consequently, observable neutrino time delays are often preferentially associated with subluminal propagation, prompting LIV interpretations of claimed time delays between high-energy cosmic neutrinos and gamma rays. However, this expectation is at odds with the observed stability of high-energy electrons. The same Lorentz-violating correction associated with subluminal neutrino propagation opens the overlooked complementary decay channel $e^- \to e^- + ν+ \barν$, leading to electron instability. We derive constraints on LIV from recent observations of TeV--PeV astrophysical electrons. These electron stability limits rule out LIV invoked to explain delays of high-energy cosmic neutrinos. Consequently, neutrino time delays are constrained on both the superluminal and subluminal sides. Therefore, observable delays require either purely astrophysical origins, a realization of LIV that affects all particle species equally, or physics beyond the standard effective-field-theory framework.

When the axion mass is generated during a first-order phase transition and becomes non-vanishing only inside expanding true-vacuum bubbles, the standard picture of misalignment production is qualitatively modified. Using lattice simulations in an expanding universe, we study dark matter production within such a framework and identify two distinct regimes. For rapid transitions, the onset of oscillations is delayed until bubble percolation, enhancing the relic abundance. For slower transitions, spatial gradients generated by expanding bubbles suppress the effective misalignment angle through the bubble misalignment mechanism. We derive a semi-analytical expression for the relic density that provides a unified description of both regimes and accurately reproduces the simulation results. Finally, we show how this mechanism also modifies isocurvature perturbations and the small-scale matter power spectrum, with important implications for axion minicluster formation.

We study the fusion of two conjugate conformal line defects on the sphere. At small separation, their spectrum is governed by a universal Fusion Master Equation. Below a critical coupling, the fused defect has two conformal fixed points; at criticality, they collide and move into the complex plane, producing walking RG behaviour. Although individual energy levels then drift with the UV scale and are scheme dependent, the $SL(2,\mathbb{R})$ Casimir continues to commute with the Hamiltonian below that scale. This organises the spectrum into conformal families and fixes a universal, scheme-independent density of states. We derive this structure in the planar ladder model and obtain an exact finite-coupling description of conjugate $1/2$-BPS Wilson-line fusion in planar ${\cal N}=4$ SYM using the Quantum Spectral Curve. We test our results against perturbation theory and semiclassical string theory.

We introduce a field-theoretic algorithm to find the $SL(2,\mathbb{Z})$ duality web of 3d $\mathcal{N}=4$ circular quiver theories with unitary gauge groups, extending the algorithm for linear quivers. Although circular and linear quivers share the same local structure, the circular topology requires additional ingredients, which we formulate in terms of topological and baryonic QFT blocks, together with new $SL(2,\mathbb{Z})$ duality moves acting on them. For good circular quivers, this provides a field-theoretic derivation of mirror symmetry and extends it to the full $SL(2,\mathbb{Z})$ duality web. We then study bad circular quivers, distinguishing between local badness, associated with under-balanced gauge nodes, and global badness, arising from the circular topology itself. In particular, we analyze the magnetic and electric dual frames of globally bad circular quivers and provide additional evidence for the proposed duality by matching the Higgs branch index with the dual Coulomb branch index. The latter exhibits a structure reminiscent of permutation-group gauging and reveals a refined relation to the ADHM quiver, flowing to the $\mathcal{N}=8$ infrared fixed point.

We initiate the study of networks of 10d string theories connected by junctions implied by the cobordism conjecture. Focusing on the recently constructed junction of the three 10d non-tachyonic heterotic theories, we generalize its $(0, 1)$ heterotic worldsheet description to construct arbitrary networks. For one-dimensional networks, we formulate their topology in terms of graph theory and provide a simple worldsheet realization for general graphs. We then extend our analysis to higher-dimensional networks, describing e.g. nucleation in a theory of bubbles of pairs of other theories. We also discuss compact configurations, which define a novel class of compactifications in which different sectors propagate on different compact spaces, in a way reminiscent of compactifications on quantum geometries like $S^1 \vee S^1$.

The nature of $φ(2170)$ remains open. We simulate its production in $e^+e^-$ collisions at $\sqrt{s}=4.95$ GeV using PACIAE 4.0, which sequentially generates the final partonic state (FPS) and the final hadronic state (FHS). While previous studies have interpreted $φ(2170)$ as an $ss\bar{s}\bar{s}$ or a $u\bar{u}s\bar{s}$ state, the $U(1)$ anomaly coupling allows non-strange quarks to couple to a vector $s\bar{s}$ component via soft-gluon interactions. This motivates us to also explore the $d\bar{d}s\bar{s}$ tetraquark configuration. In addition, we consider $φ(2170)$ as an excited strangeonium state, an $s\bar{s}g$ hybrid state, a $\barΛΛ$ bound state, and a $φK^+K^-$ resonance state. The strangeonium, hybrid, and tetraquark candidates are formed by coalescing their constituent partons in the FPS using the dynamically constrained phase-space coalescence model. The $\barΛΛ$ and $φK^+K^-$ states are produced via recombination of their constituent hadrons in the FHS. We calculate the orbital angular momentum quantum number of each candidate in its rest frame and perform spectral classification. Given $J^{PC}=1^{--}$, $φ(2170)$ can be interpreted as a $D$-wave $s\bar{s}$, a $P$-wave $s\bar{s}g$, a $P$-wave $u\bar{u}s\bar{s}/d\bar{d}s\bar{s}/ss\bar{s}\bar{s}$, an $S$-wave $\barΛΛ$, or an $S$-wave $φK^+K^-$ state. The yields of the $D$-wave $s\bar{s}$, $P$-wave $s\bar{s}g$, $u\bar{u}s\bar{s}$ and $d\bar{d}s\bar{s}$ states are of order $10^{-4}$; those for the $S$-wave $\barΛΛ$ and $φK^+K^-$ states are of order $10^{-5}$; while the $P$-wave $ss\bar{s}\bar{s}$ yield is of order $10^{-6}$. Moreover, significant discrepancies are observed in the rapidity distributions and the $p_T$ spectra among the various candidates. These discrepancies could serve as valuable criteria for unraveling the nature of $φ(2170)$.

