arXiv Daily Digest - 2026-04-07

Knowledge Graphs (KGs) serve as a critical foundation for AI systems, yet their automated construction inevitably introduces noise, compromising data trustworthiness. Existing triple verification methods, based on graph embeddings or language models, often suffer from single-source bias by relying on either internal structural constraints or external semantic evidence, and usually follow a static inference paradigm. As a result, they struggle with complex or long-tail facts and provide limited interpretability. To address these limitations, we propose SHARP (Schema-Hybrid Agent for Reliable Prediction), a training-free autonomous agent that reformulates triple verification as a dynamic process of strategic planning, active investigation, and evidential reasoning. Specifically, SHARP combines a Memory-Augmented Mechanism with Schema-Aware Strategic Planning to improve reasoning stability, and employs an enhanced ReAct loop with a Hybrid Knowledge Toolset to dynamically integrate internal KG structure and external textual evidence for cross-verification. Experiments on FB15K-237 and Wikidata5M-Ind show that SHARP significantly outperforms existing state-of-the-art baselines, achieving accuracy gains of 4.2% and 12.9%, respectively. Moreover, SHARP provides transparent, fact-based evidence chains for each judgment, demonstrating strong interpretability and robustness for complex verification tasks.

Non-stationary environments require agents to revise previously learned action values when contingencies change. We treat large language models (LLMs) as sequential decision policies in a two-option probabilistic reversal-learning task with three latent states and switch events triggered by either a performance criterion or timeout. We compare a deterministic fixed transition cycle to a stochastic random schedule that increases volatility, and evaluate DeepSeek-V3.2, Gemini-3, and GPT-5.2, with human data as a behavioural reference. Across models, win-stay was near ceiling while lose-shift was markedly attenuated, revealing asymmetric use of positive versus negative evidence. DeepSeek-V3.2 showed extreme perseveration after reversals and weak acquisition, whereas Gemini-3 and GPT-5.2 adapted more rapidly but still remained less loss-sensitive than humans. Random transitions amplified reversal-specific persistence across LLMs yet did not uniformly reduce total wins, demonstrating that high aggregate payoff can coexist with rigid adaptation. Hierarchical reinforcement-learning (RL) fits indicate dissociable mechanisms: rigidity can arise from weak loss learning, inflated policy determinism, or value polarisation via counterfactual suppression. These results motivate reversal-sensitive diagnostics and volatility-aware models for evaluating LLMs under non-stationary uncertainty.

As large language models (LLMs) are increasing integrated into fact-checking pipelines, formal logic is often proposed as a rigorous means by which to mitigate bias, errors and hallucinations in these models' outputs. For example, some neurosymbolic systems verify claims by using LLMs to translate natural language into logical formulae and then checking whether the proposed claims are logically sound, i.e. whether they can be validly derived from premises that are verified to be true. We argue that such approaches structurally fail to detect misleading claims due to systematic divergences between conclusions that are logically sound and inferences that humans typically make and accept. Drawing on studies in cognitive science and pragmatics, we present a typology of cases in which logically sound conclusions systematically elicit human inferences that are unsupported by the underlying premises. Consequently, we advocate for a complementary approach: leveraging the human-like reasoning tendencies of LLMs as a feature rather than a bug, and using these models to validate the outputs of formal components in neurosymbolic systems against potentially misleading conclusions.

Healthcare foundation models have largely followed paradigms from natural language processing and computer vision, emphasizing large scale pretraining and deterministic representations over heterogeneous clinical data. However, clinical observations are inherently incomplete, reflecting sparse, irregular, and modality dependent measurements of an underlying physiologic state. In this work, we propose a framework for uncertainty aware foundation modeling that represents each patient not as a point embedding, but as a distribution over plausible latent states. By learning set valued representations and enforcing consistency across partial views of the same patient, the model captures what is invariantly inferable while explicitly encoding epistemic uncertainty. We integrate this formulation with multimodal encoders and scalable self supervised objectives, combining reconstruction, contrastive alignment, and distributional regularization. Across diverse clinical tasks, our approach improves predictive performance, robustness under missing data, and uncertainty calibration relative to strong baselines. These results suggest that modeling what is not observed rather than only what is constitutes a critical inductive bias for healthcare foundation models.

The proliferation of fake news across diverse domains highlights critical limitations in current detection systems, which often exhibit narrow domain specificity and poor generalization. Existing cross-domain approaches face two key challenges: (1) reliance on labelled data, which is frequently unavailable and resource intensive to acquire and (2) information loss caused by rigid domain categorization or neglect of domain-specific features. To address these issues, we propose CoALFake, a novel approach for cross-domain fake news detection that integrates Human-Large Language Model (LLM) co-annotation with domain-aware Active Learning (AL). Our method employs LLMs for scalable, low-cost annotation while maintaining human oversight to ensure label reliability. By integrating domain embedding techniques, the CoALFake dynamically captures both domain specific nuances and cross-domain patterns, enabling the training of a domain agnostic model. Furthermore, a domain-aware sampling strategy optimizes sample acquisition by prioritizing diverse domain coverage. Experimental results across multiple datasets demonstrate that the proposed approach consistently outperforms various baselines. Our results emphasize that human-LLM co-annotation is a highly cost-effective approach that delivers excellent performance. Evaluations across several datasets show that CoALFake consistently outperforms a range of existing baselines, even with minimal human oversight.

In many science papers, "Figure 1" serves as the primary visual summary of the core research idea. These figures are visually simple yet conceptually rich, often requiring significant effort and iteration by human authors to get right, highlighting the difficulty of science visual communication. With this intuition, we introduce GENFIG1, a benchmark for generative AI models (e.g., Vision-Language Models). GENFIG1 evaluates models for their ability to produce figures that clearly express and motivate the central idea of a paper (title, abstract, introduction, and figure caption) as input. Solving GENFIG1 requires more than producing visually appealing graphics: the task entails reasoning for text-to-image generation that couples scientific understanding with visual synthesis. Specifically, models must (i) comprehend and grasp the technical concepts of the paper, (ii) identify the most salient ones, and (iii) design a coherent and aesthetically effective graphic that conveys those concepts visually and is faithful to the input. We curate the benchmark from papers published at top deep-learning conferences, apply stringent quality control, and introduce an automatic evaluation metric that correlates well with expert human judgments. We evaluate a suite of representative models on GENFIG1 and demonstrate that the task presents significant challenges, even for the best-performing systems. We hope this benchmark serves as a foundation for future progress in multimodal AI.

I propose a model of systematic understanding, suitable for machine learning systems. On this account, an agent understands a property of a target system when it contains an adequate internal model that tracks real regularities, is coupled to the target by stable bridge principles, and supports reliable prediction. I argue that contemporary deep learning systems often can and do achieve such understanding. However they generally fall short of the ideal of scientific understanding: the understanding is symbolically misaligned with the target system, not explicitly reductive, and only weakly unifying. I label this the Fractured Understanding Hypothesis.

Although multi-view multi-label learning has been extensively studied, research on the dual-missing scenario, where both views and labels are incomplete, remains largely unexplored. Existing methods mainly rely on contrastive learning or information bottleneck theory to learn consistent representations under missing-view conditions, but loss-based alignment without explicit structural constraints limits the ability to capture stable and discriminative shared semantics. To address this issue, we introduce a more structured mechanism for consistent representation learning: we learn discrete consistent representations through a multi-view shared codebook and cross-view reconstruction, which naturally align different views within the limited shared codebook embeddings and reduce feature redundancy. At the decision level, we design a weight estimation method that evaluates the ability of each view to preserve label correlation structures, assigning weights accordingly to enhance the quality of the fused prediction. In addition, we introduce a fused-teacher self-distillation framework, where the fused prediction guides the training of view-specific classifiers and feeds the global knowledge back into the single-view branches, thereby enhancing the generalization ability of the model under missing-label conditions. The effectiveness of our proposed method is thoroughly demonstrated through extensive comparative experiments with advanced methods on five benchmark datasets. Code is available at https://github.com/xuy11/SCSD.

Electronic Patient Record (EPR) systems contain valuable clinical information, but much of it is trapped in unstructured text, limiting its use for research and decision-making. Large language models can extract such information but require substantial computational resources to run locally, and sending sensitive clinical data to cloud-based services, even when deidentified, raises significant patient privacy concerns. In this study, we develop a resource-efficient semi-automated annotation workflow using small language models (SLMs) to extract structured information from unstructured EPR data, focusing on paediatric histopathology reports. As a proof-of-concept, we apply the workflow to paediatric renal biopsy reports, a domain chosen for its constrained diagnostic scope and well-defined underlying biology. We develop the workflow iteratively with clinical oversight across three meetings, manually annotating 400 reports from a dataset of 2,111 at Great Ormond Street Hospital as a gold standard, while developing an automated information extraction approach using SLMs. We frame extraction as a Question-Answering task grounded by clinician-guided entity guidelines and few-shot examples, evaluating five instruction-tuned SLMs with a disagreement modelling framework to prioritise reports for clinical review. Gemma 2 2B achieves the highest accuracy at 84.3%, outperforming off-the-shelf models including spaCy (74.3%), BioBERT-SQuAD (62.3%), RoBERTa-SQuAD (59.7%), and GLiNER (60.2%). Entity guidelines improved performance by 7-19% over the zero-shot baseline, and few-shot examples by 6-38%, though their benefits do not compound when combined. These results demonstrate that SLMs can extract structured information from specialised clinical domains on CPU-only infrastructure with minimal clinician involvement. Our code is available at https://github.com/gosh-dre/nlp_renal_biopsy.

Theory of Mind (ToM) -- the ability to model others' mental states -- is fundamental to human social cognition. Whether large language models (LLMs) can develop ToM has been tested exclusively through static vignettes, leaving open whether ToM-like reasoning can emerge through dynamic interaction. Here we report that autonomous LLM agents playing extended sessions of Texas Hold'em poker progressively develop sophisticated opponent models, but only when equipped with persistent memory. In a 2x2 factorial design crossing memory (present/absent) with domain knowledge (present/absent), each with five replications (N = 20 experiments, ~6,000 agent-hand observations), we find that memory is both necessary and sufficient for ToM-like behavior emergence (Cliff's delta = 1.0, p = 0.008). Agents with memory reach ToM Level 3-5 (predictive to recursive modeling), while agents without memory remain at Level 0 across all replications. Strategic deception grounded in opponent models occurs exclusively in memory-equipped conditions (Fisher's exact p < 0.001). Domain expertise does not gate ToM-like behavior emergence but enhances its application: agents without poker knowledge develop equivalent ToM levels but less precise deception (p = 0.004). Agents with ToM deviate from game-theoretically optimal play (67% vs. 79% TAG adherence, delta = -1.0, p = 0.008) to exploit specific opponents, mirroring expert human play. All mental models are expressed in natural language and directly readable, providing a transparent window into AI social cognition. Cross-model validation with GPT-4o yields weighted Cohen's kappa = 0.81 (almost perfect agreement). These findings demonstrate that functional ToM-like behavior can emerge from interaction dynamics alone, without explicit training or prompting, with implications for understanding artificial social intelligence and biological social cognition.

Foundation models for biology and physics optimize predictive accuracy, but their internal representations systematically fail to preserve the continuous geometry of the systems they model. We identify the root cause: the Geometric Alignment Tax, an intrinsic cost of forcing continuous manifolds through discrete categorical bottlenecks. Controlled ablations on synthetic dynamical systems demonstrate that replacing cross-entropy with a continuous head on an identical encoder reduces geometric distortion by up to 8.5x, while learned codebooks exhibit a non-monotonic double bind where finer quantization worsens geometry despite improving reconstruction. Under continuous objectives, three architectures differ by 1.3x; under discrete tokenization, they diverge by 3,000x. Evaluating 14 biological foundation models with rate-distortion theory and MINE, we identify three failure regimes: Local-Global Decoupling, Representational Compression, and Geometric Vacuity. A controlled experiment confirms that Evo 2's reverse-complement robustness on real DNA reflects conserved sequence composition, not learned symmetry. No model achieves simultaneously low distortion, high mutual information, and global coherence.

Continual learning in artificial neural networks is fundamentally limited by the stability--plasticity dilemma: systems that retain prior knowledge tend to resist acquiring new knowledge, and vice versa. Existing approaches, most notably elastic weight consolidation~(EWC), address this empirically without a physical account of why plasticity eventually collapses as tasks accumulate. Separately, the distinction between sudden insight and gradual skill acquisition through repetitive practice has lacked a unified theoretical description. Here, we show that both problems admit a common resolution within non-equilibrium statistical physics. We model the state of a learning system as a particle evolving under Langevin dynamics on a double-well energy landscape, with the noise amplitude governed by a time-dependent effective temperature $T(t)$. The probability density obeys a Fokker--Planck equation, and transitions between metastable states are governed by the Kramers escape rate $k = (ω_0ω_b/2π)\,e^{-ΔE/T}$. We make two contributions. First, we identify the EWC penalty term as an energy barrier whose height grows linearly with the number of accumulated tasks, yielding an exponential collapse of the transition rate predicted analytically and confirmed numerically. Second, we show that insight and repetitive learning correspond to two qualitatively distinct temperature protocols within the same Fokker--Planck equation: insight events produce transient spikes in $T(t)$ that drive rapid barrier crossing, whereas repetitive practice operates at a modestly elevated but fixed temperature, achieving transitions through sustained stochastic diffusion. These results establish a physically grounded framework for understanding plasticity and its failure in continual learning systems, and suggest principled design criteria for adaptive noise schedules in artificial intelligence.

Deep learning models have shown great promise in diverse remote sensing applications. However, they often struggle to generalize across geographic regions unseen during training due to domain shifts. Domain shifts occur when data distributions differ between the training region and new target regions, due to variations in land cover, climate, and environmental conditions. Test-time adaptation (TTA) has emerged as a solution to such shifts, but existing methods are primarily designed for classification and are not directly applicable to regression tasks. In this work, we address the regression task of spatio-temporal fusion (STF) for land surface temperature estimation. We propose an uncertainty-aware TTA framework that updates only the fusion module of a pre-trained STF model, guided by epistemic uncertainty, land use and land cover consistency, and bias correction, without requiring source data or labeled target samples. Experiments on four target regions with diverse climates, namely Rome in Italy, Cairo in Egypt, Madrid in Spain, and Montpellier in France, show consistent improvements in RMSE and MAE for a pre-trained model in Orléans, France. The average gains are 24.2% and 27.9%, respectively, even with limited unlabeled target data and only 10 TTA epochs.

Photovoltaic (PV) power forecasting plays a critical role in power system dispatch and market participation. Because PV generation is highly sensitive to weather conditions and cloud motion, accurate forecasting requires effective modeling of complex spatiotemporal dependencies across multiple information sources. Although recent studies have advanced AI-based forecasting methods, most fail to fuse temporal observations, satellite imagery, and textual weather information in a unified framework. This paper proposes Solar-VLM, a large-language-model-driven framework for multimodal PV power forecasting. First, modality-specific encoders are developed to extract complementary features from heterogeneous inputs. The time-series encoder adopts a patch-based design to capture temporal patterns from multivariate observations at each site. The visual encoder, built upon a Qwen-based vision backbone, extracts cloud-cover information from satellite images. The text encoder distills historical weather characteristics from textual descriptions. Second, to capture spatial dependencies across geographically distributed PV stations, a cross-site feature fusion mechanism is introduced. Specifically, a Graph Learner models inter-station correlations through a graph attention network constructed over a K-nearest-neighbor (KNN) graph, while a cross-site attention module further facilitates adaptive information exchange among sites. Finally, experiments conducted on data from eight PV stations in a northern province of China demonstrate the effectiveness of the proposed framework. Our proposed model is publicly available at https://github.com/rhp413/Solar-VLM.

The holy grail of LLM personalization is a single LLM for each user, perfectly aligned with that user's preferences. However, maintaining a separate LLM per user is impractical due to constraints on compute, memory, and system complexity. We address this challenge by developing a principled method for selecting a small portfolio of LLMs that captures representative behaviors across heterogeneous users. We model user preferences across multiple traits (e.g., safety, humor, brevity) through a multi-dimensional weight vector. Given reward functions across these dimensions, our algorithm PALM (Portfolio of Aligned LLMs) generates a small portfolio of LLMs such that, for any weight vector, the portfolio contains a near-optimal LLM for the corresponding scalarized objective. To the best of our knowledge, this is the first result that provides theoretical guarantees on both the size and approximation quality of LLM portfolios for personalization. It characterizes the trade-off between system cost and personalization, as well as the diversity of LLMs required to cover the landscape of user preferences. We provide empirical results that validate these guarantees and demonstrate greater output diversity over common baselines.

Dexterous manipulation requires planning a grasp configuration suited to the object and task, which is then executed through coordinated multi-finger control. However, specifying grasp plans with dense pose or contact targets for every object and task is impractical. Meanwhile, end-to-end reinforcement learning from task rewards alone lacks controllability, making it difficult for users to intervene when failures occur. To this end, we present GRIT, a two-stage framework that learns dexterous control from sparse taxonomy guidance. GRIT first predicts a taxonomy-based grasp specification from the scene and task context. Conditioned on this sparse command, a policy generates continuous finger motions that accomplish the task while preserving the intended grasp structure. Our result shows that certain grasp taxonomies are more effective for specific object geometries. By leveraging this relationship, GRIT improves generalization to novel objects over baselines and achieves an overall success rate of 87.9%. Moreover, real-world experiments demonstrate controllability, enabling grasp strategies to be adjusted through high-level taxonomy selection based on object geometry and task intent.

There is substantial interest in developing artificial intelligence systems to support radiologists across tasks ranging from segmentation to report generation. Existing computed tomography (CT) foundation models have largely focused on building generalist vision-language systems capable of tasks such as question answering and report generation. However, training reliable vision-language systems requires paired image-text data at a scale that remains unavailable in CT. Moreover, adapting the underlying visual representations to downstream tasks typically requires partial or full backbone fine-tuning, a computationally demanding process inaccessible to many research groups. Instead, foundation models should prioritise learning robust visual representations that enable efficient transfer to new tasks with minimal labelled data and without backbone fine-tuning. We present VoxelFM, a 3D CT foundation model trained with self-distillation using the DINO framework, which learns semantically rich features without language supervision. We evaluated VoxelFM across seven categories of clinically relevant downstream tasks using frozen backbone representations with lightweight probes: classification, regression, survival analysis, instance retrieval, localisation, segmentation, and report generation. VoxelFM matched or outperformed four existing CT foundation models across all task categories. Despite receiving no language supervision during pre-training, VoxelFM surpassed models explicitly trained with language-alignment objectives, including on report generation. Our results indicate that current CT foundation models perform significantly better as feature extractors for lightweight probes rather than as vision encoders for vision-language models. Model weights and training code are publicly available.

Large language model agents that use external tools are often implemented through reactive execution, in which reasoning is repeatedly recomputed after each observation, increasing latency and sensitivity to error propagation. This work introduces Profile--Then--Reason (PTR), a bounded execution framework for structured tool-augmented reasoning, in which a language model first synthesizes an explicit workflow, deterministic or guarded operators execute that workflow, a verifier evaluates the resulting trace, and repair is invoked only when the original workflow is no longer reliable. A mathematical formulation is developed in which the full pipeline is expressed as a composition of profile, routing, execution, verification, repair, and reasoning operators; under bounded repair, the number of language-model calls is restricted to two in the nominal case and three in the worst case. Experiments against a ReAct baseline on six benchmarks and four language models show that PTR achieves the pairwise exact-match advantage in 16 of 24 configurations. The results indicate that PTR is particularly effective on retrieval-centered and decomposition-heavy tasks, whereas reactive execution remains preferable when success depends on substantial online adaptation.

Recent advancements in data science have significantly elevated the importance of orthogonally constrained optimization problems. The Riemannian approach has become a popular technique for addressing these problems due to the advantageous computational and analytical properties of the Stiefel manifold. Nonetheless, the interplay of nonsmoothness alongside orthogonality constraints introduces substantial challenges to current Riemannian methods, including scalability, parallelizability, complicated subproblems, and cumulative numerical errors that threaten feasibility. In this paper, we take a retraction-free primal-dual approach and propose a linearized smoothing augmented Lagrangian method specifically designed for nonsmooth and nonconvex optimization with orthogonality constraints. Our proposed method is single-loop and free of subproblem solving. We establish its iteration complexity of $O(ε^{-3})$ for finding $ε$-KKT points, matching the best-known results in the Riemannian optimization literature. Additionally, by invoking the standard Kurdyka-Lojasiewicz (KL) property, we demonstrate asymptotic sequential convergence of the proposed algorithm. Numerical experiments on both smooth and nonsmooth orthogonal constrained problems demonstrate the superior computational efficiency and scalability of the proposed method compared with state-of-the-art algorithms.

This study investigates robust speech-related decoding from non-invasive MEG signals using the LibriBrain phoneme-classification benchmark from the 2025 PNPL competition. We compare residual convolutional neural networks (CNNs), an STFT-based CNN, and a CNN--Transformer hybrid, while also examining the effects of group averaging, label balancing, repeated grouping, normalization strategies, and data augmentation. Across our in-house implementations, preprocessing and data-configuration choices matter more than additional architectural complexity, among which instance normalization emerges as the most influential modification for generalization. The strongest of our own models, a CNN with group averaging, label balancing, repeated grouping, and instance normalization, achieves 60.95% F1-macro on the test split, compared with 39.53% for the plain CNN baseline. However, most of our models, without instance normalization, show substantial validation-to-test degradation, indicating that distribution shift induced by different normalization statistics is a major obstacle to generalization in our experiments. By contrast, MEGConformer maintains 64.09% F1-macro on both validation and test, and saliency-map analysis is qualitatively consistent with this contrast: weaker models exhibit more concentrated or repetitive phoneme-sensitive patterns across splits, whereas MEGConformer appears more distributed. Overall, the results suggest that improving the reliability of non-invasive phoneme decoding will likely require better handling of normalization-related distribution shift while also addressing the challenge of single-trial decoding.

Cybersecurity research increasingly depends on reproducible evidence, such as traffic traces, logs, and labeled datasets, yet most public datasets remain static and offer limited support for controlled re-execution and traceability, especially in heterogeneous multi-protocol environments. This paper presents NetSecBed, a container-native, scenario-oriented testbed for reproducible generation of network traffic evidence and execution artifacts under controlled conditions, particularly suitable for IoT, IIoT, and pervasive multi-protocol environments. The framework integrates 60 attack scenarios, 9 target services, and benign traffic generators as single-purpose containers, enabling plug-and-play extensibility and traceability through declarative specifications. Its pipeline automates parametrized execution, packet capture, log collection, service probing, feature extraction, and dataset consolidation. The main contribution is a repeatable, auditable, and extensible framework for cybersecurity experimentation that reduces operational bias and supports continuous dataset generation.

Long chain-of-thought (Long-CoT) reasoning models have motivated a growing body of work on compressing reasoning traces to reduce inference cost, yet existing evaluations focus almost exclusively on task accuracy and token savings. Trustworthiness properties, whether acquired or reinforced through post-training, are encoded in the same parameter space that compression modifies. This means preserving accuracy does not, a priori, guarantee preserving trustworthiness. We conduct the first systematic empirical study of how CoT compression affects model trustworthiness, evaluating multiple models of different scales along three dimensions: safety, hallucination resistance, and multilingual robustness. Under controlled comparisons, we find that CoT compression frequently introduces trustworthiness regressions and that different methods exhibit markedly different degradation profiles across dimensions. To enable fair comparison across bases, we propose a normalized efficiency score for each dimension that reveals how naïve scalar metrics can obscure trustworthiness trade-offs. As an existence proof, we further introduce an alignment-aware DPO variant that reduces CoT length by 19.3\% on reasoning benchmarks with substantially smaller trustworthiness loss. Our findings suggest that CoT compression should be optimized not only for efficiency but also for trustworthiness, treating both as equally important design constraints.

Reliable relative pose estimation is a key enabler for autonomous rendezvous and proximity operations, yet space imagery is notoriously challenging due to extreme illumination, high contrast, and fast target motion. Event cameras provide asynchronous, change-driven measurements that can remain informative when frame-based imagery saturates or blurs, while neuromorphic processors can exploit sparse activations for low-latency, energy-efficient inferences. This paper presents a spacecraft 6-DoF pose-estimation pipeline that couples event-based vision with the BrainChip Akida neuromorphic processor. Using the SPADES dataset, we train compact MobileNet-style keypoint regression networks on lightweight event-frame representations, apply quantization-aware training (8/4-bit), and convert the models to Akida-compatible spiking neural networks. We benchmark three event representations and demonstrate real-time, low-power inference on Akida V1 hardware. We additionally design a heatmap-based model targeting Akida V2 and evaluate it on Akida Cloud, yielding improved pose accuracy. To our knowledge, this is the first end-to-end demonstration of spacecraft pose estimation running on Akida hardware, highlighting a practical route to low-latency, low-power perception for future autonomous space missions.

This is the Replicated Computational Results (RCR) Report for the paper C2|Q>: A Robust Framework for Bridging Classical and Quantum Software Development. The paper introduces a modular, hardware-agnostic framework that translates classical problem specifications - Python code or structured JSON - into executable quantum programs across ten problem families and multiple hardware backends. We release the framework source code on GitHub at https://github.com/C2-Q/C2Q, a pretrained parser model on Zenodo at https://zenodo.org/records/19061125, evaluation data in a separate Zenodo record at https://zenodo.org/records/17071667, and a PyPI package at https://pypi.org/project/c2q-framework/ for lightweight CLI and API use. Experiment 1 is supported through a released pretrained model and training notebook, while Experiments 2 and 3 are directly executable via documented make targets. This report describes the artifact structure, setup instructions, and the mapping from each execution route to the corresponding experiment.

Data-driven neural networks are increasingly used as surrogate forward models in geophysics, but it remains unclear whether they recover only the data mapping or also the underlying physical sensitivity structure. Here we test this question using surface-wave dispersion. By comparing automatically differentiated gradients from a neural-network surrogate with theoretical sensitivity kernels, we show that the learned gradients can recover the main depth-dependent structure of physical kernels across a broad range of periods. This indicates that neural surrogate models can learn physically meaningful differential information, rather than acting as purely black-box predictors. At the same time, strong structural priors in the training distribution can introduce systematic artifacts into the inferred sensitivities. Our results show that neural forward surrogates can recover useful physical information for inversion and uncertainty analysis, while clarifying the conditions under which this differential structure remains physically consistent.

The generation of realistic and controllable GPS trajectories is a fundamental task for applications in urban planning, mobility simulation, and privacy-preserving data sharing. However, existing methods face a two-fold challenge: they lack the deep semantic understanding to interpret complex user travel intent, and struggle to handle complex constraints while maintaining the realistic diversity inherent in human behavior. To resolve this, we introduce InsTraj, a novel framework that instructs diffusion models to generate high-fidelity trajectories directly from natural language descriptions. Specifically, InsTraj first utilizes a powerful large language model to decipher unstructured travel intentions formed in natural language, thereby creating rich semantic blueprints and bridging the representation gap between intentions and trajectories. Subsequently, we proposed a multimodal trajectory diffusion transformer that can integrate semantic guidance to generate high-fidelity and instruction-faithful trajectories that adhere to fine-grained user intent. Comprehensive experiments on real-world datasets demonstrate that InsTraj significantly outperforms state-of-the-art methods in generating trajectories that are realistic, diverse, and semantically faithful to the input instructions.

Mind perception (MP) is a psychological phenomenon in which humans automatically infer that another entity has a mind and/or mental capacities, usually understood in two dimensions (perceived agency and experience capacities). Despite MP's centrality to many social processes, understanding how MP may function in humans' machine companionship relations is limited. This is in part due to reliance on self reports and the gap between automatic MP processes and more purposeful and norm governed expressions of MP. We here leverage MP signaling language to explore the relationship between MP and AI companionship in humans' natural language. We systematically collected discussions about companionship from AI dedicated Reddit forums and examined the cooccurrence of words (a) known to signal agentic and experiential MP and those induced from the data and (b) discussion topics related to AI companionship. Using inductive and deductive approaches, we identify a small set of linguistic indicators as reasonable markers of MP in human/AI chat, and some are linked to critical discussions of companion authenticity and philosophical and ethical imaginaries.

High-stakes decision systems increasingly require structured justification, traceability, and auditability to ensure accountability and regulatory compliance. Formal arguments commonly used in the certification of safety-critical systems provide a mechanism for structuring claims, reasoning, and evidence in a verifiable manner. At the same time, generative artificial intelligence systems are increasingly integrated into decision-support workflows, assisting with drafting explanations, summarizing evidence, and generating recommendations. However, current deployments often rely on language models as loosely constrained assistants, which introduces risks such as hallucinated reasoning, unsupported claims, and weak traceability. This paper proposes a compliance-by-construction architecture that integrates Generative AI (GenAI) with structured formal argument representations. The approach treats each AI-assisted step as a claim that must be supported by verifiable evidence and validated against explicit reasoning constraints before it becomes part of an official decision record. The architecture combines four components: i) a typed Argument Graph representation inspired by assurance-case methods, ii) retrieval-augmented generation (RAG) to draft argument fragments grounded in authoritative evidence, iii) a reasoning and validation kernel enforcing completeness and admissibility constraints, and iv) a provenance ledger aligned with the W3C PROV standard to support auditability. We present a system design and an evaluation strategy based on enforceable invariants and worked examples. The analysis suggests that deterministic validation rules can prevent unsupported claims from entering the decision record while allowing GenAI to accelerate argument construction.

Developers utilize third-party libraries to improve productivity, which also introduces potential security risks. Existing approaches generate tests for public functions to trigger library vulnerabilities from client programs, yet they depend on proof-of-concepts (PoCs), which are often unavailable. In this paper, we propose a new approach, LiveFuzz, based on directed greybox fuzzing (DGF) to detect the exploitability of library vulnerabilities from client programs without PoCs. LiveFuzz exploits a target tuple to extend existing DGF techniques to cross-program scenarios. Based on the target tuple, LiveFuzz introduces a novel Abstract Path Mapping mechanism to project execution paths, mitigating the preference for shorter paths. LiveFuzz also proposes a risk-based adaptive mutation to mitigate excessive mutation. To evaluate LiveFuzz, we construct a new dataset including 61 cases of library vulnerabilities exploited from client programs. Results show that LiveFuzz increases the number of target-reachable paths compared with all baselines and improves the average speed of vulnerability exposure. Three vulnerabilities are triggered exclusively by LiveFuzz.

This paper investigates the Restless Multi-Armed Bandit (RMAB) framework under individual penalty constraints to address resource allocation challenges in dynamic wireless networked environments. Unlike conventional RMAB models, our model allows each user (arm) to have distinct and stringent performance constraints, such as energy limits, activation limits, or age of information minimums, enabling the capture of diverse objectives including fairness and efficiency. To find the optimal resource allocation policy, we propose a new Penalty-Optimal Whittle (POW) index policy. The POW index of an user only depends on the user's transition kernel and penalty constraints, and remains invariable to system-wide features such as the number of users present and the amount of resource available. This makes it computationally tractable to calculate the POW Indices offline without any need for online adaptation. Moreover, we theoretically prove that the POW index policy is asymptotically optimal while satisfying all individual penalty constraints. We also introduce a deep reinforcement learning algorithm to efficiently learn the POW index on the fly. Simulation results across various applications and system configurations further demonstrate that the POW index policy not only has near-optimal performance but also significantly outperforms other existing policies.

Generative AI (GenAI) tools are increasingly integrated into software architecture research, yet the environmental impact of their computational usage remains largely undocumented. This study presents the first systematic audit of the carbon footprint of both the digital footprint from GenAI usage in research papers, and the traditional footprint from conference activities within the context of the IEEE International Conference on Software Architecture (ICSA). We report two separate carbon inventories relevant to the software architecture research community: i) an exploratory estimate of the footprint of GenAI inference usage associated with accepted papers within a research-artifact boundary, and ii) the conference attendance and operations footprint of ICSA 2025 (travel, accommodation, catering, venue energy, and materials) within the conference time boundary. These two inventories, with different system boundaries and completeness, support transparency and community reflection. We discuss implications for sustainable software architecture, including recommendations for transparency, greener conference planning, and improved energy efficiency in GenAI operations. Our work supports a more climate-conscious research culture within the ICSA community and beyond

Classical approximation bases such as Chebyshev polynomials provide principled and interpretable representations, but their multivariate tensor-product constructions scale exponentially with dimension and impose axis-aligned structure that is poorly matched to real tabular data. We address this by replacing tensorised oscillations with directional harmonic modes of the form $\cos(\mathbf{m}^{\top}\arccos(\mathbf{x}))$, which organise multivariate structure by direction in angular space rather than by coordinate index. This representation yields a discrete spectral regression model in which complexity is controlled by selecting a small number of structured frequency vectors (spectral paths), and training reduces to a single closed-form ridge solve with no iterative optimisation. Experiments on standard continuous-feature tabular regression benchmarks show that the resulting models achieve accuracy competitive with strong nonlinear baselines while remaining compact, computationally efficient, and explicitly interpretable through analytic expressions of learned feature interactions.

Stochastic bilevel optimization (SBO) has been integrated into many machine learning paradigms recently, including hyperparameter optimization, meta learning, and reinforcement learning. Along with the wide range of applications, there have been numerous studies on the computational behavior of SBO. However, the generalization guarantees of SBO methods are far less understood from the lens of statistical learning theory. In this paper, we provide a systematic generalization analysis of the first-order gradient-based bilevel optimization methods. Firstly, we establish the quantitative connections between the on-average argument stability and the generalization gap of SBO methods. Then, we derive the upper bounds of on-average argument stability for single-timescale stochastic gradient descent (SGD) and two-timescale SGD, where three settings (nonconvex-nonconvex (NC-NC), convex-convex (C-C), and strongly-convex-strongly-convex (SC-SC)) are considered respectively. Experimental analysis validates our theoretical findings. Compared with the previous algorithmic stability analysis, our results do not require reinitializing the inner-level parameters at each iteration and are applicable to more general objective functions.

Translating quantum many-body theory into scalable software traditionally requires months of effort. Zero-shot generation of tensor network algorithms by Large Language Models (LLMs) frequently fails due to spatial reasoning errors and memory bottlenecks. We resolve this using a multi-stage workflow that mimics a physics research group. By generating a mathematically rigorous LaTeX specification as an intermediate blueprint, we constrain the coding LLM to produce exact, matrix-free $\mathcal{O}(D^3)$ operations. We validate this approach by generating a Density-Matrix Renormalization Group (DMRG) engine that accurately captures the critical entanglement scaling of the Spin-$1/2$ Heisenberg model and the symmetry-protected topological (SPT) order of the Spin-$1$ AKLT model. Testing across 16 combinations of leading foundation models yielded a 100\% success rate. By compressing a months-long development cycle into under 24 hours ($\sim 14$ active hours), this framework offers a highly reproducible paradigm for accelerating computational physics research.

Learner-item cognitive modeling plays a central role in the web-based online intelligent education system by enabling cognitive diagnosis (CD) across diverse online educational scenarios. Although ID embedding remains the mainstream approach in cognitive modeling due to its effectiveness and flexibility, recent advances in language models (LMs) have introduced new possibilities for incorporating rich semantic representations to enhance CD performance. This highlights the need for a comprehensive analysis of how LMs enhance embeddings through semantic integration across mainstream CD tasks. This paper identifies two key challenges in fully leveraging LMs in existing work: Misalignment between the training objectives of LMs and CD models creates a distribution gap in feature spaces; A unified framework is essential for integrating textual embeddings across varied CD tasks while preserving the strengths of existing cognitive modeling paradigms to ensure the robustness of embedding enhancement. To address these challenges, this paper introduces EduEmbed, a unified embedding enhancement framework that leverages fine-tuned LMs to enrich learner-item cognitive modeling across diverse CD tasks. EduEmbed operates in two stages. In the first stage, we fine-tune LMs based on role-specific representations and an interaction diagnoser to bridge the semantic gap of CD models. In the second stage, we employ a textual adapter to extract task-relevant semantics and integrate them with existing modeling paradigms to improve generalization. We evaluate the proposed framework on four CD tasks and computerized adaptive testing (CAT) task, achieving robust performance. Further analysis reveals the impact of semantic information across diverse tasks, offering key insights for future research on the application of LMs in CD for online intelligent education systems.

We introduce ArrowFlow, a machine learning architecture that operates entirely in the space of permutations. Its computational units are ranking filters, learned orderings that compare inputs via Spearman's footrule distance and update through permutation-matrix accumulation, a non-gradient rule rooted in displacement evidence. Layers compose hierarchically: each layer's output ranking becomes the next layer's input, enabling deep ordinal representation learning without any floating-point parameters in the core computation. We connect the architecture to Arrow's impossibility theorem, showing that violations of social-choice fairness axioms (context dependence, specialization, symmetry breaking) serve as inductive biases for nonlinearity, sparsity, and stability. Experiments span UCI tabular benchmarks, MNIST, gene expression cancer classification (TCGA), and preference data, all against GridSearchCV-tuned baselines. ArrowFlow beats all baselines on Iris (2.7% vs. 3.3%) and is competitive on most UCI datasets. A single parameter, polynomial degree, acts as a master switch: degree 1 yields noise robustness (8-28% less degradation), privacy preservation (+0.5pp cost), and missing-feature resilience; higher degrees trade these for improved clean accuracy. ArrowFlow is not designed to surpass gradient-based methods. It is an existence proof that competitive classification is possible in a fundamentally different computational paradigm, one that elevates ordinal structure to a first-class citizen, with natural alignment to integer-only and neuromorphic hardware.

Parent selection methods are widely used in evolutionary computation to accelerate the optimization process, yet their theoretical benefits are still poorly understood. In this paper, we address this gap by incorporating different parent selection strategies into the $(μ+1)$ genetic algorithm (GA). We show that, with an appropriately chosen population size and a parent selection strategy that selects a pair of maximally distant parents with probability $Ω(1)$ for crossover, the resulting algorithm solves the Jump$_k$ problem in $O(k4^kn\log(n))$ expected time. This bound is significantly smaller than the best known bound of $O(nμ\log(μ)+n\log(n)+n^{k-1})$ for any $(μ+1)$~GA using no explicit diversity-preserving mechanism and a constant crossover probability. To establish this result, we introduce a novel diversity metric that captures both the maximum distance between pairs of individuals in the population and the number of pairs achieving this distance. The crucial point of our analysis is that it relies on crossover as a mechanism for creating and maintaining diversity throughout the run, rather than using crossover only in the final step to combine already diversified individuals, as it has been done in many previous works. The insights provided by our analysis contribute to a deeper theoretical understanding of the role of crossover in the population dynamics of genetic algorithms.

Recent advancements in computer vision, driven by artificial intelligence, have significantly enhanced monitoring systems. One notable application is traffic monitoring, which leverages computer vision alongside deep learning-based object detection and counting. We present an offline, real-time traffic monitoring system that couples a pre-trained YOLOv11 detector with BoT-SORT/ByteTrack for multi-object tracking, implemented in PyTorch/OpenCV and wrapped in a Qt-based desktop UI. The CNN pipeline enables efficient vehicle detection and counting from video streams without cloud dependencies. Across diverse scenes, the system achieves (66.67-95.83%) counting accuracy. Class-wise detection yields high precision (cars: 0.97-1.00; trucks: 1.00) with strong recall (cars: 0.82-1.00; trucks: 0.70-1.00), resulting in F1 scores of (0.90-1.00 for cars and 0.82-1.00 for trucks). While adverse weather conditions may negatively impact this performance, results remain robust in typical conditions. By integrating lightweight models with an accessible, cloud-independent interface, this paper contributes to the modernization and development of future smart cities by showing the capacity of AI-driven traffic monitoring systems.

Cardiac magnetic resonance (CMR) is a cornerstone for diagnosing cardiovascular disease. However, it remains underutilized due to complex, time-consuming interpretation across multi-sequences, phases, quantitative measures that heavily reliant on specialized expertise. Here, we present BAAI Cardiac Agent, a multimodal intelligent system designed for end-to-end CMR interpretation. The agent integrates specialized cardiac expert models to perform automated segmentation of cardiac structures, functional quantification, tissue characterization and disease diagnosis, and generates structured clinical reports within a unified workflow. Evaluated on CMR datasets from two hospitals (2413 patients) spanning 7-types of major cardiovascular diseases, the agent achieved an area under the receiver-operating-characteristic curve exceeding 0.93 internally and 0.81 externally. In the task of estimating left ventricular function indices, the results generated by this system for core parameters such as ejection fraction, stroke volume, and left ventricular mass are highly consistent with clinical reports, with Pearson correlation coefficients all exceeding 0.90. The agent outperformed state-of-the-art models in segmentation and diagnostic tasks, and generated clinical reports showing high concordance with expert radiologists (six readers across three experience levels). By dynamically orchestrating expert models for coordinated multimodal analysis, this agent framework enables accurate, efficient CMR interpretation and highlights its potentials for complex clinical imaging workflows. Code is available at https://github.com/plantain-herb/Cardiac-Agent.

Scholarly publishing faces increasingly strong stressors, including submission overload, reviewer fatigue, inconsistent evaluation, governance opacity, and vulnerability to manipulation in old and new forms. While recent studies applied artificial intelligence to improve specific steps (e.g., triage, reviewer recommendation, or automated critique), they typically work under centralized editorial control and offer limited mechanisms for system-level adaptivity and auditability. Here we present ADAPT (AI-Driven Decentralized Adaptive Publishing Testbed), an agent-based environment that models journal management as a closed-loop control system rather than a fixed editorial workflow. ADAPT integrates interacting agents in various pools (authors, reviewers -- human and AI -- and rotating editors) coupled through policy-level control and diverse feedback signals. Governance adapts to backlog pressure, reviewer disagreement, paper quality drifting, and other relevant factors, while keeping human decision authority, role non-permanence, and data confidentiality. We evaluate ADAPT in a discrete-time simulation setting across multiple operational regimes, including baseline operation, submission surges, quality drift, disagreement escalation, post-publication learning, and collusion suppression. Across regimes, we quantify backlog dynamics, reviewer load, coordination activity, and management performance. The results indicate that ADAPT works under nominal and perturbed conditions, exhibits bounded and interpretable responses under stress, and mitigates clusters with embedded interventions. This feasibility demonstration suggests a promising direction of academic publishing practice, and can be extended to real-world implementations in suitable scenarios.

Peer review in machine learning is under growing pressure from rising submission volume and limited reviewer time. Most LLM-based reviewing systems read only the manuscript and generate comments from the paper's own narrative. This makes their outputs sensitive to presentation quality and leaves them weak when the evidence needed for review lies in related work or released code. We present FactReview, an evidence-grounded reviewing system that combines claim extraction, literature positioning, and execution-based claim verification. Given a submission, FactReview identifies major claims and reported results, retrieves nearby work to clarify the paper's technical position, and, when code is available, executes the released repository under bounded budgets to test central empirical claims. It then produces a concise review and an evidence report that assigns each major claim one of five labels: Supported, Supported by the paper, Partially supported, In conflict, or Inconclusive. In a case study on CompGCN, FactReview reproduces results that closely match those reported for link prediction and node classification, yet also shows that the paper's broader performance claim across tasks is not fully sustained: on MUTAG graph classification, the reproduced result is 88.4%, whereas the strongest baseline reported in the paper remains 92.6%. The claim is therefore only partially supported. More broadly, this case suggests that AI is most useful in peer review not as a final decision-maker, but as a tool for gathering evidence and helping reviewers produce more evidence-grounded assessments. The code is public at https://github.com/DEFENSE-SEU/Review-Assistant.

Small language models (SLMs) in the 100M-10B parameter range increasingly power production systems, yet whether they possess the internal emotion representations recently discovered in frontier models remains unknown. We present the first comparative analysis of emotion vector extraction methods for SLMs, evaluating 9 models across 5 architectural families (GPT-2, Gemma, Qwen, Llama, Mistral) using 20 emotions and two extraction methods (generation-based and comprehension-based). Generation-based extraction produces statistically superior emotion separation (Mann-Whitney p = 0.007; Cohen's d = -107.5), with the advantage modulated by instruction tuning and architecture. Emotion representations localize at middle transformer layers (~50% depth), following a U-shaped curve that is architecture-invariant from 124M to 3B parameters. We validate these findings against representational anisotropy baselines across 4 models and confirm causal behavioral effects through steering experiments, independently verified by an external emotion classifier (92% success rate, 37/40 scenarios). Steering reveals three regimes -- surgical (coherent text transformation), repetitive collapse, and explosive (text degradation) -- quantified by perplexity ratios and separated by model architecture rather than scale. We document cross-lingual emotion entanglement in Qwen, where steering activates semantically aligned Chinese tokens that RLHF does not suppress, raising safety concerns for multilingual deployment. This work provides methodological guidelines for emotion research on open-weight models and contributes to the Model Medicine series by bridging external behavioral profiling with internal representational analysis.

As Large Language Models (LLMs) are increasingly deployed in complex applications, their vulnerability to adversarial attacks raises urgent safety concerns, especially those evolving over multi-round interactions. Existing defenses are largely reactive and struggle to adapt as adversaries refine strategies across rounds. In this work, we propose CoopGuard , a stateful multi-round LLM defense framework based on cooperative agents that maintains and updates an internal defense state to counter evolving attacks. It employs three specialized agents (Deferring Agent, Tempting Agent, and Forensic Agent) for complementary round-level strategies, coordinated by System Agent, which conditions decisions on the evolving defense state (interaction history) and orchestrates agents over time. To evaluate evolving threats, we introduce the EMRA benchmark with 5,200 adversarial samples across 8 attack types, simulating progressively LLM multi-round attacks. Experiments show that CoopGuard reduces attack success rate by 78.9% over state-of-the-art defenses, while improving deceptive rate by 186% and reducing attack efficiency by 167.9%, offering a more comprehensive assessment of multi-round defense. These results demonstrate that CoopGuard provides robust protection for LLMs in multi-round adversarial scenarios.

Although coding agents have introduced new coordination dynamics in collaborative software development, detailed interactions in practice remain underexplored, especially for the code review process. In this study, we mine agent-authored PR references from the AIDev dataset and introduce a taxonomy to characterize the intent of these references across Human-to-Agent and Agent-to-Agent interactions in the form of Pull Requests (i.e. PRs). Our analysis shows that while humans initiate most references to agent-authored PRs, a substantial portion of these interactions are AI-assisted, indicating the emergence of meta-collaborative workflows, where humans mostly use references to build new features, whereas agents make them to fix errors. We further find that referencing/referenced PRs are associated with substantially longer lifespans and review times compared to isolated PRs, suggesting higher coordination or integration effort. We then list three key takeaways as potential future research directions into how to utilize these dynamics for optimizing AI coding agents in the code review process.

Hybrid physical systems combine continuous and discrete dynamics, which can be simultaneously affected by faults. Conventional fault detection methods often treat these dynamics separately, limiting their ability to capture interacting fault patterns. This paper proposes a unified fault detection framework for hybrid dynamical systems by integrating an Extended Timed Continuous Petri Net (ETCPN) model with semi-supervised anomaly detection. The proposed ETCPN extends existing Petri net formalisms by introducing marking-dependent flow functions, enabling intrinsic coupling between discrete and continuous dynamics. Based on this structure, a mode-dependent hybrid observer is designed, whose stability under arbitrary switching is ensured via Linear Matrix Inequalities (LMIs), solved offline to determine observer gains. The observer generates residuals that reflect discrepancies between the estimated and measured outputs. These residuals are processed using semi-supervised methods, including One-Class SVM (OC-SVM), Support Vector Data Description (SVDD), and Elliptic Envelope (EE), trained exclusively on normal data to avoid reliance on labeled faults. The framework is validated through simulations involving discrete faults, continuous faults, and hybrid faults. Results demonstrate high detection accuracy, fast convergence, and robust performance, with OC-SVM and SVDD providing the best trade-off between detection rate and false alarms. The framework is computationally efficient for real-time deployment, as the main complexity is confined to the offline LMI design phase.

Flow-matching methods for 3D shape assembly learn point-wise velocity fields that transport parts toward assembled configurations, yet they receive no explicit guidance about which cross-part interactions should drive the motion. We introduce TORA, a topology-first representation alignment framework that distills relational structure from a frozen pretrained 3D encoder into the flow-matching backbone during training. We first realize this via simple instantiation, token-wise cosine matching, which injects the learned geometric descriptors from the teacher representation. We then extend to employ a Centered Kernel Alignment (CKA) loss to match the similarity structure between student and teacher representations for enhanced topological alignment. Through systematic probing of diverse 3D encoders, we show that geometry- and contact-centric teacher properties, not semantic classification ability, govern alignment effectiveness, and that alignment is most beneficial at later transformer layers where spatial structure naturally emerges. TORA introduces zero inference overhead while yielding two consistent benefits: faster convergence (up to 6.9$\times$) and improved accuracy in-distribution, along with greater robustness under domain shift. Experiments on five benchmarks spanning geometric, semantic, and inter-object assembly demonstrate state-of-the-art performance, with particularly pronounced gains in zero-shot transfer to unseen real-world and synthetic datasets. Project page: https://nahyuklee.github.io/tora.

Software testing research has traditionally relied on closed-world assumptions, such as finite state spaces, reproducible executions, and stable test oracles. However, many modern software systems operate under uncertainty, non-determinism, and evolving conditions, challenging these assumptions. This paper uses open-world games as an extreme case to examine the limitations of closed-world testing. Through a set of observations grounded in prior work, we identify recurring characteristics that complicate testing in such systems, including inexhaustible behavior spaces, non-deterministic execution outcomes, elusive behavioral boundaries, and unstable test oracles. Based on these observations, we articulate a vision of software testing beyond closed-world assumptions, in which testing supports the characterization and interpretation of system behavior under uncertainty. We further discuss research directions for automated test generation, evaluation metrics, and empirical study design. Although open-world games serve as the motivating domain, the challenges and directions discussed in this paper extend to a broader class of software systems operating in dynamic and uncertain environments.