We investigate the charmonium-nucleon interaction, focusing on its connection with the internal structure of the nucleon encoded in the gravitational form factors. To describe this interaction, we employ an effective potential based on the QCD multipole expansion within the leading chromoelectric dipole approximation. In this framework, the potential is expressed in terms of the energy and pressure distributions inside the nucleon. We first construct these distributions from the gravitational form factors fitted to lattice-QCD data. The remaining model parameters are then fixed by requiring that the resulting $J/ψ$-$N$ potential reproduce the HAL QCD potential outside the short-distance region, as well as the scattering phase shift estimated from the HAL QCD data. Based on this potential, we evaluate the $J/ψ$-$N$ correlation function and further investigate the $ψ(2S)$-$N$ system. We then examine the sensitivity of these correlation functions to the nucleon $D$-form factor.

Bayesian analyses of the full-shape clustering of Dark Energy Spectroscopic Instrument (DESI) Data Release 1 (DR1) exhibit prior-volume projection effects, whereby weakly constrained nuisance parameters of the Effective Field Theory of Large Scale Structure (EFTofLSS) shift marginalized cosmological posteriors away from the posterior maximum. We reanalyze DESI DR1 power spectrum multipoles using two complementary mitigation strategies: (i) nonlinear orthogonalization to decorrelate nuisance and cosmological parameter priors, and (ii) a fully reparameterization-invariant Jeffreys prior over all EFTofLSS coefficients, evaluated on-the-fly via closed-form Jacobians. Including data from DESI, Big-Bang Nuclesynthesis and a constraint on $n_{\mathrm{s}}$, baseline priors lead to multi-$σ$ projection in the Hubble parameter $H_{0}$ and dark energy equation of state parameters $w_{0}$ and $w_{a}$; the Jeffreys prior successfully recenters these posteriors to enclose the maximum a posteriori estimate within the 68\% credible regions, demonstrating clear mitigation of projection effects for these late-time expansion parameters. A hybrid Jeffreys+baseline-Gaussian configuration controls residual over-broad tails in the physical cold dark matter density $ω_{\mathrm{c}}$ while preserving the volume correction, and is our favoured approach. We compare the credible intervals derived using our methodology to those obtained using Halo Occupation Distribution (HOD)-informed priors and to confidence intervals derived using frequentist profile likelihood analyses, finding agreement in both central values and degeneracy directions in the $w_{0}$--$w_{a}$ plane. This demonstrates that, once projection effects are properly controlled, we can make robust inferences about the late-time cosmological expansion independent of the statistical framework adopted.

The closest giant spiral, the Andromeda galaxy (M31), shows compelling evidence for a recent, gas-rich major merger event. Pronounced substructures in its inner halo and a kinematically hot stellar disc, whose star formation history shows a widespread star formation episode 2.5 Gyr ago, are telltale evidence that may be directly linked to a major (mass ratio 1 to 4) merger event that took place 2-4 Gyr ago. Spectroscopy of resolved giant stars in the remote outskirts of M31's disc revealed a vast extended structure that rotates with a circular velocity close to the HI gas. In addition, the spatial distribution and significant prograde rotation of two distinct, compact groups of globular clusters (GCs) in the disc outskirts are unusual for typical inner halo GCs. We employ an available N-body hydrodynamical simulation of a major merger that reproduces the morphology of the inner halo substructures, the age-velocity dispersion relation, and the star formation history in the disc. We compare model particles with resolved tracers in the M31 disc. To examine the evolution of the progenitor M31 disc -- that appears to get stretched, distorted, and warped due to the gravitational perturbation inflicted by the major merger -- we investigate the properties of the pre- versus post-merger discs of the simulated analog. The merger transforms the disc of the progenitor galaxy, which becomes kinematically hot and asymmetric. In addition, the post-merger disc gets stretched by almost a factor of 2, and its extent spans distances greater than 40 kpc. The stellar warp in populations older than 2 Gyr is characterized by a monotonic decrease of inclination with radius, with the outer stellar distribution appearing less edge-on at larger galactic radii. These results provide a comprehensive picture of the evolution of the giant disc of M31, the closest merger-inflicted massive galaxy.

The Epoch of Reionization (EoR) is one of the most pivotal frontiers in modern astrophysics, marking the emergence of the first galaxies, stars, and supermassive black holes (SMBHs). Despite insights from the Atacama Large Millimetre/submillimetre Array and the James Webb Space Telescope, we still struggle to explain how $\sim10^{9}$ M$_\odot$ SMBHs powering luminous active galactic nuclei (AGN) already exist by $z\sim7$. The recent discovery of powerful radio emission from some of these early AGN is notable, offering new constraints on early black-hole accretion and, with the Square Kilometre Array Observatory (SKAO), the prospect of directly probing neutral hydrogen through 21-cm absorption studies. Yet progress remains slow: only a few radio-powerful AGN are known at $z>6$, far fewer than theoretical predictions suggest, raising questions about whether this reflects intrinsic properties or selection biases and incomplete spectral information. In this chapter we synthesise predictions from state-of-the-art hydrodynamical and semi-analytic simulations with observational constraints from SKAO pathfinder facilities. These models suggest the existence of a substantial, still-undetected population of radio-powerful AGN in the EoR, but show that present surveys are limited by selection biases and incomplete radio spectral information. We discuss a physically motivated strategy for identifying high-redshift radio AGN, based on broadband radio spectral energy distributions, spectral curvature, dynamical jet evolution, and radio-only redshift estimation, offering a transformative alternative to traditional empirical approaches. Finally, we justify how the sensitivity and spectral coverage of the SKAO will allow fine-frequency sampling across the 50 MHz - 15 GHz range, revolutionising our ability to identify the earliest radio-powerful AGN and probe the earliest SMBHs.