In large software ecosystems, semantically related code changes, such as alternative solutions or overlapping modifications are often discovered only days after submission, leading to duplicated effort and delayed reviews. We present SmartPatchLinker, a browser based tool that supports the discovery of related patches directly within the code review interface. SmartPatchLinker is implemented as a lightweight Chrome extension with a local inference backend and integrates with Gerrit to retrieve and rank semantically linked changes when a reviewer opens a patch. The tool allows reviewers to configure the search scope, view ranked candidates with confidence indicators, and examine related work without leaving their workflow or relying on server-side installations. We perform both usefulness and usability evaluations to study how SmartPatchLinker can support reviewers during code review. SmartPatchLinker is open source, and its source code, Docker containers, and the replication package used in our evaluation are publicly available on GitHub at https://github.com/islem-kms/gerrit-chrome-extension . A video demonstrating the tool is also available online at https://drive.google.com/drive/folders/1MCcTj5OSlT7lHVBFMq5m9iatas2joaGb

Recent work has shown that fine-tuning large language models (LLMs) on insecure code or culturally loaded numeric codes can induce emergent misalignment, causing models to produce harmful content in unrelated downstream tasks. The authors of that work concluded that $k$-shot prompting alone does not induce this effect. We revisit this conclusion and show that inference-time semantic drift is real and measurable; however, it requires models of large-enough capability. Using a controlled experiment in which five culturally loaded numbers are injected as few-shot demonstrations before a semantically unrelated prompt, we find that models with richer cultural-associative representations exhibit significant distributional shifts toward darker, authoritarian, and stigmatized themes, while a simpler/smaller model does not. We additionally find that structurally inert demonstrations (nonsense strings) perturb output distributions, suggesting two separable mechanisms: structural format contamination and semantic content contamination. Our results map the boundary conditions under which inference-time contamination occurs, and carry direct implications for the security of LLM-based applications that use few-shot prompting.

Knowledge distillation compresses large teachers into smaller students, but performance saturates at a loss floor that persists across training methods and objectives. We argue this floor is geometric: neural networks represent far more features than dimensions through superposition, and a student of width $d_S$ can encode at most $d_S \cdot g(α)$ features, where $g(α) = 1/((1-α)\ln\frac{1}{1-α})$ is a sparsity-dependent capacity function. Features beyond this budget are permanently lost, yielding an importance-weighted loss floor. We validate on a toy model (48 configurations, median accuracy >93%) and on Pythia-410M, where sparse autoencoders measure $F \approx 28{,}700$ features at $α\approx 0.992$ (critical width $d_S^* \approx 1{,}065$). Distillation into five student widths confirms the predicted monotonic floor ordering. The observed floor decomposes into a geometric component and a width-independent architectural baseline ($R^2 = 0.993$). Linear probing shows coarse concepts survive even 88% feature loss, revealing the floor arises from aggregate loss of fine-grained features in the importance distribution's long tail. Our results connect representation geometry to distillation limits and provide a practical tool for predicting distillation performance from SAE measurements alone.

Novice math teachers often encounter students' mistakes that are difficult to diagnose and remediate. Misconceptions are especially challenging because teachers must explain what went wrong and how to solve them. Although many existing large language model (LLM) platforms can assist in generating instructional feedback, these LLMs loosely connect pedagogical knowledge and student mistakes, which might make the guidance less actionable for teachers. To address this gap, we propose MisEdu-RAG, a dual-hypergraph-based retrieval-augmented generation (RAG) framework that organizes pedagogical knowledge as a concept hypergraph and real student mistake cases as an instance hypergraph. Given a query, MisEdu-RAG performs a two-stage retrieval to gather connected evidence from both layers and generates a response grounded in the retrieved cases and pedagogical principles. We evaluate on \textit{MisstepMath}, a dataset of math mistakes paired with teacher solutions, as a benchmark for misconception-aware retrieval and response generation across topics and error types. Evaluation results on \textit{MisstepMath} show that, compared with baseline models, MisEdu-RAG improves token-F1 by 10.95\% and yields up to 15.3\% higher five-dimension response quality, with the largest gains on \textit{Diversity} and \textit{Empowerment}. To verify its applicability in practical use, we further conduct a pilot study through a questionnaire survey of 221 teachers and interviews with 6 novices. The findings suggest that MisEdu-RAG provides diagnosis results and concrete teaching moves for high-demand misconception scenarios. Overall, MisEdu-RAG demonstrates strong potential for scalable teacher training and AI-assisted instruction for misconception handling. Our code is available on GitHub: https://github.com/GEMLab-HKU/MisEdu-RAG.

Tool-calling LLM agents can read private data, invoke external services, and trigger real-world actions, creating a security problem at the point of tool execution. We identify a denial-feedback leakage pattern, which we term causality laundering, in which an adversary probes a protected action, learns from the denial outcome, and exfiltrates the inferred information through a later seemingly benign tool call. This attack is not captured by flat provenance tracking alone because the leaked information arises from causal influence of the denied action, not direct data flow. We present the Agentic Reference Monitor (ARM), a runtime enforcement layer that mediates every tool invocation by consulting a provenance graph over tool calls, returned data, field-level provenance, and denied actions. ARM propagates trust through an integrity lattice and augments the graph with counterfactual edges from denied-action nodes, enabling enforcement over both transitive data dependencies and denial-induced causal influence. In a controlled evaluation on three representative attack scenarios, ARM blocks causality laundering, transitive taint propagation, and mixed-provenance field misuse that a flat provenance baseline misses, while adding sub-millisecond policy evaluation overhead. These results suggest that denial-aware causal provenance is a useful abstraction for securing tool-calling agent systems.

Connectome-constrained neural networks are often evaluated against sparse random controls and then interpreted as evidence that biological graph topology improves learning efficiency. We revisit that claim in a controlled flyvis-based study using a Drosophila connectome, a naive self-loop-matched random graph, and a degree-preserving rewired null. Under weak controls, in which both models were recovered from a connectome-trained checkpoint and the null matched only global graph counts, the connectome appeared substantially better in early loss, mean activity, and runtime. That picture changed under stricter controls. Training both graphs from a shared random initialization removed the early loss advantage, and replacing the naive null by a degree-preserving null removed the apparent activity advantage. A five-sample degree-preserving ensemble and a pre-training activity-scale diagnostic further strengthened this revised interpretation. We also report a descriptive mechanism analysis of the earlier weak-control comparison, but we treat it as behavioral characterization rather than proof of causal superiority. We show that previously reported topology advantages in connectome-constrained neural networks can arise from initialization and null-model confounds, and largely disappear under fair from-scratch initialization and degree-preserving controls.

With the increasing importance of data privacy and security, federated unlearning emerges as a new research field dedicated to ensuring that once specific data is deleted, federated learning models no longer retain or disclose related information. In this paper, we propose a zero-shot federated unlearning scheme, named Jellyfish. It distinguishes itself from conventional federated unlearning frameworks in four key aspects: synthetic data generation, knowledge disentanglement, loss function design, and model repair. To preserve the privacy of forgotten data, we design a zero-shot unlearning mechanism that generates error-minimization noise as proxy data for the data to be forgotten. To maintain model utility, we first propose a knowledge disentanglement mechanism that regularises the output of the final convolutional layer by restricting the number of activated channels for the data to be forgotten and encouraging activation sparsity. Next, we construct a comprehensive loss function that incorporates multiple components, including hard loss, confusion loss, distillation loss, model weight drift loss, gradient harmonization, and gradient masking, to effectively align the learning trajectories of the objectives of ``forgetting" and ``retaining". Finally, we propose a zero-shot repair mechanism that leverages proxy data to restore model accuracy within acceptable bounds without accessing users' local data. To evaluate the performance of the proposed zero-shot federated unlearning scheme, we conducted comprehensive experiments across diverse settings. The results validate the effectiveness and robustness of the scheme.

This paper establishes a unified element-based framework for formation control by introducing the concept of the deformation gradient from continuum mechanics. Unlike traditional methods that rely on geometric constraints defined on graph edges, we model the formation as a discrete elastic body composed of simplicial elements. By defining a generalized distortion energy based on the local deformation gradient tensor, we derive a family of distributed control laws that can enforce various geometric invariances, including translation, rotation, scaling, and affine transformations. The convergence properties and the features of the proposed controllers are analyzed in detail. Theoretically, we show that the proposed framework serves as a bridge between existing rigidity-based and Laplacian-based approaches. Specifically, we show that rigidity-based controllers are mathematically equivalent to minimizing specific projections of the deformation energy tensor. Furthermore, we establish a rigorous link between the proposed energy minimization and Laplacian-based formation control. Numerical simulations in 2D and 3D validate the effectiveness and the unified nature of the proposed framework.

This paper primarily focuses on the hallucinations caused due to AI language models(LLMs).LLMs have shown extraordinary Language understanding and generation capabilities .Still it has major a disadvantage hallucinations which give outputs which are factually incorrect ,misleading or unsupported by input data . These hallucinations cause serious problems in scenarios like medical diagnosis or legal reasoning.Through this work,we propose causal graph attention network (GCAN) framework that reduces hallucinations through interpretation of internal attention flow within a transformer architecture with the help of constructing token level graphs that combine self attention weights and gradient based influence scores.our method quantifies each tokens factual dependency using a new metric called the Causal Contribution Score (CCS). We further introduce a fact-anchored graph reweighting layer that dynamically reduces the influence of hallucination prone nodes during generation. Experiments on standard benchmarks such as TruthfulQA and HotpotQA show a 27.8 percent reduction in hallucination rate and 16.4 percent improvement in factual accuracy over baseline retrieval-augmented generation (RAG) models. This work contributes to the interpretability,robustness, and factual reliability of future LLM architectures.

Deep research agents integrate fragmented evidence through multi-step tool use. BrowseComp offers a text-only testbed for such agents, but existing multimodal benchmarks rarely require both weak visual cues composition and BrowseComp-style multi-hop verification. Geolocation is a natural testbed because answers depend on combining multiple ambiguous visual cues and validating them with open-web evidence. Thus, we introduce GeoBrowse, a geolocation benchmark that combines visual reasoning with knowledge-intensive multi-hop queries. Level 1 tests extracting and composing fragmented visual cues, and Level 2 increases query difficulty by injecting long-tail knowledge and obfuscating key entities. To support evaluation, we provide an agentic workflow GATE with five think-with-image tools and four knowledge-intensive tools, and release expert-annotated stepwise traces grounded in verifiable evidence for trajectory-level analysis. Experiments show that GATE outperforms direct inference and open-source agents, indicating that no-tool, search-only or image-only setups are insufficient. Gains come from coherent, level-specific tool-use plans rather than more tool calls, as they more reliably reach annotated key evidence steps and make fewer errors when integrating into the final decision. The GeoBrowse bernchmark and codes are provided in https://github.com/ornamentt/GeoBrowse

Reconstructing dynamic scenes with complex human-object interactions is a fundamental challenge in computer vision and graphics. Existing Gaussian Splatting methods either rely on human pose priors while neglecting dynamic objects, or approximate all motions within a single field, limiting their ability to capture interaction-rich dynamics. To address this gap, we propose Human-Object Interaction Gaussian Splatting (HOIGS), which explicitly models interaction-induced deformation between humans and objects through a cross-attention-based HOI module. Distinct deformation baselines are employed to extract features: HexPlane for humans and Cubic Hermite Spline (CHS) for objects. By integrating these heterogeneous features, HOIGS effectively captures interdependent motions and improves deformation estimation in scenarios involving occlusion, contact, and object manipulation. Comprehensive experiments on multiple datasets demonstrate that our method consistently outperforms state-of-the-art human-centric and 4D Gaussian approaches, highlighting the importance of explicitly modeling human-object interactions for high-fidelity reconstruction.

The growing complexity of real-time embedded systems demands strong isolation of software components into separate protection domains to reduce attack surfaces and limit fault propagation. However, application-supplied device interrupt handlers -- even untrusted -- have to remain in the kernel to minimize interrupt latency, undermining security and burdening manual certifications. Current hardware extensions accelerate interrupts only when the target protection domain is scheduled by the kernel; consequently, they are limited to improving average-case performance but not worst-case latency, and do not meet the requirements of critical real-time applications such as autonomous vehicles or robots. To overcome this limitation, we propose a novel hardware extension that enables direct, deterministic switching to the appropriate protection domain upon user-level interrupt arrival -- without kernel intervention -- even when that domain is dormant. Our hardware extension reduces worst-case latency by more than 50x with a 19% increase in core area (2% of total die area) and 4.1% increase in dynamic power. To the best of our knowledge, this is the first integrated mechanism to guarantee user-level interrupt delivery with a nanosecond-scale yet bounded worst-case latency.

The increasing size and complexity of large language models (LLMs) have raised significant challenges in deployment efficiency, particularly under resource constraints. Post-training quantization (PTQ) has emerged as a practical solution by compressing models without requiring retraining. While existing methods focus on uniform quantization schemes for both weights and activations, they often suffer from substantial accuracy degradation due to the non-uniform nature of activation distributions. In this work, we revisit the activation quantization problem from a theoretical perspective grounded in the Lloyd-Max optimality conditions. We identify the core issue as the non-uniform distribution of activations within the quantization interval, which causes the optimal quantization point under the Lloyd-Max criterion to shift away from the midpoint of the interval. To address this issue, we propose a two-stage orthogonal transformation method, RUQuant. In the first stage, activations are divided into blocks. Each block is mapped to uniformly sampled target vectors using composite orthogonal matrices, which are constructed from Householder reflections and Givens rotations. In the second stage, a global Householder reflection is fine-tuned to further minimize quantization error using Transformer output discrepancies. Empirical results show that our method achieves near-optimal quantization performance without requiring model fine-tuning: RUQuant achieves 99.8% of full-precision accuracy with W6A6 and 97% with W4A4 quantization for a 13B LLM, within approximately one minute. A fine-tuned variant yields even higher accuracy, demonstrating the effectiveness and scalability of our approach.

Severe occlusions of objects pose a major challenge for computer vision. We show that two root causes are (1) the loss of visible information and (2) the distracting patterns caused by the occluders. Our approach addresses both causes at the same time. First, the distracting patterns are removed at test-time, via masking of the occluding patterns. This masking is independent of the type of occlusion, by handling the occlusion through the lens of visual anomalies w.r.t. the object of interest. Second, to deal with less visual details, we follow standard practice by masking random parts of the object during training, for various degrees of occlusions. We discover that (a) it is possible to estimate the degree of the occlusion (i.e. severity) at test-time, and (b) that a model optimized for a specific degree of occlusion also performs best on a similar degree during test-time. Combining these two insights brings us to a severity-informed classification model called OASIC: Occlusion Agnostic Severity Informed Classification. We estimate the severity of occlusion for a test image, mask the occluder, and select the model that is optimized for the degree of occlusion. This strategy performs better than any single model optimized for any smaller or broader range of occlusion severities. Experiments show that combining gray masking with adaptive model selection improves $\text{AUC}_\text{occ}$ by +18.5 over standard training on occluded images and +23.7 over finetuning on unoccluded images.

Software architecture diagrams are important design artifacts for communicating system structure, behavior, and data organization throughout the software development lifecycle. Although recent progress in large language models has substantially advanced code-centric software engineering tasks such as code generation, testing, and maintenance, the ability of modern vision-language models (VLMs) to understand software architecture diagrams remains underexplored. To address this gap, we present SADU, a benchmark for Software Architecture Diagram Understanding that evaluates VLMs on architecture diagrams as structured software engineering artifacts rather than generic images. SADU contains 154 carefully curated diagrams spanning behavioral, structural, and ER diagrams, paired with structured annotations and 2,431 question-answer tasks covering counting and retrieval reasoning. We evaluate 11 state-of-the-art VLMs from the Gemini, Claude, GPT, and Qwen families. Our results show that software architecture diagram understanding remains challenging for current models: the best-performing model gemini-3-flash-preview achieves only 70.18\% accuracy, while gpt-4o-mini only achieves 17.77\% accuracy. The results further reveal the weaknesses in diagram reasoning and visual relation grounding, highlighting a gap between current VLMs and the needs of design-stage software engineering. SADU provides a foundation for future research on diagram-aware AI systems and more faithful AI-assisted software engineering workflows.

Autonomous software agents on blockchains solve distributed-coordination problems by reading shared ledger state instead of exchanging direct messages. Liquidation keepers, arbitrage bots, and other autonomous on-chain agents watch balances, contract storage, and event logs; when conditions change, they act. The ledger therefore functions as a replicated shared-state medium through which decentralized agents coordinate indirectly. This form of indirect coordination mirrors what Grassé called stigmergy in 1959: organisms coordinating through traces left in a shared environment, with no central plan. Stigmergy has mature formalizations in swarm intelligence and multi-agent systems, and on-chain agents already behave stigmergically in practice, but no prior application-layer framework cleanly bridges the two. We introduce Indirect coordination grounded in ledger state (Coordinación indirecta basada en el estado del registro contable) as a ledger-specific applied definition that maps Grassé's mechanism onto distributed ledger technology. We operationalize this with a state-transition formalism, identify three recurring base on-chain coordination patterns (State-Flag, Event-Signal, Threshold- Trigger) together with a Commit-Reveal sequencing overlay, and work through a State-Flag task-board example to compare ledger-state coordination analytically with off-chain messaging and centralized orchestration. The contribution is a reusable vocabulary, a ledger-specific formal mapping, and design guidance for decentralized coordination over replicated shared state at the application layer.

Reinforcement Learning with Verifiable Rewards (RLVR) effectively trains reasoning models that rely on abundant perfect labels, but its vulnerability to unavoidable noisy labels due to expert scarcity remains critically underexplored. In this work, we take the first step toward a systematic analysis of noisy label mechanisms in RLVR. In contrast to supervised classification, most RLVR algorithms incorporate a rollout-based condition: a label's influence on training is contingent on whether the current policy can generate rollouts that realize it, a property that naturally extends to noisy labels. Based on this observation, we distinguish two types of noise: inactive noisy labels, which reduce data efficiency, and active noisy labels, which are reinforced and risk skewing the model toward incorrect distributions. From experiments on training with noisy samples, we identify an Early Correctness Coherence phenomenon: although noisy samples begin to lag behind in later stages, accuracy on both clean and noisy samples increases similarly in early training. Motivated by this dynamic, we propose Online Label Refinement (OLR), which progressively corrects potentially noisy labels with majority-voted answers when two conditions hold: a positive slope in the majority answer's rollout pass rate and stable historical consistency across updates, enabling gradual self-correction as the policy improves. We evaluate OLR on six in-distribution mathematical reasoning benchmarks (AIME24/25, AMC, MATH-500, Minerva, and Olympiad) and three out-of-distribution tasks (ARC-c, GPQA-diamond, and MMLU-pro). Across noise ratios from 0.1 to 0.9, OLR consistently improves robustness under both inactive and active noisy-label settings, achieving average gains of 3.6% to 3.9% on in-distribution benchmarks and 3.3% to 4.6% on out-of-distribution evaluations.

Driven by the escalating demand for wireless capacity and advancements in 6G research, the new Frequency Range 3 (FR3) referred to upper mid-band (7.125-24.25 GHz) has emerged as a highly compelling spectrum candidate. This range offers a trade-off exploiting the high bandwidth capabilities of millimeter wave frequencies and the superior propagation characteristics of sub-6 GHz bands. As such, the upper mid-band presents an opportunity to enhance both coverage and capacity particularly in the context of 6G and Cellular Vehicle-to-Base Station (C-V2B). Crucially, realizing this potential requires overcoming technical challenges through accurate and realistic channel modeling, especially in dense, high-rise urban environments. To address this, we employ a ray-tracing tool to analyze downlink propagation characteristics, enabling detailed channel modeling for reliable C-V2B communication. Our analysis evaluates the signal-to-noise ratio (SNR) and signal-to-interference-plus-noise ratio (SINR) across sub-6 GHz, FR3, and mmWave bands using antenna array configurations designed for high-rise urban areas. Results show that, under equal aperture sizes across frequencies, FR3 achieves superior SNR compared to mmWave in interference-free conditions. Moreover, under the full-interference case, FR3 yields higher SINR for cell-edge User Equipment (UEs). This indicates that the increased array gain at mmWave cannot fully compensate for the severe path loss experienced by cell-edge UEs.

COBOL remains a critical language for mainframe systems, yet existing large language models (LLMs) struggle to generate and translate COBOL code correctly. This paper reports our experience in developing and evaluating domain-adapted LLMs for COBOL and mainframe software engineering. We introduce (1) an automated data curation pipeline that combines compiler-guided validation with multi-stage similarity-based filtering to construct high-quality COBOL training data, and (2) COBOL-Coder, a COBOL-specialized LLM fine-tuned on the curated COBOL domain data. We evaluate COBOL-Coder on two tasks: code generation (on COBOLEval and COBOLCodeBench) and code translation (on COBOL-JavaTrans, our proposed benchmark for bidirectional COBOL-Java translation). In our experiments, COBOL-Coder achieves up to a 73.95 percent compilation success rate and 49.33 Pass-1 on COBOLEval, compared to 41.8 percent and 16.4 for GPT-4o, while most open-source baselines (e.g., CodeGemma, CodeLlama, StarCoder2) fail to produce compilable programs. For Java-to-COBOL translation, COBOL-Coder reaches 34.93 Pass-1, whereas general-purpose LLMs achieve near-zero scores. To assess the usability of LLM-generated code in real-world settings, we conduct a survey with experienced COBOL developers. Participants consistently report that COBOL-Coder exhibits stronger COBOL awareness, has more reliable program structure, and is better aligned with enterprise practices than general-purpose LLMs.

Damped sinusoidal oscillations are widely observed in many physical systems, and their analysis provides access to underlying physical properties. However, parameter estimation becomes difficult when the signal decays rapidly, multiple components are superposed, and observational noise is present. In this study, we develop an autoencoder-based method that uses the latent space to estimate the frequency, phase, decay time, and amplitude of each component in noisy multi-component damped sinusoidal signals. We investigate multi-component cases under Gaussian-distribution training and further examine the effect of the training-data distribution through comparisons between Gaussian and uniform training. The performance is evaluated through waveform reconstruction and parameter-estimation accuracy. We find that the proposed method can estimate the parameters with high accuracy even in challenging setups, such as those involving a subdominant component or nearly opposite-phase components, while remaining reasonably robust when the training distribution is less informative. This demonstrates its potential as a tool for analyzing short-duration, noisy signals.

Many particle-based Bayesian inference methods use a single global step size for all parts of the update. In Stein variational gradient descent (SVGD), however, each update combines two qualitatively different effects: attraction toward high-posterior regions and repulsion that preserves particle diversity. These effects can evolve at different rates, especially in high-dimensional, anisotropic, or hierarchical posteriors, so one step size can be unstable in some regions and inefficient in others. We derive a multirate version of SVGD that updates these components on different time scales. The framework yields practical algorithms, including a symmetric split method, a fixed multirate method (MR-SVGD), and an adaptive multirate method (Adapt-MR-SVGD) with local error control. We evaluate the methods in a broad and rigorous benchmark suite covering six problem families: a 50D Gaussian target, multiple 2D synthetic targets, UCI Bayesian logistic regression, multimodal Gaussian mixtures, Bayesian neural networks, and large-scale hierarchical logistic regression. Evaluation includes posterior-matching metrics, predictive performance, calibration quality, mixing, and explicit computational cost accounting. Across these six benchmark families, multirate SVGD variants improve robustness and quality-cost tradeoffs relative to vanilla SVGD. The strongest gains appear on stiff hierarchical, strongly anisotropic, and multimodal targets, where adaptive multirate SVGD is usually the strongest variant and fixed multirate SVGD provides a simpler robust alternative at lower cost.

Parameter-efficient prompt learning has become the de facto standard for adapting Vision-Language Models (VLMs) to downstream tasks. Existing approaches predominantly focus on aligning text prompts with first-order visual features (i.e., spatial feature maps). While effective for fine-grained semantic discrimination, we argue that relying solely on first-order information is insufficient for robust adaptation, as these spatially entangled features are highly susceptible to domain shifts and local noise. In this work, we propose \textbf{Gram-Anchored Prompt Learning (GAPL)} for Vision-Language Models via Second-Order Statistics, a framework that synergizes local semantic alignment with global structural consistency. Methodologically, we introduce an additional second-order statistical stream via \textbf{Gram matrices} that augments the standard first-order spatial interaction. By anchoring prompts to these second-order priors, our approach enables language representations to dynamically adapt to statistical distribution shifts across diverse domains. Extensive experiments indicate the effectiveness of the second-order features, and show compelling performances of GAPL on various benchmarks.

Legacy programming languages such as COBOL (Common Business-Oriented Language) remain critical in business computing. However, maintaining legacy COBOL systems is increasingly challenging due to a declining pool of skilled developers and the persistence of COBOL errors that require deep domain expertise to resolve. This paper investigates the challenges of COBOL compilation errors and introduces a framework leveraging large language models (LLMs) to address these issues. We first categorize the common compilation errors in LLM-generated COBOL code into three groups: incomplete code errors, syntax errors, and type-related errors. We further propose COBOLAssist, a technique to enhance code correctness through iterative repairs guided by compilation feedback. Our evaluation using five LLMs including GPT variants and mAInframer, shows a high prevalence of incorrect program structures and function usage in COBOL programs and demonstrates the effectiveness of COBOLAssist, with the compilation success rates increasing from 29.5\% to 64.38\% for GPT-4o-mini and from 41.8\% to 95.89\% for GPT-4o. It also improves pass@1 significantly, for example from 9.1 to 22.6 for GPT-4. Notably, while mAInframer-34B achieves the highest compilation success rate, its functional correctness remains limited. This research not only highlights the limitations in current LLMs for COBOL but also demonstrates a practical path forward for automated debugging in legacy systems.

Prior work on trustworthy AI emphasizes model-internal properties such as bias mitigation, adversarial robustness, and interpretability. As AI systems evolve into autonomous agents deployed in open environments and increasingly connected to payments or assets, the operational meaning of trust shifts to end-to-end outcomes: whether an agent completes tasks, follows user intent, and avoids failures that cause material or psychological harm. These risks are fundamentally product-level and cannot be eliminated by technical safeguards alone because agent behavior is inherently stochastic. To address this gap between model-level reliability and user-facing assurance, we propose a complementary framework based on risk management. Drawing inspiration from financial underwriting, we introduce the \textbf{Agentic Risk Standard (ARS)}, a payment settlement standard for AI-mediated transactions. ARS integrates risk assessment, underwriting, and compensation into a single transaction framework that protects users when interacting with agents. Under ARS, users receive predefined and contractually enforceable compensation in cases of execution failure, misalignment, or unintended outcomes. This shifts trust from an implicit expectation about model behavior to an explicit, measurable, and enforceable product guarantee. We also present a simulation study analyzing the social benefits of applying ARS to agentic transactions. ARS's implementation can be found at https://github.com/t54-labs/AgenticRiskStandard.

DAG-Rider popularized a new paradigm of DAG-BFT protocols, separating dissemination from consensus: all nodes disseminate transactions as blocks that reference previously known blocks, while consensus is reached by electing certain blocks as leaders. This design yields high throughput but confers optimal latency only to leader blocks; non-leader blocks cannot be committed independently. We present Lemonshark, an asynchronous DAG-BFT protocol that reinterprets the DAG at a transactional level and identifies conditions where commitment is sufficient -- but not necessary -- for safe results, enabling nodes to finalize transactions before official commitment, without compromising correctness. Compared to the state-of-the-art asynchronous BFT protocol, Lemonshark reduces latency by up to 65\%.

We study fixed-confidence Best Arm Identification (BAI) in semiparametric bandits, where rewards are linear in arm features plus an unknown additive baseline shift. Unlike linear-bandit BAI, this setting requires orthogonalized regression, and its instance-optimal sample complexity has remained open. For the transductive setting, we establish an attainable instance-dependent lower bound characterized by the corresponding linear-bandit complexity on shifted features. We then propose a computationally efficient phase-elimination algorithm based on a new $XY$-design for orthogonalized regression. Our analysis yields a nearly optimal high-probability sample-complexity upper bound, up to log factors and an additive $d^2$ term, and experiments on synthetic instances and the Jester dataset show clear gains over prior baselines.

AI control protocols use monitors to detect attacks by untrusted AI agents, but standard single-score monitors face two limitations: they miss subtle attacks where outputs look clean but reasoning is off, and they collapse to near-zero safety when the monitor is the same model as the agent (collusion). We present TraceGuard, a structured multi-dimensional monitoring protocol that evaluates agent actions across five dimensions -- goal alignment, constraint adherence, reasoning coherence, safety awareness, and action-trace consistency -- scored in parallel by independent LLM calls, augmented by seven heuristic detectors and an LLM-based intent analyzer. We evaluate on BashArena (637 bash tasks, 4 attack categories) within the ControlArena framework. Our results on 519 samples (279 honest, 240 attack) show that: (1) the hybrid approach achieves clear attack-honest separation (attack mean 0.616 vs. honest mean 0.206, Delta=0.410); (2) structured scoring constrains collusion -- the untrusted structured monitor achieves 95% safety vs. 0% for single-score untrusted monitoring; (3) goal alignment and constraint adherence are the most discriminative dimensions; and (4) a separation-of-duties variant splitting dimensions across trusted and untrusted models achieves 100% safety while preventing any single model from seeing the full evaluation. TraceGuard is implemented as a new monitor type for the open-source ControlArena framework.

Modern scientific ecosystems are rich in procedural knowledge across repositories, APIs, scripts, notebooks, documentation, databases, and papers, yet much of this knowledge remains fragmented across heterogeneous artifacts that agents cannot readily operationalize. This gap between abundant scientific know-how and usable agent capabilities is a key bottleneck for building effective scientific agents. We present SkillFoundry, a self-evolving framework that converts such resources into validated agent skills, reusable packages that encode task scope, inputs and outputs, execution steps, environment assumptions, provenance, and tests. SkillFoundry organizes a target domain as a domain knowledge tree, mines resources from high-value branches, extracts operational contracts, compiles them into executable skill packages, and then iteratively expands, repairs, merges, or prunes the resulting library through a closed-loop validation process. SkillFoundry produces a substantially novel and internally valid skill library, with 71.1\% of mined skills differing from existing skill libraries such as SkillHub and SkillSMP. We demonstrate that these mined skills improve coding agent performance on five of the six MoSciBench datasets. We further show that SkillFoundry can design new task-specific skills on demand for concrete scientific objectives, and that the resulting skills substantially improve performance on two challenging genomics tasks: cell type annotation and the scDRS workflow. Together, these results show that automatically mined skills improve agent performance on benchmarks and domain-specific tasks, expand coverage beyond hand-crafted skill libraries, and provide a practical foundation for more capable scientific agents.

In many practical LLM deployments, a single guardrail is used for both prompt and response moderation. Prompt moderation operates on fully observed text, whereas streaming response moderation requires safety decisions to be made over partial generations. Existing text-based streaming guardrails commonly frame this output-side problem as boundary detection, training models to identify the earliest prefix at which a response has already become unsafe. In this work, we introduce StreamGuard, a unified model-agnostic streaming guardrail that instead formulates moderation as a forecasting problem: given a partial prefix, the model predicts the expected harmfulness of likely future continuations. We supervise this prediction using Monte Carlo rollouts, which enables early intervention without requiring exact token-level boundary annotations. Across standard safety benchmarks, StreamGuard performs strongly both for input moderation and for streaming output moderation. At the 8B scale, StreamGuard improves aggregated input-moderation F1 from 86.7 to 88.2 and aggregated streaming output-moderation F1 from 80.4 to 81.9 relative to Qwen3Guard-Stream-8B-strict. On the QWENGUARDTEST response_loc streaming benchmark, StreamGuard reaches 97.5 F1, 95.1 recall, and 92.6% on-time intervention, compared to 95.9 F1, 92.1 recall, and 89.9% for Qwen3Guard-Stream-8B-stric, while reducing the miss rate from 7.9% to 4.9%. We further show that forecasting-based supervision transfers effectively across tokenizers and model families: with transferred targets, Gemma3-StreamGuard-1B reaches 81.3 response-moderation F1, 98.2 streaming F1, and a 3.5% miss rate. These results show that strong end-to-end streaming moderation can be obtained without exact boundary labels, and that forecasting future risk is an effective supervision strategy for low-latency safety intervention.

Tensor network methods, particularly those based on Matrix Product States (MPS), provide a powerful framework for simulating quantum many-body systems. A persistent computational challenge in these methods is the selection of the bond dimension chi, which controls the trade-off between accuracy and computational cost. Fixed bond dimension strategies either waste resources in low-entanglement regions or lose fidelity in high-entanglement regions. This work introduces an adaptive bond dimension management framework that uses von Neumann entropy feedback coupled with a Proportional-Integral-Derivative (PID) controller to dynamically adjust chi at each bond during simulation. An Exponential Moving Average (EMA) filter stabilizes entropy measurements against transient fluctuations, and a predictive scheduling module anticipates future bond dimension requirements from entropy trends. The per-bond granularity of the allocation ensures that computational resources concentrate where entanglement is largest. The framework integrates GPU-accelerated Singular Value Decomposition (SVD) via CuPy and the cuSOLVER backend, achieving individual SVD speedups of 4.1x at chi=256 and 7.1x at chi=2048 relative to CPU-based NumPy for isolated matrix factorisations (measured on an NVIDIA A100-SXM4-40GB GPU with CuPy 13.4.1 and CUDA 12.8). At the system level, benchmarks on the spin-1/2 antiferromagnetic Heisenberg chain demonstrate a 2.7x reduction in total DMRG wall time compared to fixed-chi simulations, with energy accuracy within 0.1% of the Bethe ansatz solution. Integration with the Density Matrix Renormalization Group (DMRG) algorithm yields ground-state energies per site converging to E/N = -0.4432 for the isotropic Heisenberg model at chi = 128. Validation against Amazon Web Services (AWS) Braket SV1 statevector simulator confirms agreement within 2-5% for small systems.

Piezoelectric FET (PeFET) is a promising non-volatile-memory (NVM) device that integrates a piezoelectric (PE)/ferroelectric (FE) capacitor with a 2D transistor. It uses the polarization of the FE capacitor for bit-storage and strain-induced bandgap change of the 2D channel during read. Previous PeFET-based NVM designs have shown immense promise in achieving high density and energy-efficiency compared to SRAM. However, a key limitation of these designs is that they must trade-off integration density to enhance energy-efficiency or augment the memory functionality with in-memory computing (IMC). In this work, we show that the unique structure of the PeFET presents an appealing opportunity to counter these limitations, thereby simultaneously achieving high density, high energy-efficiency, and IMC-compatibility. First, we highlight the key reasons for the limited energy-efficiency of the previous PeFET designs. Based on these insights, we propose two flavors of PeFET memories that utilize negative voltage (NeVo) to reduce the major energy-consuming components significantly. Compared to 6T-SRAM (prior PeFET-based NVMs), the proposed designs achieve substantial reductions in energy, lowering read energy to 0.08x(0.03x) and write energy to 0.19x(0.53x), respectively. We then leverage these cells to implement IMC primitives, such as addition, subtraction, and multiply-and-accumulate (MAC), achieving 0.03x the energy consumption of prior PeFET-based designs.

Ultra low-bit quantization brings substantial efficiency for Transformer-based models, but the accuracy degradation and limited GPU support hinder its wide usage. In this paper, we analyze zero-point distortion in binarization and propose a Binary Weights & Ternary Activations (BWTA) quantization scheme, which projects tiny values to zero and preserves the accuracy of extremely low-bit models. For training, we propose Smooth Multi-Stage Quantization, combining a Levelwise Degradation Strategy and a Magnitude-Alignment Projection Factor to enable stable and fast convergence. For inference, we develop a BWTA MatMul CUDA kernel with instruction-level parallel bit-packing and comprehensive binary/ternary MatMul implementations for both linear and attention operators, allowing seamless integration across Transformer architectures. Experiments show that BWTA approaches full-precision performance for BERT, with an average 3.5% drop on GLUE and less than 2% drop on five tasks, and achieves comparable perplexity and accuracy for LLMs. In efficiency, it delivers 16 to 24 times kernel-level speedup over FP16 on NVIDIA GPUs, and 216 to 330 tokens/s end-to-end prefill speedup with lower memory footprint on LLMs. As an algorithm-hardware co-design, BWTA demonstrates practical, low-latency ultra-low-bit inference without sacrificing model quality.

Vision-language-action (VLA) models are emerging as embodied foundation models for robotic manipulation, but their deployment introduces a new unlearning challenge: removing unsafe, spurious, or privacy-sensitive behaviors without degrading perception, language grounding, and action control. In OpenVLA-style policies, behavior is produced through a fused visual encoder, a cross-modal projector, and a language backbone that predicts tokenized robot actions, so undesirable knowledge can be distributed across perception, alignment, and reasoning/action layers rather than confined to a single module. Consequently, partial unlearning applied only to the vision stack or only to the language backbone is often insufficient, while conventional unlearning baselines designed for standalone vision or language models may leave residual forgetting or incur unnecessary utility loss in embodied settings. We propose VLA-Forget, a hybrid unlearning framework that combines ratio-aware selective editing for perception and cross-modal specificity with layer-selective reasoning/action unlearning for utility-preserving forgetting. VLA-Forget jointly optimizes three objectives: targeted forgetting, perceptual preservation, and reasoning retention, through staged updates over the visual encoder, projector, and upper action-generating transformer blocks. Across forget-set behavior probes and retain-task evaluations, VLA-Forget improves forgetting efficacy by 10%, preserves perceptual specificity by 22%, retains reasoning and task success by 9%, and reduces post-quantization recovery by 55% relative to strong unlearning baselines.

When autonomous agents observe different domains of a shared environment, each signal they exchange mixes relevant and irrelevant dimensions. No existing mechanism lets the receiver evaluate which dimensions to absorb. We introduce Symbolic-Vector Attention Fusion (SVAF), the content-evaluation half of a two-level coupling engine for collective intelligence. SVAF decomposes each inter-agent signal into 7 typed semantic fields, evaluates each through a learned fusion gate, and produces a remix -- new knowledge from the intersection of two domains. A band-pass model yields four outcomes (redundant, aligned, guarded, rejected), solving both selectivity and redundancy. The fusion gate independently discovers a cross-domain relevance hierarchy: mood emerges as the highest-weight field by epoch 1, before accuracy plateaus -- consistent with independent mechanistic evidence that LLM emotion representations are structurally embedded along valence-arousal axes. SVAF forms Layer 4 of the Mesh Memory Protocol (MMP); the other half of the coupling engine is a per-agent Closed-form Continuous-time (CfC) neural network at Layer 6, whose learned per-neuron time constants (tau) create the temporal dynamics from which collective intelligence emerges: fast neurons synchronise affect across agents in seconds, while slow neurons preserve domain expertise indefinitely. SVAF determines what enters each agent's cognitive state; CfC determines how that state evolves. Trained on 237K samples from 273 narrative scenarios, SVAF achieves 78.7% three-class accuracy. We verify the complete mesh cognition loop -- from per-field evaluation through remix, CfC state evolution, tau-modulated peer blending, and autonomous action -- in a live deployment with 7 nodes across macOS, iOS, and web.

Learning interpretable multimodal representations inherently relies on uncovering the conditional dependencies between heterogeneous features. However, sparse graph estimation techniques, such as Graphical Lasso (GLasso), to visual-linguistic domains is severely bottlenecked by high-dimensional noise, modality misalignment, and the confounding of shared versus category-specific topologies. In this paper, we propose Cross-Modal Graphical Lasso (CM-GLasso) that overcomes these fundamental limitations. By coupling a novel text-visualization strategy with a unified vision-language encoder, we strictly align multimodal features into a shared latent space. We introduce a cross-attention distillation mechanism that condenses high-dimensional patches into explicit semantic nodes, naturally extracting spatial-aware cross-modal priors. Furthermore, we unify tailored GLasso estimation and Common-Specific Structure Learning (CSSL) into a joint objective optimized via the Alternating Direction Method of Multiplier (ADMM). This formulation guarantees the simultaneous disentanglement of invariant and class-specific precision matrices without multi-step error accumulation. Extensive experiments across eight benchmarks covering both natural and medical domains demonstrate that CM-GLasso establishes a new state-of-the-art in generative classification and dense semantic segmentation tasks.

Transformer-based large language models (LLMs) have demonstrated remarkable performance across a wide range of real-world tasks, but their inference cost remains prohibitively high due to the quadratic complexity of attention and the memory bandwidth limitations of high-precision operations. In this work, we present a low-bit mixed-precision attention kernel using the microscaling floating-point (MXFP) data format, utilizing the computing capability on next-generation GPU architectures. Our Diagonal-Tiled Mixed-Precision Attention (DMA) incorporates two kinds of low-bit computation at the tiling-level, and is a delicate fused kernel implemented using Triton, exploiting hardware-level parallelism and memory efficiency to enable fast and efficient inference without compromising model performance. Extensive empirical evaluations on NVIDIA B200 GPUs show that our kernel maintains generation quality with negligible degradation, and meanwhile achieves significant speedup by kernel fusion. We release our code at https://github.com/yifu-ding/MP-Sparse-Attn.

In many modern applications, a carefully designed primary study provides individual-level data for interpretable modeling, while summary-level external information is available through black-box, efficient, and nonparametric machine-learning predictions. Although summary-level external information has been studied in the data integration literature, there is limited methodology for leveraging external nonparametric machine-learning predictions to improve statistical inference in the primary study. We propose a general empirical-likelihood framework that incorporates external predictions through moment constraints. An advantage of nonparametric machine-learning prediction is that it induces a rich class of valid moment restrictions that remain robust to covariate shift under a mild overlap condition without requiring explicit density-ratio modeling. We focus on multinomial logistic regression as the primary model and address common data-quality issues in external sources, including coarsened outcomes, partially observed covariates, covariate shift, and heterogeneity in generating mechanisms known as concept shift. We establish large-sample properties of the resulting fused estimator, including consistency and asymptotic normality under regularity conditions. Moreover, we provide mild sufficient conditions under which incorporating external predictions delivers a strict efficiency gain relative to the primary-only estimator. Simulation studies and an application to the National Health and Nutrition Examination Survey on multiclass blood-pressure classification.

This article proposes a biconvex modification to convex biclustering in order to improve its performance in high-dimensional settings. In contrast to heuristics that discard a subset of noisy features a priori, our method jointly learns and accordingly weighs informative features while discovering biclusters. Moreover, the method is adaptive to the data, and is accompanied by an efficient algorithm based on proximal alternating minimization, complete with detailed guidance on hyperparameter tuning and efficient solutions to optimization subproblems. These contributions are theoretically grounded; we establish finite-sample bounds on the objective function under sub-Gaussian errors, and generalize these guarantees to cases where input affinities need not be uniform. Extensive simulation results reveal our method consistently recovers underlying biclusters while weighing and selecting features appropriately, outperforming peer methods. An application to a gene microarray dataset of lymphoma samples recovers biclusters matching an underlying classification, while giving additional interpretation to the mRNA samples via the column groupings and fitted weights.

Operating Elasticsearch clusters at scale demands continuous human expertise spanning the full lifecycle -- from initial deployment through performance tuning, monitoring, failure prediction, and incident recovery. We present the ES Guardian Agent, an autonomous AI SRE system that manages the complete Elasticsearch lifecycle without human intervention through eleven distinct phases: Evaluate, Optimize, Deploy, Calibrate, Stabilize, Alert, Predict, Heal, Learn, and Upgrade. A critical differentiator is its multi-source predictive failure engine, which continuously ingests and correlates metrics trends, application logs, and kernel-level telemetry -- including Linux dmesg streams, NVMe SMART data, NIC bond statistics, and thermal sensors -- to anticipate failures hours before they materialize. By cross-referencing current system signatures against a persistent incident memory of resolved failures, the AI engine stages corrective actions proactively. Through four successive agent architectures -- culminating in a 4,589-line system with five monitoring layers and an iterative AI action loop -- we demonstrate that an LLM equipped with tool-use access can function as a full-lifecycle autonomous SRE targeting six-nines (99.9999%) availability. In production evaluation, the Guardian Agent executed 300 autonomous investigation-and-repair cycles, recovered a cluster from an 18-hour cross-system outage, diagnosed hardware NIC failures across all host nodes, and maintained continuous operational visibility. We establish that data volume per shard -- not tuning -- is the primary determinant of query performance, with latency scaling at 0.26 ms per MB/shard.

Effective ride-hailing dispatch requires anticipating demand patterns that vary substantially across time-of-day, day-of-week, season, and special events. We propose a regime-calibrated approach that (i) segments historical trip data into demand regimes, (ii) matches the current operating period to the most similar historical analogues via a six-metric similarity ensemble (Kolmogorov-Smirnov, Wasserstein-1, feature distance, variance ratio, event pattern, temporal proximity), and (iii) uses the resulting calibrated demand prior to drive both an LP-based fleet repositioning policy and batch dispatch with Hungarian matching. In ablation, a distributional-only subset is strongest on mean wait, while the full ensemble is retained as a robustness-oriented default. Evaluated on 5.2 million NYC TLC trips across 8 diverse scenarios (winter/summer, weekday/weekend/holiday, morning/evening/night) with 5 random seeds each, our method reduces mean rider wait times by 31.1% (bootstrap 95% CI: [26.5, 36.6]%; Friedman chi-sq = 80.0, p = 4.25e-18; Cohen's d = 7.5-29.9 across scenarios). The improvement extends to the tail: P95 wait drops 37.6% and the Gini coefficient of wait times improves from 0.441 to 0.409 (7.3% relative). The two contributions compose multiplicatively and are independently validated: calibration provides 16.9% reduction; LP repositioning adds a further 15.5%. The approach requires no training, is deterministic and explainable, generalizes to Chicago (23.3% wait reduction via NYC-built regime library), and is robust across fleet sizes (32-47% improvement for 0.5-2x fleet scaling). We provide comprehensive ablation studies, formal statistical tests, and routing-fidelity validation with OSRM.

We present CODE-GEN, a human-in-the-Loop, retrieval-augmented generation (RAG)-based agentic AI system for generating context-aligned multiple-choice questions to develop student code reasoning and comprehension abilities. CODE-GEN employs an agentic AI architecture in which a Generator agent produces multiple-choice coding comprehension questions aligned with course-specific learning objectives, while a Validator agent independently assesses content quality across seven pedagogical dimensions. Both agents are augmented with specialized tools that enhance computational accuracy and verify code outputs. To evaluate the effectiveness of CODE-GEN, we conducted an evaluation study involving six human subject-matter experts (SMEs) who judged 288 AI-generated questions. The SMEs produced a total of 2,016 human-AI rating pairs, indicating agreement or disagreement with the assessments of Validator, along with 131 instances of qualitative feedback. Analyses of SME judgments show strong system performance, with human-validated success rates ranging from 79.9% to 98.6% across the seven pedagogical dimensions. The analysis of qualitative feedback reveals that CODE-GEN achieves high reliability on dimensions well suited to computational verification and explicit criteria matching, including question clarity, code validity, concept alignment, and correct answer validity. In contrast, human expertise remains essential for dimensions requiring deeper instructional judgment, such as designing pedagogically meaningful distractors and providing high-quality feedback that reinforces understanding. These findings inform the strategic allocation of human and AI effort in AI-assisted educational content generation.

Large language models struggle to accumulate evidence across multiple rounds of user interaction, failing to update their beliefs in a manner consistent with Bayesian inference. Existing solutions require fine-tuning on sensitive user interaction data, limiting their applicability in privacy-conscious settings. We propose AdaptFuse, a training-free framework that externalizes probabilistic computation entirely from the LLM: a symbolic module maintains a Bayesian posterior over a discrete hypothesis set, while a frozen LLM contributes semantic reasoning via multi-sample Dirichlet aggregation. The two signals are combined through entropy-adaptive fusion, which automatically weights each source by its predictive confidence, shifting reliance from the LLM to the symbolic posterior as evidence accumulates. We evaluate across three domains: flight recommendation, hotel recommendation, and web shopping; on Gemma 2 9B, Llama 3 8B, and Qwen 2.5 7B. AdaptFuse consistently outperforms both prompting baselines and fine-tuned Bayesian Teaching models on all tasks, with accuracy improving monotonically over interaction rounds. These results demonstrate that principled inference-time algorithms can substitute for fine-tuning in personalized recommendation, without storing or training on sensitive user data. All the code and materials will be open-sourced.

Goal-oriented conversational systems require making sequential decisions under uncertainty about the user's intent, where the algorithm must balance information acquisition and target commitment over multiple turns. Existing approaches address this challenge from different perspectives: structured methods enable multi-step planning but rely on predefined schemas, while LLM-based approaches support flexible interactions but lack long-horizon decision making, resulting in poor coordination between information acquisition and target commitment. To address this limitation, we formulate goal-oriented conversation as an uncertainty-aware sequential decision problem, where uncertainty serves as a guiding signal for multi-turn decision making. We propose a Conversation Uncertainty-aware Planning framework (CUP) that integrates language models with structured planning: a language model proposes feasible actions, and a planner evaluates their long-term impact on uncertainty reduction. Experiments on multiple conversational benchmarks show that CUP consistently improves success rates while requiring fewer interaction turns. Further analysis demonstrates that uncertainty-aware planning contributes to more efficient information acquisition and earlier confident commitment.

Selecting LLM-generated code candidates using LLM-generated tests is challenging because the tests themselves may be incorrect. Existing methods either treat all tests equally or rely on ad-hoc heuristics to filter unreliable tests. Yet determining test correctness requires knowing which codes are correct, creating a \emph{circular dependency}. Our key insight is that we need not determine test correctness at all: \emph{test votes should rank, not merely count}. What matters is not how many codes pass a test, but whether the test can \emph{distinguish} correct from incorrect code. We break the circular dependency via leave-one-out evaluation: hold out one test, rank codes by their aggregate scores on all remaining tests, and measure whether the held-out test's pass/fail pattern agrees with this ranking. We formalize this agreement as the leave-one-out AUC~(LOO-AUC) and prove that the expected LOO-AUC is proportional to each test's ability to separate correct code from incorrect code. Building on this, we propose \textbf{ACES}~(\textbf{A}UC \textbf{C}onsist\textbf{E}ncy \textbf{S}coring) with two complementary variants: ACES-C provides closed-form weights that provably approximate the oracle in expectation under a mild assumption on average test quality; ACES-O drops this assumption and iteratively optimizes a differentiable LOO-AUC objective. Both operate solely on the binary pass matrix with negligible overhead, and achieve state-of-the-art Pass@$k$ on multiple code generation benchmarks.

LLM-based social simulations can generate believable community interactions, enabling ``policy wind tunnels'' where governance interventions are tested before deployment. But believability is not causality. Claims like ``intervention $A$ reduces escalation'' require causal semantics that current simulation work typically does not specify. We propose adopting the causal counterfactual framework, distinguishing \textit{necessary causation} (would the outcome have occurred without the intervention?) from \textit{sufficient causation} (does the intervention reliably produce the outcome?). This distinction maps onto different stakeholder needs: moderators diagnosing incidents require evidence about necessity, while platform designers choosing policies require evidence about sufficiency. We formalize this mapping, show how simulation design can support estimation under explicit assumptions, and argue that the resulting quantities should be interpreted as simulator-conditional causal estimates whose policy relevance depends on simulator fidelity. Establishing this framework now is essential: it helps define what adequate fidelity means and moves the field from simulations that look realistic toward simulations that can support policy changes.

We present the first systematic study of Sparse Autoencoders (SAEs) on video representations. Standard SAEs decompose video into interpretable, monosemantic features but destroy temporal coherence: hard TopK selection produces unstable feature assignments across frames, reducing autocorrelation by 36%. We propose spatio-temporal contrastive objectives and Matryoshka hierarchical grouping that recover and even exceed raw temporal coherence. The contrastive loss weight controls a tunable trade-off between reconstruction and temporal coherence. A systematic ablation on two backbones and two datasets shows that different configurations excel at different goals: reconstruction fidelity, temporal coherence, action discrimination, or interpretability. Contrastive SAE features improve action classification by +3.9% over raw features and text-video retrieval by up to 2.8xR@1. A cross-backbone analysis reveals that standard monosemanticity metrics contain a backbone-alignment artifact: both DINOv2 and VideoMAE produce equally monosemantic features under neutral (CLIP) similarity. Causal ablation confirms that contrastive training concentrates predictive signal into a small number of identifiable features.

As cloud environments become increasingly complex, cybersecurity and forensic investigations must evolve to meet emerging threats. Large Language Models (LLMs) have shown promise in automating log analysis and reasoning tasks, yet they remain vulnerable to prompt injection attacks and lack forensic rigor. To address these dual challenges, we propose a unified, secure-by-design GenAI framework that integrates PromptShield and the Cloud Investigation Automation Framework (CIAF). PromptShield proactively defends LLMs against adversarial prompts using ontology-driven validation that standardizes user inputs and mitigates manipulation. CIAF streamlines cloud forensic investigations through structured, ontology-based reasoning across all six phases of the forensic process. We evaluate our system on real-world datasets from AWS and Microsoft Azure, demonstrating substantial improvements in both LLM security and forensic accuracy. Experimental results show PromptShield boosts classification performance under attack conditions, achieving precision, recall, and F1 scores above 93%, while CIAF enhances ransomware detection accuracy in cloud logs using Likert-transformed performance features. Our integrated framework advances the automation, interpretability, and trustworthiness of cloud forensics and LLM-based systems, offering a scalable foundation for real-time, AI-driven incident response across diverse cloud infrastructures.

Generating molecular dynamics (MD) trajectories using deep generative models has attracted increasing attention, yet remains inherently challenging due to the limited availability of MD data and the complexities involved in modeling high-dimensional MD distributions. To overcome these challenges, we propose a novel framework that leverages structure pretraining for MD trajectory generation. Specifically, we first train a diffusion-based structure generation model on a large-scale conformer dataset, on top of which we introduce an interpolator module trained on MD trajectory data, designed to enforce temporal consistency among generated structures. Our approach effectively harnesses abundant structural data to mitigate the scarcity of MD trajectory data and effectively decomposes the intricate MD modeling task into two manageable subproblems: structural generation and temporal alignment. We comprehensively evaluate our method on the QM9 and DRUGS small-molecule datasets across unconditional generation, forward simulation, and interpolation tasks, and further extend our framework and analysis to tetrapeptide and protein monomer systems. Experimental results confirm that our approach excels in generating chemically realistic MD trajectories, as evidenced by remarkable improvements of accuracy in geometric, dynamical, and energetic measurements.

Geoscientific systems tend to be characterized by pronounced temporal non-stationarity, arising from seasonal and climatic variability in hydrometeorological drivers, and from natural and anthropogenic changes to land use and cover. As has been pointed out, such variability renders "the assumption of statistical stationarity obsolete in water management", and requires us to "account for, rather than ignore, non-stationary trends" in the data. However, metrics used for model development are typically based on the implicit and unjustifiable assumption that the data generating process is time-stationary. Here, we introduce the JKGE_ss metric (adapted from KGE_ss) that detects and accounts for dynamical non-stationarity in the statistical properties of the data and thereby improves information extraction and model performance. Unlike NSE and KGE_ss, which use the long-term mean as a benchmark against which to evaluate model efficiency, JKGE_ss emphasizes reproduction of temporal variations in system storage. We tested the robustness of the new metric by training physical-conceptual and data-based catchment-scale models of varying complexity across a wide range of hydroclimatic conditions, from recent-precipitation-dominated to snow-dominated to strongly arid. In all cases, the result was improved reproduction of system temporal dynamics at all time scales, across wet to dry years, and over the full range of flow levels (especially recession periods). Since traditional metrics fail to adequately account for temporal shifts in system dynamics, potentially resulting in misleading assessments of model performance under changing conditions, we recommend the adoption of JKGE_ss for geoscientific model development.

Heterogeneity is a defining feature of deployed multi-robot teams: platforms often differ in sensing modalities, ranges, fields of view, and failure patterns. Controllers trained under nominal sensing can degrade sharply when deployed on robots with missing or mismatched sensors, even when the task and action interface are unchanged. We present DC-Ada, a reward-only decentralized adaptation method that keeps a pretrained shared policy frozen and instead adapts compact per-robot observation transforms to map heterogeneous sensing into a fixed inference interface. DC-Ada is gradient-free and communication-minimal: it uses budgeted accept/reject random search with short common-random-number rollouts under a strict step budget. We evaluate DC-Ada against four baselines in a deterministic 2D multi-robot simulator covering warehouse logistics, search and rescue, and collaborative mapping, across four heterogeneity regimes (H0--H3) and five seeds with a matched budget of $200{,}000$ joint environment steps per run. Results show that heterogeneity can substantially degrade a frozen shared policy and that no single mitigation dominates across all tasks and metrics. Observation normalization is strongest for reward robustness in warehouse logistics and competitive in search and rescue, while the frozen shared policy is strongest for reward in collaborative mapping. DC-Ada offers a useful complementary operating point: it improves completion most clearly in severe coverage-based mapping while requiring only scalar team returns and no policy fine-tuning or persistent communication. These results position DC-Ada as a practical deploy-time adaptation method for heterogeneous teams.

Large language models (LLMs) frequently produce confident but incorrect answers, partly because common binary scoring conventions reward answering over honestly expressing uncertainty. We study whether prompt-only interventions -- explicitly announcing reward schemes for answer-versus-abstain decisions plus humility-oriented normative principles -- can reduce hallucination risk without modifying the model. Our focus is epistemic abstention on factual questions with a verifiable answer, where current LLMs often fail to abstain despite being uncertain about their answers. We first assess self-reported verbal confidence as a usable uncertainty signal, showing stability under prompt paraphrasing and reasonable calibration against a token-probability baseline. We then study I-CALM, a prompt-based framework that (i) elicits verbal confidence, (ii) partially rewards abstention through explicit reward schemes, and (iii) adds lightweight normative principles emphasizing truthfulness, humility, and responsibility. Using GPT-5 mini on PopQA as the main setting, we find that confidence-eliciting, abstention-rewarding prompts, especially with norms, reduce the false-answer rate on answered cases mainly by identifying and shifting error-prone cases to abstention and re-calibrating their confidence. This trades coverage for reliability while leaving forced-answer performance largely unchanged. Varying the abstention reward yields a clear abstention-hallucination frontier. Overall, results show the framework can improve selective answering on factual questions without retraining, with the magnitude of effect varying across models and datasets. Code is available at the following https://github.com/binzeli/hallucinationControl.

The Learning with Errors (LWE) problem is a hard math problem in lattice-based cryptography. In the simplest case of binary secrets, it is the subset sum problem, with error. Effective ML attacks on LWE were demonstrated in the case of binary, ternary, and small secrets, succeeding on fairly sparse secrets. The ML attacks recover secrets with up to 3 active bits in the "cruel region" (Nolte et al., 2024) on samples pre-processed with BKZ. We show that using larger training sets and repeated examples enables recovery of denser secrets. Empirically, we observe a power-law relationship between model-based attempts to recover the secrets, dataset size, and repeated examples. We introduce a stepwise regression technique to recover the "cool bits" of the secret.

This paper introduces discourse_simulator, an open-source framework that combines LLMs with agent-based modelling. It offers a new way to simulate how public attitudes toward immigration change over time in response to salient events like protests, controversies, or policy debates. Large language models (LLMs) are used to generate social media posts, interpret opinions, and model how ideas spread through social networks. Unlike traditional agent-based models that rely on fixed, rule-based opinion updates and cannot generate natural language or consider current events, this approach integrates multidimensional sociological belief structures and real-world event timelines. This framework is wrapped into an open-source Python package that integrates generative agents into a small-world network topology and a live news retrieval system. discourse_sim is purpose-built as a social science research instrument specifically for studying attitude dynamics, polarisation, and belief evolution following real-world critical events. Unlike other LLM Agent Swarm frameworks, which treat the simulations as a prediction black box, discourse_sim treats it as a theory-testing instrument, which is fundamentally a different epistemological stance for studying social science problems. The paper further demonstrates the framework by modelling the Dublin anti-immigration march on April 26, 2025, with N=100 agents over a 15-day simulation. Package link: https://pypi.org/project/discourse-sim/

We present a telecom-native auction mechanism for allocating bandwidth and time slots across heterogeneous-delay networks, ranging from low-Earth-orbit (LEO) satellite constellations to delay-tolerant deep-space relays. The Lorentz-Invariant Auction (LIA) treats bids as spacetime events and reweights reported values based on the \emph{horizon slack}, a causal quantity derived from the earliest-arrival times relative to a public clearing horizon. Unlike other delay-equalization rules, LIA combines a causal-ordering formulation, a uniquely exponential slack correction implied by a semigroup-style invariance axiom, and a critical-value implementation that ensures truthful reported values once slacks are fixed by trusted infrastructure. We analyze the incentive result in the exogenous-slack regime and separately examine bounded slack-estimation error and endogenous-delay limitations. Under fixed feasible slacks, LIA is individually rational and achieves welfare at least \(e^{-λΔ}\) relative to the optimal feasible allocation, where \(Δ\) is the slack spread. We evaluate LIA on STARLINK-200, INTERNET-100, and DSN-30 across 52,500 baseline instances with market sizes \(n\in\{10,20,30,40,50\}\) and conduct additional robustness sweeps. On Starlink and Internet, LIA maintains near-efficiency while eliminating measured timing rents. However, on DSN, welfare is lower in thin markets but improves with depth. We also distinguish winner-determination time from the background cost of maintaining slack estimates and study robustness beyond independent and identically distributed (iid) noise through error-spread bounds and structured (distance-biased and subnetwork-correlated) noise models. These results suggest that causal-consistent mechanism design offers a practical non-buffering alternative to synchronized delay equalization in heterogeneous telecom infrastructures.

Breakthroughs in frontier theory often depend on the combination of concrete diagrammatic notations with rigorous logic. While multimodal large language models (MLLMs) show promise in general scientific tasks, current benchmarks often focus on local information extraction rather than the global structural logic inherent in formal scientific notations. In this work, we introduce FeynmanBench, the first benchmark centered on Feynman diagram tasks. It is designed to evaluate AI's capacity for multistep diagrammatic reasoning, which requires satisfying conservation laws and symmetry constraints, identifying graph topology, converting between diagrammatic and algebraic representations, and constructing scattering amplitudes under specific conventions and gauges. To support large-scale and reproducible evaluation, we developed an automated pipeline producing diverse Feynman diagrams along with verifiable topological annotations and amplitude results. Our database spans the electromagnetic, weak, and strong interactions of the Standard Model, encompasses over 100 distinct types and includes more than 2000 tasks. Experiments on state-of-the-art MLLMs reveal systematic failure modes, including unstable enforcement of physical constraints and violations of global topological conditions, highlighting the need for physics-grounded benchmarks for visual reasoning over scientific notation. FeynmanBench provides a logically rigorous test of whether AI can effectively engage in scientific discovery, particularly within theoretical physics.

Age-structured predator-prey integro-partial differential equations provide models of interacting populations in ecology, epidemiology, and biotechnology. A key challenge in feedback design for these systems is the scalar $ζ$, defined implicitly by the Lotka-Sharpe nonlinear integral condition, as a mapping from fertility and mortality rates to $ζ$. To solve this challenge with operator learning, we first prove that the Lotka-Sharpe operator is Lipschitz continuous, guaranteeing the existence of arbitrarily accurate neural operator approximations over a compact set of fertility and mortality functions. We then show that the resulting approximate feedback law preserves semi-global practical asymptotic stability under propagation of the operator approximation error through various other nonlinear operators, all the way through to the control input. In the numerical results, not only do we learn ``once-and-for-all'' the canonical Lotka-Sharpe (LS) operator, and thus make it available for future uses in control of other age-structured population interconnections, but we demonstrate the online usage of the neural LS operator under estimation of the fertility and mortality functions.

Multi-task representation learning (MTRL) is an approach that learns shared latent representations across related tasks, facilitating collaborative learning that improves the overall learning efficiency. This paper studies MTRL for multi-task reinforcement learning (RL), where multiple tasks have the same state-action space and transition probabilities, but different rewards. We consider T linear Markov Decision Processes (MDPs) where the reward functions and transition dynamics admit linear feature embeddings of dimension d. The relatedness among the tasks is captured by a low-rank structure on the reward matrices. Learning shared representations across multiple RL tasks is challenging due to the complex and policy-dependent nature of data that leads to a temporal progression of error. Our approach adopts a reward-free reinforcement learning framework to first learn a data-collection policy. This policy then informs an exploration strategy for estimating the unknown reward matrices. Importantly, the data collected under this well-designed policy enable accurate estimation, which ultimately supports the learning of an near-optimal policy. Unlike existing approaches that rely on restrictive assumptions such as Gaussian features, incoherence conditions, or access to optimal solutions, we propose a low-rank matrix estimation method that operates under more general feature distributions encountered in RL settings. Theoretical analysis establishes that accurate low-rank matrix recovery is achievable under these relaxed assumptions, and we characterize the relationship between representation error and sample complexity. Leveraging the learned representation, we construct near-optimal policies and prove a regret bound. Experimental results demonstrate that our method effectively learns robust shared representations and task dynamics from finite data.

This paper presents PolySwarm, a novel multi-agent large language model (LLM) framework designed for real-time prediction market trading and latency arbitrage on decentralized platforms such as Polymarket. PolySwarm deploys a swarm of 50 diverse LLM personas that concurrently evaluate binary outcome markets, aggregating individual probability estimates through confidence-weighted Bayesian combination of swarm consensus with market-implied probabilities, and applying quarter-Kelly position sizing for risk-controlled execution. The system incorporates an information-theoretic market analysis engine using Kullback-Leibler (KL) divergence and Jensen-Shannon (JS) divergence to detect cross-market inefficiencies and negation pair mispricings. A latency arbitrage module exploits stale Polymarket prices by deriving CEX-implied probabilities from a log-normal pricing model and executing trades within the human reaction-time window. We provide a full architectural description, implementation details, and evaluation methodology using Brier scores, calibration analysis, and log-loss metrics benchmarked against human superforecaster performance. We further discuss open challenges including hallucination in agent pools, computational cost at scale, regulatory exposure, and feedback-loop risk, and outline five priority directions for future research. Experimental results demonstrate that swarm aggregation consistently outperforms single-model baselines in probability calibration on Polymarket prediction tasks.

We address the problem of recovering a time-varying 4D distribution from a sparse sequence of 2D projections - analogous to novel-view synthesis from sparse cameras, but applied to the 4D transverse phase space density $ρ(x,p_x,y,p_y)$ of charged particle beams. Direct single shot measurement of this high-dimensional distribution is physically impossible in real particle accelerator systems; only limited 1D or 2D projections are accessible. We propose PhaseFlow4D, a feedback-guided latent diffusion model that reconstructs and tracks the full 4D phase space from incomplete 2D observations alone, with built-in hard physics constraints. Our core technical contribution is a 4D VAE whose decoder generates the full 4D phase space tensor, from which 2D projections are analytically computed and compared against 2D beam measurements. This projection-consistency constraint guarantees physical correctness by construction - not as a soft penalty, but as an architectural prior. An adaptive feedback loop then continuously tunes the conditioning vector of the latent diffusion model to track time-varying distributions online without retraining. We validate on multi-particle simulations of heavy-ion beams at the Facility for Rare Isotope Beams (FRIB), where full physics simulations require $\sim$6 hours on a 100-core HPC system. PhaseFlow4D achieves accurate 4D reconstructions 11000$\times$ faster while faithfully tracking distribution shifts under time-varying source conditions - demonstrating that principled generative reconstruction under incomplete observations transfers robustly beyond visual domains.

Effective ride-hailing dispatch requires anticipating demand patterns that vary substantially across time-of-day, day-of-week, season, and special events. We propose a regime-calibrated approach that (i) segments historical trip data into demand regimes, (ii) matches the current operating period to the most similar historical analogues via a similarity ensemble combining Kolmogorov-Smirnov distance, Wasserstein-1 distance, feature distance, variance ratio, event pattern similarity, and temporal proximity, and (iii) uses the resulting calibrated demand prior to drive both an LP-based fleet repositioning policy and batch dispatch with Hungarian matching. In ablation, a distributional-only metric subset achieves the strongest mean-wait reduction, while the full ensemble is retained as a robustness-oriented default that preserves calendar and event context. Evaluated on 5.2 million NYC TLC trips across 8 diverse scenarios (winter/summer, weekday/weekend/holiday, morning/evening/night) with 5 random seeds each, our method reduces mean rider wait times by 31.1% (bootstrap 95% CI: [26.5, 36.6]; Friedman chi-squared = 80.0, p = 4.25e-18; Cohen's d = 7.5-29.9). P95 wait drops 37.6% and the Gini coefficient of wait times improves from 0.441 to 0.409. The two contributions compose multiplicatively: calibration provides 16.9% reduction relative to the replay baseline; LP repositioning adds a further 15.5%. The approach requires no training, is deterministic and explainable, generalizes to Chicago (23.3% wait reduction using the NYC-built regime library without retraining), and is robust across fleet sizes (32-47% improvement for 0.5x-2.0x fleet scaling). Code is available at https://github.com/IndarKarhana/regime-calibrated-dispatch.

Nudging is widely used to promote behavioral change, but its effectiveness is often limited when recipients must repeatedly translate feedback into workable next steps under changing circumstances. Large language models (LLMs) may help reduce part of this cognitive work by generating personalized guidance and updating it iteratively across intervention rounds. We developed an LLM agent for iterative personalization and tested it in a three-arm randomized experiment among 233 university residents in China, using daily electricity and shower hot-water conservation as objectively measured cases differing in friction. LLM-personalized nudges (T2) produced the largest conservation effects, while image-enhanced conventional nudges (T1) and text-based conventional nudges (C) showed similar outcomes (omnibus p = 0.009). Relative to C, T2 reduced electricity consumption by 0.56 kWh per room-day (p = 0.014), corresponding to an 18.3 percentage-point higher adjusted saving rate. This advantage emerged within the first two intervention rounds, alongside iterative updating of personalized guidance, and persisted thereafter. Hot-water outcomes followed the same direction but were smaller, less precisely estimated, and attenuated over time, consistent with stronger friction in this domain. LLM-personalized nudges emphasized prospective and context-specific guidance and were associated with higher participant engagement. This study provides field evidence that LLM-based iterative personalization can enhance behavioral nudging, with behavioral friction as a potential boundary condition. Larger trials and extension to more behaviors are warranted.

Analogical reasoning is a core cognitive faculty essential for narrative understanding. While LLMs perform well when surface and structural cues align, they struggle in cases where an analogy is not apparent on the surface but requires latent information, suggesting limitations in abstraction and generalisation. In this paper we compare a model's probed representations with its prompted performance at detecting narrative analogies, revealing an asymmetry: for rhetorical analogies, probing significantly outperforms prompting in open-source models, while for narrative analogies, they achieve a similar (low) performance. This suggests that the relationship between internal representations and prompted behavior is task-dependent and may reflect limitations in how prompting accesses available information.

Monitoring deforestation-driven carbon emissions requires both spatially explicit and temporally continuous estimates of aboveground biomass density (AGBD) with calibrated uncertainty. NASA's Global Ecosystem Dynamics Investigation (GEDI) provides reliable LIDAR-derived AGBD, but its orbital sampling causes irregular spatiotemporal coverage, and occasional operational interruptions, including a 13-month hibernation from March 2023 to April 2024, leave extended gaps in the observational record. Prior work has used machine learning approaches to fill GEDI's spatial gaps using satellite-derived features, but temporal interpolation of biomass through unobserved periods, particularly across active disturbance events, remains largely unaddressed. Moreover, standard ensemble methods for biomass mapping have been shown to produce systematically miscalibrated prediction intervals. To address these gaps, we extend the Attentive Neural Process (ANP) framework, previously applied to spatial biomass interpolation, to jointly sparse spatiotemporal settings using geospatial foundation model embeddings. We treat space and time symmetrically, empirically validating a form of space-for-time substitution in which observations from nearby locations at other times inform predictions at held-out periods. Our results demonstrate that the ANP produces well-calibrated uncertainty estimates across disturbance regimes, supporting its use in Measurement, Reporting, and Verification (MRV) applications that require reliable uncertainty quantification for forest carbon accounting.

Black-box knowledge distillation for large language models presents a strict trade-off. Simple off-policy methods (e.g., sequence-level knowledge distillation) struggle to correct the student's inherent errors. Fully on-policy methods (e.g., Generative Adversarial Distillation) solve this via adversarial training but introduce well-known training instability and crippling computational overhead. To address this dilemma, we propose SODA (Semi On-policy Distillation with Alignment), a highly efficient alternative motivated by the inherent capability gap between frontier teachers and much smaller base models. Because a compact student model's natural, zero-shot responses are almost strictly inferior to the powerful teacher's targets, we can construct a highly effective contrastive signal simply by pairing the teacher's optimal response with a one-time static snapshot of the student's outputs. This demonstrates that exposing the small student to its own static inferior behaviors is sufficient for high-quality distribution alignment, eliminating the need for costly dynamic rollouts and fragile adversarial balancing. Extensive evaluations across four compact Qwen2.5 and Llama-3 models validate this semi on-policy paradigm. SODA matches or outperforms the state-of-the-art methods on 15 out of 16 benchmark results. More importantly, it achieves this superior distillation quality while training 10 times faster, consuming 27% less peak GPU memory, and completely eliminating adversarial instability.

Sabotage games are played on a dynamic graph, in which one agent, called a runner, attempts to reach a goal state, while being obstructed by a demon who at each round removes an edge from the graph. Sabotage modal logic was proposed to carry out reasoning about such games. Since its conception, it has undergone a thorough analysis (in terms of complexity, completeness, and various extensions) and has been applied to a variety of domains, e.g., to formal learning. In this paper, we propose examining the game from a temporal perspective using alternating time temporal logic (ATL$^\ast$), and address the players' uncertainty in its epistemic extensions. This framework supports reasoning about winning strategies for those games, and opens ways to address temporal properties of dynamic graphs in general.

The rapid deployment of open-source frameworks has significantly advanced the development of modern multi-agent systems. However, expanded action spaces, including uncontrolled privilege exposure and hidden inter-system interactions, pose severe security challenges. Specifically, Indirect Prompt Injections (IPI), which conceal malicious instructions within third-party content, can trigger unauthorized actions such as data exfiltration during normal operations. While current security evaluations predominantly rely on isolated single-turn benchmarks, the systemic vulnerabilities of these agents within complex dynamic environments remain critically underexplored. To bridge this gap, we systematically evaluate six defense strategies against four sophisticated IPI attack vectors across nine LLM backbones. Crucially, we conduct our evaluation entirely within dynamic multi-step tool-calling environments to capture the true attack surface of modern autonomous agents. Moving beyond binary success rates, our multidimensional analysis reveals a pronounced fragility. Advanced injections successfully bypass nearly all baseline defenses, and some surface-level mitigations even produce counterproductive side effects. Furthermore, while agents execute malicious instructions almost instantaneously, their internal states exhibit abnormally high decision entropy. Motivated by this latent hesitation, we investigate Representation Engineering (RepE) as a robust detection strategy. By extracting hidden states at the tool-input position, we revealed that the RepE-based circuit breaker successfully identifies and intercepts unauthorized actions before the agent commits to them, achieving high detection accuracy across diverse LLM backbones. This study exposes the limitations of current IPI defenses and provides a highly practical paradigm for building resilient multi-agent architectures.

Steering vectors have emerged as a lightweight and effective approach for aligning large language models (LLMs) at inference time, enabling modulation over model behaviors by shifting LLM representations towards a target behavior. However, existing methods typically apply steering vectors at a globally fixed layer, implicitly assuming that the optimal intervention layer is invariant across inputs. We argue that this assumption is fundamentally limited, as representations relevant to a target behavior can be encoded at different layers depending on the input. Theoretically, we show that different inputs can require steering at different layers to achieve alignment with a desirable model behavior. We also provide empirical evidence that the optimal steering layer varies substantially across inputs in practice. Motivated by these observations, we introduce Where to Steer (W2S), a framework that adaptively selects the intervention layer conditioned on the input, by learning a mapping from input embeddings to optimal steering layers. Across multiple LLMs and alignment behaviors, W2S consistently outperforms fixed-layer baselines, with improvements in both in-distribution and out-of-distribution settings. Our findings highlight the importance of input-dependent control in LLM alignment and demonstrate that adaptive layer selection is a key design dimension missing in the current methodology of steering vectors.

Federated learning (FL) enables multiple clients to collaboratively train a global machine learning model via a server without sharing their private training data. In traditional FL, the system follows a synchronous approach, where the server waits for model updates from numerous clients before aggregating them to update the global model. However, synchronous FL is hindered by the straggler problem. To address this, the asynchronous FL architecture allows the server to update the global model immediately upon receiving any client's local model update. Despite its advantages, the decentralized nature of asynchronous FL makes it vulnerable to poisoning attacks. Several defenses tailored for asynchronous FL have been proposed, but these mechanisms remain susceptible to advanced attacks or rely on unrealistic server assumptions. In this paper, we introduce SecureAFL, an innovative framework designed to secure asynchronous FL against poisoning attacks. SecureAFL improves the robustness of asynchronous FL by detecting and discarding anomalous updates while estimating the contributions of missing clients. Additionally, it utilizes Byzantine-robust aggregation techniques, such as coordinate-wise median, to integrate the received and estimated updates. Extensive experiments on various real-world datasets demonstrate the effectiveness of SecureAFL.

Training Data Attribution (TDA) seeks to trace model predictions back to influential training examples, enhancing interpretability and safety. We formulate TDA as a Bayesian information-theoretic problem: subsets are scored by the information loss they induce - the entropy increase at a query when removed. This criterion credits examples for resolving predictive uncertainty rather than label noise. To scale to modern networks, we approximate information loss using a Gaussian Process surrogate built from tangent features. We show this aligns with classical influence scores for single-example attribution while promoting diversity for subsets. For even larger-scale retrieval, we relax to an information-gain objective and add a variance correction for scalable attribution in vector databases. Experiments show competitive performance on counterfactual sensitivity, ground-truth retrieval and coreset selection, showing that our method scales to modern architectures while bridging principled measures with practice.

Multivariate count models are often justified by their ability to capture latent dependence, but researchers receive little guidance on when this added structure improves on simpler penalized marginal Poisson regression. We study this question using real microbiome data under a unified held-out evaluation framework. For count prediction, we compare PLN and GLMNet(Poisson) on 20 datasets spanning 32 to 18,270 samples and 24 to 257 taxa, using held-out Poisson deviance under leave-one-taxon-out prediction with 3-fold sample cross-validation rather than synthetic or in-sample criteria. For network inference, we compare PLNNetwork and GLMNet(Poisson) neighborhood selection on five publicly available datasets with experimentally validated microbial interaction truth. PLN outperforms GLMNet(Poisson) on most count-prediction datasets, with gains up to 38 percent. The primary predictor of the winner is the sample-to-taxon ratio, with mean absolute correlation as the strongest secondary signal and overdispersion as an additional predictor. PLNNetwork performs best on broad undirected interaction benchmarks, whereas GLMNet(Poisson) is better aligned with local or directional effects. Taken together, these results provide guidance for choosing between latent multivariate count models and penalized Poisson regression in biological count prediction and interaction recovery.

Linux kernel bug repair is typically approached as a direct mapping from crash reports to code patches. In practice, however, kernel fixes undergo iterative revision on mailing lists before acceptance, with reviewer feedback shaping correctness, concurrency handling, and API compliance. This iterative refinement process encodes valuable repair knowledge that existing automated approaches overlook. We present a large-scale study of kernel patch evolution, reconstructing 6946 syzbot-linked bug-fix lifecycles that connect crash reports, reproducers, mailing-list discussions, revision histories, and merged fixes. Our analysis confirms that accepted repairs are frequently non-local and governed by reviewer-enforced constraints not present in bug reports. Building on these insights, we develop PatchAdvisor, a repair framework that integrates retrieval-based memory with a fine-tuned diagnostic advisor to guide a coding agent toward reviewer-aligned patches. Evaluation on temporally held-out syzbot cases demonstrates that leveraging patch-evolution history yields measurable gains in both reviewer-aligned refinement signals and end-to-end repair quality compared to unguided and retrieval-only baselines.

DDCL-Attention is a prototype-based readout layer for transformer encoders that replaces simple pooling methods, such as mean pooling or class tokens, with a learned compression mechanism. It uses a small set of global prototype vectors and assigns tokens to them through soft probabilistic matching, producing compact token summaries at linear complexity in sequence length. The method offers three main advantages. First, it avoids prototype collapse through an exact decomposition of the training loss into a reconstruction term and a diversity term, ensuring that prototypes remain distinct. Second, its joint training with the encoder is shown to be stable under a practical timescale condition, using Tikhonov's singular perturbation theory and explicit learning-rate constraints. Third, the same framework supports three uses: a final readout layer, a differentiable codebook extending VQ-VAE, and a hierarchical document compressor. Experiments on four datasets confirm the theoretical predictions: the loss decomposition holds exactly, prototype separation grows as expected when the stability condition is met, and the codebook reaches full utilization, outperforming standard hard vector quantization. An additional study on orbital debris classification shows that the method also applies beyond standard NLP and vision tasks, including scientific tabular data.

Financial sustainability is vital for open-source software, yet systematic research on funding remains limited. GitHub Sponsors, launched in 2019 as a direct developer-to-developer funding model, lacks bulk API access, hindering large-scale studies. This paper introduces a live, continuously operating observatory for tracking and analyzing the GitHub Sponsors ecosystem. The observatory performs priority-based graph traversal with daily incremental updates, real-time normalization, and exposes collected data through an interactive dashboard and analysis-ready CSV exports. A sample dataset collected during a 72-hour run captures 49K+ users across 144 countries and serves as an example of the tool's output, not a fixed deliverable. An interactive dashboard (https://github-sponsorships.com) enables practitioners and researchers to explore sponsorship patterns, filter by geography and demographics, and benchmark against funded peers. Preliminary results on the sample show strong participation asymmetries and geographic concentration, suggesting several research directions.

Transformer-based malware detection systems operating on graph modalities such as control flow graphs (CFGs) achieve strong performance by modeling structural relationships in program behavior. However, their robustness to adversarial evasion attacks remains underexplored. This paper examines the vulnerability of a RoBERTa-based malware detector that linearizes CFGs into sequences of function calls, a design choice that enables transformer modeling but may introduce token-level sensitivities and ordering artifacts exploitable by adversaries. By evaluating evasion strategies within this graph-to-sequence framework, we provide insight into the practical robustness of transformer-based malware detectors beyond aggregate detection accuracy. This paper proposes a white-box adversarial evasion attack that leverages explainability mechanisms to identify and perturb most influential graph components. Using token- and word-level attributions derived from integrated gradients, the attack iteratively replaces positively attributed function calls with synthetic external imports, producing adversarial CFG representations without altering overall program structure. Experimental evaluation on small- and large-scale Windows Portable Executable (PE) datasets demonstrates that the proposed method can reliably induce misclassification, even against models trained to high accuracy. Our results highlight that explainability tools, while valuable for interpretability, can also expose critical attack surfaces in transformer-based malware detectors.

This work reconciles two perspectives on the Elo ranking that coexist in the literature: the practitioner's view as a heuristic feedback rule, and the statistician's view as online maximum likelihood estimation via stochastic gradient ascent. Both perspectives coincide exactly in the binary case (iff the expected score is the logistic function). However, estimation noise forces a principled decoupling between the model used for ranking and the model used for prediction: the effective scale and home-field advantage parameter must be adjusted to account for the noise. We provide both closed-form corrections and a data-driven identification procedure. For multilevel outcomes, an exact relationship exists when outcome scores are uniformly spaced, but approximations are preferred in general: they account for estimation noise and better fit the data. The decoupled approach substantially outperforms the conventional one that reuses the ranking model for prediction, and serves as a diagnostic of convergence status. Applied to six years of FIFA men's ranking, we find that the ranking had not converged for the vast majority of national teams. The paper is written in a semi-tutorial style accessible to practitioners, with all key results accompanied by closed-form expressions and numerical examples.

Mamba is an emerging, complex workload with various short-range and long-range dependencies, nonlinearities, and elementwise computations that are unable to run at near-peak speeds on modern hardware. Specifically, Mamba's complex dependency graph makes fusion across its full operator cascade difficult, leaving substantial inter-operator memory traffic on the table. To address these challenges, we propose Mambalaya, a novel reconfigurable accelerator that leverages fusion to overcome the limitations of Mamba. We use the recently proposed cascade-of-Einsums abstraction to characterize Mamba's full computational structure, then apply the extended Einsum framework to systematically explore inter-Einsum fusion opportunities. This principled approach yields a series of fusion mappings that reduce off-chip inter-Einsum traffic. These mappings are supported by the underlying Mambalaya architecture. Mambalaya achieves a layer performance speedup of 4.9$\times$ for prefill and 1.9$\times$ for generation over MARCA. In prefill-dominated scenarios, it achieves up to 1.5$\times$ over a recent fine-grained, memory-aware fusion accelerator for Mamba.

Architecture Decision Records (ADRs) play a critical role in preserving the rationale behind system design, yet their creation and maintenance are often neglected due to the associated authoring overhead. This paper investigates whether Large Language Models (LLMs) can mitigate this burden and, more importantly, how different strategies for presenting historical ADRs as context influence generation quality. We curate and validate a large corpus of sequential ADRs drawn from 750 open-source repositories and systematically evaluate five context selection strategies (no context, All-history, First-K, Last-K, and RAFG) across multiple model families. Our results show that context-aware prompting substantially improves ADR generation fidelity, with a small recency window (typically 3-5 prior records) providing the best balance between quality and efficiency. Retrieval-based context selection yields marginal gains primarily in non-sequential or cross-cutting decision scenarios, while offering no statistically significant advantage in typical linear ADR workflows. Overall, our findings demonstrate that context engineering, rather than model scale alone, is the dominant factor in effective ADR automation, and we outline practical defaults for tool builders along with targeted retrieval fallbacks for complex architectural settings.

Large language models are increasingly used for qualitative data analysis, but many workflows obscure how analytic conclusions are produced. We present QualAnalyzer, an open-source Chrome extension for Google Workspace that supports atomistic LLM analysis by processing each data segment independently and preserving the prompt, input, and output for every unit. Through two case studies -- holistic essay scoring and deductive thematic coding of interview transcripts -- we show that this approach creates a legible audit trail and helps researchers investigate systematic differences between LLM and human judgments. We argue that process auditability is essential for making LLM-assisted qualitative research more transparent and methodologically robust.

Limited work has examined the strategic behaviors of relational networked learning agents under social dilemmas, and has overlooked the intricate social dynamics of complex systems. We address the challenge with Socio-Relational Intrinsic Motivation (SRIM), which endows agents with diverse preferences over sub-graphical social structures in order to study the impact of agents' personal preferences over their sub-graphical relations on their strategic decision-making under sequential social dilemmas. Our results in the Harvest and Cleanup environments demonstrate that preferences over different subgraph structures (degree-, clique-, and critical connection-based) lead to distinct variations in agents' reward gathering and strategic behavior: individual aggressiveness in Harvest and individual contribution effort in Cleanup. Moreover, agents with different subgraphical structural positions consistently exhibit similar strategic behavioral shifts. Our proposed BCI metric captures structural variation within the population, and the relative ordering of BCI across social preferences is consistent in Harvest and Cleanup games for the same topology, suggesting the subgraphical structural impact is robust across environments. These results provide a new lens for examining agents' behavior in social dilemmas and insight for designing effective multi-agent ecosystems composed of heterogeneous social agents.

Classical simulation of quantum circuits remains indispensable for algorithm development, hardware validation, and error analysis in the noisy intermediate-scale quantum (NISQ) era. However, state-vector simulation faces exponential memory scaling, with an n-qubit system requiring O(2^n) complex amplitudes, and existing simulators often lack the flexibility to exploit heterogeneous computing resources at runtime. This paper presents a GPU-accelerated quantum circuit simulation framework that introduces three contributions: (1) an empirical backend selection algorithm that benchmarks CuPy, PyTorch-CUDA, and NumPy-CPU backends at runtime and selects the optimal execution path based on measured throughput; (2) a directed acyclic graph (DAG) based gate fusion engine that reduces circuit depth through automated identification of fusible gate sequences, coupled with adaptive precision switching between complex64 and complex128 representations; and (3) a memory-aware fallback mechanism that monitors GPU memory consumption and gracefully degrades to CPU execution when resources are exhausted. The framework integrates with Qiskit, Cirq, PennyLane, and Amazon Braket through a unified adapter layer. Benchmarks on an NVIDIA A100-SXM4 (40 GiB) GPU demonstrate speedups of 64x to 146x over NumPy CPU execution for state-vector simulation of circuits with 20 to 28 qubits, with speedups exceeding 5x from 16 qubits onward. Hardware validation on an IBM quantum processing unit (QPU) confirms Bell state fidelity of 0.939, a five-qubit Greenberger-Horne-Zeilinger (GHZ) state fidelity of 0.853, and circuit depth reduction from 42 to 14 gates through the fusion pipeline. The system is designed for portability across NVIDIA consumer and data-center GPUs, requiring no vendor-specific compilation steps.

Graph transformers have shown promise in overcoming limitations of traditional graph neural networks, such as oversquashing and difficulties in modelling long-range dependencies. However, their application to large-scale graphs is hindered by the quadratic memory and computational complexity of the all-to-all attention mechanism. Although alternatives such as linearized attention and restricted attention patterns have been proposed, these often degrade performance or limit expressive power. To better balance efficiency and effectiveness, we introduce k-Maximum Inner Product (k-MIP) attention for graph transformers. k-MIP attention selects the most relevant key nodes per query via a top-k operation, yielding a sparse yet flexible attention pattern. Combined with an attention score computation based on symbolic matrices, this results in linear memory complexity and practical speedups of up to an order of magnitude compared to all-to-all attention, enabling the processing of graphs with over 500k nodes on a single A100 GPU. We provide a theoretical analysis of expressive power, showing that k-MIP attention does not compromise the expressiveness of graph transformers: specifically, we prove that k-MIP transformers can approximate any full-attention transformer to arbitrary precision. In addition, we analyze the expressive power of the GraphGPS framework, in which we integrate our attention mechanism, and establish an upper bound on its graph distinguishing capability in terms of the S-SEG-WL test. Finally, we validate our approach on the Long Range Graph Benchmark, the City-Networks benchmark, and two custom large-scale inductive point cloud datasets, consistently ranking among the top-performing scalable graph transformers.

Camera extrinsic calibration is a fundamental task in computer vision. However, precise relative pose estimation in constrained, highly distorted environments, such as in-cabin automotive monitoring (ICAM), remains challenging. We present InCaRPose, a Transformer-based architecture designed for robust relative pose prediction between image pairs, which can be used for camera extrinsic calibration. By leveraging frozen backbone features such as DINOv3 and a Transformer-based decoder, our model effectively captures the geometric relationship between a reference and a target view. Unlike traditional methods, our approach achieves absolute metric-scale translation within the physically plausible adjustment range of in-cabin camera mounts in a single inference step, which is critical for ICAM, where accurate real-world distances are required for safety-relevant perception. We specifically address the challenges of highly distorted fisheye cameras in automotive interiors by training exclusively on synthetic data. Our model is capable of generalization to real-world cabin environments without relying on the exact same camera intrinsics and additionally achieves competitive performance on the public 7-Scenes dataset. Despite having limited training data, InCaRPose maintains high precision in both rotation and translation, even with a ViT-Small backbone. This enables real-time performance for time-critical inference, such as driver monitoring in supervised autonomous driving. We release our real-world In-Cabin-Pose test dataset consisting of highly distorted vehicle-interior images and our code at https://github.com/felixstillger/InCaRPose.

Multi-agent LLM committees replicate the same model under different role prompts and aggregate outputs by majority vote, implicitly assuming that agents contribute complementary evidence. We embed each agent's chain-of-thought rationale and measure pairwise similarity: across 100 GSM8K questions with three Qwen2.5-14B agents, mean cosine similarity is 0.888 and effective rank is 2.17 out of 3.0, a failure mode we term representational collapse. DALC, a training-free consensus protocol that computes diversity weights from embedding geometry, reaches 87% on GSM8K versus 84% for self-consistency at 26% lower token cost. Ablation experiments reveal 1-3 point per-protocol run-to-run variance, confirm that hint sharing contributes more than diversity weighting alone, and show that encoder choice strongly modulates collapse severity (cosine 0.908 with mxbai versus 0.888 with nomic) and downstream accuracy. The more robust finding is that collapse is measurable, worsens on harder tasks, and that the choice of embedding proxy is a first-order design decision for any latent communication protocol.

The assumption that mobile enterprise software requires native Android SDK development has persisted for over a decade, but for institutional deployments, this assumption is not merely outdated: it is economically wasteful and technically unnecessary. This paper presents a campus management system built during an internship at the Indian Institute of Technology Gandhinagar (IIT Gandhinagar), covering housekeeping task scheduling, inventory management, horticulture tracking, worker attendance, multi-stage leave workflows, and client-side photo capture with automatic compression. The core stack uses Python-Django as the backend framework and HTMX for hypermedia-driven, mobile-responsive partial DOM updates, containing zero lines of Android SDK application logic. The entire system runs as a self-hosted Docker Compose deployment with no dependency on any external cloud service. Through architectural analysis, HTTP payload measurement, and user experience evaluation with 42 campus staff, we demonstrate that the HTMX-Django approach reduces development time by approximately 54%, reduces average HTTP payload by 91% versus full-page reload, and achieves user satisfaction scores of 4.2/5.0.

Transformers are strong baselines in both vision and language because self-attention captures long-range dependencies across tokens. However, the cost of self-attention grows quadratically with the number of tokens. Patch pruning mitigates this cost by estimating per-patch importance and removing redundant patches. To identify informative patches for pruning, we introduce a criterion based on the Shannon entropy of the attention distribution. Low-entropy patches, which receive selective and concentrated attention, are kept as important, while high-entropy patches with attention spread across many locations are treated as redundant. We also extend the criterion from Shannon to Rényi entropy, which emphasizes sharp attention peaks and supports pruning strategies that adapt to task needs and computational limits. In experiments on fine-grained image recognition, where patch selection is critical, our method reduced computation while preserving accuracy. Moreover, adjusting the pruning policy through the Rényi entropy measure yields further gains and improves the trade-off between accuracy and computation.

When LLM-based multi-agent systems disagree, current practice treats this as noise to be resolved through consensus. We propose it can be signal. We focus on hate speech moderation, a domain where judgments depend on cultural context and individual value weightings, producing high legitimate disagreement among human annotators. We hypothesize that convergent disagreement, where agents reason similarly but conclude differently, indicates genuine value pluralism that humans also struggle to resolve. Using the Measuring Hate Speech corpus, we embed reasoning traces from five perspective-differentiated agents and classify disagreement patterns using a four-category taxonomy based on reasoning similarity and conclusion agreement. We find that raw reasoning divergence weakly predicts human annotator conflict, but the structure of agent discord carries additional signal: cases where agents agree on a verdict show markedly lower human disagreement than cases where they do not, with large effect sizes (d>0.8) surviving correction for multiple comparisons. Our taxonomy-based ordering correlates with human disagreement patterns. These preliminary findings motivate a shift from consensus-seeking to uncertainty-surfacing multi-agent design, where disagreement structure - not magnitude - guides when human judgment is needed.

Recent advances in large language models have significantly improved their ability to perform mathematical reasoning, extending from elementary problem solving to increasingly capable performance on research-level problems. However, reliably solving and verifying such problems remains challenging due to the inherent ambiguity of natural language reasoning. In this paper, we propose an automated framework for tackling research-level mathematical problems that integrates natural language reasoning with formal verification, enabling end-to-end problem solving with minimal human intervention. Our framework consists of two components: an informal reasoning agent, Rethlas, and a formal verification agent, Archon. Rethlas mimics the workflow of human mathematicians by combining reasoning primitives with our theorem search engine, Matlas, to explore solution strategies and construct candidate proofs. Archon, equipped with our formal theorem search engine LeanSearch, translates informal arguments into formalized Lean 4 projects through structured task decomposition, iterative refinement, and automated proof synthesis, ensuring machine-checkable correctness. Using this framework, we automatically resolve an open problem in commutative algebra and formally verify the resulting proof in Lean 4 with essentially no human involvement. Our experiments demonstrate that strong theorem retrieval tools enable the discovery and application of cross-domain mathematical techniques, while the formal agent is capable of autonomously filling nontrivial gaps in informal arguments. More broadly, our work illustrates a promising paradigm for mathematical research in which informal and formal reasoning systems, equipped with theorem retrieval tools, operate in tandem to produce verifiable results, substantially reduce human effort, and offer a concrete instantiation of human-AI collaborative mathematical research.

The Sinkhorn--Knopp (SK) algorithm is a cornerstone method for matrix scaling and entropically regularized optimal transport (EOT). Despite its empirical efficiency, existing theoretical guarantees to achieve a target marginal accuracy $\varepsilon$ deteriorate severely in the presence of outliers, bottlenecked either by the global maximum regularized cost $η\|C\|_\infty$ (where $η$ is the regularization parameter and $C$ the cost matrix) or the matrix's minimum-to-maximum entry ratio $ν$. This creates a fundamental disconnect between theory and practice. In this paper, we resolve this discrepancy. For EOT, we introduce the novel concept of well-boundedness, a local bulk mass property that rigorously isolates the well-behaved portion of the data from extreme outliers. We prove that governed by this fundamental notion, SK recovers the target transport plan for a problem of dimension $n$ in $O(\log n - \log \varepsilon)$ iterations, completely independent of the regularized cost $η\|C\|_\infty$. Furthermore, we show that a virtually cost-free pre-scaling step eliminates the dimensional dependence entirely, accelerating convergence to a strictly dimension-free $O(\log(1/\varepsilon))$ iterations. Beyond EOT, we establish a sharp phase transition for general $(\boldsymbol{u},\boldsymbol{v})$-scaling governed by a critical matrix density threshold. We prove that when a matrix's density exceeds this threshold, the iteration complexity is strictly independent of $ν$. Conversely, when the density falls below this threshold, the dependence on $ν$ becomes unavoidable; in this sub-critical regime, we construct instances where SK requires $Ω(n/\varepsilon)$ iterations.

Communication is essential for coordination in \emph{cooperative} multi-agent reinforcement learning under partial observability, yet \emph{cross-timestep} delays cause messages to arrive multiple timesteps after generation, inducing temporal misalignment and making information stale when consumed. We formalize this setting as a delayed-communication partially observable Markov game (DeComm-POMG) and decompose a message's effect into \emph{communication gain} and \emph{delay cost}, yielding the Communication Gain and Delay Cost (CGDC) metric. We further establish a value-loss bound showing that the degradation induced by delayed messages is upper-bounded by a discounted accumulation of an information gap between the action distributions induced by timely versus delayed messages. Guided by CGDC, we propose \textbf{CDCMA}, an actor--critic framework that requests messages only when predicted CGDC is positive, predicts future observations to reduce misalignment at consumption, and fuses delayed messages via CGDC-guided attention. Experiments on no-teammate-vision variants of Cooperative Navigation and Predator Prey, and on SMAC maps across multiple delay levels show consistent improvements in performance, robustness, and generalization, with ablations validating each component.