Ultra-cool dwarf stars (UCDs) are a common source of contamination in high-redshift galaxy searches as both sources are red and these early-forming galaxies can have sizes that are difficult to resolve even with space telescopes. Standard selection techniques, including photometric redshift estimation and color-color criteria, cannot fully eliminate this contamination. We develop \textbf{F}oreground \textbf{C}ontamination \textbf{E}valuator of \textbf{N}earby dwarf stars in high-\textbf{Z} photometrically selected \textbf{O}bjects (FC-ENZO), a code that predicts the number of dwarf stars misidentified as high-redshift galaxies for a given survey setup. FC-ENZO models the number of UCDs and evaluates the fraction of synthesized dwarf stars that passes user-specified selection methods. We compare two synthetic spectral energy distribution libraries and find that the ELF OWL library, which relaxes the assumption of chemical equilibrium, predicts larger contaminant fractions than the BOBCAT library, because of stronger absorption features around $ 1 $ \micron. The contamination fraction increases with metallicity and also depends on the adopted stellar number-density model. The dominant contaminants are T to early Y-type UCDs, which are most commonly misclassified as galaxies at $z \sim 8$. Comparing deep surveys from different space telescopes, we find similar overall contamination levels within the same redshift range. However, the contamination is concentrated near the limiting magnitude of each survey. At brighter magnitudes, the relative contamination is highest for HST (COSMOS), followed by Roman deep-tier survey, and JWST. Although the predicted contaminant numbers remain sensitive to model assumptions, FC-ENZO provides a practical tool for survey design and for identifying optimal fields for spectroscopic follow-up.

In the early phases of star formation, properties of prestellar cores are commonly inferred from observations of thermal dust emission and thus depend on dust properties, which must be carefully characterized. Our target, L1506C, is part of the filament L1506 in the Taurus molecular cloud. The spectral energy distributions over the whole spectral range (from 160 μm to 2 mm), built from Herschel PACS and SPIRE and IRAM-NIKA2 data, have been fitted with a modified blackbody. These data were also modelled using the 3D radiative transfer code SOC and the latest THEMIS 2 dust model using extinction observations from WIRCam at CFHT and from Spitzer as additional constraints. The MBB modeling reveals that L1506C is fragmented into two low density cores with masses smaller than their Jeans masses. The dust color temperature and the emissivity spectral index show clear anti-correlation and change in grain properties. Grains more evolved than the diffuse interstellar medium are needed to model the densest part showing that grain growth already occurs at very early stage of star formation, even before the onset of gravitational collapse.

The paper analyzes formation conditions for globular clusters (GCs) in circumgalactic clouds. The similarity between the metallicity distributions of GCs in the nearby Universe and of circumgalactic clouds is substantiated in detail over a wide range of redshifts: from \mbox{0.2} to \mbox{5.9}. The distributions of the number of circumgalactic clouds and GCs both contain a sequence of four local maxima at the metallicity values: \mbox{$[\rm{X/H}]\simeq -2.6, -2.0, -1.4,-0.5$}. The sequential enrichment of a circumgalactic cloud with a mass of $10^{8}\,M_{\odot}$ is calculated starting the extremely low metallicity \mbox{$ [\rm{X/H}] <-2.3$}, then following through the stages of \mbox{$-2.3 \le [\rm{X/H}]<-1.7$} and \mbox{$-1.7 \le [\rm{X/H}] < -0.9$} to the high metallicity \mbox{$[\rm{X/H}] \ge -0.9$}, where the boundaries of these ranges coincide with the local minima of the number of objects in the distributions. It is shown that for the reproduction of such distributions, it is sufficient that at each stage of enrichment of a part of a cloud in metals, one or more GCs with a total mass of \mbox{$3 \times 10^{6}\,M_{\odot}$} are formed. It is shown that the maximum mass of stars capable of leading to supernova explosions increases with the increase of metallicity. Possible values of this mass are calculated for the metallicities corresponding to the maxima in the distributions of clouds and GCs.

The exploration of the faint polarised radio source population is essential for interpreting the nature and evolution of magnetic fields in galaxies. While recent studies have provided insight into source counts for the $μ$Jy polarised source population at 1.4 GHz, higher frequency surveys may be more sensitive to new populations that are depolarised at lower frequencies (i.e. due to internal or external depolarisation effects). We present the deepest polarised source counts at 3 GHz to date, at an angular resolution of $1.5''$. With these relatively higher frequency observations, we aim to probe the faint polarised star-forming galaxy (SFG) population. Furthermore, through spectral modelling, we aim to provide further insight into the frequency evolution of polarised source counts. We processed the polarisation data of the VLA-COSMOS 3 GHz Large Project, one of the deepest high-resolution radio continuum surveys. We produced Stokes Q and U mosaicked channel maps. After selecting known sources in total intensity, we performed 3D rotation measure synthesis and searched for polarised emission using an empirically determined threshold. With a sensitivity of 2.6 $μ$Jy/beam in Faraday depth, we detect 65 polarised sources (51 deg$^{-2}$) above our threshold. We find that our cumulative and Euclidean-normalised source counts at 3 GHz are consistent with those in the literature at 1.4 GHz, which we attribute to the combined effect of spectral index and depolarisation in the detected sources. We detect no SFGs in our sample and derive a 2$σ$ upper limit on the density of polarised SFGs of $<2.0~\mathrm{deg}^{-2}$. This implies that significantly deeper observations will be required to readily detect this population in the SKA-era.

Mapping how the explosion properties of Type II supernovae (SNe II) relate to the properties of their progenitors can provide strong constraints for understanding the final evolutionary stages of massive stars. Type IIP SNe, linked to the explosions of single red super-giant (RSG) stars, have recently been found to require some form of interaction with circumstellar material (CSM) to reproduce the rapid rise to the plateau often seen in their light curves. In this work, we present observations and analysis of the Type IIP SN 2020bij, characterized by a slow rise to its plateau as well as high expansion velocities. We identify four other SNe IIP from the literature (ASASSN-14kg, SN 2018fif, SN 2021yja and SN 2023axu) with similarly slowly rising light curves and find that they also show high expansion velocities. Using both analytical and numerical models, all five events can be explained with weak to no CSM interaction. We therefore propose that these events constitute a new subclass of Type IIP SNe which could be associated with relatively confined CSM. Early and dense photometric coverage of future SNe IIP together with early spectroscopic observations will further map this subclass and its physical properties. Understanding such rare events could be key to constraining the diversity of late-stage mass-loss in RSGs.