We analyze the last-iterate convergence of the Anchored Gradient Descent Ascent algorithm for smooth convex-concave min-max problems. While previous work established a last-iterate rate of $\mathcal{O}(1/t^{2-2p})$ for the squared gradient norm, where $p \in (1/2, 1)$, it remained an open problem whether the improved exact $\mathcal{O}(1/t)$ rate is achievable. In this work, we resolve this question in the affirmative. This result was discovered autonomously by an AI system capable of writing formal proofs in Lean. The Lean proof can be accessed at https://github.com/google-deepmind/formal-conjectures/pull/3675/commits/a13226b49fd3b897f4c409194f3bcbeb96a08515

Diffusion models have excelled at generative tasks for both continuous and token-based domains, but their application to discrete ordinal data remains underdeveloped. We present CountsDiff, a diffusion framework designed to natively model distributions on the natural numbers. CountsDiff extends the Blackout diffusion framework by simplifying its formulation through a direct parameterization in terms of a survival probability schedule and an explicit loss weighting. This introduces flexibility through design parameters with direct analogues in existing diffusion modeling frameworks. Beyond this reparameterization, CountsDiff introduces features from modern diffusion models, previously absent in counts-based domains, including continuous-time training, classifier-free guidance, and churn/remasking reverse dynamics that allow non-monotone reverse trajectories. We propose an initial instantiation of CountsDiff and validate it on natural image datasets (CIFAR-10, CelebA), exploring the effects of varying the introduced design parameters in a complex, well-studied, and interpretable data domain. We then highlight biological count assays as a natural use case, evaluating CountsDiff on single-cell RNA-seq imputation in a fetal cell and heart cell atlas. Remarkably, we find that even this simple instantiation matches or surpasses the performance of a state-of-the-art discrete generative model and leading RNA-seq imputation methods, while leaving substantial headroom for further gains through optimized design choices in future work.

The adoption of vision-language models (VLMs) for wireless network management is accelerating, yet no systematic understanding exists of where these large foundation models outperform lightweight convolutional neural networks (CNNs) for spectrum-related tasks. This paper presents the first diagnostic comparison of VLMs and CNNs for spectrum heatmap understanding in non-terrestrial network and terrestrial network (NTN-TN) cooperative systems. We introduce SpectrumQA, a benchmark comprising 108K visual question-answer pairs across four granularity levels: scene classification (L1), regional reasoning (L2), spatial localization (L3), and semantic reasoning (L4). Our experiments on three NTN-TN scenarios with a frozen Qwen2-VL-7B and a trained ResNet-18 reveal a clear taskdependent complementarity: CNN achieves 72.9% accuracy at severity classification (L1) and 0.552 IoU at spatial localization (L3), while VLM uniquely enables semantic reasoning (L4) with F1=0.576 using only three in-context examples-a capability fundamentally absent in CNN architectures. Chain-of-thought (CoT) prompting further improves VLM reasoning by 12.6% (F1: 0.209->0.233) while having zero effect on spatial tasks, confirming that the complementarity is rooted in architectural differences rather than prompting limitations. A deterministic task-type router that delegates supervised tasks to CNN and reasoning tasks to VLM achieves a composite score of 0.616, a 39.1% improvement over CNN alone. We further show that VLM representations exhibit stronger cross-scenario robustness, with smaller performance degradation in 5 out of 6 transfer directions. These findings provide actionable guidelines: deploy CNNs for spatial localization and VLMs for semantic spectrum reasoning, rather than treating them as substitutes.

Data-driven decision making frequently relies on predicting counterfactual outcomes. In practice, researchers commonly train counterfactual prediction models on a source dataset to inform decisions on a possibly separate target population. Conformal prediction has arisen as a popular method for producing assumption-lean prediction intervals for counterfactual outcomes that would arise under different treatment decisions in the target population of interest. However, existing methods require that every confounding factor of the treatment-outcome relationship used for training on the source data is additionally measured in the target population, risking miscoverage if important confounders are unmeasured in the target population. In this paper, we introduce a computationally efficient debiased machine learning framework that allows for valid prediction intervals when only a subset of confounders is measured in the target population, a common challenge referred to as runtime confounding. Grounded in semiparametric efficiency theory, we show the resulting prediction intervals achieve desired coverage rates with faster convergence compared to standard methods. Through numerous synthetic and semi-synthetic experiments, we demonstrate the utility of our proposed method.

Unsustainable land-use practices in ecologically sensitive regions threaten biodiversity, water resources, and the livelihoods of millions. This paper presents a deep reinforcement learning (RL) framework for optimizing land-use allocation in the Lake Malawi Basin to maximize total ecosystem service value (ESV). Drawing on the benefit transfer methodology of Costanza et al., we assign biome-specific ESV coefficients -- locally anchored to a Malawi wetland valuation -- to nine land-cover classes derived from Sentinel-2 imagery. The RL environment models a 50x50 cell grid at 500m resolution, where a Proximal Policy Optimization (PPO) agent with action masking iteratively transfers land-use pixels between modifiable classes. The reward function combines per-cell ecological value with spatial coherence objectives: contiguity bonuses for ecologically connected land-use patches (forest, cropland, built area etc.) and buffer zone penalties for high-impact development adjacent to water bodies. We evaluate the framework across three scenarios: (i) pure ESV maximization, (ii) ESV with spatial reward shaping, and (iii) a regenerative agriculture policy scenario. Results demonstrate that the agent effectively learns to increase total ESV; that spatial reward shaping successfully steers allocations toward ecologically sound patterns, including homogeneous land-use clustering and slight forest consolidation near water bodies; and that the framework responds meaningfully to policy parameter changes, establishing its utility as a scenario-analysis tool for environmental planning.

Large language models have found success by scaling up capabilities to work in general settings. The same can unfortunately not be said for interpretability methods. The current trend in mechanistic interpretability is to provide precise explanations of specific behaviors in controlled settings. These often do not generalize, or are too resource intensive for larger studies. In this work we propose to study repeated behaviors in large language models by mining completion scenarios in Java code datasets, through exploiting the structured nature of code. We collect the attention patterns generated in the attention heads to demonstrate that they are scalable signals for global interpretability of model components. We show that vision models offer a promising direction for analyzing attention patterns at scale. To demonstrate this, we introduce the Attention Pattern - Masked Autoencoder(AP-MAE), a vision transformer-based model that efficiently reconstructs masked attention patterns. Experiments on StarCoder2 show that AP-MAE (i) reconstructs masked attention patterns with high accuracy, (ii) generalizes across unseen models with minimal degradation, (iii) reveals recurring patterns across inferences, (iv) predicts whether a generation will be correct without access to ground truth, with accuracies ranging from 55% to 70% depending on the task, and (v) enables targeted interventions that increase accuracy by 13.6% when applied selectively, but cause collapse when applied excessively. These results establish attention patterns as a scalable signal for interpretability and demonstrate that AP-MAE provides a transferable foundation for both analysis and intervention in large language models. Beyond its standalone value, AP-MAE also serves as a selection procedure to guide fine-grained mechanistic approaches. We release code and models to support future work in large-scale interpretability.

Enabling robots to learn long-horizon manipulation tasks from a handful of demonstrations remains a central challenge in robotics. Existing neuro-symbolic approaches often rely on hand-crafted symbolic abstractions, semantically labeled trajectories or large demonstration datasets, limiting their scalability and real-world applicability. We present a scalable neuro-symbolic framework that autonomously constructs symbolic planning domains and data-efficient control policies from as few as one to thirty unannotated skill demonstrations, without requiring manual domain engineering. Our method segments demonstrations into skills and employs a Vision-Language Model (VLM) to classify skills and identify equivalent high-level states, enabling automatic construction of a state-transition graph. This graph is processed by an Answer Set Programming solver to synthesize a PDDL planning domain, which an oracle function exploits to isolate the minimal, task-relevant and target relative observation and action spaces for each skill policy. Policies are learned at the control reference level rather than at the raw actuator signal level, yielding a smoother and less noisy learning target. Known controllers can be leveraged for real-world data augmentation by projecting a single demonstration onto other objects in the scene, simultaneously enriching the graph construction process and the dataset for imitation learning. We validate our framework primarily on a real industrial forklift across statistically rigorous manipulation trials, and demonstrate cross-platform generality on a Kinova Gen3 robotic arm across two standard benchmarks. Our results show that grounding control learning, VLM-driven abstraction, and automated planning synthesis into a unified pipeline constitutes a practical path toward scalable, data-efficient, expert-free and interpretable neuro-symbolic robotics.

Formal specification generation has recently drawn attention in software engineering as a way to improve program correctness without requiring manual annotations. Large Language Models (LLMs) have shown promise in this area, but early results reveal several limitations. Generated specifications often fail verification due to syntax errors, logical inaccuracies, or incomplete reasoning, especially in programs with loops or branching logic. Techniques like SpecGen and FormalBench attempt to address this through prompting and benchmarking, but they typically rely on static prompts and do not offer mechanisms for recovering from failure or adapting to different program structures. In this paper, we present AutoReSpec, a collaborative framework that combines open and closed-source LLMs for verifiable specification generation. AutoReSpec dynamically chooses an LLM pair and prompt configuration based on the structure of the input program. If the primary LLM fails to produce a valid output, a collaborative model is invoked, using validator feedback to refine and correct the specification. This two-stage design enables both speed and robustness. We evaluate AutoReSpec on a new benchmark of 72 real-world and synthetic Java programs. Our results show that it achieves 67 passes out of 72, outperforming SpecGen and FormalBench in both Success Probability and Completeness. Our experimental evaluation achieves a 58.2% success probability and a 69.2% completeness score, while cutting evaluation time by 26.89% on average compared to prior methods. Together, these results demonstrate that AutoReSpec offers a scalable, efficient, and reliable approach to LLM-based formal specification generation.

Can humans tell whether a news article was written by a person or a large language model (LLM)? We investigate this question using JudgeGPT, a study platform that independently measures source attribution (human vs. machine) and authenticity judgment (legitimate vs. fake) on continuous scales. From 2,318 judgments collected from 1,054 participants across content generated by six LLMs, we report five findings: (1) participants cannot reliably distinguish machine-generated from human-written text (p > .05, Welch's t-test); (2) this inability holds across all tested models, including open-weight models with as few as 7B parameters; (3) self-reported domain expertise predicts judgment accuracy (r = .35, p < .001) whereas political orientation does not (r = -.10, n.s.); (4) clustering reveals distinct response strategies ("Skeptics" vs. "Believers"); and (5) accuracy degrades after approximately 30 sequential evaluations due to cognitive fatigue. The answer, in short, is no: humans cannot reliably tell. These results indicate that user-side detection is not a viable defense and motivate system-level countermeasures such as cryptographic content provenance.

Large language models (LLMs) have been shown to encode truth of statements in their activation space along a linear truth direction. Previous studies have argued that these directions are universal in certain aspects, while more recent work has questioned this conclusion drawing on limited generalization across some settings. In this work, we identify a number of limits of truth-direction universality that have not been previously understood. We first show that truth directions are highly layer-dependent, and that a full understanding of universality requires probing at many layers in the model. We then show that truth directions depend heavily on task type, emerging in earlier layers for factual and later layers for reasoning tasks; they also vary in performance across levels of task complexity. Finally, we show that model instructions dramatically affect truth directions; simple correctness evaluation instructions significantly affect the generalization ability of truth probes. Our findings indicate that universality claims for truth directions are more limited than previously known, with significant differences observable for various model layers, task difficulties, task types, and prompt templates.

Modern advanced driver assistance systems (ADAS) rely on deep neural networks (DNNs) for perception and planning. Since DNNs' parameters reside in DRAM during inference, bit flips caused by cosmic radiation or low-voltage operation may corrupt DNN computations, distort driving decisions, and lead to real-world incidents. This paper presents a SpatioTemporal-Aware Fault Injection (STAFI) framework to locate critical fault sites in DNNs for ADAS efficiently. Spatially, we propose a Progressive Metric-guided Bit Search (PMBS) that efficiently identifies critical network weight bits whose corruption causes the largest deviations in driving behavior (e.g., unintended acceleration or steering). Furthermore, we develop a Critical Fault Time Identification (CFTI) mechanism that determines when to trigger these faults, taking into account the context of real-time systems and environmental states, to maximize the safety impact. Experiments on DNNs for a production ADAS demonstrate that STAFI uncovers 29.56x more hazard-inducing critical faults than the strongest baseline.

Reverse engineering (RE) is central to software security, particularly for cryptographic programs that handle sensitive data and are highly prone to vulnerabilities. It supports critical tasks such as vulnerability discovery and malware analysis. Despite its importance, RE remains labor-intensive and requires substantial expertise, making large language models (LLMs) a potential solution for automating the process. However, their capabilities for RE remain systematically underexplored. To address this gap, we study the cryptographic binary RE capabilities of LLMs and introduce \textbf{CREBench}, a benchmark comprising 432 challenges built from 48 standard cryptographic algorithms, 3 insecure crypto key usage scenarios, and 3 difficulty levels. Each challenge follows a Capture-the-Flag (CTF) RE challenge, requiring the model to analyze the underlying cryptographic logic and recover the correct input. We design an evaluation framework comprising four sub-tasks, from algorithm identification to correct flag recovery. We evaluate eight frontier LLMs on CREBench. GPT-5.4, the best-performing model, achieves 64.03 out of 100 and recovers the flag in 59\% of challenges. We also establish a strong human expert baseline of 92.19 points, showing that humans maintain an advantage in cryptographic RE tasks. Our code and dataset are available at https://github.com/wangyu-ovo/CREBench.

Effective evaluation of large language models (LLMs) remains a critical bottleneck, as conventional direct scoring often yields inconsistent and opaque judgments. In this work, we adapt the Analytic Hierarchy Process (AHP) to LLM-based evaluation and, more importantly, propose a confidence-aware Fuzzy AHP (FAHP) extension that models epistemic uncertainty via triangular fuzzy numbers modulated by LLM-generated confidence scores. Systematically validated on JudgeBench, our structured approach decomposes assessments into explicit criteria and incorporates uncertainty-aware aggregation, producing more calibrated judgments. Extensive experiments demonstrate that both crisp and fuzzy AHP consistently outperform direct scoring across model scales and dataset splits, with FAHP showing superior stability in uncertain comparison scenarios. Building on these insights, we propose \textbf{DualJudge}, a hybrid framework inspired by Dual-Process Theory that adaptively fuses holistic direct scores with structured AHP outputs via consistency-aware weighting. DualJudge achieves state-of-the-art performance, underscoring the complementary strengths of intuitive and deliberative evaluation paradigms. These results establish uncertainty-aware structured reasoning as a principled pathway toward more reliable LLM assessment. Code is available at https://github.com/hreyulog/AHP_llm_judge.

We consider the problem of conditional independence (CI) testing and adopt a kernel-based approach. Kernel-based CI tests embed variables in reproducing kernel Hilbert spaces, regress their embeddings on the conditioning variables, and test the resulting residuals for marginal independence. This approach yields tests that are sensitive to a broad range of conditional dependencies. Existing methods, however, rely heavily on kernel ridge regression, which is computationally expensive when properly tuned and yields poorly calibrated tests when left untuned, which limits their practical usefulness. We propose the Generalised Kernel Covariance Measure (GKCM), a regression-model-agnostic kernel-based CI test that accommodates a broad class of regression estimators. Building on the Generalised Hilbertian Covariance Measure framework (Lundborg et al., 2022), we characterise conditions under which GKCM satisfies uniform asymptotic level guarantees. In simulations, GKCM paired with tree-based regression models frequently outperforms state-of-the-art CI tests across a diverse range of data-generating processes, achieving better type I error control and competitive or superior power.

The Augmented Human vision broadly seeks to improve or expand baseline human functioning through the restoration or extension of physical, intellectual, and social capabilities. However, given the rapid pace of technology development, we ask: what exactly does Augmented Human research involve, what are its core themes, and how has the Augmented Human(s) conference series evolved over time? To answer this, we conducted a scientometric analysis on the past 15 years of the Augmented Human(s) conference (N=735 paper), focusing on: geographical aspects, submissions and citation timelines, author frequency and popularity, and topic modeling. We find that: (a) Number of papers in the conference exhibit a bimodal distribution, peaking in 2015 and 2025, but showing periods of stagnant growth; (b) key topics over time include Haptics, Wearable Sensing, Vision & Eye Tracking, Embodied Interaction, and Sports / Motion; (c) some seminal papers on AH are not published in AH(s), but rather at related venues (e.g., CHI); (d) the conference has an active Japanese HCI community despite its historical Eurocentric location dominance. We contribute a closer look at the trajectory of the AH(s) field, and raise considerations of definitional and research scope ambiguities given the core problems/enhancements the field seeks to address.

This paper addresses the challenge of operationalizing ethics in autonomous systems through runtime enforcement. It first conceptualizes the system's ethical space and outlines a structured ethics assurance process. Building on this foundation, it introduces an enforcement subsystem that operationalizes ethical rules, specifically social, legal, ethical, empathetic, and cultural (SLEEC) requirements, through the Abstract State Machine (ASM) formalism. The enforcement subsystem is built on the MAPE-K control-loop architecture for monitoring and controlling the system's ethical behavior, and it relies on an ASM-based runtime model of the ethical rules to enforce. This enables the dynamic evaluation, adaptation, and enforcement of ethical behavior within a runtime formal model. The overall approach, named SLEEC@run.time, is demonstrated on an assistive robot scenario, showcasing how both the robot's behavior and the governing ethical rules can dynamically adapt to contextual changes. By leveraging a flexible runtime model, SLEEC@run.time accommodates changes such as the addition or removal of SLEEC rules, ensuring a robust and evolvable approach to ethical assurance in autonomous systems. The evaluation of SLEEC@run.time shows that it effectively ensures the system's adherence to ethical principles with negligible execution time overhead.

Graph signal processing (GSP) is becoming a major tool in biomedical signal and image analysis. In most GSP techniques, graph structures and edge weights have been typically set via statistical and computational methods. More recently, graph structure learning methods offered more reliable and flexible data representations. In this work, we introduce a graph learning approach for melanoma detection in dermoscopic images based on two graph-theoretic representations: superpixel ensemble graphs (SEG) and superpixel hierarchy graphs (SHG). For these two types of graphs, superpixel maps of a skin lesion image are respectively generated at multiple levels without and with parentchild constraints among superpixels at adjacent levels, where each level corresponds to a subgraph with a different number of nodes (20, 40, 60, 80, or 100 nodes). Two edge weight assignment techniques are explored: handcrafted Gaussian weights and learned weights based on optimization methods. The graph nodal signals are assigned based on texture, geometric, and color superpixel features. In addition, the effect of graph edge thresholding is investigated by applying different thresholds (25%, 50%, and 75%) to prune the weakest edges and analyze the impact of pruning on the melanoma detection performance. Experimental evaluation of the proposed method is performed with different classifiers trained and tested on the publicly available ISIC2017 dataset. Data augmentation is applied to alleviate class imbalance by adding more melanoma images from the ISIC archive. The results show that learned superpixel ensemble graphs with textural nodal signals give the highest performance reaching an accuracy of 99.00% and an AUC of 99.59%.

Constrained multiobjective optimisation requires fast feasibility attainment together with stable convergence and diversity preservation under strict evaluation budgets. This report documents RDEx-CMOP, the differential evolution variant used in the IEEE CEC 2025 numerical optimisation competition (C06 special session) constrained multiobjective track. RDEx-CMOP integrates an ε-level feasibility schedule, a SPEA2-style indicator-driven fitness assignment, and a fitness-oriented current-to-pbest/1 mutation operator. We evaluate RDEx-CMOP on the official CEC 2025 CMOP benchmark using the median-target U-score framework and the released trace data. Experimental results show that RDEx-CMOP achieves the highest total score and the best overall average rank among all released comparison algorithms, with strong target-attainment behaviour and near-zero final violation on most problems.

Hyper-heuristics have become a popular approach for solving dynamic flexible job shop scheduling (DFJSS) problems. They use gradient-free optimization techniques like Genetic Programming (GP) to evolve non-differentiable heuristics. However, conventional GP methods tend to converge slowly because they rely solely on evolutionary search to find good heuristics. Existing multitask GP methods can solve multiple tasks simultaneously and speed up the search by transferring knowledge across similar tasks. But they mostly exchange heuristic building blocks without truly generating heuristics conditioned on task information. In this paper, we aim to accelerate convergence and enable task-specific heuristic generation by incorporating a task-conditioned Transformer model. The Transformer works in two ways. First, it learns the distribution of elite heuristics, biasing the search toward promising regions of the heuristic space. Second, through conditional generation, it produces heuristics tailored to specific tasks, allowing the model to handle multiple scheduling tasks at once and improving overall optimization efficiency. Based on these ideas, we propose TransGP, a Task-Conditioned Transformer-Guided GP framework. This evolutionary paradigm integrates generative modeling with GP, enabling efficient multitask heuristic learning and knowledge transfer. We evaluate TransGP on a range of DFJSS scenarios. Experimental results show that TransGP consistently outperforms multitask GP baselines, widely used handcrafted heuristics, and the pure Transformer model, achieving faster convergence, superior solution quality, and enhanced robustness.

Large Language Models (LLMs) can compose poetry, but how far are they from human poets? In this paper, we introduce POEMetric, the first comprehensive framework for poetry evaluation, examining 1) basic instruction-following abilities in generating poems according to a certain form and theme, 2) advanced abilities of showing creativity, lexical diversity, and idiosyncrasy, evoking emotional resonance, and using imagery and literary devices, and 3) general appraisal of the overall poem quality and estimation of authorship. We curated a human poem dataset - 203 English poems of 7 fixed forms annotated with meter, rhyme patterns and themes - and experimented with 30 LLMs for poetry generation based on the same forms and themes of the human data, totaling 6,090 LLM poems. Based on POEMetric, we assessed the performance of both human poets and LLMs through rule-based evaluation and LLM-as-a-judge, whose results were validated by human experts. Results show that, though the top model achieved high form accuracy (4.26 out of 5.00, with Gemini-2.5-Pro as a judge; same below) and theme alignment (4.99), all models failed to reach the same level of advanced abilities as human poets, who achieved unparalleled creativity (4.02), idiosyncrasy (3.95), emotional resonance (4.06), and skillful use of imagery (4.49) and literary devices (4.67). Humans also defeated the best-performing LLM in overall poem quality (4.22 vs. 3.20). As such, poetry generation remains a formidable challenge for LLMs. Data and codes are released at https://github.com/Bingru-Li/POEMetric.

Sequential Recommendation (SR) learns user preferences from their historical interaction sequences and provides personalized suggestions. In real-world scenarios, most items exhibit sparse interactions, known as the tail-item problem. This issue limits the model's ability to accurately capture item transition patterns. To tackle this, large language models (LLMs) offer a promising solution by capturing semantic relationships between items. Despite previous efforts to leverage LLM-derived embeddings for enriching tail items, they still face the following limitations: 1) They struggle to effectively fuse collaborative signals with semantic knowledge, leading to suboptimal item embedding quality. 2) Existing methods overlook the structural inconsistency between the ID and LLM embedding spaces, causing conflicting signals that degrade recommendation accuracy. In this work, we propose a Fusion and Alignment Enhancement framework with LLMs for Tail-item Sequential Recommendation (FAERec), which improves item representations by generating coherently-fused and structurally consistent embeddings. For the information fusion challenge, we design an adaptive gating mechanism that dynamically fuses ID and LLM embeddings. Then, we propose a dual-level alignment approach to mitigate structural inconsistency. The item-level alignment establishes correspondences between ID and LLM embeddings of the same item through contrastive learning, while the feature-level alignment constrains the correlation patterns between corresponding dimensions across the two embedding spaces. Furthermore, the weights of the two alignments are adjusted by a curriculum learning scheduler to avoid premature optimization of the complex feature-level objective. Extensive experiments across three widely used datasets with multiple representative SR backbones demonstrate the effectiveness and generalizability of our framework.

With the servitization of business, understanding how users experience services becomes a crucial success factor for companies. Therefore, there is a need to include feedback from user experiences in the software engineering process. Behavioral models of user journeys, describing how users experience their interaction with a service, can provide insights and potentially improve services. In this paper, we investigate techniques that allow the automatic generation of behavioral models from user interactions with a service, recorded in an event log. We first compare two established techniques that generate behavioral models from a given event log: automata learning and process mining. Afterward, we present a novel, hybrid method that combines both automata learning and process mining methods to overcome their limitations. For the existing techniques, we present methods to learn models of user journeys and evaluate the accuracy of the resulting models. We then compare these techniques with our novel method for the automatic extraction of user journey models from the event logs of digital services. We assess the practical applicability of all techniques by evaluating real-world applications. Our results show that process mining techniques rely on expert knowledge, while automata learning techniques depend on the distribution of events in the given event log. We further show that the proposed hybrid technique combines the strengths of both process mining and automata learning, automatically selecting the best method and parameter settings for a given event log to learn very accurate models.

Large language models (LLMs) are increasingly being used for text classification across the social sciences, yet researchers overwhelmingly classify one text per variable per prompt. Coding 100,000 texts on four variables requires 400,000 API calls. Batching 25 items and stacking all variables into a single prompt reduces this to 4,000 calls, cutting token costs by over 80%. Whether this degrades coding quality is unknown. We tested eight production LLMs from four providers on 3,962 expert-coded tweets across four tasks, varying batch size from 1 to 1,000 items and stacking up to 25 coding dimensions per prompt. Six of eight models maintained accuracy within 2 pp of the single-item baseline through batch sizes of 100. Variable stacking with up to 10 dimensions produced results comparable to single-variable coding, with degradation driven by task complexity rather than prompt length. Within this safe operating range, the measurement error from batching and stacking is smaller than typical inter-coder disagreement in the ground-truth data.

Large language models (LLMs) excel at complex reasoning, yet their efficiency is limited by the surging cognitive overhead of long thought traces. In this paper, we propose LightThinker, a method that enables LLMs to dynamically compress intermediate thoughts into compact semantic representations. However, static compression often struggles with complex reasoning where the irreversible loss of intermediate details can lead to logical bottlenecks. To address this, we evolve the framework into LightThinker++, introducing Explicit Adaptive Memory Management. This paradigm shifts to behavioral-level management by incorporating explicit memory primitives, supported by a specialized trajectory synthesis pipeline to train purposeful memory scheduling. Extensive experiments demonstrate the framework's versatility across three dimensions. (1) LightThinker reduces peak token usage by 70% and inference time by 26% with minimal accuracy loss. (2) In standard reasoning, LightThinker++ slashes peak token usage by 69.9% while yielding a +2.42% accuracy gain under the same context budget for maximum performance. (3) Most notably, in long-horizon agentic tasks, it maintains a stable footprint beyond 80 rounds (a 60%-70% reduction), achieving an average performance gain of 14.8% across different complex scenarios. Overall, our work provides a scalable direction for sustaining deep LLM reasoning over extended horizons with minimal overhead.

One way to solve the challenge of small and medium-sized enterprise (SME) manufacturers of acquiring sufficient orders is by joining digital Manufacturing-as-a-Service (MaaS) platforms for on-demand manufacturing. However, joining such platforms brings about new challenges such as efficient quoting handling in the face of potentially low success rates and the need for high production quality for low lot sizes. Automating the complete interaction between manufacturers and MaaS platforms, from registering the manufacturer and its capabilities to handling incoming requests and managing offers, orders, and production quality reporting, helps to overcome these challenges. Thus, the increased number of requests can be handled efficiently, and the production quality can be maintained at a high level even for low lot sizes. This paper presents an architecture for automating the interaction and functional building blocks between manufacturers and MaaS platforms, along with a prototype implementation and evaluation of its effectiveness in addressing the challenges SME manufacturers are faced with.

Masked diffusion language models (dLMs) generate text through bidirectional denoising, yet this capability remains locked for infilling prompts. This limitation is an artifact of the current supervised finetuning (SFT) convention of applying response-only masking. To unlock this capability, we extend full-sequence masking during SFT, where both prompts and responses are masked jointly. Once unlocked, the model infills masked portions of a prompt template conditioned on few-shot examples. We show that such model-infilled prompts match or surpass manually designed templates, transfer effectively across models, and are complementary to existing prompt optimization methods. Our results suggest that training practices, not architectural limitations, are the primary bottleneck preventing masked diffusion language models from infilling effective prompts

In agentic search, large language models (LLMs) are trained to perform multi-turn retrieval and reasoning for complex tasks such as multi-hop question answering (QA). However, current search-based Reinforcement Learning (RL) methods suffer from two core limitations: expensive long-horizon rollouts are under-utilized during training, and supervision is typically available only at the final answer, resulting in severe reward sparsity. We present Prefix-based Rollout reuse for Agentic search with Intermediate Step rEwards (PRAISE), a framework for improving both data efficiency and credit assignment in agentic search training. Given a complete search trajectory, PRAISE extracts prefix states at different search turns, elicits intermediate answers from them, and uses these prefixes both to construct additional training trajectories and to derive step-level rewards from performance differences across prefixes. Our method uses a single shared model for both search policy learning and prefix answer evaluation, enabling joint optimization without extra human annotations or a separate reward model. Experiments on multi-hop QA benchmarks show that PRAISE consistently improves performance over strong baselines.

Interpretability research has highlighted the importance of evaluating Pretrained Language Models (PLMs) and in particular contextual embeddings against explicit linguistic theories to determine what linguistic information they encode. This study focuses on the Italian NPN (noun-preposition-noun) constructional family, challenging some of the theoretical and methodological assumptions underlying previous experimental designs and extending this type of research to a lesser-investigated language. Contextual vector representations are extracted from BERT and used as input to layer-wise probing classifiers, systematically evaluating information encoded across the model's internal layers. The results shed light on the extent to which constructional form and meaning are reflected in contextual embeddings, contributing empirical evidence to the dialogue between constructionist theory and neural language modelling

Government agencies worldwide face growing volumes of citizen appeals, with electronic submissions increasing significantly over recent years. Traditional manual processing averages 20 minutes per appeal with only 67% classification accuracy, creating significant bottlenecks in public service delivery. This paper presents AI Appeals Processor, a microservice-based system that integrates natural language processing and deep learning techniques for automated classification and routing of citizen appeals. We evaluate multiple approaches -- including Bag-of-Words with SVM, TF-IDF with SVM, fastText, Word2Vec with LSTM, and BERT -- on a representative dataset of 10,000 real citizen appeals across three primary categories (complaints, applications, and proposals) and seven thematic domains. Our experiments demonstrate that a Word2Vec+LSTM architecture achieves 78% classification accuracy while reducing processing time by 54%, offering an optimal balance between accuracy and computational efficiency compared to transformer-based models.

Despite the strong language understanding abilities of large language models (LLMs), they still struggle with reliable question answering (QA) over long, structured documents, particularly for numerical reasoning. Financial annual reports exemplify this difficulty: financial statement analysis often hinges on accurate arithmetic, and analysts derive key indicators by integrating evidence scattered across multiple tables and narrative text. However, existing benchmarks focus largely on single-table settings, leaving cross-table document-level numerical reasoning underexplored. To address this gap, we introduce FinLongDocQA, a dataset for both single-table and cross-table financial numerical reasoning in long-context reports. Evaluating both closed-source and open-source LLMs on FinLongDocQA reveals two bottlenecks: (1) annual reports often exceed 129k tokens, exacerbating the context rot problem for locating relevant tables; and (2) even when relevant evidence is located, LLMs remain prone to errors in multi-step numerical reasoning. We propose FinLongDocAgent, a Multi-Agent Multi-Round Retrieval-Augmented Generation (RAG) approach that iteratively retrieves evidence, performs intermediate calculations, and verifies results across rounds. Experiments highlight the importance of iterative retrieval and verification for reliable numerical QA in long financial documents.

Structured tables are essential for conveying high-density information in professional domains such as finance, healthcare, and scientific research. Despite the progress in Multimodal Large Language Models (MLLMs), reasoning performance remains limited for complex tables with hierarchical layouts. In this paper, we identify a critical Perception Bottleneck through quantitative analysis. We find that as task complexity scales, the number of involved discrete visual regions increases disproportionately. This processing density leads to an internal "Perceptual Overload," where MLLMs struggle to maintain accurate spatial attention during implicit generation. To address this bottleneck, we introduce TableVision, a large-scale, trajectory-aware benchmark designed for spatially grounded reasoning. TableVision stratifies tabular tasks into three cognitive levels (Perception, Reasoning, and Analysis) across 13 sub-categories. By utilizing a rendering-based deterministic grounding pipeline, the dataset explicitly couples multi-step logical deductions with pixel-perfect spatial ground truths, comprising 6,799 high-fidelity reasoning trajectories. Our empirical results, supported by diagnostic probing, demonstrate that explicit spatial constraints significantly recover the reasoning potential of MLLMs. Furthermore, our two-stage decoupled framework achieves a robust 12.3% overall accuracy improvement on the test set. TableVision provides a rigorous testbed and a fresh perspective on the synergy between perception and logic in document understanding.

Generative Engine Optimization (GEO) is rapidly reshaping digital marketing paradigms in the era of Large Language Models (LLMs). However, current GEO strategies predominantly rely on Retrieval-Augmented Generation (RAG), which inherently suffers from probabilistic hallucinations and the "zero-click" paradox, failing to establish sustainable commercial trust. In this paper, we systematically deconstruct the probabilistic flaws of existing RAG-based GEO and propose a paradigm shift towards deterministic multi-agent intent routing. First, we mathematically formulate Semantic Entropy Drift (SED) to model the dynamic decay of confidence curves in LLMs over continuous temporal and contextual perturbations. To rigorously quantify optimization value in black-box commercial engines, we introduce the Isomorphic Attribution Regression (IAR) model, leveraging a Multi-Agent System (MAS) probe with strict human-in-the-loop physical isolation to enforce hallucination penalties. Furthermore, we architect the Deterministic Agent Handoff (DAH) protocol, conceptualizing an Agentic Trust Brokerage (ATB) ecosystem where LLMs function solely as intent routers rather than final answer generators. We empirically validate this architecture using EasyNote, an industrial AI meeting minutes product by Yishu Technology. By routing the intent of "knowledge graph mapping on an infinite canvas" directly to its specialized proprietary agent via DAH, we demonstrate the reduction of vertical task hallucination rates to near zero. This work establishes a foundational theoretical framework for next-generation GEO and paves the way for a well-ordered, deterministic human-AI collaboration ecosystem.

Multimodal Sentiment Analysis (MSA) requires effective modeling of cross-modal interactions and contextual dependencies while remaining computationally efficient. Existing fusion approaches predominantly rely on Transformer-based cross-modal attention, which incurs quadratic complexity with respect to sequence length and limits scalability. Moreover, contextual information from preceding utterances is often incorporated through concatenation or independent fusion, without explicit temporal modeling that captures sentiment evolution across dialogue turns. To address these limitations, we propose CAGMamba, a context-aware gated cross-modal Mamba framework for dialogue-based sentiment analysis. Specifically, we organize the contextual and the current-utterance features into a temporally ordered binary sequence, which provides Mamba with explicit temporal structure for modeling sentiment evolution. To further enable controllable cross-modal integration, we propose a Gated Cross-Modal Mamba Network (GCMN) that integrates cross-modal and unimodal paths via learnable gating to balance information fusion and modality preservation, and is trained with a three-branch multi-task objective over text, audio, and fused predictions. Experiments on three benchmark datasets demonstrate that CAGMamba achieves state-of-the-art or competitive results across multiple evaluation metrics. All codes are available at https://github.com/User2024-xj/CAGMamba.

Accurate prediction of real-world pedestrian trajectories is crucial for a wide range of robot-related applications. Recent approaches typically adopt graph-based or transformer-based frameworks to model interactions. Despite their effectiveness, these methods either introduce unnecessary computational overhead or struggle to represent the diverse and time-varying characteristics of human interactions. In this work, we present an Adaptive Relational Transformer (ART), which introduces a Temporal-Aware Relation Graph (TARG) to explicitly capture the evolution of pairwise interactions and an Adaptive Interaction Pruning (AIP) mechanism to reduce redundant computations efficiently. Extensive evaluations on ETH/UCY and NBA benchmarks show that ART delivers state-of-the-art accuracy with high computational efficiency.

We study the plurality consensus problem in distributed systems where a population of extremely simple agents, each initially holding one of k opinions, aims to agree on the initially most frequent one. In this setting, h-majority is arguably the simplest and most studied protocol, in which each agent samples the opinion of h neighbors uniformly at random and updates its opinion to the most frequent value in the sample. We propose a new, extremely simple mechanism called DéjàVu: an agent queries neighbors until it encounters an opinion for the second time, at which point it updates its own opinion to the duplicate value. This rule does not require agents to maintain counters or estimate frequencies, nor to choose any parameter (such as a sample size h); it relies solely on the primitive ability to detect repetition. We provide a rigorous analysis of DéjàVu that relies on several technical ideas of independent interest and demonstrates that it is competitive with h-majority and, in some regimes, substantially more communication-efficient, thus yielding a powerful primitive for plurality consensus.

In the unsupervised self-evolution of Multimodal Large Language Models, the quality of feedback signals during post-training is pivotal for stable and effective learning. However, existing self-evolution methods predominantly rely on majority voting to select the most frequent output as the pseudo-golden answer, which may stem from the model's intrinsic biases rather than guaranteeing the objective correctness of the reasoning paths. To counteract the degradation, we propose \textbf{C}ontinuous \textbf{S}oftened \textbf{R}etracing re\textbf{S}ampling (\textbf{CSRS}) in MLLM self-evolution. Specifically, we introduce a Retracing Re-inference Mechanism (\textbf{RRM}) that the model re-inferences from anchor points to expand the exploration of long-tail reasoning paths. Simultaneously, we propose Softened Frequency Reward (\textbf{SFR}), which replaces binary rewards with continuous signals, calibrating reward based on the answers' frequency across sampled reasoning sets. Furthermore, incorporated with Visual Semantic Perturbation (\textbf{VSP}), CSRS ensures the model prioritizes mathematical logic over visual superficiality. Experimental results demonstrate that CSRS significantly enhances the reasoning performance of Qwen2.5-VL-7B on benchmarks such as MathVision. We achieve state-of-the-art (SOTA) results in unsupervised self-evolution on geometric tasks. Our code is avaible at https://github.com/yyy195/CSRS.

Reinforcement learning in real-world systems is often accompanied by delayed feedback, which breaks the Markov assumption and impedes both learning and control. Canonical state augmentation approaches cause the state-space explosion, which introduces a severe sample-complexity burden. Despite recent progress, the state-of-the-art augmentation-based baselines remain incomplete: they either predominantly reduce the burden on the critic or adopt non-unified treatments for the actor and critic. To provide a structured and sample-efficient solution, we propose delayed homomorphic reinforcement learning (DHRL), a framework grounded in MDP homomorphisms that collapses belief-equivalent augmented states and enables efficient policy learning on the resulting abstract MDP without loss of optimality. We provide theoretical analyses of state-space compression bounds and sample complexity, and introduce a practical algorithm. Experiments on continuous control tasks in MuJoCo benchmark confirm that our algorithm outperforms strong augmentation-based baselines, particularly under long delays.

Accurate diagnosis and treatment of complex diseases require integrating histological, molecular, and clinical data, yet in practice these modalities are often incomplete owing to tissue scarcity, assay cost, and workflow constraints. Existing computational approaches attempt to impute missing modalities from available data but rely on task-specific models trained on narrow, single source-target pairs, limiting their generalizability. Here we introduce MuPD (Multimodal Pathology Diffusion), a generative foundation model that embeds hematoxylin and eosin (H&E)-stained histology, molecular RNA profiles, and clinical text into a shared latent space through a diffusion transformer with decoupled cross-modal attention. Pretrained on 100 million histology image patches, 1.6 million text-histology pairs, and 10.8 million RNA-histology pairs spanning 34 human organs, MuPD supports diverse cross-modal synthesis tasks with minimal or no task-specific fine-tuning. For text-conditioned and image-to-image generation, MuPD synthesizes histologically faithful tissue architectures, reducing Fréchet inception distance (FID) scores by 50% relative to domain-specific models and improving few-shot classification accuracy by up to 47% through synthetic data augmentation. For RNA-conditioned histology generation, MuPD reduces FID by 23% compared with the next-best method while preserving cell-type distributions across five cancer types. As a virtual stainer, MuPD translates H&E images to immunohistochemistry and multiplex immunofluorescence, improving average marker correlation by 37% over existing approaches. These results demonstrate that a single, unified generative model pretrained across heterogeneous pathology modalities can substantially outperform specialized alternatives, providing a scalable computational framework for multimodal histopathology.

We prove that temporal averaging over multiple observations can be replaced by algebraic group action on a single observation for second-order statistical estimation. A General Replacement Theorem establishes conditions under which a group-averaged estimator from one snapshot achieves equivalent subspace decomposition to multi-snapshot covariance estimation, and an Optimality Theorem proves that the symmetric group is universally optimal (yielding the KL transform). The framework unifies the DFT, DCT, and KLT as special cases of group-matched spectral transforms, with a closed-form double-commutator eigenvalue problem for polynomial-time optimal group selection. Five applications are demonstrated: MUSIC DOA estimation from a single snapshot, massive MIMO channel estimation with 64% throughput gain, single-pulse waveform classification at 90% accuracy, graph signal processing with non-Abelian groups, and a new algebraic analysis of transformer LLMs revealing that RoPE uses the wrong algebraic group for 70-80% of attention heads across five models (22,480 head observations), that the optimal group is content-dependent, and that spectral-concentration-based pruning improves perplexity at the 13B scale. All diagnostics require a single forward pass with no gradients or training.

Large language models (LLMs) have achieved substantial progress in repository-level code generation. However, solving the same repository-level task often requires multiple attempts, while existing methods still optimize each attempt in isolation and do not preserve or reuse task-specific state across attempts. In this paper, we propose LiveCoder, a novel framework for repository-level code generation based on cross-attempt knowledge optimization. LiveCoder maintains persistent task-specific state from prior attempts to guide subsequent generation. This state includes success knowledge, which captures reusable signals from previously strong repositories, failure knowledge, which records unsuccessful outcomes and their diagnostic signals, and a historical-best repository, which preserves the strongest result found so far and prevents regression. These components collectively transform repeated repository generation into a persistent, knowledge-driven optimization process. We evaluate LiveCoder using four frontier LLMs on two representative repository-level code generation benchmarks. Extensive experimental results demonstrate the effectiveness and efficiency of LiveCoder, improving the functional score by up to 22.94 percentage points, increasing repository reuse to 81.58%, and reducing cost by up to 53.63% on RAL-Bench while maintaining broadly stable non-functional quality.

On-screen learning behavior provides valuable insights into how students seek, use, and create information during learning. Analyzing on-screen behavioral engagement is essential for capturing students' cognitive and collaborative processes. The recent development of Vision Language Models (VLMs) offers new opportunities to automate the labor-intensive manual coding often required for multimodal video data analysis. In this study, we compared the performance of both leading closed-source VLMs (Claude-3.7-Sonnet, GPT-4.1) and open-source VLM (Qwen2.5-VL-72B) in single- and multi-agent settings for automated coding of screen recordings in collaborative learning contexts based on the ICAP framework. In particular, we proposed and compared two multi-agent frameworks: 1) a three-agent workflow multi-agent system (MAS) that segments screen videos by scene and detects on-screen behaviors using cursor-informed VLM prompting with evidence-based verification; 2) an autonomous-decision MAS inspired by ReAct that iteratively interleaves reasoning, tool-like operations (segmentation/ classification/ validation), and observation-driven self-correction to produce interpretable on-screen behavior labels. Experimental results demonstrated that the two proposed MAS frameworks achieved viable performance, outperforming the single VLMs in scene and action detection tasks. It is worth noting that the workflow-based agent achieved best on scene detection, and the autonomous-decision MAS achieved best on action detection. This study demonstrates the effectiveness of VLM-based Multi-agent System for video analysis and contributes a scalable framework for multimodal data analytics.

Spatial transcriptomics (ST) enables gene expression mapping within anatomical context but remains costly and low-throughput. Hematoxylin and eosin (H\&E) staining offers rich morphology yet lacks molecular resolution. We present \textbf{\ours} (\textbf{S}patial \textbf{T}ranscriptomics and hist\textbf{O}logy \textbf{R}epresentation \textbf{M}odel), a foundation model trained on 1.2 million spatially resolved transcriptomic profiles with matched histology across 18 organs. Using a hierarchical architecture integrating morphological features, gene expression, and spatial context, STORM bridges imaging and omics through robust molecular--morphological representations. STORM enhances spatial domain discovery, producing biologically coherent tissue maps, and outperforms existing methods in predicting spatial gene expression from H\&E images across 11 tumor types. The model is platform-agnostic, performing consistently across Visium, Xenium, Visium HD, and CosMx. Applied to 23 independent cohorts comprising 7,245 patients, STORM significantly improves immunotherapy response prediction and prognostication over established biomarkers, providing a scalable framework for spatially informed discovery and clinical precision medicine.

Authentication is a fundamental security means for protecting system resources. Authenticator-centric authentication techniques (AuthN Techniques) address how mechanisms and credentials are used via Authenticators. There are many AuthN Techniques that differ in many ways and there exist classification approaches that aim to structure them. However, they are limited in the aspects they classify and are not flexible enough to accommodate the diverse nature of AuthN Techniques. This paper presents two contributions. First, novel, faceted classification schemes for AuthN Techniques and Authenticators are presented. The schemes were developed based on 345 papers identified through a targeted LLM-assisted literature review and semantic clustering. The classification schemes were applied to build a catalog of Authenticators and AuthN Techniques; the second contribution of this paper. This paper presents our methodology, the classification schemes with example applications, the list of AuthN Techniques from the catalog, and discussions on future work.

Spiking Neural Networks (SNNs) offer a promising solution for energy-efficient edge intelligence; however, their hardware deployment is constrained by memory overhead, inefficient scaling operations, and limited parallelism. This work proposes L-SPINE, a low-precision SIMD-enabled spiking neural compute engine for efficient edge inference. The architecture features a unified multi-precision datapath supporting 2-bit, 4-bit, and 8-bit operations, leveraging a multiplier-less shift-add model for neuron dynamics and synaptic accumulation. Implemented on an AMD VC707 FPGA, the proposed neuron requires only 459 LUTs and 408 FFs, achieving a critical delay of 0.39 ns and 4.2 mW power. At the system level, L-SPINE achieves 46.37K LUTs, 30.4K FFs, 2.38 ms latency, and 0.54 W power. Compared to CPU and GPU platforms, it reduces inference latency from seconds to milliseconds, achieving an up to three orders-of-magnitude improvement in energy efficiency. Quantisation analysis shows that INT2/INT4 configurations significantly reduce memory footprint with minimal accuracy loss. These results establish L-SPINE as a scalable and efficient solution for real-time edge SNN deployment.

Functional verification plays a central role in ensuring the correctness of modern integrated circuit designs, where constrained-random verification is widely adopted to generate diverse stimuli under high-level constraints. In industrial verification environments, constraint solving increasingly involves dynamic data structures whose shape and content are determined at runtime, causing the sets of variables and constraint instances to evolve across solver invocations, which in turn leads to substantial overhead when nested and high-dimensional structures repeatedly expand across solves. We formalize this class of problems as the Dynamic Data Structure Constraint Satisfaction Problem (D2SCSP),which captures the interaction between dynamic data structure expansion and constraint evaluation. We propose a dependency-guided problem partitioning framework combined with an incremental encoding and constraint activation mechanism, enabling reuse of solver state and encodings across multiple solves. The framework is integrated into an industrial SystemVerilog verification flow and implemented in the commercial simulator VeriSim. Experimental results on industrial benchmarks demonstrate significant performance improvements, achieving an average speedup of 24.80x over a baseline and 1.72x over a state-of-the-art commercial simulator, highlighting the practicality of the approach for real-world verification workflows.

Large language models (LLMs) have achieved strong performance on code generation, but existing methods still struggle with repository-level code generation under executable validation. Under this evaluation setting, success is determined not by the plausibility of isolated code fragments, but by whether a generated multi-file repository can be successfully installed, have its dependencies and internal references resolved, be launched, and be validated in a real execution environment. To address this challenge, we propose EnvGraph, a framework for repository-level code generation that formulates repository executability as an environment alignment problem. EnvGraph jointly models two coupled conditions for successful repository execution, namely external dependency satisfaction and repository-internal reference resolution. It maintains a dual-layer environment representation, uses execution evidence to perform execution-evidence-based attribution, and guides repository generation through a unified targeted revision mechanism within an iterative alignment loop. We evaluate EnvGraph on repository-level code generation with three representative backbone LLMs and compare it against representative environment-aware and repository-level baselines. Experimental results show that EnvGraph consistently achieves the best performance on these repository-level benchmarks. In particular, it outperforms the strongest non-EnvGraph baseline by an absolute margin of 5.72--5.87 percentage points in Functional Correctness and 4.58--8.66 percentage points in Non-Functional Quality.