We analyzed \textit{NICER} observations of the 2025 outburst of the black hole X-ray binary 4U~1630$-$47 to investigate the spectral--timing properties of its transient low-frequency quasi-periodic oscillations (QPOs) and millihertz-scale quasi-regular modulation (QRM). During the rising phase of the outburst, the QPO centroid frequency increased from $\sim 0.24$ Hz to $\sim 3.43$ Hz. Wavelet-based state separation shows that the with-QPO intervals are associated with a higher inner disk temperature and a lower \texttt{diskbb} normalization than the without-QPO intervals, while the photon index ($Γ$) shows weaker changes within the uncertainties. Near the outburst peak, the source displayed a weak QRM at $\sim 0.07$ Hz with a fractional rms amplitude of $\sim 4.7\%$, lower than that of the heartbeat state observed in 2023. Phase-resolved Hilbert--Huang analysis shows that the inner disk temperature is positively correlated with the X-ray flux, the \texttt{diskbb} normalization is anticorrelated, and $Γ$ varies only weakly. Overall, the short-timescale spectral--timing variability is expressed most clearly through the disk-related parameters. The transient QPOs are therefore consistent with short-timescale disk-related variability during the rising phase, whereas the millihertz-scale QRM may represent a weaker heartbeat-like variability mode appearing near the outburst peak.

Spatially resolved, deep HI observations from SKA precursors and pathfinders such as MeerKAT, FAST, and ASKAP have demonstrated their ability to reveal the complex interactions between galaxies and their environments. These include, but are not limited to, recent observations of the Virgo cluster showing that the hydrodynamical effects of ram pressure stripping can operate effectively at unexpectedly large cluster-centric distances. In the Fornax cluster, the discovery of long HI tails with mixed tidal-ram-pressure origins indicates the interplay between gravitational and hydrodynamical mechanisms. Similar HI features in nearby filaments and galaxy groups, where ram pressure is expected to be weak, highlight the influence of hydrodynamical processes even in low-density environments. Multi-resolution studies have further revealed signs of cold gas accretion and HI replenishment driven by tidal interactions. While highly informative, these studies remain limited to small, specific regions of the sky. With SKA-mid AA4, it will become possible to carry out deep, spatially resolved HI imaging over hundreds of square degrees, covering environments from isolated galaxies to filaments. By reaching column-density sensitivities between $1.0 \times 10^{18}$ and $\sim 1.0 \times 10^{19}~\mathrm{cm^{-2}}$ at physical resolutions of $\sim$10 and $\sim$1 - 2 kpc, respectively, and by enabling sensitive, contiguous observations of wide areas within short integrations, SKA-mid AA4 will allow the construction of large, statistically representative samples of galaxies and detailed studies of environmental mechanisms operating across the full range of these less-studied environments at resolved scales.

We investigate black hole accretion in self-interacting dark matter (SIDM) halos using a self-gravitating fluid model with thermal conduction. We develop a robust one-dimensional spherically symmetric hydrodynamic code based on an operator-splitting finite-volume method. Simulating both Singular Isothermal Sphere (SIS) and Navarro-Frenk-White (NFW) profiles, we find that black hole growth is regulated by the competition between gravity-driven inflow and SIDM heat transport. Our results demonstrate that an SIS-like environment facilitates rapid accretion, allowing a $100\,\mathrm{M_{\odot}}$ seed to grow to $10^4\,\mathrm{M_{\odot}}$ within $2\,\mathrm{Myr}$. Furthermore, we show that larger initial black hole masses, steeper density profiles, and higher scattering cross sections significantly enhance the accretion rate. This study provides a comprehensive fluid-dynamical picture of black hole growth in SIDM halos.

Context: In collisionless shocks, energetic particles can carry sufficient pressure to modify the upstream plasma and the shock structure itself, a regime often invoked in theories of cosmic-ray acceleration but rarely observed in the heliosphere. Aims: We find and characterize {interplanetary} IP shocks where energetic particles dynamically dominate the upstream pressure. Methods: We analyze IP shocks observed by Solar Orbiter inside 1 au and compute the energetic particle pressure $P_{EP}$ from proton measurements above 10\,keV, comparing it with the upstream thermal $P_{Th}$ and magnetic $P_{B}$ pressures. Results: We identify four shocks for which $P_{EP} \geq P_{Th} + P_B $. These events correspond to strong and fast shocks in the high-Mach-number tail of the Solar Orbiter shock population. In several cases the $P_{EP}$ increase coincides with a decreasing upstream bulk flow speed in the shock frame, and the resulting particle-mediated foreshocks extend up to $\sim10^5$ {ion inertial lengths} $d_i$. The extent of such energetic particle dominated region depends on shock geometry. Conclusions: These observations provide evidence that accelerated particles can dynamically modify interplanetary shocks. They highlight the importance of the coupling between energetic particles, upstream fluctuations, and shock structure for understanding particle acceleration at collisionless shocks.

We implemented a data acquisition system for 21 CentiMeter Array (21CMA), enabling baseband observations targeting pulsars and fast radio bursts. Based on the Radio Frequency System-on-Chip (RFSoC) platform, the new backend is capable of instantaneously covering the effective bandwidth from 50 to 350 MHz, with multi-board synchronization achieved at the timescale of the sampling clock. We observed PSR B0329+54 with a single station to verify the signal path integrity; then solved phase relations of multiple station pairs using bright persistent radio sources like Cas A and Cyg A; using these phase solutions, a multiple-station coherently beamformed observation of PSR B0329+54 was carried out, showing a signal-to-noise ratio of 699.09 for a 2.5-hour observation with eight stations, opening up a possibility of tied-array low-frequency pulsar observations on 21CMA.

We present the observing campaign of the afterglow of GRB241025A, a gamma-ray burst (GRB) whose prompt emission has been simultaneously detected by Swift, Einstein Probe, Fermi/GBM, SVOM, Konus-Wind and VZLUSAT-2 3U CubeSat. Our multi-wavelength campaign comprises radio, near-infrared, Optical and X-ray observations. The afterglow was clearly detected in all bands. We performed a semi-empirical fit of the data, showing that the afterglow behaviour can be reasonably reproduced by a single component, i.e. an ultra-relativistic shock. However, the results from the semi-empirical fit are inconsistent with the predicted evolution from the standard afterglow model in the slow cooling regime. Specifically, we found that at early times the synchrotron self-absorption frequency $ν_a$ should be at higher frequencies with respect to the ones sampled by our campaign, in order to explain the observed colour evolution in radio, namely the spectral evolution in time. To reconcile the prediction from the standard model with the observed data set, we fit the observations with a semi-analytical model, including a multiplicative factor $τ_{enh}$ to the optical depth which, in turn, artificially increases $ν_a$. We found that the radio colour evolution, together with the near-infrared, optical and X-ray emission, can be described reasonably well by a forward shock from a structured jet, provided that the optical depth in the shocked material is enhanced by a factor $τ_{enh}=500$. We suggest that such enhancement in the optical depth can result from a population of cold electrons in the downstream material, i.e. electrons that were not accelerated by Fermi I process at the shock front, in agreement with the theoretical expectations previously reported in the literature. Overall, our work underscores the importance of systematic, multi-frequency, multi-epoch radio follow-ups of these extreme events.