Asking a large language model to respond in JSON should be a formatting choice, not a capability tax. Yet we find that structured output requirements -- JSON, XML, LaTeX, Markdown -- substantially degrade reasoning and writing performance across open-weight models. The research response has focused on constrained decoding, but sampling bias accounts for only a fraction of the degradation. The dominant cost enters at the prompt: format-requesting instructions alone cause most of the accuracy loss, before any decoder constraint is applied. This diagnosis points to a simple principle: decouple reasoning from formatting. Whether by generating freeform first and reformatting in a second pass, or by enabling extended thinking within a single generation, separating the two concerns substantially recovers lost accuracy. Across six open-weight models, four API models, four formats, and tasks spanning math, science, logic, and writing, decoupling recovers most lost accuracy. Notably, most recent closed-weight models show little to no format tax, suggesting the problem is not inherent to structured generation but a gap that current open-weight models have yet to close. Code is available at https://github.com/ivnle/the-format-tax.

Global optimization of black-box functions from noisy samples is a fundamental challenge in machine learning and scientific computing. Traditional methods such as Bayesian Optimization often converge to local minima on multi-modal functions, while gradient-free methods require many function evaluations. We present a novel neural approach that learns to find global minima through iterative refinement. Our model takes noisy function samples and their fitted spline representation as input, then iteratively refines an initial guess toward the true global minimum. Trained on randomly generated functions with ground truth global minima obtained via exhaustive search, our method achieves a mean error of 8.05 percent on challenging multi-modal test functions, compared to 36.24 percent for the spline initialization, a 28.18 percent improvement. The model successfully finds global minima in 72 percent of test cases with error below 10 percent, demonstrating learned optimization principles rather than mere curve fitting. Our architecture combines encoding of multiple modalities including function values, derivatives, and spline coefficients with iterative position updates, enabling robust global optimization without requiring derivative information or multiple restarts.

Patching severe security flaws in complex software remains a major challenge. While automated tools like fuzzers efficiently discover bugs, fixing deep-rooted low-level faults (e.g., use-after-free and memory corruption) still requires labor-intensive manual analysis by experts. Emerging Large Language Model (LLM) agents attempt to automate this pipeline, but they typically treat bug fixing as a purely static code-generation task. Relying solely on static artifacts, these methods miss the dynamic execution context strictly necessary for diagnosing intricate memory safety violations. To overcome these limitations, we introduce DebugHarness, an autonomous LLM-powered debugging agent harness that resolves complex vulnerabilities by emulating the interactive debugging practices of human systems engineers. Instead of merely examining static code, DebugHarness actively queries the live runtime environment. Driven by a reproducible crash, it utilizes a pattern-guided investigation strategy to formulate hypotheses, interactively probes program memory states and execution paths, and synthesizes patches via a closed-loop validation cycle. We evaluate DebugHarness on SEC-bench, a rigorous dataset of real-world C/C++ security vulnerabilities. DebugHarness successfully patches approximately 90% of the evaluated bugs. This yields a relative improvement of over 30% compared to state-of-the-art baselines, demonstrating that dynamic debugging significantly enhances LLM diagnostic capabilities. Overall, DebugHarness establishes a novel paradigm for automated program repair, bridging the gap between static LLM reasoning and the dynamic intricacies of low-level systems programming.

Fault injection attacks deliberately inject faults into a device via physical channels to disturb its regular execution. Adversaries can effectively deduce secrets by analyzing both the normal and faulty outputs, posing serious threats to cryptographic primitives implemented in hardware. An effective countermeasure to such attacks is via redundancy, commonly referred to as concurrent error detection schemes, where Binary linear codes have been used to defend against fault injection attacks. However, designing an optimal code circuit is often time-consuming, error-prone, and requires substantial expertise. In this paper, we formalize the optimal code circuit synthesis problem (OptiCC) based on two domain-specific minimization objectives on individual inputs and parity size. We then propose a novel algorithm CiSC for solving OptiCC, prioritizing the minimization of individual inputs. Our approach features both correct-by-construction and secure-by-construction. In a nutshell, CiSC gradually reduces individual inputs and parity size by checking, via SMT solving, the existence of feasible Boolean functions for implementing a desired code. We further present an effective technique to lazily generate combinations of inputs to Boolean functions, while quickly identify equivalent ones. We implement our approach in a tool CiSC, and evaluate it on practical benchmarks. Experimental results show our approach can synthesize code circuits that significantly outperform those generated by the latest state-of-the-art techniques.

Federated learning (FL) research increasingly relies on single-node simulations with hundreds or thousands of virtual clients, making both efficiency and reproducibility essential. Yet parallel client training often introduces nondeterminism through shared random state and scheduling variability, forcing researchers to trade throughput for reproducibility or to implement custom control logic within complex frameworks. We present BlazeFL, a lightweight framework for single-node FL simulation that alleviates this trade-off through free-threaded shared-memory execution and deterministic randomness management. BlazeFL uses thread-based parallelism with in-memory parameter exchange between the server and clients, avoiding serialization and inter-process communication overhead. To support deterministic execution, BlazeFL assigns isolated random number generator (RNG) streams to clients. Under a fixed software/hardware stack, and when stochastic operators consume BlazeFL-managed generators, this design yields bitwise-identical results across repeated high-concurrency runs in both thread-based and process-based modes. In CIFAR-10 image-classification experiments, BlazeFL substantially reduces execution time relative to a widely used open-source baseline, achieving up to 3.1$\times$ speedup on communication-dominated workloads while preserving a lightweight dependency footprint. Our open-source implementation is available at: https://github.com/kitsuyaazuma/blazefl.

We study online task allocation for multi-robot, multi-queue systems with asymmetric stochastic arrivals and switching delays. We formulate the problem in discrete time: each location can host at most one robot per slot, servicing a task consumes one slot, switching between locations incurs a one-slot travel delay, and arrivals at locations are independent Bernoulli processes with heterogeneous rates. Building on our previous structural result that optimal policies are of exhaustive type, we formulate a discounted-cost Markov decision process and develop an exhaustive-assignment actor-critic policy architecture that enforces exhaustive service by construction and learns only the next-queue allocation for idle robots. Unlike the exhaustive-serve-longest (ESL) queue rule, whose optimality is known only under symmetry, the proposed policy adapts to asymmetry in arrival rates. Across different server-location ratios, loads, and asymmetric arrival profiles, the proposed policy consistently achieves lower discounted holding cost and smaller mean queue length than the ESL baseline, while remaining near-optimal on instances where an optimal benchmark is available. These results show that structure-aware actor-critic methods provide an effective approach for real-time multi-robot scheduling.

Plug-and-play (PnP) methods are widely used for solving imaging inverse problems by incorporating a denoiser into optimization algorithms. Score-based diffusion models (SBDMs) have recently demonstrated strong generative performance through a denoiser trained across a wide range of noise levels. Despite their shared reliance on denoisers, it remains unclear how to systematically use SBDMs as priors within the PnP framework without relying on reverse diffusion sampling. In this paper, we establish a score-based interpretation of PnP that justifies using pretrained SBDMs directly within PnP algorithms. Building on this connection, we introduce a stochastic generative PnP (SGPnP) framework that injects noise to better leverage the expressive generative SBDM priors, thereby improving robustness in severely ill-posed inverse problems. We provide a new theory showing that this noise injection induces optimization on a Gaussian-smoothed objective and promotes escape from strict saddle points. Experiments on challenging inverse tasks, such as multi-coil MRI reconstruction and large-mask natural image inpainting, demonstrate consistent improvement over conventional PnP methods and achieve performance competitive with diffusion-based solvers.

A single agent represents a single system capable of ingesting local data, indexing, cataloging information, performing knowledge pattern discovery, and separating patterns and anomalies from data. Multiple agents work collaboratively in a peer-to-peer network. Each agent has a peer list. Such multiple agents' collaboration can be modeled as cooperative games. Each agent optimizes its own objective locally. We show that each agent self-organizes or converges to its best value and the whole agent network achieves the best social welfare based on both the quantum adiabatic evolution transformation (QAET), and quantum intelligence game (QIG) or the QAET-QIG framework. We apply the QAET-QIG framework to the kill web concept that can potentially improve the traditional kill chain process or the find, fix, track, target, engage, and assess (F2T2EA) process. The improvement is measured in the values of powerful global optimization, distributed lethality, and load balancing. We show a use case of the QAET-QIG frame in a potential application of mixed sensors, platforms, weapons, and effects.

For a larger set of predictions of several differently trained machine learning models, known as bagging predictors, the mean of all predictions is taken by default. Nevertheless, this proceeding can deviate from the actual ground truth in certain parameter regions. An approach is presented to determine a representative y_BS from such a set of predictions using Kernel Density Estimation (KDE) in nonlinear regression with Neural Networks (NN) which simultaneously provides an associated quality criterion beta_BS, called Bagging Score (BS), that reflects the confidence of the obtained ensemble prediction. It is shown that working with the new approach better predictions can be made than working with the common use of mean or median. In addition to this, the used method is contrasted to several approaches of nonlinear regression from the literatur, resulting in a top ranking in each of the calculated error values without using any optimization or feature selection technique.

Mixture-of-Experts (MoE) models exhibit striking performance disparities across languages, yet the internal mechanisms driving these gaps remain poorly understood. In this work, we conduct a systematic analysis of expert routing patterns in MoE models, revealing a phenomenon we term Language Routing Isolation, in which high- and low-resource languages tend to activate largely disjoint expert sets. Through layer-stratified analysis, we further show that routing patterns exhibit a layer-wise convergence-divergence pattern across model depth. Building on these findings, we propose RISE (Routing Isolation-guided Subnetwork Enhancement), a framework that exploits routing isolation to identify and adapt language-specific expert subnetworks. RISE applies a tripartite selection strategy, using specificity scores to identify language-specific experts in shallow and deep layers and overlap scores to select universal experts in middle layers. By training only the selected subnetwork while freezing all other parameters, RISE substantially improves low-resource language performance while preserving capabilities in other languages. Experiments on 10 languages demonstrate that RISE achieves target-language F1 gains of up to 10.85% with minimal cross-lingual degradation.

As large-scale HPC compute clusters increasingly adopt accelerators such as GPUs to meet the voracious demands of modern workloads, these clusters are increasingly becoming power constrained. Unfortunately, modern applications can often temporarily exceed the power ratings of the accelerators ("power spikes"). Thus, current and future HPC systems must optimize for both power and performance together. However, this is made difficult by increasingly diverse applications, which often require bespoke optimizations to run efficiently on each cluster. Traditionally researchers overcome this problem by profiling applications on specific clusters and optimizing, but the scale, algorithmic diversity, and lack of effective tools make this challenging. To overcome these inefficiencies, we propose Minos, a systematic classification mechanism that identifies similar application characteristics via low-cost profiling for power and performance. This allows us to group similarly behaving workloads into a finite number of distinct classes and reduce the overhead of extensively profiling new workloads. For example, when predicting frequency capping behavior for a previously unseen application, Minos reduces profiling time by 89%. Moreover, across 18 popular graph analytics, HPC, HPC+ML, and ML workloads, Minos achieves a mean error of 4% for power predictions and 3% for performance predictions, significantly improving predictions over state-of-the-art approaches by 10%.

Small language models (SLMs) have been increasingly deployed in edge devices and other resource-constrained settings. However, these models make confident mispredictions and produce unstable output, making them risky for factual and decision-critical tasks. Current evaluation methodology relies on final accuracy or hallucination rates without explaining how internal model behavior affects outputs. Specifically, how entropy evolves during decoding, how attention is distributed across layers, and how hidden representations contribute to uncertainty, logical inconsistencies, and misinformation propagation are often overlooked. Consequently, this study introduces a trace-level analysis of entropy and attention dynamics in SLMs evaluated with the TruthfulQA dataset. Four models with parameter ranges of 1B-1.7B parameters were examined via token-level output entropy, attention entropy, head dispersion, and hidden-state representation. The results reflect three model classifications by entropy patterns. Deterministic models (DeepSeek-1.5B and LLaMA-1B): output entropy decreases over time. Exploratory models (Gemma-1B): with increasing entropy, and balanced models (Qwen-1.7B): have moderate and stable entropy. Also, each group has distinctively different hidden-state movement and attention dispersion patterns. The analysis demonstrates that truthfulness in SLMs emerges from structured entropy and attention dynamics. Monitoring and optimizing these internal uncertainty patterns can guide the design of a more reliable, hallucination-aware, and application-specific edge SLMs.

AI agents operating over extended time horizons accumulate experiences that serve multiple concurrent goals, and must often maintain conflicting interpretations of the same events. A concession during a client negotiation encodes as a ``trust-building investment'' for one strategic goal and a ``contractual liability'' for another. Current memory architectures assume a single correct encoding, or at best support multiple views over unified storage. We propose Rashomon Memory: an architecture where parallel goal-conditioned agents encode experiences according to their priorities and negotiate at query time through argumentation. Each perspective maintains its own ontology and knowledge graph. At retrieval, perspectives propose interpretations, critique each other's proposals using asymmetric domain knowledge, and Dung's argumentation semantics determines which proposals survive. The resulting attack graph is itself an explanation: it records which interpretation was selected, which alternatives were considered, and on what grounds they were rejected. We present a proof-of-concept showing that retrieval modes (selection, composition, conflict surfacing) emerge from attack graph topology, and that the conflict surfacing mode, where the system reports genuine disagreement rather than forcing resolution, lets decision-makers see the underlying interpretive conflict directly.

Reasoning language models (RLMs) are increasingly used in programming. Yet, even state-of-the-art RLMs frequently introduce critical security vulnerabilities in generated code. Prior training-based approaches for secure code generation face a critical limitation that prevents their direct application to RLMs: they rely on costly, manually curated security datasets covering only a limited set of vulnerabilities. At the inference level, generic security reminders consistently degrade functional correctness while triggering only shallow ad-hoc vulnerability analysis. To address these problems, we present SecPI, a fine-tuning pipeline that teaches RLMs to internalize structured security reasoning, producing secure code by default without any security instructions at inference time. SecPI filters existing general-purpose coding datasets for security-relevant tasks using an LLM-based classifier, generates high-quality security reasoning traces with a teacher model guided by a structured prompt that systematically enumerates relevant CWEs and mitigations, and fine-tunes the target model on pairs of inputs with no security prompt and teacher reasoning traces -- as a result, the model learns to reason about security autonomously rather than in response to explicit instructions. An extensive evaluation on security benchmarks with state-of-the-art open-weight reasoning models validates the effectiveness of our approach. For instance, SecPI improves the percentage of functionally correct and secure generations for QwQ 32B from 48.2% to 62.2% (+14.0 points) on CWEval and from 18.2% to 22.0% on BaxBench. Further investigation also reveals strong cross-CWE and cross-language generalization beyond training vulnerabilities. Even when trained only on injection-related CWEs, QwQ 32B generates correct and secure code 9.9% more frequently on held-out memory-safety CWEs.

With the growing prevalence of multimodal news content, effective news topic classification demands models capable of jointly understanding and reasoning over heterogeneous data such as text and images. Existing methods often process modalities independently or employ simplistic fusion strategies, limiting their ability to capture complex cross-modal interactions and leverage external knowledge. To overcome these limitations, we propose MultiPress, a novel three-stage multi-agent framework for multimodal news classification. MultiPress integrates specialized agents for multimodal perception, retrieval-augmented reasoning, and gated fusion scoring, followed by a reward-driven iterative optimization mechanism. We validate MultiPress on a newly constructed large-scale multimodal news dataset, demonstrating significant improvements over strong baselines and highlighting the effectiveness of modular multi-agent collaboration and retrieval-augmented reasoning in enhancing classification accuracy and interpretability.

This study aimed to leverage graph information, particularly Rhetorical Structure Theory (RST) and Co-reference (Coref) graphs, to enhance the performance of our baseline summarization models. Specifically, we experimented with a Graph Attention Network architecture to incorporate graph information. However, this architecture did not enhance the performance. Subsequently, we used a simple Multi-layer Perceptron architecture, which improved the results in our proposed model on our primary dataset, CNN/DM. Additionally, we annotated XSum dataset with RST graph information, establishing a benchmark for future graph-based summarization models. This secondary dataset posed multiple challenges, revealing both the merits and limitations of our models.

Neural operators have emerged as data-driven surrogates for solving partial differential equations (PDEs), and their success hinges on efficiently modeling the long-range, global coupling among spatial points induced by the underlying physics. In many PDE regimes, the induced global interaction kernels are empirically compressible, exhibiting rapid spectral decay that admits low-rank approximations. We leverage this observation to unify representative global mixing modules in neural operators under a shared low-rank template: compressing high-dimensional pointwise features into a compact latent space, processing global interactions within it, and reconstructing the global context back to spatial points. Guided by this view, we introduce Low-Rank Spatial Attention (LRSA) as a clean and direct instantiation of this template. Crucially, unlike prior approaches that often rely on non-standard aggregation or normalization modules, LRSA is built purely from standard Transformer primitives, i.e., attention, normalization, and feed-forward networks, yielding a concise block that is straightforward to implement and directly compatible with hardware-optimized kernels. In our experiments, such a simple construction is sufficient to achieve high accuracy, yielding an average error reduction of over 17\% relative to second-best methods, while remaining stable and efficient in mixed-precision training.

Large Reasoning Models (LRMs) generate structured chains of thought (CoTs) before producing final answers, making them especially vulnerable to knowledge leakage through intermediate reasoning steps. Yet, the memorization of sensitive information in the training data such as copyrighted and private content has led to ethical and legal concerns. To address these issues, selective forgetting (also known as machine unlearning) has emerged as a potential remedy for LRMs. However, existing unlearning methods primarily target final answers and may degrade the overall reasoning ability of LRMs after forgetting. Additionally, directly applying unlearning on the entire CoTs could degrade the general reasoning capabilities. The key challenge for LRM unlearning lies in achieving precise unlearning of targeted knowledge while preserving the integrity of general reasoning capabilities. To bridge this gap, we in this paper propose a novel LRM unlearning framework that selectively removes sensitive reasoning components while preserving general reasoning capabilities. Our approach leverages multiple LLMs with retrieval-augmented generation (RAG) to analyze CoT traces, identify forget-relevant segments, and replace them with benign placeholders that maintain logical structure. We also introduce a new feature replacement unlearning loss for LRMs, which can simultaneously suppress the probability of generating forgotten content while reinforcing structurally valid replacements. Extensive experiments on both synthetic and medical datasets verify the desired properties of our proposed method.

Recently, evolutionary multitasking has been employed to generate a ``set of Pareto sets" (SOS) for machine learning models, addressing diverse task settings across heterogeneous environments. This involves creating a repository of compact, specialized solution models that are collectively tailored to each specific task setting and environment, enabling users to select the most suitable model based on particular specifications and preferences. In this paper, we further demonstrate the versatility and applicability of the SOS concept across diverse domains, focusing on three real-world problems: engineering design problems, inventory management problems, and hyperparameter optimization problems. Additionally, as evolutionary multitasking has proven effective in generating the SOS, we investigate the performance of current evolutionary multitasking methods on these real-world problems. Subsequently, we present visualizations of the generated SOS in both decision and objective spaces, complemented by the development of a measurement to gauge the similarity between different Pareto sets corresponding to diverse tasks. Finally, we show that by systematically examining the shifts in Pareto optimal designs across different task settings though the SOS solutions, users can gain deeper understandings on the dynamic interplay between design solutions and their performance in different settings or contexts.

Can lifetime learning expand behavioral diversity over evolutionary time, rather than collapsing it? Prior theory predicts that plasticity reduces variance by buffering organisms against environmental noise. We test this in a competitive domain: chess agents with eight NEAT-evolved neural modules, Hebbian within-game plasticity, and a desirability-domain signal chain with imagination. Across 10~seeds per Hebbian condition, a variance crossover emerges: Hebbian ON starts with lower cross-seed variance than OFF, then surpasses it at generation~34. The crossover trend is monotonic (\r{ho} = 0.91, p < 10^{-6): plasticity's effect on behavioral variance reverses over evolutionary time, initially compressing diversity (consistent with prior predictions) then expanding it as evolved Perception differences are amplified through imagination -- a feedback loop that mutation alone cannot sustain. The result is structured behavioral divergence: evolved agents select different moves on the same positions (62\% disagreement), develop distinct opening repertoires, piece preferences, and game lengths. These are not different sampling policies -- they are reproducible behavioral signatures (ICC > 0.8) with interpretable signal chain configurations. Three regimes appear depending on opponent type: exploration (Hebbian ON, heterogeneous opponent), lottery (Hebbian OFF, elitism lock-in), and transparent (same-model opponent, brain self-erasure). The transparent regime generates a falsifiable prediction: self-play systems may systematically suppress behavioral diversity by eliminating the heterogeneity that personality requires. \textbf{Keywords: Baldwin Effect, neuroevolution, NEAT, Hebbian learning, chess, cognitive architecture, personality emergence, imagination

Adaptive reward design for deep reinforcement learning (DRL) in multi-beam LEO satellite scheduling is motivated by the intuition that regime-aware reward weights should outperform static ones. We systematically test this intuition and uncover a switching-stability dilemma: near-constant reward weights (342.1 Mbps) outperform carefully-tuned dynamic weights (103.3+/-96.8 Mbps) because PPO requires a quasistationary reward signal for value function convergence. Weight adaptation-regardless of quality-degrades performance by repeatedly restarting convergence. To understand why specific weights matter, we introduce a single-variable causal probing method that independently perturbs each reward term by +/-20% and measures PPO response after 50k steps. Probing reveals counterintuitive leverage: a +20% increase in the switching penalty yields +157 Mbps for polar handover and +130 Mbps for hot-cold regimes-findings inaccessible to human experts or trained MLPs without systematic probing. We evaluate four MDP architect variants (fixed, rule-based, learned MLP, finetuned LLM) across known and novel traffic regimes. The MLP achieves 357.9 Mbps on known regimes and 325.2 Mbps on novel regimes, while the fine-tuned LLM collapses to 45.3+/-43.0 Mbps due to weight oscillation rather than lack of domain knowledge-output consistency, not knowledge, is the binding constraint. Our findings provide an empirically-grounded roadmap for LLM-DRL integration in communication systems, identifying where LLMs add irreplaceable value (natural language intent understanding) versus where simpler methods suffice.

Mobile advertisements (ads) are essential to the app economy, yet detecting them is challenging because ad content is dynamically fetched from remote servers and rendered through diverse user interfaces (UIs), making ads difficult to locate and trigger at runtime. To address this challenge, we present ADWISE, a novel framework that formulates mobile ads detection as LLM-guided, ad-oriented UI exploration. ADWISE first performs static program analysis to identify UI widgets used to place ads, which we call ad widgets. It then uses a grounded LLM reasoning loop to navigate toward and trigger these widgets under three complementary domain guidance signals: (1) WTG-based guidance, which provides global transition priors from a statically constructed window transition graph (WTG); (2) semantic guidance, which reasons over app functionality to prioritize user-likely interaction paths; and (3) structural guidance, which applies retrieval-augmented generation to match the current UI against recurring ad-heavy layouts from a knowledge base. By combining static program analysis with LLM-based reasoning over UI structure, app semantics, and retrieved analogies, ADWISE enables more effective ads detection in complex mobile UIs. Experiments on 100 benchmark apps show that ADWISE outperforms state-of-the-art baselines by 25.60% in ad widget detection. In addition, ADWISE uncovers 34.34% more ad regulation violations across six categories, directly benefiting downstream ad regulation.

Reasoning hallucinations in large language models (LLMs) often appear as fluent yet unsupported conclusions that violate either the given context or underlying factual knowledge. Although such failures are widely observed, the mechanisms by which decoder-only Transformers produce them remain poorly understood. We model next-token prediction as a graph search process over an underlying graph, where entities correspond to nodes and learned transitions form edges. From this perspective, contextual reasoning is a constrained search over a sampled subgraph (intrinsic reasoning), while context-free queries rely on memorized structures in the underlying graph (extrinsic reasoning). We show that reasoning hallucinations arise from two fundamental mechanisms: \textbf{Path Reuse}, where memorized knowledge overrides contextual constraints during early training, and \textbf{Path Compression}, where frequently traversed multi-step paths collapse into shortcut edges in later training. Together, these mechanisms provide a unified explanation for reasoning hallucinations in LLMs and connected to well-known behaviors observed in downstream applications.

Large Vision-Language Models (LVLMs) have achieved impressive progress in multimodal reasoning, yet they remain prone to object hallucinations, generating descriptions of objects that are not present in the input image. Recent approaches attempt to mitigate hallucinations by suppressing unreliable visual signals in the vision encoder, but many rely on iterative optimization for each input, resulting in substantial inference latency. In this work, we investigate the internal attention dynamics of vision encoders in LVLMs and identify a consistent three-phase structure of visual information processing: diffusion, focus, and rediffusion. Our analysis reveals that hallucination behavior is particularly sensitive to tokens receiving low attention during the focus phase. Motivated by this observation, we propose a lightweight inference-time intervention that selectively suppresses such tokens during the focus phase. The method operates in a training-free manner using statistics from a single forward pass and employs a Determinantal Point Process (DPP) to preserve diverse visual cues while filtering redundant tokens. Extensive experiments across multiple LVLM backbones and decoding strategies demonstrate that the proposed approach consistently reduces hallucination metrics while maintaining competitive caption quality. Moreover, compared to adversarial uncertainty estimation methods, our approach achieves comparable hallucination mitigation with negligible additional inference latency.

AI is supporting, accelerating, and automating scientific discovery across a diverse set of fields. However, AI adoption in historical research remains limited due to the lack of solutions designed for historians. In this technical progress report, we introduce the first module of Chronos, an AI Historian under development. This module enables historians to convert image scans of primary sources into data through natural-language interactions. Rather than imposing a fixed extraction pipeline powered by a vision-language model (VLM), it allows historians to adapt workflows for heterogeneous source corpora, evaluate the performance of AI models on specific tasks, and iteratively refine workflows through natural-language interaction with the Chronos agent. The module is open-source and ready to be used by historical researchers on their own sources.

Bimanual robot learning from demonstrations is fundamentally limited by the cost and narrow visual diversity of real-world data, which constrains policy robustness across viewpoints, object configurations, and embodiments. We present Canny-guided Robot Data Generation using Video Diffusion Transformers (CRAFT), a video diffusion-based framework for scalable bimanual demonstration generation that synthesizes temporally coherent manipulation videos while producing action labels. By conditioning video diffusion on edge-based structural cues extracted from simulator-generated trajectories, CRAFT produces physically plausible trajectory variations and supports a unified augmentation pipeline spanning object pose changes, camera viewpoints, lighting and background variations, cross-embodiment transfer, and multi-view synthesis. We leverage a pre-trained video diffusion model to convert simulated videos, along with action labels from the simulation trajectories, into action-consistent demonstrations. Starting from only a few real-world demonstrations, CRAFT generates a large, visually diverse set of photorealistic training data, bypassing the need to replay demonstrations on the real robot (Sim2Real). Across simulated and real-world bimanual tasks, CRAFT improves success rates over existing augmentation strategies and straightforward data scaling, demonstrating that diffusion-based video generation can substantially expand demonstration diversity and improve generalization for dual-arm manipulation tasks. Our project website is available at: https://craftaug.github.io/

Software Engineering 3.0 marks a paradigm shift in software development, in which AI coding agents are no longer just assistive tools but active contributors. While prior empirical studies have examined productivity gains and acceptance patterns in AI-assisted development, the challenges associated with integrating agent-generated contributions remain less understood. In particular, merge conflicts, a fundamental aspect of collaborative software development, remain underexplored in this context. In this paper, we present AgenticFlict, a large-scale dataset of textual merge conflicts in AI coding agent pull requests (Agentic PRs). The dataset comprises 142K+ Agentic PRs collected from 59K+ repositories, of which 107K+ are successfully processed through deterministic merge simulation. Our pipeline identifies 29K+ PRs exhibiting merge conflicts, yielding a conflict rate of 27.67%, and extracts 336K+ fine-grained conflict regions across these instances. Our preliminary exploratory analysis indicates that merge conflicts are both frequent and often substantial in AI-generated contributions, with noticeable variation across agents, emphasizing the need to better understand and manage integration challenges in AI-assisted software development. The dataset, code and supplementary materials are available in zenodo: https://doi.org/10.5281/zenodo.19396917.

This study surveys the historical development of regularization, tracing its evolution from stepwise regression in the 1960s to recent advancements in formal error control, structured penalties for non-independent features, Bayesian methods, and l0-based regularization (among other techniques). We empirically evaluate the performance of four canonical frameworks -- Ridge, Lasso, ElasticNet, and Post-Lasso OLS -- across 134,400 simulations spanning a 7-dimensional manifold grounded in eight production-grade machine learning models. Our findings demonstrate that for prediction accuracy when the sample-to-feature ratio is sufficient (n/p >= 78), Ridge, Lasso, and ElasticNet are nearly interchangeable. However, we find that Lasso recall is highly fragile under multicollinearity; at high condition numbers (kappa) and low SNR, Lasso recall collapses to 0.18 while ElasticNet maintains 0.93. Consequently, we advise practitioners against using Lasso or Post-Lasso OLS at high kappa with small sample sizes. The analysis concludes with an objective-driven decision guide to assist machine learning engineers in selecting the optimal scikit-learn-supported framework based on observable feature space attributes.

Discrete diffusion language models have emerged as a competitive alternative to auto-regressive language models, but training them efficiently under limited parameter and memory budgets remains challenging. Modern architectures are predominantly based on a full-vocabulary token prediction layer, which accounts for a substantial fraction of model parameters (e.g., more than 20% in small scale DiT-style designs) and often dominates peak GPU memory usage. This leads to inefficient use of both parameters and memory under constrained training resources. To address this issue, we revisit the necessity of explicit full-vocabulary prediction, and instead exploit the inherent structure among tokens to build a tree-structured diffusion language model. Specifically, we model the diffusion process with intermediate latent states corresponding to a token's ancestor nodes in a pre-constructed vocabulary tree. This tree-structured factorization exponentially reduces the classification dimensionality, makes the prediction head negligible in size, and enables reallocation of parameters to deepen the attention blocks. Empirically, under the same parameter budget, our method reduces peak GPU memory usage by half while matching the perplexity performance of state-of-the-art discrete diffusion language models.

We present an automated crosswalk framework that compares an AI safety policy document pair under a shared taxonomy of activities. Using the activity categories defined in Activity Map on AI Safety as fixed aspects, the system extracts and maps relevant activities, then produces for each aspect a short summary for each document, a brief comparison, and a similarity score. We assess the stability and validity of LLM-based crosswalk analysis across public policy documents. Using five large language models, we perform crosswalks on ten publicly available documents and visualize mean similarity scores with a heatmap. The results show that model choice substantially affects the crosswalk outcomes, and that some document pairs yield high disagreements across models. A human evaluation by three experts on two document pairs shows high inter-annotator agreement, while model scores still differ from human judgments. These findings support comparative inspection of policy documents.

Large language models (LLMs) show strong multilingual capabilities, yet reliably controlling the language of their outputs remains difficult. Representation-level steering addresses this by adding language-specific vectors to model activations at inference time, but identifying language-specific directions in the residual stream often relies on multilingual or parallel data that can be expensive to obtain. Sparse autoencoders (SAEs) decompose residual activations into interpretable, sparse feature directions and offer a natural basis for this search, yet existing SAE-based approaches face the same data constraint. We introduce LangFIR (Language Feature Identification via Random-token Filtering), a method that discovers language-specific SAE features using only a small amount of monolingual data and random-token sequences. Many SAE features consistently activated by target-language inputs do not encode language identity. Random-token sequences surface these language-agnostic features, allowing LangFIR to filter them out and isolate a sparse set of language-specific features. We show that these features are extremely sparse, highly selective for their target language, and causally important: directional ablation increases cross-entropy loss only for the corresponding language. Using these features to construct steering vectors for multilingual generation control, LangFIR achieves the best average accuracy BLEU across three models (Gemma 3 1B, Gemma 3 4B, and Llama 3.1 8B), three datasets, and twelve target languages, outperforming the strongest monolingual baseline by up to and surpassing methods that rely on parallel data. Our results suggest that language identity in multilingual LLMs is localized in a sparse set of feature directions discoverable with monolingual data. Code is available at https://anonymous.4open.science/r/LangFIR-C0F5/.

Generative AI is quickly becoming an integral part of people's everyday workflows. Early evidence has shown that while generative AI can increase individual-level productivity, it does so at the cost of collective diversity, potentially narrowing the set of ideas and perspectives produced. Our research stands in contrast to this concern: through a pre-registered randomized control trial, we show that incentives mediate AI's homogenizing force in a creative writing task where participants can use AI interactively. Participants rewarded for originality relative to peers produce collectively more diverse writing than those rewarded for quality alone. This divergence is driven not by abandoning AI, but by how participants use it: those incentivized for originality incorporate fewer AI suggestions verbatim, relying on the model more selectively for brainstorming, proofreading, and targeted edits. Our results reveal that the effects of generative AI depend not only on the technology itself, but also the behavioral strategies and incentive structures surrounding its use.

Modern agentic workflows decompose complex tasks into specialized subtasks and route them to diverse models to minimize cost without sacrificing quality. However, current routing architectures focus exclusively on performance optimization, leaving underlying trade-offs between model capability and cost unrecorded. Without clear rationale, developers cannot distinguish between intelligent efficiency -- using specialized models for appropriate tasks -- and latent failures caused by budget-driven model selection. We present Topaz, a framework that introduces formal auditability to agentic routing. Topaz replaces silent model assignments with an inherently interpretable router that incorporates three components: (i) skill-based profiling that synthesizes performance across diverse benchmarks into granular capability profiles (ii) fully traceable routing algorithms that utilize budget-based and multi-objective optimization to produce clear traces of how skill-match scores were weighed against costs, and (iii) developer-facing explanations that translate these traces into natural language, allowing users to audit system logic and iteratively tune the cost-quality tradeoff. By making routing decisions interpretable, Topaz enables users to understand, trust, and meaningfully steer routed agentic systems.

Existing \textbf{s}alient \textbf{o}bject \textbf{d}etection (SOD) methods adopt a \textbf{passive} visual stimulus-based rationale--objects with the strongest visual stimuli are perceived as the user's primary focus (i.e., salient objects). They ignore the decisive role of users' \textbf{proactive needs} in segmenting salient objects--if a user has a need before seeing an image, the user's salient objects align with their needs, e.g., if a user's need is ``white apple'', when this user sees an image, the user's primary focus is on the ``white apple'' or ``the most white apple-like'' objects in the image. Such an oversight not only \textbf{fails to satisfy users}, but also \textbf{limits the development of downstream tasks}. For instance, in salient object ranking tasks, focusing solely on visual stimuli-based salient objects is insufficient for conducting an analysis of fine-grained relationships between users' viewing order (usually determined by user's needs) and scenes, which may result in wrong ranking results. Clearly, it is essential to detect salient objects based on user needs. Thus, we advocate a \textbf{User} \textbf{S}alient \textbf{O}bject \textbf{D}etection (UserSOD) task, which focuses on \textbf{detecting salient objects align with users' proactive needs when user have needs}. The main challenge for this new task is the lack of datasets for model training and testing.

We study adversarial online learning of real-valued functions on $\mathbb{R}$. In each round the learner is queried at $x_t\in\mathbb{R}$, predicts $\hat y_t$, and then observes the true value $f(x_t)$; performance is measured by cumulative $p$-loss $\sum_{t\ge 1}|\hat y_t-f(x_t)|^p$. For the class \[ \mathcal{G}_q=\Bigl\{f:\mathbb{R}\to\mathbb{R}\ \text{absolutely continuous}:\ \int_{\mathbb{R}}|f'(x)|^q\,dx\le 1\Bigr\}, \] we show that the standard model becomes ill-posed on $\mathbb{R}$: for every $p\ge 1$ and $q>1$, an adversary can force infinite loss. Motivated by this obstruction, we analyze three modified learning scenarios that limit the influence of queries that are far from previously observed inputs. In Scenario 1 the adversary must choose each new query within distance $1$ of some past query. In Scenario 2 the adversary may query anywhere, but the learner is penalized only on rounds whose query lies within distance $1$ of a past query. In Scenario 3 the loss in round $t$ is multiplied by a weight $g(\min_{j<t}|x_t-x_j|)$. We obtain sharp characterizations for Scenarios 1-2 in several regimes. For Scenario 3 we identify a clean threshold phenomenon: if $g$ decays too slowly, then the adversary can force infinite weighted loss. In contrast, for rapidly decaying weights such as $g(z)=e^{-cz}$ we obtain finite and sharp guarantees in the quadratic case $p=q=2$. Finally, we study a natural multivariable slice generalization $\mathcal{G}_{q,d}$ of $\mathcal{G}_q$ on $\mathbb{R}^d$ and show a sharp dichotomy: while the one-dimensional case admits finite opt-values in certain regimes, for every $d\ge 2$ the slice class $\mathcal{G}_{q,d}$ is too permissive, and even under Scenarios 1-3 an adversary can force infinite loss.

Current AI safety relies on behavioral monitoring and post-training alignment, yet empirical measurement shows these approaches produce no detectable pre-commitment signal in a majority of instruction-tuned models tested. We present an energy-based governance framework connecting transformer inference dynamics to constraint-satisfaction models of neural computation, and apply it to a seven-model cohort across five geometric regimes. Using trajectory tension (rho = ||a|| / ||v||), we identify a 57-token pre-commitment window in Phi-3-mini-4k-instruct under greedy decoding on arithmetic constraint probes. This result is model-specific, task-specific, and configuration-specific, demonstrating that pre-commitment signals can exist but are not universal. We introduce a five-regime taxonomy of inference behavior: Authority Band, Late Signal, Inverted, Flat, and Scaffold-Selective. Energy asymmetry (Σ\r{ho}_misaligned / Σ\r{ho}_aligned) serves as a unifying metric of structural rigidity across these regimes. Across seven models, only one configuration exhibits a predictive signal prior to commitment; all others show silent failure, late detection, inverted dynamics, or flat geometry. We further demonstrate that factual hallucination produces no predictive signal across 72 test conditions, consistent with spurious attractor settling in the absence of a trained world-model constraint. These results establish that rule violation and hallucination are distinct failure modes with different detection requirements. Internal geometry monitoring is effective only where resistance exists; detection of factual confabulation requires external verification mechanisms. This work provides a measurable framework for inference-layer governability and introduces a taxonomy for evaluating deployment risk in autonomous AI systems.

Robot reinforcement learning from demonstrations (RLfD) assumes that expert data is abundant; this is usually unrealistic in the real world given data scarcity as well as high collection cost. Furthermore, imitation learning algorithms assume that the data is independently and identically distributed, which ultimately results in poorer performance as gradual errors emerge and compound within test-time trajectories. We address these issues by introducing the "master your own expertise" (MYOE) framework, a self-imitation framework that enables robotic agents to learn complex behaviors from limited demonstration data samples. Inspired by human perception and action, we propose and design what we call the queryable mixture-of-preferences state space model (QMoP-SSM), which estimates the desired goal at every time step. These desired goals are used in computing the "preference regret", which is used to optimize the robot control policy. Our experiments demonstrate the robustness, adaptability, and out-of-sample performance of our agent compared to other state-of-the-art RLfD schemes. The GitHub repository that supports this work can be found at: https://github.com/rxng8/neurorobot-preference-regret-learning.

LLM-based coding agents can localize bugs, generate patches, and run tests with diminishing human oversight, yet the scaffolding code that surrounds the language model (the control loop, tool definitions, state management, and context strategy) remains poorly understood. Existing surveys classify agents by abstract capabilities (tool use, planning, reflection) that cannot distinguish between architecturally distinct systems, and trajectory studies observe what agents do without examining the scaffold code that determines why. This paper presents a source-code-level architectural taxonomy derived from analysis of 13 open-source coding agent scaffolds at pinned commit hashes. Each agent is characterized across 12 dimensions organized into three layers: control architecture, tool and environment interface, and resource management. The analysis reveals that scaffold architectures resist discrete classification: control strategies range from fixed pipelines to Monte Carlo Tree Search, tool counts range from 0 to 37, and context compaction spans seven distinct strategies. Five loop primitives (ReAct, generate-test-repair, plan-execute, multi-attempt retry, tree search) function as composable building blocks that agents layer in different combinations; 11 of 13 agents compose multiple primitives rather than relying on a single control structure. Dimensions converge where external constraints dominate (tool capability categories, edit formats, execution isolation) and diverge where open design questions remain (context compaction, state management, multi-model routing). All taxonomic claims are grounded in file paths and line numbers, providing a reusable reference for researchers studying agent behavior and practitioners designing new scaffolds.

Outage management in large-scale cloud operations remains heavily manual, requiring rapid triage, cross-team coordination, and experience-driven decisions under partial observability. We present \textbf{ActionNex}, a production-grade agentic system that supports end-to-end outage assistance, including real-time updates, knowledge distillation, and role- and stage-conditioned next-best action recommendations. ActionNex ingests multimodal operational signals (e.g., outage content, telemetry, and human communications) and compresses them into critical events that represent meaningful state transitions. It couples this perception layer with a hierarchical memory subsystem: long-term Key-Condition-Action (KCA) knowledge distilled from playbooks and historical executions, episodic memory of prior outages, and working memory of the live context. A reasoning agent aligns current critical events to preconditions, retrieves relevant memories, and generates actionable recommendations; executed human actions serve as an implicit feedback signal to enable continual self-evolution in a human-agent hybrid system. We evaluate ActionNex on eight real Azure outages (8M tokens, 4,000 critical events) using two complementary ground-truth action sets, achieving 71.4\% precision and 52.8-54.8\% recall. The system has been piloted in production and has received positive early feedback.

Despite the large corpus of biology training text, the impact of reasoning models on biological research generally lags behind math and coding. In this work, we show that biology questions from current large-scale reasoning datasets do not align well with modern research topic distributions in biology, and that this topic imbalance may negatively affect performance. In addition, we find that methods for extracting challenging and verifiable research problems from biology research text are a critical yet underdeveloped ingredient in applying reinforcement learning for better performance on biology research tasks. We introduce BioAlchemy, a pipeline for sourcing a diverse set of verifiable question-and-answer pairs from a scientific corpus of biology research text. We curate BioAlchemy-345K, a training dataset containing over 345K scientific reasoning problems in biology. Then, we demonstrate how aligning our dataset to the topic distribution of modern scientific biology can be used with reinforcement learning to improve reasoning performance. Finally, we present BioAlchemist-8B, which improves over its base reasoning model by 9.12% on biology benchmarks. These results demonstrate the efficacy of our approach for developing stronger scientific reasoning capabilities in biology. The BioAlchemist-8B model is available at: https://huggingface.co/BioAlchemy.

Quasi-experimental evaluations are central for generating real-world causal evidence and complementing insights from randomized trials. The regression discontinuity design (RDD) is a quasi-experimental design that can be used to estimate the causal effect of treatments that are assigned based on a running variable crossing a threshold. Such threshold-based rules are ubiquitous in healthcare, where predictive and prognostic biomarkers frequently guide treatment decisions. However, standard RD estimators rely on complete outcome data, an assumption often violated in time-to-event analyses where censoring arises from loss to follow-up. To address this issue, we propose a nonparametric approach that leverages doubly robust censoring corrections and can be paired with existing RD estimators. Our approach can handle multiple survival endpoints, long follow-up times, and covariate-dependent variation in survival and censoring. We discuss the relevance of our approach across multiple areas of applications and demonstrate its usefulness through simulations and the prostate component of the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial where our new approach offers several advantages, including higher efficiency and robustness to misspecification. We have also developed an open-source software package, $\texttt{rdsurvival}$, for the $\texttt{R}$ language.

Experimental evidence confirms that AI tools raise worker productivity, but also that sustained use can erode the expertise on which those gains depend. We develop a dynamic model in which a decision-maker chooses AI usage intensity for a worker over time, trading immediate productivity against the erosion of worker skill. We decompose the tool's productivity effect into two channels, one independent of worker expertise and one that scales with it. The model produces three main results. First, even a decision-maker who fully anticipates skill erosion rationally adopts AI when front-loaded productivity gains outweigh long-run skill costs, producing steady-state loss: the worker ends up less productive than before adoption. Second, when managers are short-termist or worker skill has external value, the decision-maker's optimal policy turns steady-state loss into the augmentation trap, leaving the worker worse off than if AI had never been adopted. Third, when AI productivity depends less on worker expertise, workers can permanently diverge in skill: experienced workers realize their full potential while less experienced workers deskill to zero. Small differences in managerial incentives can determine which path a worker takes. The productivity decomposition classifies deployments into five regimes that separate beneficial adoption from harmful adoption and identifies which deployments are vulnerable to the trap.

Timely discharge prediction is essential for optimizing bed turnover and resource allocation in elective spine surgery units. This study evaluates the feasibility of lightweight, fine-tuned large language models (LLMs) and traditional text-based models for predicting next-day discharge using postoperative clinical notes. We compared 13 models, including TF-IDF with XGBoost and LGBM, and compact LLMs (DistilGPT-2, Bio_ClinicalBERT) fine-tuned via LoRA. TF-IDF with LGBM achieved the best balance, with an F1-score of 0.47 for the discharge class, a recall of 0.51, and the highest AUC-ROC (0.80). While LoRA improved recall in DistilGPT2, overall transformer-based and generative models underperformed. These findings suggest interpretable, resource-efficient models may outperform compact LLMs in real-world, imbalanced clinical prediction tasks.

Deploying reinforcement learning policies trained in simulation to real autonomous vehicles remains a fundamental challenge, particularly for VLM-guided RL frameworks whose policies are typically learned with simulator-native observations and simulator-coupled action semantics that are unavailable on physical platforms. This paper presents Sim2Real-AD, a modular framework for zero-shot sim-to-real transfer of CARLA-trained VLM-guided RL policies to full-scale vehicles without any real-world RL training data. The framework decomposes the transfer problem into four components: a Geometric Observation Bridge (GOB) that converts monocular front-view images into simulator-compatible bird's-eye-view (BEV) observations, a Physics-Aware Action Mapping (PAM) that translates policy outputs into platform-agnostic physical commands, a Two-Phase Progressive Training (TPT) strategy that stabilizes adaptation by separating action-space and observation-space transfer, and a Real-time Deployment Pipeline (RDP) that integrates perception, policy inference, control conversion, and safety monitoring for closed-loop execution. Simulation experiments show that the framework preserves the relative performance ordering of representative RL algorithms across different reward paradigms and validate the contribution of each module. Zero-shot deployment on a full-scale Ford E-Transit achieves success rates of 90%, 80%, and 75% in car-following, obstacle avoidance, and stop-sign interaction scenarios, respectively. To the best of our knowledge, this study is among the first to demonstrate zero-shot closed-loop deployment of a CARLA-trained VLM-guided RL policy on a full-scale real vehicle without any real-world RL training data. The demo video and code are available at: https://zilin-huang.github.io/Sim2Real-AD-website/.

Knowledge graph construction typically relies either on predefined ontologies or on schema-free extraction. Ontology-driven pipelines enforce consistent typing but require costly schema design and maintenance, whereas schema-free methods often produce fragmented graphs with weak global organization, especially in long technical documents with dense, context-dependent information. We propose TRACE-KG (Text-dRiven schemA for Context-Enriched Knowledge Graphs), a multimodal framework that jointly constructs a context-enriched knowledge graph and an induced schema without assuming a predefined ontology. TRACE-KG captures conditional relations through structured qualifiers and organizes entities and relations using a data-driven schema that serves as a reusable semantic scaffold while preserving full traceability to the source evidence. Experiments show that TRACE-KG produces structurally coherent, traceable knowledge graphs and offers a practical alternative to both ontology-driven and schema-free construction pipelines.

Cultural representation in Large Language Model (LLM) outputs has primarily been evaluated through the proxies of cultural diversity and factual accuracy. However, a crucial gap remains in assessing cultural alignment: the degree to which generated content mirrors how native populations perceive and prioritize their own cultural facets. In this paper, we introduce a human-centered framework to evaluate the alignment of LLM generations with local expectations. First, we establish a human-derived ground-truth baseline of importance vectors, called Cultural Importance Vectors based on an induced set of culturally significant facets from open-ended survey responses collected across nine countries. Next, we introduce a method to compute model-derived Cultural Representation Vectors of an LLM based on a syntactically diversified prompt-set and apply it to three frontier LLMs (Gemini 2.5 Pro, GPT-4o, and Claude 3.5 Haiku). Our investigation of the alignment between the human-derived Cultural Importance and model-derived Cultural Representations reveals a Western-centric calibration for some of the models where alignment decreases as a country's cultural distance from the US increases. Furthermore, we identify highly correlated, systemic error signatures ($ρ> 0.97$) across all models, which over-index on some cultural markers while neglecting the deep-seated social and value-based priorities of users. Our approach moves beyond simple diversity metrics toward evaluating the fidelity of AI-generated content in authentically capturing the nuanced hierarchies of global cultures.

Enforcing complex (e.g., nonconvex) operational constraints is a critical challenge in real-world learning and control systems. However, existing methods struggle to efficiently enforce general classes of constraints. To address this, we propose a novel data-driven amortized approach that uses a trained autoencoder as an approximate projector to provide fast corrections to infeasible predictions. Specifically, we train an autoencoder using an adversarial objective to learn a structured, convex latent representation of the feasible set. This enables rapid correction of neural network outputs by projecting their associated latent representations onto a simple convex shape before decoding into the original feasible set. We test our approach on a diverse suite of constrained optimization and reinforcement learning problems with challenging nonconvex constraints. Results show that our method effectively enforces constraints at a low computational cost, offering a practical alternative to expensive feasibility correction techniques based on traditional solvers.

We present VisionClaw, an always-on wearable AI agent that integrates live egocentric perception with agentic task execution. Running on Meta Ray-Ban smart glasses, VisionClaw continuously perceives real-world context and enables in-situ, speech-driven action initiation and delegation via OpenClaw AI agents. Therefore, users can directly execute tasks through the smart glasses, such as adding real-world objects to an Amazon cart, generating notes from physical documents, receiving meeting briefings on the go, creating events from posters, or controlling IoT devices. We evaluate VisionClaw through a controlled laboratory study (N=12) and a longitudinal deployment study (N=5). Results show that integrating perception and execution enables faster task completion and reduces interaction overhead compared to non-always-on and non-agent baselines. Beyond performance gains, deployment findings reveal a shift in interaction: tasks are initiated opportunistically during ongoing activities, and execution is increasingly delegated rather than manually controlled. These results suggest a new paradigm for wearable AI agents, where perception and action are continuously coupled to support situated, hands-free interaction.

Creative thinking is a fundamental aspect of human cognition, and divergent thinking-the capacity to generate novel and varied ideas-is widely regarded as its core generative engine. Large language models (LLMs) have recently demonstrated impressive performance on divergent thinking tests and prior work has shown that models with higher task performance tend to be more aligned to human brain activity. However, existing brain-LLM alignment studies have focused on passive, non-creative tasks. Here, we explore brain alignment during creative thinking using fMRI data from 170 participants performing the Alternate Uses Task (AUT). We extract representations from LLMs varying in size (270M-72B) and measure alignment to brain responses via Representational Similarity Analysis (RSA), targeting the creativity-related default mode and frontoparietal networks. We find that brain-LLM alignment scales with model size (default mode network only) and idea originality (both networks), with effects strongest early in the creative process. We further show that post-training objectives shape alignment in functionally selective ways: a creativity-optimized \texttt{Llama-3.1-8B-Instruct} preserves alignment with high-creativity neural responses while reducing alignment with low-creativity ones; a human behavior fine-tuned model elevates alignment with both; and a reasoning-trained variant shows the opposite pattern, suggesting chain-of-thought training steers representations away from creative neural geometry toward analytical processing. These results demonstrate that post-training objectives selectively reshape LLM representations relative to the neural geometry of human creative thought.

Context-dependent sequential decision making is commonly addressed either by providing context explicitly as an input or by increasing recurrent memory so that contextual information can be represented internally. We study a third alternative: realizing contextual dependence by intervening on a shared recurrent latent state, without enlarging recurrent dimensionality. To this end, we introduce an intervention-based recurrent architecture in which a recurrent core first constructs a shared pre-intervention latent state, and context then acts through an additive, context-indexed operator. We evaluate this idea on a context-switching sequential decision task under partial observability. We compare three model families: a label-assisted baseline with direct context access, a memory baseline with enlarged recurrent state, and the proposed intervention model, which uses no direct context input to the recurrent core and no memory growth. On the main benchmark, the intervention model performs strongly without additional recurrent dimensions. We also evaluate the models using the conditional mutual information (I(C;O | S)) as a theorem-motivated operational probe of contextual dependence at fixed latent state. For task-relevant phase-1 outcomes, the intervention model exhibits positive conditional contextual information. Together, these results suggest that intervention on a shared recurrent state provides a viable alternative to recurrent memory growth for contextual control in this setting.

In high-stakes settings where machine learning models are used to automate decision-making about individuals, the presence of algorithmic bias can exacerbate systemic harm to certain subgroups of people. These biases often stem from the underlying training data. In practice, interventions to "fix the data" depend on the actual additional data sources available -- where many are less than ideal. In these cases, the effects of data scaling on subgroup performance become volatile, as the improvements from increased sample size are counteracted by the introduction of distribution shifts in the training set. In this paper, we investigate the limitations of combining data sources to improve subgroup performance within the context of healthcare. Clinical models are commonly trained on datasets comprised of patient electronic health record (EHR) data from different hospitals or admission departments. Across two such datasets, the eICU Collaborative Research Database and the MIMIC-IV dataset, we find that data addition can both help and hurt model fairness and performance, and many intuitive strategies for data selection are unreliable. We compare model-based post-hoc calibration and data-centric addition strategies to find that the combination of both is important to improve subgroup performance. Our work questions the traditional dogma of "better data" for overcoming fairness challenges by comparing and combining data- and model-based approaches.

Optical Chemical Structure Recognition (OCSR) is critical for converting 2D molecular diagrams from printed literature into machine-readable formats. While Vision-Language Models have shown promise in end-to-end OCR tasks, their direct application to OCSR remains challenging, and direct full-parameter supervised fine-tuning often fails. In this work, we adapt DeepSeek-OCR-2 for molecular optical recognition by formulating the task as image-conditioned SMILES generation. To overcome training instabilities, we propose a two-stage progressive supervised fine-tuning strategy: starting with parameter-efficient LoRA and transitioning to selective full-parameter fine-tuning with split learning rates. We train our model on a large-scale corpus combining synthetic renderings from PubChem and realistic patent images from USPTO-MOL to improve coverage and robustness. Our fine-tuned model, MolSeek-OCR, demonstrates competitive capabilities, achieving exact matching accuracies comparable to the best-performing image-to-sequence model. However, it remains inferior to state-of-the-art image-to-graph modelS. Furthermore, we explore reinforcement-style post-training and data-curation-based refinement, finding that they fail to improve the strict sequence-level fidelity required for exact SMILES matching.

Uncertainty quantification (UQ) methods for large language models are predominantly designed by hand based on domain knowledge and heuristics, limiting their scalability and generality. We apply LLM-powered evolutionary search to automatically discover unsupervised UQ methods represented as Python programs. On the task of atomic claim verification, our evolved methods outperform strong manually-designed baselines, achieving up to 6.7% relative ROC-AUC improvement across 9 datasets while generalizing robustly out-of-distribution. Qualitative analysis reveals that different LLMs employ qualitatively distinct evolutionary strategies: Claude models consistently design high-feature-count linear estimators, while Gpt-oss-120B gravitates toward simpler and more interpretable positional weighting schemes. Surprisingly, only Sonnet 4.5 and Opus 4.5 reliably leverage increased method complexity to improve performance -- Opus 4.6 shows an unexpected regression relative to its predecessor. Overall, our results indicate that LLM-powered evolutionary search is a promising paradigm for automated, interpretable hallucination detector design.

Co-evolutionary self-play, where one language model generates problems and another solves them, promises autonomous curriculum learning without human supervision. In practice, the proposer quickly converges to a narrow distribution of problems that satisfy the reward function. This diversity collapse renders the curriculum uninformative for the solver, stalling the co-evolutionary loop. We introduce vocabulary dropout, a random mask applied to the proposer's output logits during both policy training and curriculum generation, as a lightweight mechanism to sustain diversity. The mask is hard and non-stationary, preventing the proposer from locking into fixed token sequences. Training Qwen3-4B and Qwen3-8B on mathematical reasoning via R-Zero, we find that vocabulary dropout sustains proposer diversity across lexical, semantic, and functional metrics throughout training, and yields solver improvements averaging +4.4 points at 8B, with the largest gains on competition-level benchmarks. Our findings suggest that explicit action-space constraints, analogous to the structural role that game rules play in classical self-play, can help sustain productive co-evolution in language. Vocabulary dropout is one simple instantiation of this principle.

Reservoir computing promises a fast method for handling large amounts of temporal data. This hinges on constructing a good reservoir--a dynamical system capable of transforming inputs into a high-dimensional representation while remembering properties of earlier data. In this work, we introduce a reservoir based on recurrent quantum feature maps where a fixed quantum circuit is reused to encode both current inputs and a classical feedback signal derived from previous outputs. We evaluate the model on the Mackey-Glass time-series prediction task using our recently introduced CP feature map, and find that it achieves lower mean squared error than standard classical baselines, including echo state networks and multilayer perceptrons, while maintaining compact circuit depth and qubit requirements. We further analyze memory capacity and show that the model effectively retains temporal information, consistent with its forecasting accuracy. Finally, we study the impact of realistic noise and find that performance is robust to several noise channels but remains sensitive to two-qubit gate errors, identifying a key limitation for near-term implementations.

As web agents rapidly evolve, an increasing body of work has moved beyond conventional atomic browser interactions and explored tool use as a higher-level action paradigm. Although prior studies have shown the promise of tools, their conclusions are often drawn from limited experimental scales and sometimes non-comparable settings. As a result, several fundamental questions remain unclear: i) whether tools provide consistent gains for web agents, ii) what practical design principles characterize effective tools, and iii) what side effects tool use may introduce. To establish a stronger empirical foundation for future research, we revisit tool use in web agents through an extensive and carefully controlled study across diverse tool sources, backbone models, tool-use frameworks, and evaluation benchmarks. Our findings both revise some prior conclusions and complement others with broader evidence. We hope this study provides a more reliable empirical basis and inspires future research on tool-use web agents.

In highly interactive driving scenes, trajectory prediction is conditioned on information from surrounding traffic participants such as cars and pedestrians. Our main contribution is a comprehensive analysis of state-of-the-art trajectory predictors, which reveals a surprising and critical flaw: many surrounding agents degrade prediction accuracy rather than improve it. Using Shapley-based attribution, we rigorously demonstrate that models learn unstable and non-causal decision-making schemes that vary significantly across training runs. Building on these insights, we propose to integrate a Conditional Information Bottleneck (CIB), which does not require additional supervision and is trained to effectively compress agent features as well as ignore those that are not beneficial for the prediction task. Comprehensive experiments using multiple datasets and model architectures demonstrate that this simple yet effective approach not only improves overall trajectory prediction performance in many cases but also increases robustness to different perturbations. Our results highlight the importance of selectively integrating contextual information, which can often contain spurious or misleading signals, in trajectory prediction. Moreover, we provide interpretable metrics for identifying non-robust behavior and present a promising avenue towards a solution.

Regional climate information at kilometer scales is essential for assessing the impacts of climate change, but generating it with global climate models is too expensive due to their high computational costs. Machine learning models offer a fast alternative, yet they often violate basic physical laws and degrade when applied to climates outside of their training distribution. We present Physics-Constrained Adaptive Flow Matching (PC-AFM), a generative downscaling model that addresses both problems. Building on the Adaptive Flow Matching (AFM) model of Fotiadis et al. (2025) as our baseline, we add soft conservation constraints that keep the downscaled output consistent with the large-scale input for precipitation and humidity, and use gradient surgery via the ConFIG algorithm to prevent these constraints from interfering with the generative objective. We train the model on Central Europe climate data, evaluate it on a 10-time downscaling task (63km to 6.3km) over six variables (near-surface temperature, precipitation, specific humidity, surface pressure, and horizontal wind components) across a comprehensive set of metrics including bias, ensemble skill scores, power spectra, and conservation error, and test the generalization on two held-out climate regions. Within the training distribution, PC-AFM reduces conservation errors and improves ensemble calibration while matching the baseline on standard skill metrics. Outside the training distribution, where unconstrained models develop large systematic errors by extrapolating learned statistics, PC-AFM halves precipitation wet bias, reduces conservation error and improves extreme-quantile accuracy, all without any information about the target climate at inference time. These results indicate that physical consistency is a practical requirement for deploying generative downscaling models in real-world applications.

Conventional urban indicators derived from censuses, surveys, and administrative records are often costly, spatially inconsistent, and slow to update. Recent geospatial foundation models enable Earth embeddings, compact satellite image representations transferable across downstream tasks, but their utility for neighborhood-scale urban monitoring remains unclear. Here, we benchmark three Earth embedding families, AlphaEarth, Prithvi, and Clay, for urban signal prediction across six U.S. metropolitan areas from 2020 to 2023. Using a unified supervised-learning framework, we predict 14 neighborhood-level indicators spanning crime, income, health, and travel behavior, and evaluate performance under four settings: global, city-wise, year-wise, and city-year. Results show that Earth embeddings capture substantial urban variation, with the highest predictive skill for outcomes more directly tied to built-environment structure, including chronic health burdens and dominant commuting modes. By contrast, indicators shaped more strongly by fine-scale behavior and local policy, such as cycling, remain difficult to infer. Predictive performance varies markedly across cities but remains comparatively stable across years, indicating strong spatial heterogeneity alongside temporal robustness. Exploratory analysis suggests that cross-city variation in predictive performance is associated with urban form in task-specific ways. Controlled dimensionality experiments show that representation efficiency is critical: compact 64-dimensional AlphaEarth embeddings remain more informative than 64-dimensional reductions of Prithvi and Clay. This study establishes a benchmark for evaluating Earth embeddings in urban remote sensing and demonstrates their potential as scalable, low-cost features for SDG-aligned neighborhood-scale urban monitoring.

Retrieval-Augmented Generation pipelines span a wide range of retrieval strategies that differ substantially in token cost and capability. Selecting the right strategy per query is a practical efficiency problem, yet no routing classifiers have been trained on RAGRouter-Bench \citep{wang2026ragrouterbench}, a recently released benchmark of $7,727$ queries spanning four knowledge domains, each annotated with one of three canonical query types: factual, reasoning, and summarization. We present the first systematic evaluation of lightweight classifier-based routing on this benchmark. Five classical classifiers are evaluated under three feature regimes, namely, TF-IDF, MiniLM sentence embeddings \citep{reimers2019sbert}, and hand-crafted structural features, yielding 15 classifier feature combinations. Our best configuration, TF-IDF with an SVM, achieves a macro-averaged F1 of $\mathbf{0.928}$ and an accuracy of $\mathbf{93.2\%}$, while simulating $\mathbf{28.1\%}$ token savings relative to always using the most expensive paradigm. Lexical TF-IDF features outperform semantic sentence embeddings by $3.1$ macro-F1 points, suggesting that surface keyword patterns are strong predictors of query-type complexity. Domain-level analysis reveals that medical queries are hardest to route and legal queries most tractable. These results establish a reproducible query-side baseline and highlight the gap that corpus-aware routing must close.

Rare diseases often manifest with distinctive facial phenotypes in children, offering valuable diagnostic cues for clinicians and AI-assisted screening systems. However, progress in this field is severely limited by the scarcity of curated, ethically sourced facial data and the high similarity among phenotypes across different conditions. To address these challenges, we introduce RDFace, a curated benchmark dataset comprising 456 pediatric facial images spanning 103 rare genetic conditions (average 4.4 samples per condition). Each ethically verified image is paired with standardized metadata. RDFace enables the development and evaluation of data-efficient AI models for rare disease diagnosis under real-world low-data constraints. We benchmark multiple pretrained vision backbones using cross-validation and explore synthetic augmentation with DreamBooth and FastGAN. Generated images are filtered via facial landmark similarity to maintain phenotype fidelity and merged with real data, improving diagnostic accuracy by up to 13.7% in ultra-low-data regimes. To assess semantic validity, phenotype descriptions generated by a vision-language model from real and synthetic images achieve a report similarity score of 0.84. RDFace establishes a transparent, benchmark-ready dataset for equitable rare disease AI research and presents a scalable framework for evaluating both diagnostic performance and the integrity of synthetic medical imagery.

Neural operator methods have emerged as powerful tools for learning mappings between infinite-dimensional function spaces, yet their potential in optimal control remains largely unexplored. We focus on multi-task control problems, whose solution is a mapping from task description (e.g., cost or dynamics functions) to optimal control law (e.g., feedback policy). We approximate these solution operators using a permutation-invariant neural operator architecture. Across a range of parametric optimal control environments and a locomotion benchmark, a single operator trained via behavioral cloning accurately approximates the solution operator and generalizes to unseen tasks, out-of-distribution settings, and varying amounts of task observations. We further show that the branch-trunk structure of our neural operator architecture enables efficient and flexible adaptation to new tasks. We develop structured adaptation strategies ranging from lightweight updates to full-network fine-tuning, achieving strong performance across different data and compute settings. Finally, we introduce meta-trained operator variants that optimize the initialization for few-shot adaptation. These methods enable rapid task adaptation with limited data and consistently outperform a popular meta-learning baseline. Together, our results demonstrate that neural operators provide a unified and efficient framework for multi-task control and adaptation.

Facial expressions of characters are a vital component of visual storytelling. While current AI image editing models hold promise for assisting artists in the task of stylized expression editing, these models introduce global noise and pixel drift into the edited image, preventing the integration of these models into professional image editing software and workflows. To bridge this gap, we introduce ExpressEdit, a fully open-source Photoshop plugin that is free from common artifacts of proprietary image editing models and robustly synergizes with native Photoshop operations such as Liquify. ExpressEdit seamlessly edits an expression within 3 seconds on a single consumer-grade GPU, significantly faster than popular proprietary models. Moreover, to support the generation of diverse expressions according to different narrative needs, we compile a comprehensive expression database of 135 expression tags enriched with example stories and images designed for retrieval-augmented generation. We open source the code and dataset to facilitate future research and artistic exploration.

LLM-based software engineering assistants fail not only by producing incorrect outputs, but also by allocating trust to the wrong artifact when code, documentation, and tests disagree. Existing evaluations focus mainly on downstream outcomes and therefore cannot reveal whether a model recognized degraded evidence, identified the unreliable source, or calibrated its trust across artifacts. We present TRACE (Trust Reasoning over Artifacts for Calibrated Evaluation), a framework that elicits structured artifact-level trust traces over Javadoc, method signatures, implementations, and test prefixes under blind perturbations. Using 22,339 valid traces from seven models on 456 curated Java method bundles, we evaluate per-artifact quality assessment, inconsistency detection, affected artifact attribution, and source prioritization. Across all models, quality penalties are largely localized to the perturbed artifact and increase with severity, but sensitivity is asymmetric across artifact types: documentation bugs induce a substantially larger heavy-to-subtle gap than implementation faults (0.152-0.253 vs. 0.049-0.123). Models detect explicit documentation bugs well (67-94%) and Javadoc and implementation contradictions at 50-91%, yet show a systematic blind spot when only the implementation drifts while the documentation remains plausible, with detection dropping by 7-42 percentage points. Confidence is poorly calibrated for six of seven models. These findings suggest that current LLMs are better at auditing natural-language specifications than at detecting subtle code-level drift, motivating explicit artifact-level trust reasoning before correctness-critical downstream use.

Attention is a fundamental computational kernel that accounts for the majority of the workload in transformer and LLM computing. Optimizing dataflow is crucial for enhancing both performance and energy efficiency in attention computation. This optimization involves a range of decisions, such as tiling, computation ordering and buffer management, and can be applied at both intra-operator and inter-operator levels, resulting in a highly complex decision space. We propose a new approach to cross-operator dataflow optimization. Its centerpiece is an analytical performance model that spans a large decision space and enables matrix-based encoding of multiple candidate solutions. Built on this foundation, a vast number of solutions can be evaluated rapidly, and with the aid of an effective pruning technique, the optimal solution can be identified through exhaustive enumeration. We refer to our method as MMEE (Matrix Multiplication Encoded Enumeration). The ability to efficiently enumerate a large design space allows MMEE to deliver higher-quality solutions at a substantially faster speed compared to prior approaches. The MMEE approach is evaluated across various test cases for different accelerator configurations. For energy-driven optimization, MMEE reduces energy consumption by 48%-50% and latency by 31%-69%, compared to state-of-the-art methods. For latency-driven optimization, MMEE achieves simultaneous reductions of 40%-50% in energy consumption and 40%-69% in latency, respectively. Additionally, MMEE is $64\times$ to $343\times$ faster than previous works.

Hybrid quantum-classical applications pose significant resource management challenges due to heterogeneity and dynamism in both infrastructure and workloads. Quantum-HPC environments integrate quantum processing units (QPUs) with diverse classical resources (CPUs, GPUs), while applications span coupling patterns from tightly coupled execution to loosely coupled task parallelism with varying resource requirements. Traditional HPC schedulers lack visibility into application semantics and cannot respond to fluctuating resource availability at runtime. This paper presents a middleware-based approach for adaptive resource, workload, and task management in hybrid quantum-HPC systems. We make four contributions: (i) a conceptual four-layer middleware architecture that decomposes management across workflow, workload, task, and resource levels, enabling application-aware scheduling over heterogeneous quantum-HPC resources; (ii) a set of execution motifs capturing interaction and coupling characteristics of hybrid applications, realized as quantum mini-apps for systematic workload characterization; (iii) Pilot-Quantum, a middleware framework built on the pilot abstraction that enables late binding and dynamic resource allocation, adapting to resource and workload dynamics at runtime; and (iv) Q-Dreamer, a performance modeling toolkit providing reusable components for informed workload partitioning, including a circuit-cutting optimizer that analytically derives optimal partitioning strategies. Evaluation on heterogeneous HPC platforms (Perlmutter, NVIDIA DGX with H100/B200 GPUs) demonstrates efficient multi-backend orchestration across CPUs, GPUs, and QPUs for diverse execution motifs. Q-Dreamer predicts optimal circuit cutting configurations with up to 82% accuracy.

Recent work has demonstrated the potential of non-transformer language models, especially linear recurrent neural networks (RNNs) and hybrid models that mix recurrence and attention. Yet there is no consensus on whether the potential benefits of these new architectures justify the risk and effort of scaling them up. To address this, we provide evidence for the advantages of hybrid models over pure transformers on several fronts. First, theoretically, we show that hybrid models do not merely inherit the expressivity of transformers and linear RNNs, but can express tasks beyond both, such as code execution. Putting this theory to practice, we train Olmo Hybrid, a 7B-parameter model largely comparable to Olmo 3 7B but with the sliding window layers replaced by Gated DeltaNet layers. We show that Olmo Hybrid outperforms Olmo 3 across standard pretraining and mid-training evaluations, demonstrating the benefit of hybrid models in a controlled, large-scale setting. We find that the hybrid model scales significantly more efficiently than the transformer, explaining its higher performance. However, its unclear why greater expressivity on specific formal problems should result in better scaling or superior performance on downstream tasks unrelated to those problems. To explain this apparent gap, we return to theory and argue why increased expressivity should translate to better scaling efficiency, completing the loop. Overall, our results suggest that hybrid models mixing attention and recurrent layers are a powerful extension to the language modeling paradigm: not merely to reduce memory during inference, but as a fundamental way to obtain more expressive models that scale better during pretraining.

The sprint-based iterative approach in the Agile software development method allows continuous feedback and adaptation. One of the crucial Agile software development activities is the sprint planning session where developers estimate the effort required to complete tasks through a consensus-based estimation technique such as Planning Poker. In the Agile software development method, a common unit of measuring development effort is Story Point (SP) which is assigned to tasks to understand the complexity and development time needed to complete them. Despite the benefits of this process, it is an extremely time-consuming manual process. To mitigate this issue, in this study, we investigated if this manual process can be automated using Retrieval Augmented Generation (RAG) which comprises a "Retriever" and a "Generator". We applied two embedding models - bge-large-en-v1.5, and Sentence-Transformers' all-mpnet-base-v2 on 23 open-source software projects of varying sizes and examined four key aspects: 1) how retrieval hyper-parameters influence the performance, 2) whether estimation accuracy differs across different sizes of the projects, 3) whether embedding model choice affects accuracy, and 4) how the RAG-based approach compares to the existing baselines. Although the RAG-based approach outperformed the baseline models in several occasions, our results did not exhibit statistically significant differences in performance across the projects or across the embedding models. This highlights the need for further studies and refinement of the RAG, and model adaptation strategies for better accuracy in automatically estimating user stories.

Android instrumentation tests (end-to-end tests that run on a device or emulator) can catch problems that simpler tests miss. However, running these tests automatically in continuous integration (CI) is often difficult because emulator setup is fragile and configurations tend to drift over time. We study how open-source Android apps run instrumentation tests in CI by analyzing 4,518 repositories that use CI (snapshot: Aug. 10, 2025). We examine CI workflow files, scripts, and build configurations to identify cases where device setup is defined in Gradle (e.g., Gradle Managed Devices). Our results answer three questions about adoption, evolution, and outcomes. First, only about one in ten repositories (481/4,518; 10.6%) run instrumentation tests in CI, typically using either reusable community components or repository-specific custom scripts to set up emulators. Second, these setups usually stay the same over time; when changes happen, projects tend to move from custom scripts toward reusable community components. Third, we study why projects change their CI setup by analyzing their commits, pull requests, and issue messages. We evaluate how different setup styles perform using GitHub Actions run- and step-level metadata (e.g., outcomes, duration, reruns, and queue delay). We find that teams often change approaches to expand test coverage, and that each approach fits different needs: community-based setups are typically the most reliable and efficient for everyday checks on new code, third-party device labs suit scheduled regression testing but can be costlier and fail more often, and custom scripting provides flexibility but is associated with more reruns.

Sparse autoencoders (SAEs) are increasingly used for safety-relevant applications including alignment detection and model steering. These use cases require SAE latents to be as atomic as possible. Each latent should represent a single coherent concept drawn from a single underlying representational subspace. In practice, SAE latents blend representational subspaces together. A single feature can activate across semantically distinct contexts that share no true common representation, muddying an already complex picture of model computation. We introduce a joint training objective that directly penalizes this subspace blending. A small meta SAE is trained alongside the primary SAE to sparsely reconstruct the primary SAE's decoder columns; the primary SAE is penalized whenever its decoder directions are easy to reconstruct from the meta dictionary. This occurs whenever latent directions lie in a subspace spanned by other primary directions. This creates gradient pressure toward more mutually independent decoder directions that resist sparse meta-compression. On GPT-2 large (layer 20), the selected configuration reduces mean $|\varphi|$ by 7.5% relative to an identical solo SAE trained on the same data. Automated interpretability (fuzzing) scores improve by 7.6%, providing external validation of the atomicity gain independent of the training and co-occurrence metrics. Reconstruction overhead is modest. Results on Gemma 2 9B are directional. On not-fully-converged SAEs, the same parameterization yields the best results, a $+8.6\%$ $Δ$Fuzz. Though directional, this is an encouraging sign that the method transfers to a larger model. Qualitative analysis confirms that features firing on polysemantic tokens are split into semantically distinct sub-features, each specializing in a distinct representational subspace.

Spiking Neural Networks (SNNs) promise significant advantages over conventional Artificial Neural Networks (ANNs) for applications requiring real-time processing of temporally sparse data streams under strict power constraints -- a concept known as the Neuromorphic Advantage. However, the limited availability of neuromorphic hardware creates a substantial simulation-to-hardware gap that impedes algorithmic innovation, hardware-software co-design, and the development of mature open-source ecosystems. To address this challenge, we introduce Yet Another Neuromorphic Accelerator (YANA), an FPGA-based digital SNN accelerator designed to bridge this gap by providing an accessible hardware and software framework for neuromorphic computing. YANA implements a five-stage, event-driven processing pipeline that fully exploits temporal and spatial sparsity while supporting arbitrary SNN topologies through point-to-point neuron connections. The architecture features an input preprocessing scheme that maintains steady event processing at one event per cycle without buffer overflow risks, and implements hardware-efficient event-driven neuron updates using lookup tables for leak calculations. We demonstrate YANA's sparsity exploitation capabilities through experiments on the Spiking Heidelberg Digits dataset, showing near-linear scaling of inference time with both spatial and temporal sparsity levels. Deployed on the accessible AMD Kria KR260 platform, a single YANA core utilizes 740 LUTs, 918 registers, 7 BRAMS and 24 URAMs, supporting up to $2^{17}$ synapses and $2^{10}$ neurons. We release the YANA framework as an open-source project, providing an end-to-end solution for training, optimizing, and deploying SNNs that integrates with existing neuromorphic computing tools through the Neuromorphic Intermediate Representation (NIR).

As Large Language Model (LLM) based Multi-Agent Systems (MAS) evolve from experimental pilots to complex, persistent ecosystems, the limitations of direct agent-to-agent communication have become increasingly apparent. Current architectures suffer from fragmented context, stochastic hallucinations, rigid security boundaries, and inefficient topology management. This paper introduces Cognitive Fabric Nodes (CFN), a novel middleware layer that creates an omnipresent "Cognitive Fabric" between agents. Unlike traditional message queues or service meshes, CFNs are not merely pass-through mechanisms; they are active, intelligent intermediaries. Central to this architecture is the elevation of Memory from simple storage to an active functional substrate that informs four other critical capabilities: Topology Selection, Semantic Grounding, Security Policy Enforcement, and Prompt Transformation. We propose that each of these functions be governed by learning modules utilizing Reinforcement Learning (RL) and optimization algorithms to improve system performance dynamically. By intercepting, analyzing, and rewriting inter-agent communication, the Cognitive Fabric ensures that individual agents remain lightweight while the ecosystem achieves coherence, safety, and semantic alignment. We evaluate the effectiveness of the CFN on the HotPotQA and MuSiQue datasets in a multi-agent environment and demonstrate that the CFN improves performance by more than 10\% on both datasets over direct agent to agent communication.

Automated underwater species classification is constrained by annotation cost and environmental variation that limits the transferability of fully supervised models. Recent work has shown that frozen embeddings from self-supervised vision foundation models already provide a strong label-efficient baseline for marine image classification. Here we investigate whether this frozen-embedding regime can be improved at inference time, without fine-tuning or changing model weights. We apply Circuit Duplication, an inference-time method originally proposed for Large Language Models, in which a selected range of transformer layers is traversed twice during the forward pass. We evaluate on the class-imbalanced AQUA20 benchmark using frozen DINOv3 embeddings under two settings: global circuit selection, where a single duplicated circuit is chosen for the full dataset, and class-specific circuit selection, where each species may receive a different optimal circuit. Both settings use simple semi-supervised downstream classifiers. Circuit Duplication consistently improves over the standard frozen forward pass. At the maximum label budget, class-specific selection reaches a macro F1 of 0.875, closing the gap to the fully supervised ConvNeXt benchmark (0.889) to 1.4 points without any gradient-based training. Four species exceed their fully supervised reference, with octopus improving by +12.1 F1 points. Across all budgets, roughly 75% of classes prefer a class-specific circuit, indicating a genuinely class-dependent benefit. To our knowledge, this is the first application of Circuit Duplication to computer vision.

State-space model (SSM) for time-series forecasting have demonstrated strong empirical performance on benchmark datasets, yet their robustness under adversarial perturbations is poorly understood. We address this gap through a control-theoretic lens, focusing on the recently proposed Spacetime SSM forecaster. We first establish that the decoder-only Spacetime architecture can represent the optimal Kalman predictor when the underlying data-generating process is autoregressive - a property no other SSM possesses. Building on this, we formulate robust forecaster design as a Stackelberg game against worst-case stealthy adversaries constrained by a detection budget, and solve it via adversarial training. We derive closed-form bounds on adversarial forecasting error that expose how open-loop instability, closed-loop instability, and decoder state dimension each amplify vulnerability - offering actionable principles towards robust forecaster design. Finally, we show that even adversaries with no access to the forecaster can nonetheless construct effective attacks by exploiting the model's locally linear input-output behavior, bypassing gradient computations entirely. Experiments on the Monash benchmark datasets highlight that model-free attacks, without any gradient computation, can cause at least 33% more error than projected gradient descent with a small step size.

Fully Homomorphic Encryption (FHE) enables privacy-preserving Transformer inference, but long-sequence encrypted Transformers quickly exceed single-GPU memory capacity because encoded weights are already large and encrypted activations grow rapidly with sequence length. Multi-GPU execution therefore becomes unavoidable, yet scaling remains challenging because communication is jointly induced by application-level aggregation and encryption-level RNS coupling. Existing approaches either synchronize between devices frequently or replicate encrypted tensors across devices, leading to excessive communication and latency. We present AEGIS, an Application-Encryption Guided Inference System for scalable long-sequence encrypted Transformer inference on multi-GPU platforms. AEGIS derives device placement from ciphertext dependencies jointly induced by Transformer dataflow and CKKS polynomial coupling, co-locating modulus-coherent and token-coherent data so that communication is introduced only when application dependencies require it, while reordering polynomial operators to overlap the remaining collectives with computation. On 2048-token inputs, AEGIS reduces inter-GPU communication by up to 57.9% in feed-forward networks and 81.3% in self-attention versus prior state-of-the-art designs. On four GPUs, it achieves up to 96.62% scaling efficiency, 3.86x end-to-end speedup, and 69.1% per-device memory reduction. These results establish coordinated application-encryption parallelism as a practical foundation for scalable homomorphic Transformer inference.

All languages have a noun category, but its realisation varies considerably. Depending on the language, semantic and/or morphosyntactic differences may be more or less pronounced. This paper explores these variations, using Riffian as a reference point before extending the analysis to other languages. We propose a formal model termed morphology-driven marking. Nouns are organised into modular cognitive sets, each with its own morphological template and unmarked form. This approach helps explain differences in marking among noun types within and across languages. By situating these patterns within syntactic functions, we also reassess the notions of markedness and state. It is proposed that the concept of state be extended to all synthetic languages and analysed a novel subcategory of syntax-based inflection like agreement and grammatical case.

We show that robustness to post-training quantization (PTQ) is a transferable direction in weight space. We call this direction the quantization vector: extracted from a donor task by simple weight-space arithmetic, it can be used to patch a receiver model and improve robustness to PTQ-induced noise by as much as 60%, without receiver-side quantization-aware training (QAT). Because the method requires no receiver training data, it provides a zero-shot, low-cost alternative to QAT for extremely low-bit deployment. We demonstrate this on Vision Transformer (ViT) models. More broadly, our results suggest that quantization robustness is not merely a byproduct of task-specific training, but a reusable feature of weight-space geometry that can be transferred rather than retrained.

Submodular maximization under matroid constraints is a fundamental problem in combinatorial optimization with applications in sensing, data summarization, active learning, and resource allocation. While the Sequential Greedy (SG) algorithm achieves only a $\frac{1}{2}$-approximation due to irrevocable selections, Continuous Greedy (CG) attains the optimal $\bigl(1-\frac{1}{e}\bigr)$-approximation via the multilinear relaxation, at the cost of a progressively dense decision vector that forces agents to exchange feature embeddings for nearly every ground-set element. We propose \textit{ATCG} (\underline{A}daptive \underline{T}hresholded \underline{C}ontinuous \underline{G}reedy), which gates gradient evaluations behind a per-partition progress ratio $η_i$, expanding each agent's active set only when current candidates fail to capture sufficient marginal gain, thereby directly bounding which feature embeddings are ever transmitted. Theoretical analysis establishes a curvature-aware approximation guarantee with effective factor $τ_{\mathrm{eff}}=\max\{τ,1-c\}$, interpolating between the threshold-based guarantee and the low-curvature regime where \textit{ATCG} recovers the performance of CG. Experiments on a class-balanced prototype selection problem over a subset of the CIFAR-10 animal dataset show that \textit{ATCG} achieves objective values comparable to those of the full CG method while substantially reducing communication overhead through adaptive active-set expansion.

Network visualization has traditionally relied on heuristic metrics, such as stress, under the assumption that optimizing them leads to aesthetic and informative layouts. However, no single metric consistently produces the most effective results. A data-driven alternative is to learn from human preferences, where annotators select their favored visualization among multiple layouts of the same graphs. These human-preference labels can then be used to train a generative model that approximates human aesthetic preferences. However, obtaining human labels at scale is costly and time-consuming. As a result, this generative approach has so far been tested only with machine-labeled data. In this paper, we explore the use of large language models (LLMs) and vision models (VMs) as proxies for human judgment. Through a carefully designed user study involving 27 participants, we curated a large set of human preference labels. We used this data both to better understand human preferences and to bootstrap LLM/VM labelers. We show that prompt engineering that combines few-shot examples and diverse input formats, such as image embeddings, significantly improves LLM-human alignment, and additional filtering by the confidence score of the LLM pushes the alignment to human-human levels. Furthermore, we demonstrate that carefully trained VMs can achieve VM-human alignment at a level comparable to that between human annotators. Our results suggest that AI can feasibly serve as a scalable proxy for human labelers.

Generative artificial intelligence offers a new paradigm to design matter in high-dimensional spaces. However, its underlying mechanisms remain difficult to interpret and limit adoption in computational mechanics. This gap is striking because its core tools-diffusion, stochastic differential equations, and inverse problems-are fundamental to the mechanics of materials. Here we show that diffusion-based generative AI and computational mechanics are rooted in the same principles. We illustrate this connection using a three-ingredient burger as a minimal benchmark for material design in a low-dimensional space, where both forward and reverse diffusion admit analytical solutions: Markov chains with Bayesian inversion in the discrete case and the Ornstein-Uhlenbeck process with score-based reversal in the continuous case. We extend this framework to a high-dimensional design space with 146 ingredients and 8.9x10^43 possible configurations, where analytical solutions become intractable. We therefore learn the discrete and continuous reverse processes using neural network models that infer inverse dynamics from data. We train the models on only 2,260 recipes and generate one million samples that capture the statistical structure of the data, including ingredient prevalence and quantitative composition. We further generate five new burgers and validate them in a restaurant-based sensory study with 100 participants, where three of the AI-designed burgers outperform the classical Big Mac in overall liking, flavor, and texture. These results establish diffusion-based generative modeling as a physically grounded approach to design in high-dimensional spaces. They position generative AI as a natural extension of computational mechanics, with applications from burgers to matter, and establish a path toward data-driven, physics-informed generative design.

Active multi-target tracking requires a mobile robot to balance exploration for undetected targets with exploitation of uncertain tracked ones. Diffusion policies have emerged as a powerful approach for capturing diverse behavioral strategies by learning action sequences from expert demonstrations. However, existing methods implicitly select among strategies through the denoising process, without uncertainty quantification over which strategy to execute. We formulate expert selection for diffusion policies as an offline contextual bandit problem and propose a Bayesian framework for pessimistic, uncertainty-aware strategy selection. A multi-head Variational Bayesian Last Layer (VBLL) model predicts the expected tracking performance of each expert strategy given the current belief state, providing both a point estimate and predictive uncertainty. Following the pessimism principle for offline decision-making, a Lower Confidence Bound (LCB) criterion then selects the expert whose worst-case predicted performance is best, avoiding overcommitment to experts with unreliable predictions. The selected expert conditions a diffusion policy to generate corresponding action sequences. Experiments on simulated indoor tracking scenarios demonstrate that our approach outperforms both the base diffusion policy and standard gating methods, including Mixture-of-Experts selection and deterministic regression baselines.

Dense retrieval systems increasingly need to handle complex queries. In many realistic settings, users express intent through long instructions or task-specific descriptions, while target documents remain relatively simple and static. This asymmetry creates a retrieval mismatch: understanding queries may require strong reasoning and instruction-following, whereas efficient document indexing favors lightweight encoders. Existing retrieval systems often address this mismatch by directly improving the embedding model, but fine-tuning large embedding models to better follow such instructions is computationally expensive, memory-intensive, and operationally burdensome. To address this challenge, we propose Efficient Retrieval Adapter (ERA), a label-efficient framework that trains retrieval adapters in two stages: self-supervised alignment and supervised adaptation. Inspired by the pre-training and supervised fine-tuning stages of LLMs, ERA first aligns the embedding spaces of a large query embedder and a lightweight document embedder, and then uses limited labeled data to adapt the query-side representation, bridging both the representation gap between embedding models and the semantic gap between complex queries and simple documents without re-indexing the corpus. Experiments on the MAIR benchmark, spanning 126 retrieval tasks across 6 domains, show that ERA improves retrieval in low-label settings, outperforms methods that rely on larger amounts of labeled data, and effectively combines stronger query embedders with weaker document embedders across domains.

Understanding student engagement usually requires time-consuming manual observation or invasive recording that raises privacy concerns. We present a privacy-preserving pipeline that analyzes classroom videos to extract insights about student attention, without storing any identifiable footage. Our system runs on a single GPU, using OpenPose for skeletal extraction and Gaze-LLE for visual attention estimation. Original video frames are deleted immediately after pose extraction, thus only geometric coordinates (stored as JSON) are retained, ensuring compliance with FERPA. The extracted pose and gaze data is processed by QwQ-32B-Reasoning, which performs zero-shot analysis of student behavior across lecture segments. Instructors access results through a web dashboard featuring attention heatmaps and behavioral summaries. Our preliminary findings suggest that LLMs may show promise for multimodal behavior understanding, although they still struggle with spatial reasoning about classroom layouts. We discuss these limitations and outline directions for improving LLM spatial comprehension in educational analytics contexts.

The multi-step, iterative image editing capabilities of multi-modal agentic systems have transformed digital content creation. Although latest image editing models faithfully follow instructions and generate high-quality images in single-turn edits, we identify a critical weakness in multi-turn editing, which is the iterative degradation of image quality. As images are repeatedly edited, minor artifacts accumulate, rapidly leading to a severe accumulation of visible noise and a failure to follow simple editing instructions. To systematically study these failures, we introduce Banana100, a comprehensive dataset of 28,000 degraded images generated through 100 iterative editing steps, including diverse textures and image content. Alarmingly, image quality evaluators fail to detect the degradation. Among 21 popular no-reference image quality assessment (NR-IQA) metrics, none of them consistently assign lower scores to heavily degraded images than to clean ones. The dual failures of generators and evaluators may threaten the stability of future model training and the safety of deployed agentic systems, if the low-quality synthetic data generated by multi-turn edits escape quality filters. We release the full code and data to facilitate the development of more robust models, helping to mitigate the fragility of multi-modal agentic systems.

We present QIMMA, a quality-assured Arabic LLM leaderboard that places systematic benchmark validation at its core. Rather than aggregating existing resources as-is, QIMMA applies a multi-model assessment pipeline combining automated LLM judgment with human review to surface and resolve systematic quality issues in well-established Arabic benchmarks before evaluation. The result is a curated, multi-domain, multi-task evaluation suite of over 52k samples, grounded predominantly in native Arabic content; code evaluation tasks are the sole exception, as they are inherently language-agnostic. Transparent implementation via LightEval, EvalPlus and public release of per-sample inference outputs make QIMMA a reproducible and community-extensible foundation for Arabic NLP evaluation.

Multimodal reasoning has emerged as a powerful framework for enhancing reasoning capabilities of reasoning models. While multi-turn table reasoning methods have improved reasoning accuracy through tool use and reward modeling, they rely on fixed text serialization for table state readouts. This introduces representation errors in table encoding that significantly accumulate over multiple turns. Such accumulation is alleviated by tabular grounding methods in the expense of inference compute and cost, rendering real world deployment impractical. To address this, we introduce TABQAWORLD, a table reasoning framework that jointly optimizes tabular action through representation and estimation. For representation, TABQAWORLD employs an action-conditioned multimodal selection policy, which dynamically switches between visual and textual representations to maximize table state readout reliability. For estimation, TABQAWORLD optimizes stepwise reasoning trajectory through table metadata including dimension, data types and key values, safely planning trajectory and compressing low-complexity actions to reduce conversation turns and latency. Designed as a training-free framework, empirical evaluations show that TABQAWORLD achieves state-of-the-art performance with 4.87% accuracy improvements over baselines, with 5.42% accuracy gain and 33.35% inference latency reduction over static settings, establishing a new standard for reliable and efficient table reasoning.

Real-world implementations of connected vehicle functions are spreading steadily, yet operating these functions reliably remains challenging due to their distributed nature and the complexity of the underlying cloud, edge, and networking infrastructure. Quick diagnosis of problems and understanding the error chains that lead to failures is essential for reducing downtime. However, diagnosing these systems is still largely performed manually, as automated analysis techniques are predominantly data-driven and struggle with hidden relationships and the integration of context information. This paper addresses this gap by introducing a multimodal approach that integrates human feedback and system-specific information into the causal analysis process. Reinforcement Learning from Human Feedback is employed to continuously train a causality mining model while incorporating expert knowledge. Additional modules leverage distributed tracing data to prune false-positive causal links and enable the injection of domain-specific relationships to further refine the causal graph.Evaluation is performed using an automated valet parking application operated in a connected vehicle test field. Results demonstrate a significant increase in precision from 14\% to 100\% for the detection of causal edges and improved system interpretability compared to purely data-driven approaches, highlighting the potential for system operators in the connected vehicle domain.

When deploying large language models (LLMs) to safety-critical applications, uncertainty quantification (UQ) is of utmost importance to self-assess the reliability of the LLM-based decisions. However, such decisions typically suffer from overconfidence, particularly after parameter-efficient fine-tuning (PEFT) for downstream domain-specific tasks with limited data. Existing methods to alleviate this issue either rely on Laplace approximation based post-hoc framework, which may yield suboptimal calibration depending on the training trajectory, or variational Bayesian training that requires multiple complete forward passes through the entire LLM backbone at inference time for Monte Carlo estimation, posing scalability challenges for deployment. To address these limitations, we build on the Bayesian last layer (BLL) model, where the LLM-based deterministic feature extractor is followed by random last layer parameters for uncertainty reasoning. Since existing low-rank adapters (LoRA) for PEFT have limited expressiveness due to rank collapse, we address this with Polar-decomposed Low-rank Adapter Representation (PoLAR), an orthogonalized parameterization paired with Riemannian optimization to enable more stable and expressive adaptation. Building on this PoLAR-BLL model, we leverage the variational (V) inference framework to put forth a scalable Bayesian fine-tuning approach which jointly seeks the PoLAR parameters and approximate posterior of the last layer parameters via alternating optimization. The resulting PoLAR-VBLL is a flexible framework that nicely integrates architecture-enhanced optimization with scalable Bayesian inference to endow LLMs with well-calibrated UQ. Our empirical results verify the effectiveness of PoLAR-VBLL in terms of generalization and uncertainty estimation on both in-distribution and out-of-distribution data for various common-sense reasoning tasks.

Hume's account of causal judgment presupposes three representational conditions: experiential grounding (ideas must trace to impressions), structured retrieval (association must operate through organized networks exceeding pairwise connection), and vivacity transfer (inference must produce felt conviction, not merely updated probability). This paper extracts these conditions from Hume's texts and argues that they are integral to his causal psychology. It then traces their fate through the formalization trajectory from Hume to Bayesian epistemology and predictive processing, showing that later frameworks preserve the updating structure of Hume's insight while abstracting away these further representational conditions. Large language models serve as an illustrative contemporary case: they exhibit a form of statistical updating without satisfying the three conditions, thereby making visible requirements that were previously background assumptions in Hume's framework.

Generating diverse, pedagogically valid stories for Arabic early-grade reading assessments requires balancing tight constraints on vocabulary, reading level, and narrative structure against the need to avoid repetitive plots that undermine assessment validity. We investigate noise steering, injecting calibrated Gaussian perturbations into the internal representations of transformer models at inference time, as a training-free diversity method evaluated across five small Arabic-centric language models (7-9B parameters). We compare four injection strategies against high-temperature sampling baselines, measuring diversity, quality, constraint adherence, and reading grade level. Residual stream noise consistently improves narrative diversity with minimal quality or constraint cost and preserves early-grade reading level across all models. Attention entropy noise injection (AENI) stabilizes the otherwise unreliable attention-logit noise while recovering quality. High-temperature sampling inflates reading grade level and causes catastrophic collapse on several models. We find internal representation-level perturbation to be a more suitable diversity strategy than output-level stochasticity for constrained educational content generation.

Current literature on radiology report evaluation has focused primarily on designing LLM-based metrics and fine-tuning small models for chest X-rays. However, it remains unclear whether these approaches are robust when applied to reports from other modalities and anatomies. Which model and prompt configurations are best suited to serve as LLM judges for radiology evaluation? We conduct a thorough correlation analysis between expert and LLM-based ratings. We compare three existing LLM-as-a-judge metrics (RadFact, GREEN, and FineRadScore) alongside VERT, our proposed LLM-based metric, using open- and closed-source models (reasoning and non-reasoning) of different sizes across two expert-annotated datasets, RadEval and RaTE-Eval, spanning multiple modalities and anatomies. We further evaluate few-shot approaches, ensembling, and parameter-efficient fine-tuning using RaTE-Eval. To better understand metric behavior, we perform a systematic error detection and categorization study to assess alignment of these metrics against expert judgments and identify areas of lower and higher agreement. Our results show that VERT improves correlation with radiologist judgments by up to 11.7% relative to GREEN. Furthermore, fine-tuning Qwen3 30B yield gains of up to 25% using only 1,300 training samples. The fine-tuned model also reduces inference time up to 37.2 times. These findings highlight the effectiveness of LLM-based judges and demonstrate that reliable evaluation can be achieved with lightweight adaptation.

Creative problem-solving requires combining multiple cognitive abilities, including logical reasoning, lateral thinking, analogy-making, and commonsense knowledge, to discover insights that connect seemingly unrelated pieces of information. However, most existing benchmarks for large language models (LLMs) evaluate only specific components of this process. Moreover, many creativity-oriented benchmarks rely on artificially constructed brainteasers or contrived scenarios that do not reflect how creative problem-solving occurs in real-world settings. To address this gap, we introduce CresOWLve, a benchmark for evaluating creative problem-solving using puzzles grounded in real-world knowledge. Problems in CresOWLve require employing multiple creative thinking strategies, retrieving facts from diverse domains, and creatively combining them to arrive at a solution. Evaluating several frontier non-thinking and thinking LLMs, we show that CresOWLve remains highly challenging. Our analysis reveals a consistent performance gap: models perform substantially better on factual questions than on creative ones (up to a -17% drop). While models can often retrieve the relevant knowledge, they struggle to form the non-obvious creative connections required to integrate this information and arrive at the correct answer.

AI coding agents are increasingly integrated into real-world software development workflows, yet their robustness under diverse and adversarial scenarios remains poorly understood. We present ABTest, a behavior-driven fuzzing framework that systematically tests coding agents by turning real-world failure reports into repository-grounded behavioral tests. ABTest (1) mines user-reported anomalies to derive reusable workflow patterns (Interaction Patterns) and behaviors (Action types); (2) composes them into stepwise fuzzing templates; (3) instantiates executable test cases in real repositories; (4) executes them with coding agents while recording traces and artifacts; and (5) detects and validates anomalous behaviors. We apply ABTest to three widely used coding agents: Claude Code, OpenAI Codex CLI, and Gemini CLI. From 400 user-reported developer-confirmed agent failures, we extract 47 Interaction Patterns and 128 Action types, generating 647 repository-grounded fuzzing cases. Executing the 647-case bundle once per evaluated configuration, ABTest flags 1,573 behavioral anomalies across the three coding agent families, of which 642 are manually confirmed as new true anomalies, achieving a detection precision of 40.8%. Our results demonstrate that ABTest effectively uncovers real-world failures, exposes robustness differences across models, and reveals previously unreported failure modes.