Accurate distance measurements to supernova remnants (SNRs) are crucial for understanding their physical properties, evolutionary processes, and role in the Galactic interstellar medium (ISM) cycle. In this study, we apply for the first time to the southern Galactic plane a distance determination method that utilizes CO emission data from the Mopra survey to identify molecular clouds (MCs) interacting with SNRs. By combining this with extinction-distance profiles from the DECaPS three-dimensional (3D) extinction map, we directly measure the distances to the associated MCs, thereby obtaining precise distances to the remnants. To overcome the extinction-missing bias in extremely dense regions where the 3D map suffers from a deficit of background stars, we supplement our analysis with two-dimensional (2D) extinction maps as cross-validation. Applying this method, we have derived precise distances for nine SNRs: G290.1-0.8 (7.32+0.60/-0.47 kpc), G292.2-0.5 (10.85+0.43/-0.68 kpc), G296.1-0.5 (4.59+0.18/-0.19 kpc), G296.8-0.3 (8.74+0.40/-0.29 kpc), G298.6-0.0 (6.50 +/- 0.21 kpc), G312.4-0.4 (3.60+0.19/-0.23 kpc), G332.4-0.4 (2.66+0.23/-0.15 kpc), G335.2+0.1 (2.76+0.37/-0.31 kpc), and G353.6-0.7 (1.81+0.18/-0.14 kpc). Additionally, we established a robust lower distance limit of 1.34 kpc for G351.7+0.8.

We present meer21cm, a comprehensive python package for cosmological data analysis of single-dish HI intensity mapping surveys. This package is simple to use, with a modularised code structure designed for interactive usage. meer21cm is designed for data analysis, with particular focus on the UHF-band observation of MeerKAT Large Area Synoptic Survey (MeerKLASS). We explicitly impose meer21cm to be survey-oriented, ensuring consistent modelling of observational effects in the clustering power spectrum with the survey specifications and data analysis choices. meer21cm covers a large range of data analysis procedures post calibration, including data read-in, foreground cleaning, power spectrum estimation, mock simulation, transfer function corrections and parameter inference. It handles both meer21cm intensity maps and overlapping galaxy catalogues, allowing for multi-tracer and cross-correlation analysis between MeerKLASS and optical galaxy surveys. Tested with a simulated survey of ten $750\,$deg$^2$ sky patches in the redshift sub-band $0.6\,{<}\,z\,{<}\,0.8$, the meer21cm pipeline achieves per-cent accuracy in the power spectrum estimation for $k \in [0.02, 0.2]\,{h{\rm Mpc}^{-1}}$, with deviations $\lesssim 0.5σ$ between the mock and the model power spectra, where $σ$ is the signal variance. The meer21cm package is publicly available and easy to install, with a comprehensive documentation website at https://meer21cm.readthedocs.io

We report the discovery of Gaia23bra b, the first gravitationally bound microlensing planet detected by the Transiting Exoplanet Survey Satellite (TESS). Initially flagged as a single-lens event by the Gaia Science Alerts system, Gaia23bra was serendipitously observed by TESS over two consecutive sectors. During those TESS sectors, the light curve of the event displayed caustic-crossing features characteristic of a binary-lens event. Joint modeling of Gaia and TESS photometry with pyLIMA, supplemented by stellar parameter inference using pyLIMASS, suggests a K dwarf ($M_L = 0.79^{+0.19}_{-0.17}\,M_\odot$) hosting a Jovian planet with $M_P = 1.63_{-0.38}^{+0.42}\,M_{\rm Jup}$ at a projected separation of $a_{\perp,\min} \approx 4.8\,\mathrm{AU}$. This result underscores the synergy between high-cadence photometry and long-baseline monitoring for robust microlensing characterization. Its location along the Galactic Plane highlights TESS's unexpected capacity for microlensing science through its all-sky coverage and its potential to detect planets in regions beyond the Galactic Bulge.

Post-starburst (PSB) galaxies, identified by strong Balmer absorption and weak nebular emission, provide a key laboratory for studying rapid quenching. Using the final data release of the SDSS-IV MaNGA survey, we follow the traditional PSB selection criteria of Chen et al. (2019) and develop a new method to identify regions that simultaneously exhibit PSB features and nuclear activities (AGN-PSBs). Our final sample comprises 48 AGN-PSBs, 92 central PSBs (CPSBs), 89 ring-like PSBs (RPSBs), and 828 irregular PSBs (IPSBs). We find the global and spatially resolved properties of CPSBs and RPSBs are consistent with the results of Chen et al. (2019). In this work, we focus on the properties of AGN-PSBs, comparing them with CPSBs, RPSBs, and control galaxies. Similar to CPSBs and RPSBs, AGN-PSBs show positive $\mathrm{D}_{n}4000$ gradients relative to negative $\mathrm{D}_{n}4000$ gradients of their controls, which indicates younger stellar populations in the central region than that in the outskirt. Among the three sub-types, high-mass CPSBs (H-CPSBs, with $\log(M_{*}/M_{\odot})>9.5$) display the highest incidence of merger remnants and gas--star kinematic misalignment, consistent with a merger/interaction-dominated origin. AGN-PSBs and RPSBs, however, show lower and comparable fractions of merger remnants and gas--star kinematic misalignment, favoring less violent external mechanisms. Based on radial profiles of mass-weighted age and $V_{\rm star}/σ_{\rm star}$, we suggest that RPSBs can evolve into AGN-PSBs, whereas H-CPSBs likely follow a distinct evolutionary pathway. The existence of RPSBs and IPSBs also indicates that AGN feedback is not a necessary condition for the formation of PSB.

3I/ATLAS, the third confirmed interstellar object discovered in the Solar System, provides a unique opportunity for targeted technosignature searches. We report a periodic radio technosignature search toward 3I/ATLAS using the Five-hundred-meter Aperture Spherical Telescope (FAST) L-band multibeam receiver. To search for periodically modulated signals and distinguish center-beam-dominated candidates from multibeam radio-frequency interference, we apply canonical polyadic decomposition (CPD) to the multibeam dynamic spectra. CPD factorizes the multibeam data tensor into a set of separable components, with associated time, frequency, and beam signatures. Candidate components are then selected through periodogram and autocorrelation diagnostics. We find no credible artificial periodic radio technosignature above 0.146 W is detected from the direction of 3I/ATLAS. This search expands the range of signal types explored for this target by including periodic modulated signal, and illustrates that CPD is a promising framework for multibeam periodic technosignature searches.

The estimate of the bolometric luminosity and the radiative efficiency are two key aspects for understanding the properties of the accretion flow around a supermassive black hole (BH). In this paper, we focus on the estimate of the bolometric luminosity and the radiative efficiency of the early super-Eddington accretion flow in tidal disruption events (TDEs). Specifically, we first perform radiation hydrodynamic simulations of super-Eddington accretion flow in TDE environment, and then calculate the corresponding emergent spectra with the method of post processing for the simulation data. Based on the emergent spectra, we calculate the isotropic-equivalent X-ray bolometric correction factor $k_\mathrm{bol}$ and the radiative efficiency $η$ of the super-Eddington accretion flow. We find that both $k_\mathrm{bol}$ and $η$ are BH mass and viewing-angle dependent. $k_\mathrm{bol}$ is in the range of about a few tens to a few thousands, and $η$ is in the range of $\sim 10^{-3}-10^{-1}$ for BH mass in the range of $10^{6-7}M_\odot$ and the viewing angle in the range of $0^{\rm o}-90^{\rm o}$. Finally, we apply the derived $k_\mathrm{bol}$ and $η$ to some specific TDEs to estimate the accreted mass during an event, which can significantly alleviate the so-called missing energy problem in TDEs.

A comprehensive analysis of the energy and waiting time distributions of the bursts from FRB 20240114A detected by the Five-hundred-meter Aperture Spherical Radio Telescope between 28 January and 29 August 2024 is presented. For the full sample, its energy distribution cannot be fitted with the simple power-law (SPL),bent power-law (BPL), thresholded power-law (TPL) or Band function models, and its waiting time distribution excluding intervals shorter than 0.5 s cannot be fitted with the Poisson or Weibull models. Nevertheless, for the subsamples with more than 50 bursts in single-day observations, their energy distributions can be fitted with the BPL or TPL models, and their waiting time distributions are better described by a Weibull model. It is noted that the best-fitting BPL parameter $β$ is approximately invariant within the epochs before and after 21 March 2024, with an average of $\bar β_b = 1.006 \pm 0.074$ and $\bar β_a = 1.236 \pm 0.183$ (one standard deviation), respectively. Most subsamples from the later epoch have a smaller burst rate parameter $r$ in the Weibull model than those from the earlier epoch. The majority of bursts with $E>10^{39}$ erg occurred in the earlier epoch. The energy distributions in the high-energy range ($> 6\times10^{37}$ erg) differ significantly between the two epochs, and power-law fits to $dN/dE$ yield indices of $-1.97_{-0.02}^{+0.02}$ and $-2.34_{-0.06}^{+0.06}$, respectively. The median of the waiting time distribution of the later epoch is larger than that in the earlier epoch. These results suggest that the two epochs may be dominated by different types of bursts, possibly attributed to changes in the physical properties of the emission region.

In the Central Molecular Zone (CMZ), shocks play a key role in triggering star formation and driving chemical enrichment. The Sgr B2 complex is a prime template, hosting massive protoclusters (N, M, S) and the northern G+0.693 cloud, which exhibits shock-induced prestellar signatures. We report on G+0.633-0.0604, a newly identified shock-dominated and chemically rich cloud at the southern edge of Sgr B2, where the next star formation episodes are proposed. We characterise its physical properties and the shocks shaping it. We present analyses on CH$_3$CCH, CH$_3$CN, HC$_3$N, HNCO and several isotopologues of CO to infer the gas $T_{\rm kin}$ and density, using high-sensitivity spectral surveys from the Yebes 40m, IRAM 30m and APEX radio telescopes that covered ~100 GHz across the 31-275 GHz range. We also used 3 mm IRAM 30m mosaics (13'$\times$13') of Sgr B2 in HC$_3$N, HNCO and C$_2$H$_5$OH to probe G+0.633 environment. We identify three velocity components: a narrow main one (C1, $v_{\rm LSR}$~48.5 km/s; FWHM~10 km/s), and two broader, fainter components at higher velocities, C2 (~61 km/s; ~13 km/s) and C3 (~89 km/s; ~18 km/s), all showing similar properties ($T_{\rm kin}$~55-90 K, $N_{\rm H_2}$~(3-7)$\times$10$^{22}$ cm$^{-2}$, $n_{\rm H_2}$~(0.5-2.5)$\times$10$^{4}$ cm$^{-3}$) and extended distributions. C1 delineates G+0.633 physically and coincides with a peak in HNCO, supporting a shock-driven origin likely rooted in the cloud-cloud collision shaping Sgr B2 and also traced by C2, which extends north to G+0.693. C3 is kinematically unlinked and related to large-scale CMZ dynamics. Of the three, C1 may represent a very early protocluster phase, yet to be confirmed. G+0.633 thus emerges as a new shock-dominated CMZ cloud resembling G+0.693, providing another unique laboratory to investigate how shocks drive molecular complexity and regulate the onset of cluster formation in the CMZ.

Broad emission lines (BELs) are a defining feature of active galactic nuclei (AGNs), yet they weaken or disappear in both very low- and very high-accretion systems. These regimes are typically treated separately, and a unified physical explanation has remained elusive. Here we show that this behavior arises if line formation is governed not by the intrinsic luminosity of the central engine, but by the ionizing radiation field that survives filtering before reaching the broad-line region (BLR). In this picture, line production depends on the product of intrinsic ionizing capability and an effective transmission. Because the former increases from low accretion rates while the latter declines at high accretion rates, the effective ionizing field naturally develops a finite and non-universal window for BEL formation. This framework unifies the absence or extreme faintness of BELs in low-luminosity AGNs, LINERs, and weak-line quasars (WLQs), and accounts for the Baldwin effect and the $R_{\rm Fe}$ trend. It also necessarily implies the breakdown of standard BLR-based scaling relations in extreme accretion regimes. We show that a minimal quantitative realization reproduces this behavior across black-hole mass, accretion rate, and radiative efficiency. These results suggest that AGN emission-line phenomenology is governed by global regulation of the ionizing radiation field rather than by mere presence or condition of local gas.