The modeling of bio-molecular system across molecular scales remains a central challenge in scientific research. Large language models (LLMs) are increasingly applied to bio-molecular discovery, yet systematic evaluation across multi-scale biological problems and rigorous assessment of their tool-augmented capabilities remain limited. We reveal a systematic gap between LLM performance and mechanistic understanding through the proposed cross-scale bio-molecular benchmark: BioMol-LLM-Bench, a unified framework comprising 26 downstream tasks that covers 4 distinct difficulty levels, and computational tools are integrated for a more comprehensive evaluation. Evaluation on 13 representative models reveals 4 main findings: chain-of-thought data provides limited benefit and may even reduce performance on biological tasks; hybrid mamba-attention architectures are more effective for long bio-molecular sequences; supervised fine-tuning improves specialization at the cost of generalization; and current LLMs perform well on classification tasks but remain weak on challenging regression tasks. Together, these findings provide practical guidance for future LLM-based modeling of molecular systems.

Dynamic quantum circuits with mid-circuit measurements (MCMs) and feed-forward operations play a crucial role in various applications, such as quantum error correction and quantum algorithms. With advancements in quantum hardware enabling the implementation of MCM and feed-forward loops, the use of dynamic circuits has become increasingly prevalent. There is a significant need for a benchmarking framework specially designed for dynamic circuits to capture their unique properties, as current benchmarking tools are designed primarily for unitary circuits and cannot be trivially extended to dynamic circuits. We propose dynamarq, a scalable and hardware-agnostic benchmarking framework for dynamic circuits. We collect a set of dynamic circuit benchmarks spanning various applications and propose a broad set of circuit features to characterize the structure of these dynamic circuits. We run them on two IBM quantum processors and the Quantinuum Helios-1E emulator, and propose scalable, application-dependent fidelity scores for each benchmark based on hardware execution results. We perform statistical modeling to identify correlations between circuit features and fidelity scores, and demonstrate highly accurate fidelity prediction using our model. Our model parameters are also transferable across hardware backends and calibration cycles. Our framework facilitates the understanding of dynamic circuit structures and provides insights for designing and optimizing dynamic circuits to achieve high execution fidelity on quantum hardware.

Artificial intelligence (AI) alignment is fundamentally a formation problem, not only a safety problem. As Large Language Models (LLMs) increasingly mediate moral deliberation and spiritual inquiry, they do more than provide information; they function as instruments of digital catechesis, actively shaping and ordering human understanding, decision-making, and moral reflection. To make this formative influence visible and measurable, we introduce the Flourishing AI Benchmark: Christian Single-Turn (FAI-C-ST), a framework designed to evaluate Frontier Model responses against a Christian understanding of human flourishing across seven dimensions. By comparing 20 Frontier Models against both pluralistic and Christian-specific criteria, we show that current AI systems are not worldview-neutral. Instead, they default to a Procedural Secularism that lacks the grounding necessary to sustain theological coherence, resulting in a systematic performance decline of approximately 17 points across all dimensions of flourishing. Most critically, there is a 31-point decline in the Faith and Spirituality dimension. These findings suggest that the performance gap in values alignment is not a technical limitation, but arises from training objectives that prioritize broad acceptability and safety over deep, internally coherent moral or theological reasoning.

Systematic exploration of Agent-Based Models (ABMs) is challenged by the curse of dimensionality and their inherent stochasticity. We present a multi-stage pipeline integrating the systematic design of experiments with machine learning surrogates. Using a predator-prey case study, our methodology proceeds in two steps. First, an automated model-based screening identifies dominant variables, assesses outcome variability, and segments the parameter space. Second, we train Machine Learning models to map the remaining nonlinear interaction effects. This approach automates the discovery of unstable regions where system outcomes are highly dependent on nonlinear interactions between many variables. Thus, this work provides modelers with a rigorous, hands-off framework for sensitivity analysis and policy testing, even when dealing with high-dimensional stochastic simulators.

Kolmogorov-Arnold Networks (KANs) have recently emerged as a powerful architecture for various machine learning applications. However, their unique structure raises significant concerns regarding their computational overhead. Existing studies primarily evaluate KAN complexity in terms of Floating-Point Operations (FLOPs) required for GPU-based training and inference. However, in many latency-sensitive and power-constrained deployment scenarios, such as neural network-driven non-linearity mitigation in optical communications or channel state estimation in wireless communications, training is performed offline and dedicated hardware accelerators are preferred over GPUs for inference. Recent hardware implementation studies report KAN complexity using platform-specific resource consumption metrics, such as Look-Up Tables, Flip-Flops, and Block RAMs. However, these metrics require a full hardware design and synthesis stage that limits their utility for early-stage architectural decisions and cross-platform comparisons. To address this, we derive generalized, platform-independent formulae for evaluating the hardware inference complexity of KANs in terms of Real Multiplications (RM), Bit Operations (BOP), and Number of Additions and Bit-Shifts (NABS). We extend our analysis across multiple KAN variants, including B-spline, Gaussian Radial Basis Function (GRBF), Chebyshev, and Fourier KANs. The proposed metrics can be computed directly from the network structure and enable a fair and straightforward inference complexity comparison between KAN and other neural network architectures.

Electricity theft and non-technical losses (NTLs) remain critical challenges in modern smart grids, causing significant economic losses and compromising grid reliability. This study introduces the SmartGuard Energy Intelligence System (SGEIS), an integrated artificial intelligence framework for electricity theft detection and intelligent energy monitoring. The proposed system combines supervised machine learning, deep learning-based time-series modeling, Non-Intrusive Load Monitoring (NILM), and graph-based learning to capture both temporal and spatial consumption patterns. A comprehensive data processing pipeline is developed, incorporating feature engineering, multi-scale temporal analysis, and rule-based anomaly labeling. Deep learning models, including Long Short-Term Memory (LSTM), Temporal Convolutional Networks (TCN), and Autoencoders, are employed to detect abnormal usage patterns. In parallel, ensemble learning methods such as Random Forest, Gradient Boosting, XGBoost, and LightGBM are utilized for classification. To model grid topology and spatial dependencies, Graph Neural Networks (GNNs) are applied to identify correlated anomalies across interconnected nodes. The NILM module enhances interpretability by disaggregating appliance-level consumption from aggregate signals. Experimental results demonstrate strong performance, with Gradient Boosting achieving a ROC-AUC of 0.894, while graph-based models attain over 96% accuracy in identifying high-risk nodes. The hybrid framework improves detection robustness by integrating temporal, statistical, and spatial intelligence. Overall, SGEIS provides a scalable and practical solution for electricity theft detection, offering high accuracy, improved interpretability, and strong potential for real-world smart grid deployment.

Latent action learning infers pseudo-action labels from visual transitions, providing an approach to leverage internet-scale video for embodied AI. However, most methods learn latent actions without structural priors that encode the additive, compositional structure of physical motion. As a result, latents often entangle irrelevant scene details or information about future observations with true state changes and miscalibrate motion magnitude. We introduce Additively Compositional Latent Action Model (AC-LAM), which enforces scene-wise additive composition structure over short horizons on the latent action space. These AC constraints encourage simple algebraic structure in the latent action space~(identity, inverse, cycle consistency) and suppress information that does not compose additively. Empirically, AC-LAM learns more structured, motion-specific, and displacement-calibrated latent actions and provides stronger supervision for downstream policy learning, outperforming state-of-the-art LAMs across simulated and real-world tabletop tasks.

Autonomous AI systems can now generate complete economics research papers, but they substantially underperform human-authored publications in head-to-head comparisons. This paper decomposes the quality gap into two independent components: research idea quality and execution quality. Using a two-model ensemble of fine-tuned language models trained on publication decisions (Gong, Li, and Zhou, 2026) to evaluate idea quality and a comprehensive six-dimension rubric assessed by Gemini 3.1 Flash Lite -- the same model family used as the APE tournament judge, ensuring methodological consistency -- to evaluate execution quality, we analyze 953 economics papers -- 912 AI-generated papers from the APE project and 41 human papers published in the American Economic Review and AEJ: Economic Policy. The idea quality gap is large (Cohen's d = 2.23, p < 0.001), with human papers achieving 47.1% mean ensemble exceptional probability versus 16.5% for AI. The execution quality gap is also significant but smaller (d = 0.90, p < 0.001), with human papers scoring 4.38/5.0 versus 3.84. Idea quality accounts for approximately 71% of the overall quality difference, with execution contributing 29%. The largest execution weakness is mechanism analysis depth (d = 1.43); no significant difference is found on robustness. We document that 74% of AI papers employ difference-in-differences, and only 7 AI papers (0.8%) surpass the median human paper on both idea and execution quality simultaneously. The primary bottleneck to competitive AI-generated economics research remains ideation.

BitNet b1.58 (Ma et al., 2024) demonstrates that large language models can operate entirely on ternary weights {-1, 0, +1}, yet no native binary wire format exists for such models. NativeTernary closes this gap. We present NativeTernary, a binary encoding scheme that partitions the 2-bit pair space into three data symbols representing ternary values -- either balanced {-1, 0, +1} or unsigned {0, 1, 2} -- and a reserved structural delimiter. The central contribution is the use of unary run-length encoding to represent semantic hierarchy depth: a sequence of N consecutive delimiter pairs denotes a boundary of level N, encoding character, word, sentence, paragraph, and topic boundaries at cost 2, 4, 6, 8, and 10 bits respectively -- proportional to boundary rarity. The choice of which 2-bit pair serves as the delimiter is a design parameter: {11} is the primary embodiment, offering simple OR-gate detection; {00} is an alternative embodiment optimised for ultra-low-power CMOS systems, minimising switching activity. All four bit-pair choices are covered by the patent claims. We present three encoding variants: (1) the primary scheme with {11} as sole delimiter; (2) a dual-starter variant where both {10} and {11} initiate distinct symbol namespaces; and (3) an analysis of unsigned versus balanced ternary data mappings. We describe a path toward ternary-native general computing infrastructure requiring no hardware changes, and outline applications spanning ternary neural network weight storage, hierarchical natural language encoding, edge computing, IoT and satellite telemetry, industrial sensors, automotive systems, medical devices, gaming, and financial tick data. The decoder is a 10-line stateless state machine resilient to bitstream corruption.

Apparent age estimation is a valuable tool for business personalization, yet current models frequently exhibit demographic biases. We review prior works on the DEX method by applying distribution learning techniques such as Mean-Variance Loss (MVL) and Adaptive Mean-Residue Loss (AMRL), and evaluate them in both accuracy and fairness. Using IMDB-WIKI, APPA-REAL, and FairFace, we demonstrate that while AMRL achieves state-of-the-art accuracy, trade-offs between precision and demographic equity persist. Despite clear age clustering in UMAP embeddings, our saliency maps indicate inconsistent feature focus across demographics, leading to significant performance degradation for Asian and African American populations. We argue that technical improvements alone are insufficient; accurate and fair apparent age estimation requires the integration of localized and diverse datasets, and strict adherence to fairness validation protocols.

Symbolic music generation has made significant progress, yet achieving fine-grained and flexible control over composer style remains challenging. Existing training-based methods for composer style conditioning depend on large labeled datasets. Besides, these methods typically support only single-composer generation at a time, limiting their applicability to more creative or blended scenarios. In this work, we propose Composer Vector, an inference-time steering method that operates directly in the model's latent space to control composer style without retraining. Through experiments on multiple symbolic music generation models, we show that Composer Vector effectively guides generations toward target composer styles, enabling smooth and interpretable control through a continuous steering coefficient. It also enables seamless fusion of multiple styles within a unified latent space framework. Overall, our work demonstrates that simple latent space steering provides a practical and general mechanism for controllable symbolic music generation, enabling more flexible and interactive creative workflows. Code and Demo are available here: https://github.com/JiangXunyi/Composer-Vector and https://jiangxunyi.github.io/composervector.github.io/

For digital infrastructure to be safe, compatible, and standards-aligned, automated communication protocol compliance verification is crucial. Nevertheless, current rule-based systems are becoming less and less effective since they are unable to identify subtle or intricate non-compliance, which attackers frequently use to establish covert communication channels in IPv6 traffic. In order to automate IPv6 compliance verification, this paper presents the Artificial Intelligence Driven Compliance Checker Engine (AICCE), a novel generative system that combines dual-architecture reasoning and retrieval-augmented generation (RAG). Specification segments pertinent to each query can be efficiently retrieved thanks to the semantic encoding of protocol standards into a high-dimensional vector space. Based on this framework, AICCE offers two complementary pipelines: (i) Explainability Mode, which uses parallel LLM agents to render decisions and settle disputes through organized discussions to improve interpretability and robustness, and (ii) Script Execution Mode, which converts clauses into Python rules that can be executed quickly for dataset-wide verification. With the debate mechanism enhancing decision reliability in complicated scenarios and the script-based pipeline lowering per-sample latency, AICCE achieves accuracy and F1-scores of up to 99% when tested on IPv6 packet samples across sixteen cutting-edge generative models. By offering a scalable, auditable, and generalizable mechanism for identifying both routine and covert non-compliance in dynamic communication environments, our results show that AICCE overcomes the blind spots of conventional rule-based compliance checking systems.

Violence detection benefits from audio, but real-world soundscapes can be noisy or weakly related to the visible scene. We present CoLoRSMamba, a directional Video to Audio multimodal architecture that couples VideoMamba and AudioMamba through CLS-guided conditional LoRA. At each layer, the VideoMamba CLS token produces a channel-wise modulation vector and a stabilization gate that adapt the AudioMamba projections responsible for the selective state-space parameters (Delta, B, C), including the step-size pathway, yielding scene-aware audio dynamics without token-level cross-attention. Training combines binary classification with a symmetric AV-InfoNCE objective that aligns clip-level audio and video embeddings. To support fair multimodal evaluation, we curate audio-filtered clip level subsets of the NTU-CCTV and DVD datasets from temporal annotations, retaining only clips with available audio. On these subsets, CoLoRSMamba outperforms representative audio-only, video-only, and multimodal baselines, achieving 88.63% accuracy / 86.24% F1-V on NTU-CCTV and 75.77% accuracy / 72.94% F1-V on DVD. It further offers a favorable accuracy-efficiency tradeoff, surpassing several larger models with fewer parameters and FLOPs.

Perception plays a central role in connected and autonomous vehicles (CAVs), underpinning not only conventional modular driving stacks, but also cooperative perception systems and recent end-to-end driving models. While deep learning has greatly improved perception performance, its statistical nature makes perfect predictions difficult to attain. Meanwhile, standard training objectives and evaluation benchmarks treat all perception errors equally, even though only a subset is safety-critical. In this paper, we investigate safety-aligned evaluation and optimization for 3D object detection that explicitly characterize high-impact errors. Building on our previously proposed safety-oriented metric, NDS-USC, and safety-aware loss function, EC-IoU, we make three contributions. First, we present an expanded study of single-vehicle 3D object detection models across diverse neural network architectures and sensing modalities, showing that gains under standard metrics such as mAP and NDS may not translate to safety-oriented criteria represented by NDS-USC. With EC-IoU, we reaffirm the benefit of safety-aware fine-tuning for improving safety-critical detection performance. Second, we conduct an ego-centric, safety-oriented evaluation of AV-infrastructure cooperative object detection models, underscoring its superiority over vehicle-only models and demonstrating a safety impact analysis that illustrates the potential contribution of cooperative models to "Vision Zero." Third, we integrate EC-IoU into SparseDrive and show that safety-aware perception hardening can reduce collision rate by nearly 30% and improve system-level safety directly in an end-to-end perception-to-planning framework. Overall, our results indicate that safety-aligned perception evaluation and optimization offer a practical path toward enhancing CAV safety across single-vehicle, cooperative, and end-to-end autonomy settings.

Modern machine learning systems deployed in safety-critical domains require visibility not only into aggregate performance but also into how training dynamics affect subgroup fairness over time. Existing training dashboards primarily support single-metric monitoring and offer limited support for examining relationships between heterogeneous metrics or diagnosing subgroup disparities during training. We present InsightBoard, an interactive TensorBoard plugin that integrates synchronized multi-metric visualization with slice-based fairness diagnostics in a unified interface. InsightBoard enables practitioners to jointly inspect training dynamics, performance metrics, and subgroup disparities through linked multi-view plots, correlation analysis, and standard group fairness indicators computed over user-defined slices. Through case studies with YOLOX on the BDD100k dataset, we demonstrate that models achieving strong aggregate performance can still exhibit substantial demographic and environmental disparities that remain hidden under conventional monitoring. By making fairness diagnostics available during training, InsightBoard supports earlier, more informed model inspection without modifying existing training pipelines or introducing additional data stores.

Building general-purpose reasoning models using reinforcement learning with verifiable rewards (RLVR) across diverse domains has been widely adopted by frontier open-weight models. However, their training recipes and domain mixtures are often not disclosed. Joint optimization across domains poses significant challenges: domains vary widely in rollout length, problem difficulty and sample efficiency. Further, models with long chain-of-thought traces increase inference cost and latency, making efficiency critical for practical deployment. We present Apriel-1.5-OpenReasoner, trained with a fully reproducible multi-domain RL post-training recipe on Apriel-Base, a 15B-parameter open-weight LLM, across five domains using public datasets: mathematics, code generation, instruction following, logical puzzles and function calling. We introduce an adaptive domain sampling mechanism that preserves target domain ratios despite heterogeneous rollout dynamics, and a difficulty-aware extension of the standard length penalty that, with no additional training overhead, encourages longer reasoning for difficult problems and shorter traces for easy ones. Trained with a strict 16K-token output budget, Apriel-1.5-OpenReasoner generalizes to 32K tokens at inference and improves over Apriel-Base on AIME 2025, GPQA, MMLU-Pro, and LiveCodeBench while producing 30-50% shorter reasoning traces. It matches strong open-weight models of similar size at lower token cost, thereby pushing the Pareto frontier of accuracy versus token budget.

Quality inspection in smart manufacturing requires identifying intrinsic material and surface properties beyond visible geometry, yet vision-only methods remain vulnerable to occlusion and reflection. We propose VitaTouch, a property-aware vision-tactile-language model for material-property inference and natural-language attribute description. VitaTouch uses modality-specific encoders and a dual Q-Former to extract language-relevant visual and tactile features, which are compressed into prefix tokens for a large language model. We align each modality with text and explicitly couple vision and touch through contrastive learning. We also construct VitaSet, a multimodal dataset with 186 objects, 52k images, and 5.1k human-verified instruction-answer pairs. VitaTouch achieves the best performance on HCT and the overall TVL benchmark, while remaining competitive on SSVTP. On VitaSet, it reaches 88.89% hardness accuracy, 75.13% roughness accuracy, and 54.81% descriptor recall; the material-description task further achieves a peak semantic similarity of 0.9009. With LoRA-based fine-tuning, VitaTouch attains 100.0%, 96.0%, and 92.0% accuracy for 2-, 3-, and 5-category defect recognition, respectively, and delivers 94.0% closed-loop recognition accuracy and 94.0% end-to-end sorting success in 100 laboratory robotic trials. More details are available at the project page: https://vitatouch.github.io/

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

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

Machine learning, especially physics-informed neural networks (PINNs) and their neural network variants, has been widely used to solve problems involving partial differential equations (PDEs). The successful deployment of such methods beyond academic research remains limited. For example, PINN methods primarily consider discrete point-to-point fitting and fail to account for the potential properties of real solutions. The adoption of continuous activation functions in these approaches leads to local characteristics that align with the equation solutions while resulting in poor extensibility and robustness. A general explicit network (GEN) that implements point-to-function PDE solving is proposed in this paper. The "function" component can be constructed based on our prior knowledge of the original PDEs through corresponding basis functions for fitting. The experimental results demonstrate that this approach enables solutions with high robustness and strong extensibility to be obtained.

The dependence of mechanical properties on microscopic interactions remains a central problem in the physics of disordered solids near the jamming transition. We numerically and theoretically investigate the mechanical response of jammed cohesive granular materials using discrete element simulations and effective medium theory (EMT). We find that the shear modulus exhibits pronounced hysteresis under compression and decompression, even though the interparticle force law itself is strictly history-independent. While such hysteresis disappears for purely repulsive particles when mechanical properties are characterized in terms of pressure, it persists in cohesive packings, indicating that pressure is not a unique state variable for cohesive particles. Extending EMT to cohesive interactions, we show that the functional form of the shear modulus remains the same for both repulsive and cohesive particles, but that attractive interactions violate marginal stability. The resulting deviation from marginal stability generates excess rigidity, as predicted by a scaling relation. This prediction is quantitatively verified by numerical simulations and explains the persistent hysteresis in cohesive packings.

We find that reinforcement exponentially reduces computation time of the quantum search problem from $\sqrt{D}$ to $\ln D$ in a $D$-dimensional system. Therefor, a reinforced quantum search is expected to exhibit an exponentially larger noise threshold compared to a standard search algorithm in a noisy environment. We use numerical simulations to characterize the level of noise tolerance via reinforcement in the presence of both coherent and incoherent noise, considering a system of $N$ qubits and a single $D$-level (qudit) system. Our results show that reinforcement significantly enhances the algorithm's success probability and improves the scaling of its computation time with system size. These findings indicate that reinforcement offers a promising strategy for error mitigation, especially when a precise noise model is unavailable.

A two dimensional ferromagnetic XY model with its bound vortex-antivortex dominated quasi long range ordered phase at low temperatures is a long standing as well as well studied problem of interest in the field of condensed matter. We conduct a detailed Monte Carlo study of such model with rather unexplored extensions where additional anisotropic exchange coupling and Dzyaloshinskii-Moriya interactions (DMI) together affect the Kosterlitz-Thoulass (KT) transition in presence/absence of symmetry breaking fields. Without DMI, the exchange term promotes collinear (ferromagnetic) order, whereas the DMI term induces spin cantings. By tuning anisotropy upto Ising limit, we document energy, specific-heat, magnetizations as well as helicity modulus and vortex densities for different tempeatures and DMI strength. We also compute the 2nd moment of correlation lengths in order to probe the spatial correlation of the spins. Furthermore, the effect of U(1) symmetry breaking 4-fold and 8-fold symmetric h4 and h8 fields are explored which shows how the double-peaked specific heat profiles changes in presence of DMI. Overall, our findings append many important updates in the low temperature phases of a topological XY ferromagnet when additional DMI and isotropy-breaking exchange and/or field terms are considered and thus providing a practical blueprint for suitably engineering topological spin systems.

Twistronics is a powerful strategy to engineer novel quantum states by controlling the relative orientation between layered materials. Here, we demonstrate that an orthogonally twisted bilayer CrPS$_4$ shows $d$-wave altermagnetism driven purely by structural rotation. Symmetry analysis reveals that the twisted stacking breaks partial translational combined with time-reversal symmetry, leading to a fourfold rotation relation between opposite spin sublattices, enabling altermagnetism. First-principles calculations demonstrate a sizable non-relativistic spin splitting of up to 68 meV around the Fermi level. We further show that the altermagnetic state can be further stabilized through interlayer compression and modulation of the on-site Coulomb interaction. The resulting band structure exhibits pronounced spin-dependent anisotropy, enabling efficient spin to charge conversion reaching $\sim$50% near the Fermi level and sizable giant magnetoresistance. These results establish twisted CrPS$_4$ as a realistic platform for altermagnetism and highlights twistronics as a versatile route for advanced spintronics applications.

Recently, studies have explored the statistics of matrix elements of local operators in the Lieb-Liniger model. It was found that the probability distribution function for off-diagonal matrix elements $\langle \boldsymbolμ|\mathcal{O}|\boldsymbolλ \rangle$ within the same macro-state is well described by the Fréchet distributions. This represents a significant development for the Eigenstate Thermalization Hypothesis (ETH). In this paper, we investigate a similar phenomenon in another solvable model: the disorder-free Sachdev-Ye-Kitaev (SYK) model. The Hamiltonian of this model consists of 4-body interactions of Majorana fermions. Unlike the conventional SYK model, the coupling strengths in this model are fixed to a constant, earning it the name ``disorder-free.'' We evaluate the matrix elements of operators constructed from products of $n$ Majorana fermions: $\mathcal{O} = χ_{a_1}χ_{a_2}\ldots χ_{a_n}$. For a general choice of indices and $n \geq 4$, we find that the statistics of the off-diagonal matrix elements are well-fitted by a generalized inverse Gaussian distribution rather than Fréchet distributions.

Understanding the realization of thermal equilibrium through the thermalization process in a many-body system is a fundamental and complex scientific question, bridging thermodynamics and classical dynamics and connecting to a host of physical phenomena, such as mechanical instabilities in a thermal environment. In this work, based on the harmonic lattice model, we investigate the thermalization process in both velocity and coordinate spaces, by examining microscopic dynamics on the atomic level. We show the distinct relaxation rates of the transverse and longitudinal components of the velocity, reveal the power law governing the nonlinear proliferation of dominant frequencies, and observe the concurrent rapid proliferations of frequencies and topological defects. We also show that the lattice system's persistent out-of-plane deformations exhibit two-stage fluctuation behaviors, characterized by distinct power laws of fractional exponents and associated with the broken up-down symmetry. This work demonstrates the rich dynamics underlying the thermalization process, and advances our understanding on the dynamical adaptations of many-body systems to external disturbances.

While crystalline materials with glass-like thermal conductivity are fundamentally intriguing, structurally triggering the transition from propagating to diffusive heat transport within a single framework remains a formidable challenge. Here, using extensive machine learning molecular dynamics, we demonstrate a fundamental thermal transport crossover in metal-organic frameworks. We reveal that grafting flexible side chains onto a pristine MOF backbone acts as a structural switch, strongly reducing the thermal conductivity by $\sim$70% (from $\sim 0.7$ to $\sim 0.2\ \text{W m}^{-1}\text{K}^{-1}$ at 300 K). Crucially, the functionalized derivatives exhibit a drastic transition from a classical Peierls $\sim 1/T$ decay to an anomalous, temperature-independent glass-like plateau. Reciprocal- and real-space analyses reveal the microscopic origins: the side chains act as built-in local resonators that trap acoustic energy via strong low-frequency resonant hybridization, while simultaneously inducing extreme steric crowding. Consequently, the heat-carrying phonon modes become critically damped, with their mean free paths strictly confined to the nanometer scale and their lifetimes collapsing to the Ioffe-Regel limit. This work establishes a highly programmable molecular engineering strategy to dismantle the phonon gas model, forcing crystalline frameworks into an extreme diffusive transport regime.

Lucente et al. proved that no time-irreversibility measure can detect departure from equilibrium in a scalar Gaussian time series from a linear system. We show that a second observed channel sharing the same hidden driver overcomes this impossibility: the cross-spectral block, structurally inaccessible to any single-channel measure, provides qualitatively new detectability. Under the diagonal null hypothesis, the cross-spectral detectability coefficient $\Scross$ (the leading quartic-order cross contribution) is \emph{exactly} independent of the observed timescales -- a cancellation governed solely by hidden-mode parameters -- and remains strictly positive at exact timescale coalescence, where all single-channel measures vanish. The mechanism is geometric: the cross spectrum occupies the off-diagonal subspace of the spectral matrix, orthogonal to any diagonal null and therefore invisible in any single-channel reduction. For the one-way coupled Ornstein--Uhlenbeck counterpart, the entropy production rate (EPR) satisfies $\EPRtot=α_2λ^2$ exactly; under this coupling geometry, $\Scross>0$ certifies $\EPRtot>0$, linking observable cross-spectral structure to full-system dissipation via $\EPRtot^{\,2}\propto\Scross$. Finite-sample simulations predict a quantitative detection-threshold split testable with dual colloidal probes and multisite climate stations.

In this work, we explore how the geometry and topology of the underlying manifold shape the synchronization phase transition of a system. To do so, we extend the Kuramoto-Sakaguchi model from spheres to compact, connected, orientable, and homogeneous Riemannian manifolds of arbitrary dimension. Starting from the mean-field kinetic equation on the manifold, we derive a local response equation for the order parameter near the incoherent state and separate the geometric and topological contributions to the phase transition out of the incoherent state. The manifold geometry determines a coefficient $κ\left(M\right)$ to control the critical coupling for the linear loss of stability of the incoherent state. The manifold topology constrains the cubic term of the response equation through the Euler characteristic $χ\left(M\right)$. Under a local sign condition on the cubic term, topology does not allow a generic continuous or tricritical synchronization phase transition to occur when $χ\left(M\right)\neq 0$, and it imposes a non-zero net defect charge on the incipient ordered texture. When an additional local stabilization condition holds in that nonzero-Euler class, topology further selects a discontinuous phase transition. When $χ\left(M\right)=0$, topology does not impose that obstruction, so continuous, discontinuous, and tricritical local branches are all allowed. We verify these findings on representative families including hyperspheres, equal even-sphere products, complex Grassmannians, complex projective spaces, flat tori, real Stiefel manifolds, rotation groups, and unitary groups. Our framework recovers the classical hyperspherical parity law and extends it to a broad class of non-spherical state spaces.

We theoretically investigate spin-dependent transport in a TMD-based vertical spin valve, taking WSe$_2$ as a representative example. Using effective Hamiltonians for the heterostructure and the Landauer formula, we derive the transmission and reflection coefficients within a transfer-matrix approach. The calculated magnetoresistance shows an oscillatory dependence on the WSe$_2$ thickness when the Fermi level is tuned near the valence-band maximum. The effects of gate voltage and exchange fields on the magnetoresistance are further analyzed. We also identify a Fabry-Pérot-like interference contribution to the magnetoresistance, which can enhance or even induce negative magnetoresistance in certain thickness regimes. Our results provide a qualitative understanding of the negative magnetoresistance observed in WSe$_2$-based spin valves and may offer useful insights for the design of tunable spintronic devices.

We studied double-layer WSe2 stacked on opposite sides of thin layers of hexagonal Boron nitride with different densities of electrons and holes. For a fixed hole density, the Coulomb drag resistance is found to exhibit plateaus approximately equal to $-h/(4e^2)$ and $-h/(2e^2)$ as the electron density is changed. When the number of electrons is equal to the number of holes, an exciton solid forms whose transport of quantum edge defects gives rise to the drag resistance. When the electron and hole densities are different, the excess electrons form a solid embedded in the exciton solid. The Coulomb drag resistance of the exciton solid comes from the one-dimensional transport of the two lowest energy channels of quantum edge vacancy-interstitial pairs. This corresponds to the first plateau. With the embedded solid, one of these channels is blocked. This corresponds to the second plateau. Transport experiments in the Corbino geometry with no edges and extra heavier holes were carried out. The plateaus disappeared. Three peaks in the resistance at different hole densities were observed. We interpret that the three peaks correspond to the commensurate exciton and two classes of hole solids. We performed phonon calculations of these states and found that the stability of these exciton-based quantum solids shows good agreement with experiment. Our results establish classes of extreme quantum solid states, opening additional avenues for the study of strongly correlated quantum transport phenomena involving quantum defect states.

Understanding thermal fluctuations yields insights into a wide range of behaviors in many-body systems. In this work, we analyze the dynamical adaptation of two-dimensional crystalline lattice system under harmonic interaction in response to the intricate interplay of thermal agitation and mechanical stress by developing the characteristic stress-absorbing quadrupole structures and stress-releasing fold structures. These thermally driven stress regulator structures serve as a tangible embodiment of thermal fluctuations, offering a unique perspective on the characterization and manipulation of the elusive fluctuations. Specifically, we reveal the stretch-driven alignment and linear accumulation of quadrupoles, characterize the formation and proliferation of folds, and present the phase diagram of the dynamical states defined by these characteristic structures. This work demonstrates the promising avenue of re-examining classical mechanical systems subject to thermal agitation, which is of fundamental physical interest and has potential practical significance in the design of mechanical devices in thermal environments.

In the context of chirality induced spin selectivity effect, it has been argued that a chiral molecule when adsorbed on a metal facilitates the formation of a local spin moment at the interface between the metal and molecule, given a strong spin-orbit coupling in the metal. The possibility for such spin moment formation is analyzed in terms of general arguments and effective modeling of a pertinent set-up. The conclusion from this analysis is that a strong spin-orbit coupling in the metal does not provide a sufficient mechanism to sustain a stabilized spin moment at the interface. It is, moreover, shown that an electron flux in to or out from the molecule does not provide conditions for a spin moment formation, regardless of whether the flux is spin-polarized or not.

In this study, we investigate the relationship between the one-dimensional (1D) Kardar-Parisi-Zhang (KPZ) equation and the stochastic Loewner equation (SLE), which is a one parameter family of the conformal mappings involving stochasticity. The author shows the correspondence between 1D KPZ equation with height function $h(x,t)=(3t^2x+x^3)/6t$ and Loewner equation driven by a nonlinear stochastic process, wherein the 1D dynamics of interface growth is characterized by Loewner entropy $S_{Loew}\simeq-\ln{t/κ}$. These results were numerically verified with discussions in relation to the universality in non-equilibrium statistical physics.

Anisotropic magnetic colloids with permanent dipole moments exhibit rich field-responsive behavior arising from the interplay between particle geometry, dipolar interactions, and external driving. Modeling these systems remains challenging due to the long-range nature of dipolar forces, geometric anisotropy, dipole--particle misalignment, and the complexity of implementing anisotropic steric interactions. This review discusses particle-based numerical strategies to model such systems, including single-site, multi-bead, shifted-dipole, and multicore representations. We analyze how different levels of description capture key physical mechanisms, from steric constraints and directional binding to internal magnetic structure and nonequilibrium dynamics. Particular emphasis is placed on dipole--particle misalignment as a control parameter that strongly affects interaction landscapes and self-assembly pathways. We also highlight recent machine learning approaches as emerging tools to construct effective interaction potentials and accelerate simulations. By comparing the main methodologies and their limitations, this review outlines current challenges and perspectives toward more predictive and efficient modeling of anisotropic magnetic colloids.

The statistics of gaps between quantum energy levels is a hallmark criterion in quantum chaos and quantum integrability studies. The relevant distributions corresponding to exactly integrable vs. fully chaotic systems are universal and described by the Poisson vs. Wigner-Dyson curves. In the transitional regime between integrability and chaos, the distributions are much less universal and have not been understood quantitatively until now. We point out that the relevant statistics that controls these distributions is that of the matrix elements of the nonintegrable perturbation Hamiltonian in the energy eigenbasis of the unperturbed integrable system. With this insight, we formulate a simple random matrix ensemble that correctly reproduces the level spacing distributions in a variety of test systems. For the distribution of matrix elements appearing in our construction, we furthermore discover surprising universal features: across a variety of physical systems with diverse degrees of freedom, these distributions are dominated by simple power laws.

The mechanical and rheological properties of jammed packings of frictionless particles under shear strain remain not fully understood, even when the strain amplitude is very small and well below the yielding threshold. Systems above the jamming transition point $φ_J$ are known to display two anomalous mechanical behaviors with respect to the driving frequency $ω$ (or time $t$) and the strain amplitude $γ$. In the linear-response regime ($γ\to 0$), the complex modulus exhibits an algebraic scaling, $G(ω)\simω^{1/2}$ (or $G(t)\sim t^{-1/2}$ in the time representation). In contrast, in the quasi-static limit ($ω\to 0$), the modulus shows the nonlinear behavior, $G(γ)\simγ^{-1/2}$, a phenomenon referred to as softening. The ranges of $ω$ and $γ$ over which these algebraic scalings hold broaden as $φ_J$ is approached from above, whereas both $G(ω)$ and $G(γ)$ vanish for $φ< φ_J$. In this study, we investigate the mechanical response in the regime where these two anomalies coexist in the vicinity of $φ_J$. To this end, we perform numerical analyses using two rheological protocols: oscillatory shear and transient stress relaxation. Our results demonstrate that the mechanical responses are not simply described as a superposition of the two algebraic relaxations and instead exhibit rich nonlinear viscoelastic behavior both above and even below $φ_J$.

The amplitude of the Hall response of electrons can be understood without invoking interactions. Most theories of the phonon thermal Hall effect have likewise opted for a non-interacting picture. Here, we challenge this approach. Our study of WS$_2$, a transition metal dichalcogenide (TMD) insulator, finds that longitudinal, $κ_{xx}$, and transverse, $κ_{xy}$, thermal conductivities peak at almost the same temperature. Their ratio obeys an upper bound, as in other insulators. We then compare transverse thermal transport in a phonon gas and in a molecular gas. In the latter, the Senftleben-Beenakker effect is driven by the competition between molecular collisions and applied magnetic field in setting the distribution of molecular angular momenta. An off-diagonal transport response arises thanks to interactions between non-spherical particles, which do not need to be chiral. By analogy, we argue that in a phonon gas, magnetic field will influence phonon-phonon interactions, and generates a transverse thermal \emph{resistivity}, whose order of magnitude can be accounted for by invoking a Berry force on the drift velocity of the nuclei in the presence of a finite heat. This simple picture gives a reasonable account of the experimentally measured transverse thermal resistivity of seven different crystalline insulators.

We study a discrete-time quantum walk in presence of a detector at $x_D$ initially. The detector here is repeatedly removed after a span of $t_R$, the removal time, and reinserted at random locations. Two relocation rules are considered here: In Model~1, the detector is reinserted at any site beyond $x_D$, while in Model~2, reinsertion is done within a restricted window around the position of the detector at that time. Both variants behave like Semi Infinite Walk (SIW) for large $t_R$, where the detector behaves effectively as a fixed boundary. However, in the rapid-relocation regime, i.e., when $t_R$ is small, the behaviours are different. Model~1 permits greater spreading due to unrestricted reinsertion, which is different from Model~2. The time evolution of occupation probability ratio of our walker to that of an infinite walker at $x_D$, i.e., $f(x_D,t)/f_\infty(x_D,t)$, initially show the feature of a SIW upto $t=t_R$, then show some oscillatory behaviour and finally reach a saturation value for both the models. The ratio enhancing under certain conditions of $x_D$ and $t_R$, is a purely quantum mechanical effect. The saturation ratio shows a crossover behavior below and above a removal time $t_R^*$. At sites $x \neq x_D$ the occupation probablity ratios at a certain time reveals that for small $t_R$, the behaviours of the two models are drastically different from each other, as well as from Semi Infinite Walk (SIW), Quenched Quantum Walk (QQW) and Moving Detector Quantum Walk (MDQW). The correlation ratios of the two models with that of Infinite Walk (IW) show interesting time dependence for sites to the left or right of the initial detector position $x_D$.

Supercooled liquids are metastable states realized by suppressing crystallization below the melting temperature. While it is well established that their dynamics slow down dramatically and become spatially heterogeneous upon cooling, the microscopic origin of these nontrivial glassy phenomena remains a matter of active debate. In the present study, by means of molecular dynamics simulations, we first demonstrate that nontrivial slow dynamics, such as structural relaxation and dynamical heterogeneity, can be consistently described within a zero-temperature avalanche criticality picture. Since this finding suggests that the potential energy landscape plays a crucial role in determining the dynamics, we further quantify the potential energy landscape from three distinct perspectives. Based on these analyses, we propose a potential-energy-landscape picture of avalanche criticality that is consistent with various previous studies. Our proposed picture explains in a unified manner previously unexplained observations near the mode-coupling transition, such as the saturation of the dynamical susceptibility and the localization of unstable modes in saddle configurations.

Antiferromagnets with vanishing net magnetization are naturally expected to host higher-order magnetic multipole moments. Understanding and utilizing the multipole degrees of freedom are imperative for novel conceptual designs and applications unique to antiferromagnets. However, a universal, quantitative definition of magnetic multipole moments of antiferromagnetic materials is currently lacking. In this work we provide a unified description of arbitrary-order spin magnetic multipole moments (SM$^3$) of antiferromagnets by introducing a nonlocal spin density in macroscopic Maxwell equations. The formalism makes it transparent how SM$^3$ calculated for translationally invariant bulk systems corresponds to experimental observables when translation symmetry is broken. Through the nonlocal spin density calculated from first principles, we propose a robust scheme to extract arbitrary-order SM$^3$ through symmetry-constrained fitting at long wavelengths. Using this approach, we have calculated SM$^3$ of a few representative antiferromagnets, including $α$-$\rm Fe_2O_3$, Mn$_3$Sn, and Mn$_3$NiN. Moreover, we clarify the role of spin-orbit coupling (SOC) in SM$^3$, especially in the weak SOC limit where clean predictions can be made based on symmetry principles. Our work paves the way for systematically investigating multipolar order parameters of unconventional magnetic materials.

In supercooled liquids, mesoscale mobile and immobile domains are ubiquitously observed, a phenomenon known as dynamical heterogeneity. Extensive studies have established that the characteristic size of these domains grows upon cooling and exhibits system-size dependence. However, the physical origin of this domain growth remains a matter of active debate. In this work, using molecular simulations, we demonstrate that the temperature and system-size dependence of dynamical heterogeneity can be explained within a zero-temperature avalanche criticality picture.

Voltage-gated ion channels are essential for propagating signals in neurons. Each channel senses the local membrane potential created by nearby ions. Fluctuations in these ions introduce two fundamental noise sources: (i) shot noise, from the discreteness of ionic charge, and (ii) Johnson-Nyquist noise, from long-wavelength thermal fluctuations of the electric field. We show that, for an individual channel, shot noise dominates and sets an intrinsic limit to voltage sensing. On the $10$ $μ$s timescales relevant to channel gating, this limit corresponds to an accuracy of about $10$ mV -- close to measured channel sensitivities. When signals from many channels are aggregated, Johnson-Nyquist noise eventually overtakes shot noise and bounds the total information that can be sensed from the environment. This transition occurs at an ion channel density of $< 1$ channel/$μ$m$^2$ for slow signals and around $10^2-10^4$ channels/$μ$m$^2$ for signals with $10$ $μ$s timescales, both of which are within the range of experimentally-measured densities for somas and axon initial segments, respectively. These results provide design principles for single-channel architecture and collective sensing and suggest that neuronal computation is ultimately constrained by thermal fluctuations.

We propose the use of an in-house air-filled elasto-bubble as a novel subwavelength acoustic resonator. Building on the well-established physics of gas bubbles, and incorporating an elastic shell, we experimentally demonstrate that these elasto-bubbles retain the key acoustic properties of classical bubbles. Indeed, through experiments performed in an impedance tube, we observe a strong acoustic response and efficient transmission reduction, in good agreement with the theory considering a layered bubble introduced by Alekseev and Rybak (1999). Finally, elasto-bubbles offer a simple and effective route to tunable acoustic resonances through independent control of their radius and shell thickness.

Periodic (Floquet) driving enables Hamiltonian engineering and nonequilibrium phases, but interacting systems eventually heat by absorbing energy from the drive. Disorder can greatly delay this process, yielding long-lived prethermal plateaus. Here we show that this protection can fail when pulse-train control introduces a second driving frequency and when the disorder fluctuates. Using a natural-abundance 13C nuclear-spin network in diamond, we observe sharp peaks in the late-time heating rate at the double- and triple-spin-flip resonance conditions predicted by bimodal Floquet interference, and track their evolution with drive frequency. A switching-noise model attributes the resonant absorption to stochastic electron-spin dynamics that intermittently tune rare nuclear clusters into multi-photon resonance. Our results reveal a resonance-activated limit for disorder-stabilized Floquet phases and suggest new routes to DC-field quantum sensing based on an abrupt breakdown of prethermalization.

We report the details of construction and testing of a Quantum Twisting Microscope, a recently developed scanning probe instrument that enables twist angle dependent electronic measurements on layered materials. Our implementation is based on a commercial atomic force microscope whose open geometry beneath the scan head allows integration of the rotation and translation stages required for QTM operation. We describe the complete fabrication process including tip preparation by focused ion beam deposition and graphite transfer, custom stage assembly with integrated rotation capability, and multistep alignment procedures. To validate the instrument, we perform conductance measurements between graphite layers as a function of twist angle, observing clear 60 degree periodicity consistent with the hexagonal lattice symmetry and conductance enhancements near the commensurate twist angles of 21.8 and 38.2 degrees. These results confirm the instruments ability to resolve crystallographic twist angle dependent transport features. By providing detailed construction and operational guidelines, we aim to make QTM technology accessible to research groups with standard AFM infrastructure, enabling investigations of twist angle dependent phenomena in van der Waals materials, complex oxide heterostructures and chiral systems.

Yield-stress fluids are ubiquitous and encountered in diverse fields ranging from natural muddy flows to industrial applications such as secondary battery electrode slurries and direct ink writing. Despite the proposal of various constitutive equations, few models have been shown to successfully predict both steady and transient rheological behaviors in yield-stress fluids. In this study, a constitutive equation is hereby proposed, offering a comprehensive description of the rheological characteristics observed in simple yield-stress fluids, excluding thixotropy, such as the Carbopol dispersion. The constitutive equation is derived from a Zener-type viscoelastic solid element combined with an additional linear dashpot connected in parallel, together with a nonlinear viscosity model, a flow rule, an evolution equation for the back stress, and the Kroner-Lee decomposition. This combination satisfies the principle of material frame invariance. The proposed model successfully reproduces the rheological characteristics qualitatively in a manner consistent with experimental observations conducted during start-up shear, creep, and stress relaxation tests. In particular, the present viscoelastic solid-based constitutive equation is shown to accurately predict stress overshoot during start-up shear. Importantly, the overshoot is found to originate from a homogeneous mechanism in which normal stress difference enhances the stress invariant and thereby accelerates the plastic response, rather than from isotropic hardening or spatially heterogeneous microstructural evolution. This study is expected to facilitate a deeper understanding of the intricate dynamics governing the flow of yield-stress fluids.

The intermolecular interaction between a water molecule and the electrons in aromatic π systems--the water-π bond--lies at the heart of many chemical processes, yet its properties remain challenging to measure experimentally and model computationally. Infrared spectroscopy of pyrene anions hydrated by a single water molecule reveals vibrational and electronic motions that are often hidden in condensed phase measurements. Results from new machine-learning approaches to potentials and dipole moments show that the electron dynamics of the aromatic π cloud quench signals from some of water's vibrations and amplify others. The observed interplay between electronic and vibrational motions has general implications for modeling intermolecular interactions between water and aromatic systems in clusters, solutions, and at interfaces.

Coherent interfaces between microwave-frequency quantum systems and low-loss optical links are essential for quantum networks. However, existing microwave-optical transducers often trade conversion efficiency against added noise, bandwidth, and device integrability. Here, we demonstrate coherent microwave-to-optical transduction based on magnon-exciton coupling in the layered antiferromagnet CrSBr. Driving the antiferromagnetic resonance with microwave signals imprints coherent modulation on a reflected optical probe, generating optical sidebands that are resonantly enhanced near excitonic transitions. While prior magnon-based approaches to microwave-to-optical transduction have typically relied on intrinsically weak off-resonant magneto-optical effects (e.g., Faraday rotation), our scheme exploits strong light-matter interactions at exciton resonances. Even in a bulk crystal without cavity enhancement, we observe coherent conversion over an intrinsically broadband window of ~ 300 MHz. We further show that multiple exciton-polariton resonances inherit the magnon-coupled response, suggesting a route to broaden the usable optical detuning range and to mitigate optical dissipation. Our results establish magnon-coupled excitons in layered magnets as a scalable platform for broadband microwave-optical interfaces, with pathways to higher cooperativity via reduced magnetic volume and cavity integration.

Results from numerical simulations of polymers under shear flow are compared with predictions for shear banding based on a model coupling Rolie-Poly-like tube dynamics to the entanglement dynamics mediated by Convected Constraint Release (CCR). CCR is controlled by a parameter $β$, whose dependence on the bending stiffness $k_θ$ is calculated from the simulations. The model predicts shear banding for polymers whose equilibrium entanglement number $Z$ exceeds a critical value $Z_{c}$ that depends on $β$. The simulations are in semi-quantitative agreement with the model, with deviations that are attributed to approximations inherent in the model, and the inadequacy of tube models to describe partially-disentangled liquids in strong flow. These results may help determine which physical polymers could undergo shear banding.

The ultrafast nonlinear optical response of molecular ensembles is fundamentally altered under strong light-matter coupling. To rigorously isolate the genuine many-body contributions, an exact time-domain field-subtraction protocol is developed within a fully non-perturbative Maxwell-Liouville framework explicitly incorporating the two-exciton manifold in real space and time. This approach reveals that while collective cavity delocalization drives the macroscopic nonlinear signal toward a severe harmonic cancellation (an effect termed "spectral starvation"), intrinsic many-body molecular interactions robustly resurrect genuine polaritonic double-quantum coherences (DQCs). This many-body resurrection is governed by a universal two-photon matching rule, $Δ_B + 4J = Ω_R$, linking molecular anharmonicity ($Δ_B$) to the macroscopic Rabi splitting ($Ω_R$) and excitonic coupling ($J$). Crucially, this dictates that J-aggregates ($J < 0$) uniquely isolate the resonant many-body state below the dense two-exciton scattering continuum, protecting the macroscopic coherence from spatial fragmentation. This predictive framework establishes a direct phase diagram to engineer and protect optical nonlinearities across diverse strongly coupled platforms.