We present the first application of Fourier-space coherence analysis between the lensing-reconstructed projected mass distribution and the X-ray-emitting intracluster medium to a sample of 49 observed galaxy clusters. Using publicly available HST convergence maps from the Hubble Frontier Fields, CLASH, and RELICS programs, together with Chandra X-ray imaging, we measure the scale-dependent coherence between the dark-matter-dominated surface mass density and the hot baryonic gas. We use the coherence length, l_CR, defined as the scale above which the two maps remain at least 90% coherent, as a diagnostic of cluster dynamical state. Across the sample, dynamically relaxed systems exhibit high coherence over a broad range of scales and small l_CR/r500, while disturbed and merging systems show a loss of coherence on intermediate and small scales, yielding larger l_CR/r500. The inferred coherence lengths show sensitivity to lens-model assumptions and to the heterogeneous extent of the available convergence maps. Nevertheless, the coherence signal remains physically interpretable and provides a stringent measure of dark-matter-gas alignment. Applying a conservative threshold, l_CR/r500 < 0.2, we find that only 16% of the sample is relaxed; this fraction rises to 41% for a more permissive threshold of l_CR/r500 < 0.4. Relative to previous X-ray and morphological classifications, we find a 24% disagreement, with the coherence method identifying more systems as dynamically disturbed. These results demonstrate that lensing-X-ray coherence provides a complementary, scale-resolved probe of cluster dynamical state, while highlighting the need for homogeneous, wide-field weak-lensing maps to control reconstruction and field-of-view systematics.

Fast Radio Bursts (FRBs) exhibit diverse spectro-temporal characteristics, which can probe vital propagation and source physics via Stokes polarimetry. We investigate whether the circular polarization (Stokes $V$) observed in some bursts is produced by Faraday conversion in the near-source wind of magnetars rather than being intrinsic to the source. Our calculation includes the increase in the effective mass of $e^\pm$ in the presence of the FRB wave. We find that Faraday conversion in the magnetar wind can explain the broad range of observed circular polarization in FRBs, including its frequent non-detection. Observationally derived upper limits on $V$ provide stringent constraints on the wind luminosity, magnetization, bulk Lorentz factor, and effective particle mass when ions are present. When available, frequency resolved Stokes spectra offer direct estimates of the wind environment. The Stokes parameters can undergo rapid oscillations with frequency in the high-wind/low-FRB-luminosity regime, resulting in Stokes-V depolarization. Bursts with significantly lower luminosities than typical FRBs can also develop measurable circular polarization, within the model framework. Additionally, separate zones are favored for significant circular polarization and rotation measure, when the model is applicable. The model constrains instantaneous wind parameters for several sources, including FRB 20201124A, FRB 20180301A, and SGR 1935+2154. This work represents the first instance in which properties of winds from compact objects associated with FRBs are inferred from polarization data.

X-ray polarimetry has opened a new window into understanding the physics around magnetized compact objects. IXPE detection of linear polarization from such systems has prompted a new spurt of theoretical modeling. Our study is based on the dominant paradigm that the observed polarization arises from the scattering of photons around highly magnetized systems. Our main focus is the dependence of the polarization of the scattered light on properties of the incoming light, i.e., geometry and the polarization state, and the determination of the spectral shape of the polarized light for a wide range of magnetic field strengths. We also analyze the impact of vacuum birefringence on photon polarization. We show that, generically, we expect a higher linear degree of polarization from magnetars as compared to normal pulsars, which is in agreement with IXPE observations. Under some conditions, our study helps to understand the observed degree of polarization from normal pulsars and low-magnetized neutron stars and their spectral dependence. However, we cannot conclusively explain the spectral shape of the observed polarization for magnetars using only a single component emission from scattering in a strong magnetic field. This probably points to the system being more complex, e.g., multi-component, than our study allows for. Upcoming X-ray polarimeters with broader energy coverage could probe some of our other predictions, e.g., the spectral shape of the polarized light close to the resonance frequency.

Short gamma-ray bursts (GRBs) at redshift z>2 remain exceptionally rare, yet they are crucial for tracing compact binary mergers in the early Universe and understanding their role in the production of r-process elements. GRB 061201 is an unusual and still debated event: although its optical afterglow was accurately localized, no secure coincident host galaxy was identified, and the proposed associations with nearby galaxies all require a large separation between the GRB and its birth site. In this work, we revisit GRB 061201 and argue that the observations are more naturally explained if the burst occurred within a faint F322W2~28.4 AB mag galaxy at z>2. By combining constraints from the afterglow and deep near-infrared imaging from JWST, we show that a distant origin provides a coherent explanation of the burst phenomenology. If confirmed, GRB 061201 would represent one of the most distant short GRBs known, extending the observed compact merger population to an epoch when the Universe was only about two billion years old.

In understanding the 3D structure of the Milky Way's Central Molecular Zone (CMZ), we are limited by our edge-on perspective. Towards addressing this problem, we introduce Imagery Reversion Informed by Simulation (IRIS). IRIS is a novel machine-learning code base featuring a deep convolutional neural network (CNN), which we have designed to translate edge-on observations of our Milky Way Galaxy into top-down images by training on data generated from AREPO galaxy simulations and synthetic observations of those simulations. We develop a large custom dataset on which we train our bespoke model, and then test the trained model on synthetic data to probe the potential of this machine-learning method, which we call supervised reversion. We then apply our trained model to real observations from the SEDIGISM 13CO(2-1) survey, yielding new top-down views of our CMZ. Though our SEDIGISM reversions are not fully consistent across model training runs, we posit that this lack of convergence can be alleviated by expansion of the training dataset. We argue that these results represent a strong proof-of-concept for the use of supervised reversion to decipher our CMZ's 3D structure. Crucial in generating our training dataset's 100k synthetic observations, we introduce IRIS Synthetic Observation (IRIS-SO), a new GPU-accelerated and fully differentiable code implemented in PyTorch for the non-LTE synthetic observation of spectral lines and dust. We find that IRIS-SO provides up to 10,000x speedups in comparison to the synthetic-observation code RADMC-3D. We release all the IRIS code open-source at https://github.com/bldubois/IRIS.

We investigate the effect of primordial mass segregation (PMS) in shaping the tidal tail structures of star clusters, searching for any trace of PMS on the tails at both early and late evolutionary stages. Through N-body simulations, we analyze clusters with two different degrees of PMS at various Galactocentric distances (R_G), considering two black hole retention scenarios. Our findings reveal that PMS influences early cluster expansion and the formation of tidal tails with a bottom-heavy stellar mass function, this being more pronounced at smaller R_G but diminishes over time. Primordially segregated clusters exhibit denser, unified, and longer tail structures compared to non-segregated clusters. The mean stellar mass distribution along the tails shows distinct patterns for primordially segregated and non-segregated clusters, converging at later evolutionary stages. The retention of stellar remnants has a weak impact on the mean mass distribution along the tails and on its morphology. We find that although mean mass differences persist along the tidal tails, the rate of change in primordially mass-segregated clusters eventually converges with that of non-segregated clusters, suggesting that the influence of primordial mass segregation on the tidal tails gradually diminishes over the course of cluster evolution.

The recently discovered low-ionisation broad absorption line (LoBAL) quasar GQ 1309$+$2904 is unusual due to its very broad, highly blueshifted absorption troughs and an absence of broad emission lines except for ${\mathrm{H} α}$. In this paper, we present observations of six quasars that appear very similar to GQ 1309$+$2904 in the rest-frame ultraviolet (UV). We measure the systemic redshifts of these quasars to be $z\approx$ 2.07--3.28 from detected ${\mathrm{H} α}$ emission lines. We confirm that all targets are quasars with highly blueshifted BALs possessing high-speed outflows with velocities up to $\sim 0.16\,c$, and five of them are confidently identified as LoBAL quasars. Based on ${\mathrm{H} α}$ emission, black hole masses and Eddington ratios of these quasars are $M_{\mathrm{BH}} \approx 10^{8.7}$--$10^{9.4}\,M_{\odot}$ and $L_{\mathrm{bol}} / L_{\mathrm{Edd}} \approx$ 0.14--0.34, indicating that their central black holes are very massive and active. Every quasar in our sample exhibits a very flat or reddened continuum. The spectral shapes of three objects are well-fitted by a normal quasar composite reddened by a Small-Magellanic-Cloud-like (SMC-like) extinction curve, while the other three require a steeper extinction law. Broad-band ($BVR$) polarimetry for two of the latter group (plus GQ 1309$+$2904) reveals their low polarisations, consistent with low inclination (more face-on) angles. We propose that these objects are weak emission-line quasars (WLQs) observed through the disc wind, caught emerging from their dust cocoons. As quasars shed their cocoons, dust grains in the disc wind are shattered into smaller particles, producing the UV-steeper extinction curve observed along the outflow. We present a schematic illustration of this shedding process that can account for the peculiar spectral features observed in our sample.

TEQUILA (Transient Event $Q$, $U$, and $I$ Light Analyzer) is an optical imaging polarimeter developed for the second Nasmyth port of the 1.3-m COLIBRÍ altitude-azimuth telescope at Observatorio Astronómico Nacional in San Pedro Mártir, México (OAN-SPM). TEQUILA uses a CMOS sensor with an on-chip wire-grid micro-polarizer array to obtain simultaneous, single-exposure measurements of the Stokes parameters $I$, $Q$, and $U$ without moving optical components. This mechanism-free instrument, built entirely from commercial components, delivers seeing-limited imaging in a fixed optical band and is optimized for early-time follow-up of transient sources, including gamma-ray burst afterglows, blazars, and variable young stellar objects. In this paper, we describe the scientific motivation, the instrument design and implementation, the calibration, and initial science results. Sensor characterization reveals a polarimetric structure in the flat field and a low quantum efficiency, which we estimate to be approximately 17%, including losses introduced by the micro-polarizer array. For point sources, TEQUILA achieves absolute polarimetry with RMS uncertainties of 0.15% in pupil-tracking observations and 0.20% in field-tracking observations. In pupil-tracking mode, the observed RMS is fully explained by the measurement and standard-star uncertainties, with no evidence for an additional calibration term. In contrast, field-tracking observations require an additional calibration uncertainty of approximately 0.10%. Calibration for resolved-source polarimetry remains in progress.

Local primordial non-Gaussianity (local PNG) induces a scale-dependent contribution to galaxy clustering proportional to $f_{\rm NL}\,b_φ$, where $f_{\rm NL}$ is the local PNG amplitude and $b_φ$ encodes the galaxy response to a long-wavelength primordial potential perturbation. Uncertainty in $b_φ$ is the dominant obstacle to precise, robust constraints on $f_{\rm NL}$ from galaxy surveys. We translate small-scale clustering constraints on the galaxy--halo connection into priors on $b_φ$: sampling the posterior of a halo occupation distribution (HOD) model fit to the DESI EDR, we generate mocks from which we measure $b_φ$ and construct its prior. Validating against additional mocks with different local PNG amplitudes, we show that the method recovers unbiased $f_{\rm NL}$, even in the presence of assembly bias.

The large instantaneous sensitivity, wide frequency coverage and flexible observation modes, with large number of beams in the sky, are the main features of the SKA observatory's two telescopes, the SKA-Low and the SKA-Mid. Owing to these capabilities, the SKAO telescopes are going to be a game-changer for radio astronomy in general and pulsar astronomy in particular. Eleven chapters in this book describe their impact on different areas of pulsar science. In this overview article each chapter is briefly summarised and the inter-relationship between different pulsar science use cases are explored: new deep surveys, covering the Galactic field, globular clusters and the Galactic centre, will discover thousands of new pulsars; these will form the backbone for studies of neutron star physics and of their environments. The enhanced understanding provided by these studies will feed into the main contributions to fundamental physics from pulsar astronomy: testing relativistic gravity, studying gravitational waves in the nano-Hz regime and studying the equation of state of nuclear matter. Synergies with other science cases are also highlighted throughout this overview.