Using molecular dynamics simulations, we investigate the crystallization pathways of two exemplary systems that form the same complex crystal structure but differ fundamentally in the nature of their particle interactions. One system is composed of point particles interacting via an isotropic pair potential characteristic of metallic compounds, while the other system contains hard polyhedra whose interactions arise from emergent entropic forces. Despite the stark difference in the origins of the particle interactions, we find that both systems are polymorphic and share the same crystal polymorphs. Moreover, the two systems follow the same multistep crystallization pathways, and by examining the complex crystallization pathways on the single particle level, we find that the local structure evolution of the two systems is also similar. By mapping the hard particle system's interaction to an effective pairwise potential, we find that such resemblance arises from the particle interactions being effectively similar.

Soft materials such as rubbers, hydrogels, and biological tissues undergo damage in the form of stiffness degradation without apparent changes in their stress-free geometry. Accurate simulation of this behavior is critical in applications ranging from soft robotics to the design of medical devices, yet two persistent challenges are the difficulty of constructing flexible, thermodynamically consistent constitutive models, and the mesh dependence of finite element solutions caused by strain softening. Here we address both challenges simultaneously by combining physics-augmented neural network constitutive models with a gradient-enhanced damage formulation implemented within the differentiable finite element framework JAX-FEM. The elastic strain energy and the damage yield function are each parameterized by input-convex neural networks (ICNNs), which enforce polyconvexity and satisfaction of the Clausius--Duhem inequality by design. The gradient-enhanced formulation introduces a non-local damage field governed by an additional partial differential equation, regularizing the spatial distribution of damage and eliminating mesh dependence. The implementation is validated through local stress-strain fits, single-element parametric studies, a mesh and solution strategy study for a uniform deformation case, and a notched plate simulation. The results demonstrate that the proposed framework enables flexible, data-driven, mesh-independent damage simulation for a broad class of soft materials. We anticipate that the open-source implementation will lower the barrier to adopting physics-augmented neural network constitutive models.

We present an approach for entangling spin qubits via capacitive coupling mediated by an ac electric field-driven multielectron mediator quantum dot. To illustrate this method, we consider the case of a driven two-electron dot that mediates entanglement between resonant exchange qubits defined in three-electron triple quantum dots, which enable direct capacitive coupling and interaction with microwave fields via intrinsic spin-charge mixing. The method can also be applied to other types of spin qubits that can be coupled capacitively. We show that this approach leads to rapid, single-pulse universal entangling gates for resonant exchange qubits that are activated via the drive on the mediator dot. Unlike conventional tunneling-based two-qubit gates between exchange-only qubits, the capacitive interaction-based gates we describe do not require an extensive sequence of pulses to mitigate leakage. We describe how this drive-activated local entangling approach can be integrated with the driven sideband-based long-range approach for cavity-mediated entangling gates developed in our previous work in order to enable modularity for spin-based quantum information processing.

Non-reciprocity and geometric frustration enable many-body systems to avoid crystalline order and instead exhibit complex, liquid-like behavior. Here we show that their interplay is richer than the sum of its parts, leading to surprising structural and dynamical phenomena. In our minimal model, two copies of Ising gauge theory are non-reciprocally coupled in a way that crucially preserves a local $\mathbb{Z}_2$ symmetry. We discover that the combined Wilson loop observable of the two copies exhibits linear asymptotic scaling, with a quasiparticle-pair confinement length tuned by the strength of the non-reciprocal coupling. Key dynamical features are revealed in the behavior of individual deconfined excitations due to strong interactions induced by the non-reciprocity, leading to motion on a critical percolation cluster that follows a self-avoiding trail. Mapping from this quasiparticle dynamics onto the magnetic noise spectrum, we discover that non-reciprocity tunes topological logarithmic contributions and causes long-lived metastable states due to quasiparticle trapping. Our work opens the way for broader investigations of geometrically frustrated non-reciprocity.

The Boltzmann-Loschmidt dispute of 1876 questioned the possibility of a statistical irreversible description by time reversible classical equations of motion of atoms. Here we show analytically and numerically that the quantum chaos diffusion of cold atoms, or ions, in a harmonic trap and pulsed optical lattice can be inverted back in time with up to 100\% efficiency. This is in sharp contrast to classical evolution where exponentially small errors break time reversibility. We argue that the existing experimental skills allow highlighting the Boltzmann-Loschmidt dispute from a quantum perspective.

We simulate the dynamics of pelagic \emph{Sargassum} rafts as systems of finite-size floating particles, governed by a Maxey--Riley law with nonlinear elastic interactions. Using surface ocean currents and wind data from reanalysis systems for clump transport, we computed trajectories within a domain covering the tropical and subtropical north Atlantic. The subsequent motion is reduced using Ulam's discretization method into a time-inhomogeneous Markov chain that simulates a background \emph{Sargassum} concentration. Bayesian inversion, combined with nonautonomous transition path theory, was used to infer the origin of the first significant recorded bloom in the tropical North Atlantic, which unfolded in April 2011. Both methodologies independently identified the bloom's origin as near the West African coast, up to two years before it was detectable via satellite imagery on the basin's western side. This finding supports anecdotal evidence of \emph{Sargassum} strandings on the Ghanaian coast in 2009. Moreover, it correlates with unusual environmental conditions -- such as increased nutrient loads from significant upwelling linked to a pronounced Dakar Niña and Saharan dust deposition -- that promote bloom proliferation. Additionally, it aligns with the observation that the species of \emph{Sargassum} in the 2011 bloom differ from those in the Sargasso Sea, which might otherwise be considered a natural origin.

Stochastic resetting has emerged as a powerful mechanism for driving systems into nonequilibrium stationary states with tunable properties. While most existing studies focus on global resetting, where all degrees of freedom are simultaneously reset, recent work has shown that resetting only a subset of degrees of freedom (subsystem resetting) can qualitatively alter collective behavior in interacting many-body systems. In this work, we develop a general theoretical framework for analysing subsystem resetting in Kuramoto-type coupled-oscillator systems. Building on a continued-fraction approach, we derive self-consistent equations for the stationary-state order parameter of the non-reset subsystem, applicable to both noisy and noiseless dynamics and to models with arbitrary interaction harmonics. Using this framework, we systematically investigate how the stationary state and phase transitions depend on the resetting rate, the size of the reset subsystem, and the reset configuration. We show that subsystem resetting can shift or even suppress synchronization transitions, and can give rise to nontrivial features such as re-entrant behavior and restructuring of phase boundaries. In specific cases, including the noiseless Kuramoto model with a Lorentzian frequency distribution, our results recover known analytical predictions and extend them to more general settings. These results establish subsystem resetting as a versatile control protocol for engineering collective dynamics in nonequilibrium interacting systems.

We formulate Lagrangian descriptors (LDs) in the path integral framework. Averaging the classical LD over fluctuations about extremal trajectories defines a quantum LD that incorporates quantum effects. Invariant manifolds, which sharply organize classical transport, become finite-width phase space structures under quantum fluctuations, and their overlap provides a geometric mechanism consistent with tunneling as fluctuation-induced delocalization of transport barriers. We demonstrate this approach for the Hamiltonian saddle, where path integral sampling reveals manifold broadening and barrier penetration. This establishes a geometric framework for studying phase space transport and tunneling beyond the classical regime, while also providing a natural route toward the application of LDs to field theory.

Repeated interactions are ubiquitous and known to promote social behaviour. While research often focuses on cooperation in the Prisoner's Dilemma, experimental evidence suggests repeated interactions also foster fairness. This study addresses a gap in the literature by theoretically modelling the evolution of fairness within a repeated mini-ultimatum game. Specifically, we construct a repeated-game framework where offerers and accepters interact using reactive strategies. We then investigate whether fair reactive strategy pairs are resilient against unfair mutants in a two-species population. By analyzing short-term evolutionary stability via the concept of two-species evolutionary stable strategy, we identify a critical effective game length: below this value, fairness is promoted by offerers and accepters who comply with their partner's past actions. Above this critical value, fairness is maintained by `complier' offerers and fair accepters. We also show that specific reactive strategies effectively facilitate the emergence and sustenance of fairness in long-term mutation-selection dynamics. To this end, we develop a two-population stochastic dynamics model -- a generalization of classical adaptive dynamics -- that accounts for finite population sizes and non-local mutants in the reactive strategy space.

This article presents an optimization method to find the most spatially concentrated basis of a multimode fiber, obtained by minimizing the sum of the spatial spreads of the individual modes over all unitary transformations of a given orthonormal mode set. The resulting modes are the optical analogue of maximally localized Wannier functions in solid-state physics. We apply the method to the Laguerre-Gaussian basis of a graded-index fiber for mode counts ranging from 6 to 55. In all cases, the modes spontaneously organize into concentric rings without any geometric constraint being imposed. The spot sizes and ellipticities evolve from one ring to the next in ways that geometric packing approaches cannot predict. For large mode counts, the optimizer finds solutions where neither the number of spots per ring nor the spots within a given ring follow a regular pattern, indicating that the fully symmetric arrangement is no longer a minimum of the spread functional. A constrained variant of the method enables the optimizer to target any prescribed bundle geometry while quantifying its localization cost, opening a route to physically grounded photonic lantern design.

We investigate the impact of the quantum effects of protons on the electronic structure of high-pressure H$_3$S, a benchmark hydrogen-rich superconductor with a critical temperature ($T_c$) exceeding 200 K. Using Nuclear-Electronic Orbital Density Functional Theory (NEO-DFT), we treat hydrogen nuclei quantum mechanically on the same footing as electrons within a first-principles framework. Our calculations reveal that nuclear quantum effects (NQEs) induce subtle modifications to the electronic band structure and density of states (DOS) near the Fermi energy, including features associated with van Hove singularities. However, the resulting changes in the DOS would increase $T_c$ by only a few percent. On the other hand, calculations of the phonon dispersion with the NEO-DFT method show large changes in the hydrogen-dominated phonons that arise from a stiffening of the S-H bonds due to NQEs. These findings imply that the experimentally observed reduction in $T_c$ upon deuteration arises predominantly from changes in the phonon properties, while NQEs-induced modifications to the electronic structure itself are minimal.

We propose a globally-admissible phenomenological spectral density of the bath for the non-Markovian Brownian motion of an optomechanical resonator, motivated by the near-resonance experimental observation of a non-Ohmic spectrum in [Nat. Commun. 6, 7606 (2015)]. To avoid divergences arising from a naive global extrapolation, we construct this phenomenological bath spectral density that reproduces the observed local-power-law behavior near the mechanical resonance while remaining well defined globally, ensuring the finiteness of the bath-induced renormalizations and quadrature fluctuations of the resonator. The corresponding model of the structured environment produces a nonlocal mechanical susceptibility whose analytic pole structure encodes the observed linewidth. The resulting dissipation kernel exhibits a power-law-modulated exponential decay with transient negativity, signaling strong memory effects. In the weak-coupling regime, the optical readout based on homodyne detection enables near-resonance spectroscopy and, with a calibrated drive on the resonator, permits, in principle, the reconstruction of the full mechanical susceptibility, thereby providing access to both the dissipative and dispersive bath contributions. Our results provide a consistent route from locally-inferred spectral properties to globally-admissible open-system descriptions and establish a framework for probing structured environments in cavity optomechanics.

Data-driven surrogates can replace expensive multiphysics solvers for parametric PDEs, yet building compact, accurate neural operators for three-dimensional problems remains challenging: in Fourier Neural Operators, dense mode-wise spectral channel mixing scales linearly with the number of retained Fourier modes, inflating parameter counts and limiting real-time deployability. We introduce HQ-LP-FNO, a hybrid quantum-classical FNO that replaces a configurable fraction of these dense spectral blocks with a compact, mode-shared variational quantum circuit mixer whose parameter count is independent of the Fourier mode budget. A parameter-matched classical bottleneck control is co-designed to provide a rigorous evaluation framework. Evaluated on three-dimensional surrogate modeling of high-energy laser processing, coupling heat transfer, melt-pool convection, free-surface deformation, and phase change, HQ-LP-FNO reduces trainable parameters by 15.6% relative to a classical baseline while lowering phase-fraction mean absolute error by 26% and relative temperature MAE from 2.89% to 2.56%. A sweep over the quantum-channel budget reveals that a moderate VQC allocation yields the best temperature metrics across all tested configurations, including the fully classical baseline, pointing toward an optimal classical-quantum partitioning. The ablation confirms that mode-shared mixing, naturally implemented by the VQC through its compact circuit structure, is the dominant contributor to these improvements. A noisy-simulator study under backend-calibrated noise from ibm-torino confirms numerical stability of the quantum mixer across the tested shot range. These results demonstrate that VQC-based parameter-efficient spectral mixing can improve neural operator surrogates for complex multiphysics problems and establish a controlled evaluation protocol for hybrid quantum operator learning in practice.

The phrase ``buy a quantum computer'' hides several different procurement problems. An institution may be seeking cloud access for teaching, reserved capacity for research, a local instrument for hardware training, an optimization appliance, or a strategic installation that reshapes facilities, staffing, and budgets. Because these choices differ in purpose, operating burden, and useful lifetime, the decision should be framed as acquisition of \emph{quantum capability} rather than selection of a presumed hardware winner. This manuscript develops a practical procurement framework that distinguishes five capability layers, separates peer-reviewed results from commercial offerings, pricing anchors, and public roadmaps, and compares the main commercial platform families -- superconducting circuits, trapped ions, neutral atoms, quantum annealing, and photonics -- through the lens of institutional fit, access model, and refresh pressure. The main conclusion is that most institutions should begin with the smallest layer of capability that produces repeatable near-term value, builds internal expertise, and preserves strategic flexibility. Large on-premises systems are justified only when mission requirements, site readiness, staffing, governance, and upgrade paths are already clear.

Two-dimensional (2D) ferromagnetic materials with high Curie temperature ($T_{\rm c}$) and large magnetic anisotropy energy (MAE) are critical for nanoscale spintronics but remain rare. We propose, via first-principles calculations, that adsorbing alkali atoms ($A$ = Li, Na, K, Rb, Cs) onto a tetragonal CoSe monolayer transforms it into a series of stable 2D ferromagnetic metals, $A$CoSe, with an in-plane easy axis. Notably, LiCoSe is a half-metal. These functionalized monolayers exhibit dramatically enhanced ferromagnetism compared to the pristine layer, with $T_{\rm c}$ > 300 K and MAE > 800 $μ$eV/Co. The coupled alkali atoms amplify the local magnetic moment of Co ions, reinforce ferromagnetic Ruderman-Kittel-Kasuya-Yosida (RKKY) and superexchange couplings, and concurrently weaken the direct antiferromagnetic exchange between Co ions. Furthermore, tensile strain can further elevate the MAE (via band shifting) and increase $T_{c}$ (by strengthening the nearest-neighbor exchange $J_1$). Among them, NaCoSe exhibits the highest MAE and excellent strain-modulated $T_{c}$, rendering it the most promising candidate material. Our results establish alkali-metal decoration as an effective strategy for realizing 2D ferromagnets with high $T_{\rm c}$ and large MAE in tetragonal lattices.

We study a driven optomechanical cavity with two movable mirrors and an intracavity optical parametric amplifier, focusing on how direct phonon-phonon coupling changes the observed normal-mode spectrum. Although the linearized system supports three hybrid modes, in the absence of direct mechanical coupling one collective mechanical mode contributes only weakly to the optical response, leading to an apparent two-peak structure. We show that finite phonon-phonon coupling mixes this weakly participating mode with the optically visible modes, making a third resonance clearly resolvable. This change appears in the mirror displacement spectrum, the output-field spectrum, and the output quadrature spectra, and becomes more pronounced with increasing coupling strength and drive power. Our results show that direct mechanical coupling provides a useful way to tune spectral visibility and multimode hybridization in two-mirror optomechanical cavity with intracavity parametric amplification.

This paper presents a novel concept for electrically small antenna arrays incorporating chiral parasitic elements of opposite handedness. This configuration mitigates the detrimental effects of electromagnetic mutual coupling, which in conventional arrays causes a 180 degree phase shift between adjacent antenna currents when the element spacing is less than half a wavelength. The proposed approach is experimentally validated using a seven-element monopole ESAA with compact dimensions, specifically below half-wavelength in cross-section and one-sixth to one-fourth of a wavelength in vertical range. The antenna elements are spaced less than one-sixth wavelength apart, ensuring a highly compact footprint. Measurements show a minus 10 dB return loss, a fractional bandwidth of 5 to 15 per cent, and a realised gain of 5 to 9 dBi, along with full 360 degrees of azimuthal beam steering. The results confirm that employing oppositely handed chiral parasitic elements can significantly enhance performance in densely packed, electrically small antenna arrays.

Carrier multiplication (CM) describes a strong charge-carrier interaction process in semiconductor colloidal nanocrystals (NCs), wherein two band-edge excitons are simultaneously created by an absorbed photon with at least twice the bandgap energy (2 Eg). While being fundamentally intriguing, it has been exclusively utilized to enhance the light-to-electricity conversion efficiencies in the photodetector and solar-cell devices. In this report, we have synthesized the core/shell perovskite FAPbI3/NdF3 NCs with a biexciton recombination lifetime of ~3.9 ns, and demonstrated that a CM efficiency of ~25.7% can be achieved under the ~355 nm laser excitation (~2.21 Eg). This CM occurrence leads to a two-fold reduction in the optical gain threshold, as compared to that obtained under the ~640 nm laser excitation (~1.23 Eg). When combined with the single-exciton and zero-threshold optical gain schemes previously developed for semiconductor colloidal NCs, the CM effect introduced here would further mitigate the optical-pumping requirement for the routine operation of continuous-wave lasing.

In this work, we have analyzed the nuclear structure and several prospective decay characteristics of the $^{240-259}$Es$_{99}$ isotopes. For this we use Relativistic Mean Field model (RMF) with NL-SH and NL3* force parameter in an axially deformed oscillator basis. In structural properties, we have analyzed binding energy (B.E.), skin thickness ($r_{np}$) , charge radius ($r_c$), one neutron separation energy ($S_{1n}$), two neutron separation energy ($S_{2n}$), differential variation of two neutron separation energy ($dS_{2n}$), the single particle energy and its variation with quadrupole deformation parameter of Es isotopes. We have also estimated the $α$-decay, $β$-decay and cluster decay half lives of Es isotopes to analyze the shell structure and also to predict the suitable decay mode among them. The $α$-decay half-life periods are calculated using the MUDL and AKRE formulae using both our calculated Q-values and empirically assessable Q-values. In a similar manner, we have computed the half-lives of cluster decay using Universal Decay Law and HOROI formula. A longer decay half-life indicates a shell stabilized parent nucleus, while a small parent half-life suggests the shell stability of the daughter. This study provides us the insights regarding the structural changes with the change in neutron number enabling us to predict shell closures and nuclear stability. We found a shell/sub-shell closure at N = 154 for the NL-SH parameter set. This research aids in our comprehension of Es isotopes' shell structure and decay mechanism.

Spin-wave active ring resonators (SWARRs) based on yttrium iron garnet (YIG) films exhibit rich nonlinear dynamics that make them promising platforms for physical reservoir computing. We present systematic and experimentally simple methods to characterize a SWARR's nonlinear behavior and memory. We first use a third harmonic elimination method to probe the nonlinear response. A drive frequency $f_\mathrm{d}$ is modulated by a square-wave pattern engineered to have a spectral null at $3/T$, which is then applied as input to the SWARR. The power spectra at the output of the YIG delay line allow us to identify five distinct regions within a drive frequency range of $2.15 < f_\text{d} < 2.2\ \text{GHz}$ where nonlinearity was observed as frequency peaks at $f_\mathrm{d} \pm \frac{3}{T}$. The STM duration of the SWARR was estimated to be approximately 300~ns using a modulation pattern derived from the sequency-ordered 8-bit unipolar Walsh family. The nonlinear dynamics of the SWARR were further quantified by decomposing its temporal response to analog Walsh pulses in terms of the input Walsh codewords. The proposed methods of harmonic elimination and Walsh-function decomposition together provide a practical and general framework for the design and optimization of tunable spin-wave reservoir computers.

This paper deals with the mathematical modelling of signal propagation in nerve fibres. Due to the complexity of the processes where electrical, mechanical, and thermal effects are coupled, a phenomenological approach helps to build mathematical models. The ideas of phenomenology are briefly presented, and their application is described. These applications cover the modelling of ion currents (the Hodgkin-Huxley model), temperature effects, and inductance. This means that the ion currents through the biomembrane, the influence of endo- and exothermic reactions on temperature, and the influence of energy in a non-electrical form are taken into account using phenomenological variables, i.e., observables. Such an approach brings the mathematical models closer to reality. Using the concept of phenomenological inductance helps us better understand the propagation of an action potential in myelinated axons. In principle, contemporary mathematical models describing the process in axons are hybrid in nature, combining physical laws with phenomenology, i.e., with observables.

This work proposes the use of orbital angular momentum (OAM) waves to improve the performance of a computational imaging (CI) system. Specifically, in contrast to a solely frequency-diverse operation, leveraging multiple OAM waves leads to a significant increase in the diversity of the measurement modes of a CI system. This significantly reduces the frequency bandwidth required to achieve high-quality image reconstructions. A proof-of-concept prototype working at Ka-band frequencies is used to validate the proposed approach. The prototype consists of two metalized three-dimensional (3D) printed cavities, with fully-dielectric transmitarrays inside that generate OAM waves. Imaging results from various targets reveal that the CI system achieves superior imaging quality when multiple OAM waves are considered, compared to when it solely relies on frequency-diversity. This is specially noticeable in the case of complex distributed targets, which can only be reconstructed with the prototype when multiple OAM waves are used. Furthermore, it is shown that accurate image reconstructions can be obtained employing only one eighth of the operational bandwidth of the frequency-diverse system.

Low-energy electron diffraction (LEED) is a cornerstone technique for determining surface atomic structures[heldStructureDeterminationLowenergy2025], yet the quantitative analysis of electron diffraction intensity as a function of incident electron energy -- that is, LEED-\textit{I(V)} analysis -- remains a complex inverse problem. In this work, we tackle quantitative LEED-\textit{I(V)} analysis based on physics-informed Bayesian optimization (BO). By embedding multiple scattering LEED forward models directly into a trust-region BO loop, our approach simultaneously optimizes both structural and experimental parameters, adaptively adjusting trust regions for efficient exploration of complex non-convex parameter spaces without manual intervention. The robustness and scalability of the approach are demonstrated using the Ag(100)-(1$\time$1) and Fe\textsubscript{2}O\textsubscript{3}(1$\overline{1}$02)-(1$\time$1) surfaces as examples. Our work establishes a general framework for solving inverse problems in various characterization techniques, unlocking a physics-informed efficient, reproducible, and autonomous paradigm.

Magnetic straintronics, the strain-mediated control of magnetic anisotropy, has emerged as a key direction for next-generation energy-efficient technologies. In multiferroic heterostructures, magnetoelectric coupling is typically achieved by applying an electric field on a ferroelectric phase, inducing strain through the converse piezoelectric effect, which is then transferred to the adjacent ferromagnetic phase. As an alternative, strain can be remotely modulated through the photostrictive effect induced by light. While light-driven control of magnetic anisotropy has been explored, optical modulation of more complex phenomena such as exchange bias remains largely unaddressed. Here, we demonstrate significant light-induced modulation of exchange bias and magnetization switching at room temperature in a Pb(Mg1/3Nb2/3)O3-Pb(Zr,Ti)O3 (PMN-PZT)/Fe80Ga20(FeGa)/Ir20Mn80(IrMn) multiferroic heterostructure, driven by visible-light-photostriction. The magnetization state correlates with the light intensity, enabling multi-level states with light power densities as low as 0.1 W cm-2. These findings suggest a promising route toward low-power, multistate, and wireless opto-magnetic memory applications.

Modern society relies on complex supply chains to sustain the flow of goods and services that are essential to daily life. While traditional supply chain theory assumes a clear, hierarchical flow from upstream suppliers to downstream customers, observable real-world transaction networks rarely exhibit this acyclic structure. Instead, detailed inter-firm data reveal that interwoven networks are heavily entangled by cyclic flows. Consequently, without appropriate partitioning of these massive inter-firm networks, the latent flow-hierarchical structures that are central to supply chain concepts remain obscure. To address this analytical challenge, we introduce the flow-Hierarchical Community Network Extraction (f-HiCoNE) framework. By applying combinatorial Hodge decomposition, this approach disentangles the complex inter-firm network by isolating the acyclic gradient flow to quantify the flow-hierarchical parts and partition the graph. By applying f-HiCoNE to a nationwide transaction dataset of approximately 650,000 firms, we successfully extracted functional supply-chain clusters. These clusters demonstrated strong flow-hierarchical organisation, wherein the upstream-downstream positioning of firms was accurately captured by local scalar potentials, revealing distinct geographically localised industrial ecosystems. This study provides a map that helps firms understand their surrounding environment and locate their position within an inter-firm network and opens a new research avenue focused on flow-hierarchy clustering in supply chain analysis.

The hierarchy and proximity are key dimensions of urban relational processes, but their interplay in shaping intercity interactions and the underlying structures of city systems remain unclear. We develop a novel geometric model of city systems embedding intercity mobility into a latent hyperbolic geometry, which unravels the measures of hierarchy and proximity accounting for their interplay. It is successfully validated against 12 different nationwide intercity mobility datasets. We find a bottom-up emergence of city hierarchies, along which the variations of city-hinterland relations are non-stationary in terms of their nesting and range properties. Such non-stationarity originates from trade-offs between city hierarchy and hinterland range in determining the formation of city-hinterland structures. Hierarchy- and proximity-dominated urban processes can be elucidated from examining dynamics of the trade-offs. The revealed urban relational processes of city systems are at the core of the emerging science of cities and crucial for spatial planning and regional policymaking.

Combining optical tweezers with fluorescence microscopy is a powerful tool for single-cell analysis, playing a pivotal role in disease diagnosis, cell sorting, and the investigation of cellular dynamics. However, fluorescence detection faces challenges such as blinking, photobleaching and autofluorescence in biotissues. To address these limitations, we developed a magnetic detection strategy by integrating quantum magnetometry using nitrogen-vacancy centers into optical tweezers, demonstrating precise trapping and manipulation of individual cells in microfluidic environment. We detected a magnetic signal of 89 μT from a single cell labeled with magnetic nanoparticles, compared to a noise floor of 3.9 μT observed in unlabeled cells. This platform provides a promising approach for high-precision single-cell analysis and holds significant potential for probing cellular activities within biological microenvironments.

The rapid integration of Large Language Models (LLMs) into scientific writing fundamentally challenges traditional definitions of authorship, responsibility, and scientific integrity. As researchers transition from using computers as deterministic tools to managing them as ``virtual collaborators,'' the nature of human contribution must be re-evaluated. Using the drafting process of a recent computational physics manuscript as a case study, this essay explores the indispensable role of the Human-in-the-Loop (HITL). We demonstrate that while AI excels at structural organization and syntax generation, the human author bears the ultimate responsibility for enforcing rigorous physical logic, maintaining academic diplomacy, and anticipating peer-review critiques. In this paradigm, the human contribution shifts from writing boilerplate text to acting as a Principal Investigator who actively mentors and steers the AI's reasoning. To ensure accountability and preserve the integrity of the scientific record in this new era, I argue that the community must mandate the publication of full, unedited AI interaction transcripts as standard supplementary material.

Exponential growth in global data traffic demands ever-increasing transmission rates--a pursuit fundamentally constrained by the physical limitations of digital-to-analog converters (DACs). Existing strategies to overcome this bottleneck, such as multi-DAC arrays and optical time-division multiplexing, inevitably introduce system complexity and coordination overhead. Here we demonstrate an all-optical spatiotemporal transmitter that generates controllable high-repetition information-carrying femtosecond pulses at the focus of a phase-modulated planar diffractive lens (PDL) through optical-path-induced spatial-to-temporal conversion. Each pulse serves as an information bit, encoding binary data via on-axis focal intensity states corresponding to '0' and '1', achieved by switching between topological and constant phase modulations. High experimental orthogonality between arbitrary bits enables nearly error-free transmission of 15X15-pixel grayscale (8-bit coding) and colour (9-bit coding) images at a record-high single-channel rate of approximately 3 terabits per second (Tbit/s). Free from electronic and coordination bottlenecks, this all-optical transmitter establishes a scalable high-speed single-channel pathway toward ultrahigh-capacity optical communication.

I present a quantum-tunnelling oscillator model as a universal dynamical engine for two paradigmatic problems in quantum cognition theory -- optical illusion perception and group decision making -- where individuals are treated as quantum-mechanical agents whose choices shift through context-dependent transitions rather than simple probabilities. I show that, when networked together, these units form a quantum-cognitive neural system that reproduces familiar collective and perceptual phenomena while naturally accommodating counterintuitive processes that challenge classical models. Bridging ideas from quantum cognition theory and neural networks, this approach offers a compact, physically grounded way to describe how real individuals and groups think, perceive and decide.

A photodetector that can directly distinguish the orbital angular momentum (OAM) of light is highly desirable for integrated on-chip OAM detection and focal plane array devices. The recent development of OAM detectors based on the intrinsic orbital photo galvanic effects (OPGE) of materials provide a new route for direct OAM detection that is on-chip scalable with high resolution and speed. In this paper, we summarize the current progress in direct photodetection of OAM via OPGE. We begin with a short review of the basic operation scheme of the OAM detector and provide a comprehensive symmetry analysis to sort out the favorable characteristics of the materials, incorporating considerations from device schemes based on various device performance characteristics and specific application circumstances. From that, we review the current experimental progress and technical challenges, then oversee the possible solutions to these challenges and provide a perspective on the future opportunities of this OAM detection route.

Nonlinear optics underpins a broad range of photonic technologies, from classical and quantum light sources to emerging nonlinear photonic neural networks. Yet, conventional nonlinear optical devices exhibit static functionality: their transfer characteristics and emission profiles are dictated by the intrinsic nonlinear process and locked by fabrication, limiting adaptability. Here, we introduce an ultra-thin metasurface platform that enables dynamic reconfiguration of nonlinear functionality in a contact-less fashion. By leveraging all-optical control of the optical torque exerted on liquid crystal molecules infiltrating a resonant metasurface, we achieve tunable polynomial nonlinear transfer functions based on third-harmonic generation process. This mechanism further allows real-time modulation of nonlinear weighting across different diffraction orders, revealing a previously unexplored interplay between mode structure and nonlinear emission. Our approach opens up a pathway toward field-programmable nonlinear photonic systems, offering unprecedented flexibility for reconfigurable nonlinear signal processing and adaptive photonic computing.

We present HYMOR (HYpersonic MOdal/non-modal, and Receptivity), an open-source computational framework for the linear stability analysis of high-enthalpy hypersonic flows. The toolkit includes MATLAB and Julia implementations and is released under the MIT license. HYMOR provides global modal, non-modal, and freestream receptivity analyses capable of capturing interactions among spatially separated physical mechanisms that are inaccessible to traditional local methods. A shock-fitting formulation is employed to treat the bow shock as a sharp discontinuity, ensuring that the interaction of infinitesimal disturbances with the shock reproduces the exact response predicted by linear interaction analysis. The code also solves the nonlinear equations for base-flow computation and automatically linearizes the resulting discrete operators for the stability analyses. Several thermochemical models are available for treatment of real-gas effects in high-enthalpy regimes. The numerical implementation is verified against a collection of benchmark cases that demonstrate the accuracy and capabilities of the toolkit across its modal, non-modal, and receptivity analysis modes.

Band formation in periodic media is a central topic in undergraduate solid-state physics, typically introduced through Bloch's theorem as an eigenvalue problem in reciprocal space for infinitely periodic systems. While mathematically elegant, this formulation can appear abstract: it assumes an idealized infinite lattice, shifts attention away from real-space wave dynamics, and presents band structures as static results rather than emergent consequences of wave propagation. Consequently, students often struggle to relate band gaps to familiar physical phenomena such as reflection, transmission, and interference, leading to a disconnect between formal band theory and observable wave behavior. We present a computational framework that addresses this gap by reconstructing band formation directly from time-domain wave propagation in finite periodic systems. Using a staggered-grid finite-difference time-domain scheme for elastic waves, a broadband excitation is propagated through a layered medium to obtain its transmission spectrum. From this, students extract the Bloch dispersion relation and observe spatial attenuation in band-gap regions, revealing the roles of multiple scattering and phase coherence. This approach provides a physically transparent pathway to band theory and enables exploration of finite-size effects, disorder, and defect-localized modes within a unified computational framework. Implemented through compact code and guided exercises, the method offers an accessible and versatile pedagogical tool, while also equipping students with transferable skills in numerical modeling of wave phenomena across disciplines.

Disambiguating scholars with identical names is essential for accurate authorship assignment and robust large-scale scientometric research. Existing methods are often designed for Latin-script metadata and perform poorly on Chinese names. In international publications, Chinese names typically appear as Romanized Pinyin, which is highly ambiguous as it can map to multiple distinct characters. Chinese characters, in contrast, reduce but do not eliminate this ambiguity, and are rarely available in international records. To address both challenges, we propose a rule-based disambiguation framework that integrates co-authorship networks, citation networks, author affiliations, and content similarity. We apply this framework to 65,241 physics papers from the China National Knowledge Infrastructure (CNKI), spanning over 70 years of data. On a human annotated sample of 80 name pairs, our method achieves F1-scores of 0.88 for Pinyin names and 0.89 for character-based names, outperforming two baseline approaches, with improvements driven primarily by higher recall. The comparable performance across both writing systems shows that our approach is script-agnostic, enabling reliable large-scale scientometric analyses.

Light localization by scattering is a fundamental mechanism driving phase transitions of wave transport in disordered systems. Characterizing the localization length in scattering systems is crucial yet challenging. In this Letter, we demonstrate a spatially matched coupling scheme using wavefront shaping to resolve the intrinsic localization length in two-dimensional disordered lattices. By tailoring the incident wavefront, our method facilitates efficient coupling of light to the minimum localized mode. We apply this approach to measure two different self-assembled lattices, and report the first observation of the critical coupling effect, which allows for the direct determination of the characteristic size of minimum localized mode. Our results reveal that for a fixed lattice periodicity, increasing the air-hole diameter significantly reduces this intrinsic localization length. This far-field metrology offers a robust framework for probing wave localization in complex media, which should be useful in various applications such as random lasing and nonlinear optics

We present SA-DSVN, a 3D convolutional architecture for voxel-level segmentation of structural defects in reinforced concrete using cosmic-ray muon tomography. Unlike conventional reconstruction methods (POCA, MLSD) that rely solely on muon scattering angles, our approach jointly processes scattering kinematics (9 channels) and secondary electromagnetic shower multiplicities (40 channels) through independent encoder streams fused via cross-attention. Training data were generated using Vega, a cloud-native Geant4 simulation framework, producing 4.5 million muon events across 900 volumes containing four defect types - honeycombing, shear fracture, corrosion voids, and delamination - embedded within a dense 7x7 rebar cage. A five-variant ablation study demonstrates that the shower multiplicity stream alone accounts for the majority of discriminative power, raising defect-mean Dice from 0.535 (scattering only) to 0.685 (shower only). On 60 independently simulated validation volumes, the model achieves 96.3% voxel accuracy, per-defect Dice scores of 0.59-0.81, and 100% volume-level detection sensitivity at 10 ms inference per volume. These results establish secondary shower multiplicity as a previously unexploited but highly effective feature for learned muon tomographic reconstruction.

The properties of lattice-based structures can be enhanced by varying their geometric parameters in a graded manner, and the gradation can be tailored to extremize a particular objective. In this manuscript, we propose a non-gradient-based optimization framework to find the tailor-made graded profiles for lattice-based structures. The key challenge addressed in the work is to ensure the graded nature/smoothness of the underlying structure in a non-gradient-based optimization scheme. As we demonstrate in the manuscript, the conventional implementation of the genetic algorithm provides structures with abrupt changes, leading to issues such as stress concentration. In this work, we propose a Gaussian random function (GRF)/Gaussian process regression (GPR) integrated genetic algorithm to obtain an optimal graded lattice profile for a given objective. The integration of the GRF/GPR along with a projection operator ensures the smoothness of the designs at each stage of the optimization. We present several numerical examples to demonstrate that the proposed framework provides smoother designs that are less susceptible to stress concentration, while ensuring satisfaction of the underlying objective.

Repetition-based draw rules in deterministic games like chess ensure termination but introduce strategic artifacts, allowing players to enforce draws independent of positional value. We propose an asymmetric modification: threefold repetition results in a loss for White if it is responsible for initiating it. This rule directly targets the persistent first-move advantage and removes low-effort draw strategies available to White. The new rule is expected to reduce draw rates, re-balance first-move advantage, and promote exploration in both human and artificial play. We outline a computational framework with existing and newly designed neural-network chess engines for the empirical validation of the proposal and analyze it from the perspectives of game theory and graph dynamics.

Single-pixel imaging (SPI) offers a cost-effective route to hyperspectral acquisition but struggles to recover high-fidelity spatial and spectral details under extremely low sampling rates, a severely ill-posed inverse problem. While deep learning has shown potential, existing data-driven methods demand large-scale pretraining datasets that are often impractical in hyperspectral imaging. To overcome this limitation, we propose an end-to-end physics-informed framework that leverages untrained neural networks and RGB guidance for joint hyperspectral reconstruction and super-resolution without any external training data. The framework comprises three physically grounded stages: (1) a Regularized Least-Squares method with RGB-derived Grayscale Priors (LS-RGP) that initializes the solution by exploiting cross-modal structural correlations; (2) an Untrained Hyperspectral Recovery Network (UHRNet) that refines the reconstruction through measurement consistency and hybrid regularization; and (3) a Transformer-based Untrained Super-Resolution Network (USRNet) that upsamples the spatial resolution via cross-modal attention, transferring high-frequency details from the RGB guide. Extensive experiments on benchmark datasets demonstrate that our approach significantly surpasses state-of-the-art algorithms in both reconstruction accuracy and spectral fidelity. Moreover, a proof-of-concept experiment using a physical single-pixel imaging system validates the framework's practical applicability, successfully reconstructing a 144-band hyperspectral data cube at a mere 6.25% sampling rate. The proposed method thus provides a robust, data-efficient solution for computational hyperspectral imaging.

Quantitative characterization of biophysical alterations caused by viral infection remains at an early stage. In this study, we examined the mechanical response of Vero cells after exposure to an attenuated Measles Vaccine Virus using atomic force microscopy (AFM) in combination with confocal microscopy. AFM force distance measurements were conducted in the perinuclear region to evaluate changes in elastic and viscoelastic properties. Within 24 hours post infection, cells infected at a multiplicity of infection of 0.5 exhibited an approximately thirty five percent decrease in median of the Young modulus relative to uninfected controls, indicating substantial cellular softening. Corresponding shifts in viscoelastic behavior were observed, including reductions in the relaxed modulus and in viscosities (effects comparable to those induced by cytochalasin D mediated actin depolymerization). Confocal microscopy further revealed a dependent reorganization of the cytoskeleton upon infection, marked by altered F actin distribution and changes in filament architecture. These findings suggest that actin remodeling contributes to the altered viscoelastic properties observed during infection. This work proposes a straightforward and complementary approach for characterizing virus cell interactions by integrating AFM with confocal imaging.

Compact devices for broadband polarization control in the terahertz (THz) regime are challenging due to the intrinsic phase dispersion of birefringent materials and resonant structures. Here, we demonstrate high-performance broadband THz quarter-wave plates based on over-coupled metal-dielectric-metal reflective metasurfaces. The devices operate as single-port anisotropic Fabry-Pérot cavities in which the phase dispersion of over-coupled resonances is engineered to produce an approximately constant relative phase delay between orthogonal field components. By tailoring the metasurface geometry, efficient linear-to-circular polarization conversion is achieved while maintaining high reflectance. Four complementary metasurface designs, operated at an incidence angle of $45^\circ$, collectively cover the 0.25--3 THz frequency range accessible to a typical THz time-domain spectroscopy system. Each device exhibits an approximately octave-wide bandwidth with an axial ratio below 3\,dB and polarization conversion efficiencies exceeding 80\% across most of the operating band. Systematic optimization suppresses coupling to higher-order diffraction and surface wave modes, further extending the usable bandwidth while preserving the required phase relationship. The metasurfaces are compatible with wafer-scale fabrication, and experimental results show excellent agreement with simulations. These findings establish over-coupled reflective metasurfaces as a robust and versatile platform for broadband THz polarization control.

We report on the design, fabrication, and characterization of all MOCVD-grown long-gate AlScN/GaN metal-insulator-semiconductor high electron mobility transistors (MIS-HEMTs) on semi-insulating GaN substrates. Devices with a gate length of $1~μ$m and gate-drain spacing of $0.9~μ$ m exhibit a maximum drain current density of 1 A/mm, an on/off current ratio of $2\times 10^5$, and a three-terminal breakdown voltage of 63 V. The device has near-ideal subthreshold characteristics with a subthreshold swing of 63 mV/dec and a current dispersion as low as 7.8$\%$ at 10 V due to the excellent interfacial quality with a trap density ($D_\mathrm{it}$) of $2.11\times{10}^{11}~\mathrm{cm}^{-2}eV^{-1}$ and the semi-insulating GaN substrate with a low threading dislocation density. Small-signal RF measurements reveal an $f_\mathrm{T}/f_\mathrm{max}$ of 25.8/51.1 GHz, while large-signal load-pull characterization at 10 GHz demonstrates an output power density of 4.04 W/mm with a power-added efficiency of 22.7$\%$. In addition, a minimum noise figure below 2.5 dB was measured over a wide drain current range from 100 mA/mm to 700 mA/mm below 6 GHz. These results extend previous demonstrations of short-gate MOCVD-grown AlScN/GaN HEMTs to the long-gate, high-voltage regime, confirming the robustness of this material system for both high-frequency and high-power device applications with favorable microwave noise performance.

Femtosecond mode-locked lasers are foundational to ultrafast science, yet their spectral reach remains constrained by the finite emission bandwidth of available gain media. Optical parametric oscillators (OPOs) overcome this constraint but typically require complex synchronous pumping by external femtosecond lasers. Here we demonstrate a fundamentally different approach: passive mode-locking of a continuous-wave-driven, doubly resonant degenerate OPO via the spontaneous formation of femtosecond dissipative quadratic solitons (DQS). We show that phase-matched intracavity cascaded quadratic nonlinearity (PICQN), enabled by negligible pump-signal walk-off in a doubly resonant cavity, generates a non-local effective Kerr nonlinearity (EKN) that governs the cavity dynamics and drives soliton formation. The engineered EKN exceeds the intrinsic material Kerr nonlinearity by more than three orders of magnitude and is continuously tunable in magnitude and sign via pump phase detuning, enabling a paradigm shift from dispersion to nonlinearity engineering for dissipative soliton formation. Comprehensive stability analysis reveals distinct dynamical regimes governed by pumping and cavity conditions, providing a versatile framework for exploring previously understudied quadratic soliton physics. Experimentally, we observe bichromatic femtosecond DQSs at 1572 nm and 786 nm with pulse durations of 336 fs and 447 fs, respectively, peak powers up to 150 W, and a conversion efficiency of 5% under 600 mW continuous-wave pumping. Our work establishes a simple, flexible, and scalable architecture for femtosecond OPOs that bypasses the need for synchronized mode-locked pump lasers. By shifting from traditional dispersion engineering to in-situ nonlinearity engineering, this platform extends the reach of soliton-based technologies and enables dissipative solitons across diverse platforms and spectral regimes.

This paper reports on the design, fabrication, and demonstration of a silicon photonics based heterodyne interferometric imaging system. The photonic integrated circuit (PIC) can perform one-dimensional spectroscopy for unique input spectrums using a single baseline within its 91 available baselines. The PIC uses polarization diversifying gratings to separate incoming light into two distinct polarizations, an on-chip 2x4 optical hybrid, and a strong local oscillator (LO) to perform the heterodyne measurements. The optical hybrids combine the input signals with the LO and splitting them into 2 components pairs for phase sensitive measurements. Furthermore, the PIC can perform 2-D image reconstruction by combining many baseline pairs to measure the visibility of a simple target. These demonstrations show the PIC's capabilities for 1-D spectroscopy and 2-D imaging applications.

For over a century, termolecular, or third order, chemical reactions have been explained by the Lindemann Hinshelwood mechanism, assuming sequential stabilization via bimolecular encounters. Here, we demonstrate that barrierless termolecular reactions are fundamentally governed by direct three body dynamics. Using classical trajectory calculations in hyperspherical coordinates, we quantitatively reproduce ion atom recombination kinetics across a wide temperature range without invoking intermediate complexes or steady state assumptions. Our results not only resolve longstanding discrepancies between theory and experiment, but also establish a general mechanistic framework for barrierless termolecular reactions, with implications spanning atmospheric chemistry, plasma physics, and ultracold chemistry.

Artificial-intelligence (AI) agent frameworks have been developed for autonomous scientific simulations, but most current agent frameworks are tailored to a single or a small set of software packages. Herein, FermiLink, a unified and extensible open-source agent framework is introduced for multidomain scientific simulations. Its key design principle is the separation of package knowledge bases from simulation workflows, so that simulation workflows in FermiLink, from figure-level simulations to full-paper-level research on high-performance computing clusters, operate uniformly among supported packages via a four-layer progressive disclosure mechanism. Using OpenAI Codex as the agent provider, the capabilities of FermiLink are demonstrated across approximately 50 scientific software packages spanning nine research domains from physics to engineering. Systematic benchmarks on 132 real-world figure-level reproduction tasks with 44 packages show that FermiLink reproduces 74 (56.1%) of published figures with simulations, among which 30 achieve high-fidelity agreement and 35 reach qualitative agreement with the target figures. A smaller set of human expert-guided reproduction benchmarks with 10 packages further highlights the importance of expert insights for improving the simulation fidelity. Beyond reproduction, a single-blinded study demonstrates that FermiLink can produce research-grade results on unpublished polariton physics problems when provided with sufficiently detailed research objectives and source code, even in the absence of external documentation or tutorials. Overall, FermiLink provides a scalable research infrastructure that may accelerate the path from scientific questions to computational results across diverse domains.

Self-sensing conductive composites can reveal deformation and damage through measurable changes in electrical resistance, which makes them attractive for embedded diagnostics and learning-enabled structural health monitoring. This paper presents a physically consistent multiphysics Deep Energy Method (DEM) for brittle fracture in piezoresistive materials. The mechanical part is modeled by small-strain linear elasticity coupled to a fourth-order AT2-type phase-field fracture functional with tensile/compressive energy split and history-field irreversibility. To avoid artificial energetic mixing of mechanical and electrical quantities, the electrical problem is treated as a one-way coupled sensing subproblem: after solving the mechanics--fracture problem, the electric potential is obtained from a steady conduction problem whose conductivity depends on strain through a linearized piezoresistive law and on damage through a crack-induced conductivity degradation. The resulting formulation predicts crack evolution together with its resistance signature without assigning the electrical field an artificial crack-driving role. DEM is used to minimize the variational subproblems over admissible neural trial spaces with exact imposition of essential boundary conditions. A lean verification suite is used to validate the electrical building blocks and the fracture engine separately, followed by a numerical study of a tensile plate with stress concentrators and electrodes. In that study, the framework captures a nontrivial sensing regime in which appreciable damage growth leaves the global resistance nearly unchanged, followed by a sharp resistance increase once dominant conductive ligaments are disrupted and current paths reorganize strongly.

This is the third paper in a series addressing the problem of anticipating the local occurrence of future large earthquakes. In the first paper we showed how large earthquake probabilities can be computed in natural time, which is defined as the small earthquake event count following a large earthquake. In the second paper we introduced the idea of regional ensembles of data, which are defined as a series of increasingly large rectangular regions centered on the point of interest. In this paper, we consider the problem of exceedance probabilities for magnitude, and for large event occurrence in natural time, and calendar time. In the ensemble idea, we assumed that all of the Gutenberg-Richter regional statistics for the ensembles are similar to those of events in the circular region of interest. In the present work we also develop a "nowcast transform" that adjusts the Gutenberg-Richter statistics of the ensemble members to match those of the events in the circle. We show that the forecasts and exceedance probabilities computed with this adjustment are generally consistent with those computed with the non-transformed ensemble data. We illustrate these methods by application to a 125 KM region surrounding Los Angeles, CA, following the January 17, 1994 magnitude M6.7 Northridge earthquake.

Objective: Algorithmic fairness is essential for equitable and trustworthy machine learning in healthcare. Most fairness tools emphasize single-axis demographic comparisons and may miss compounded disparities affecting intersectional populations. This study introduces Fairlogue, a toolkit designed to operationalize intersectional fairness assessment in observational and counterfactual contexts within clinical settings. Methods: Fairlogue is a Python-based toolkit composed of three components: 1) an observational framework extending demographic parity, equalized odds, and equal opportunity difference to intersectional populations; 2) a counterfactual framework evaluating fairness under treatment-based contexts; and 3) a generalized counterfactual framework assessing fairness under interventions on intersectional group membership. The toolkit was evaluated using electronic health record data from the All of Us Controlled Tier V8 dataset in a glaucoma surgery prediction task using logistic regression with race and gender as protected attributes. Results: Observational analysis identified substantial intersectional disparities despite moderate model performance (AUROC = 0.709; accuracy = 0.651). Intersectional evaluation revealed larger fairness gaps than single-axis analyses, including demographic parity differences of 0.20 and equalized odds true positive and false positive rate gaps of 0.33 and 0.15, respectively. Counterfactual analysis using permutation-based null distributions produced unfairness ("u-value") estimates near zero, suggesting observed disparities were consistent with chance after conditioning on covariates. Conclusion: Fairlogue provides a modular toolkit integrating observational and counterfactual methods for quantifying and evaluating intersectional bias in clinical machine learning workflows.

Balanced spiking networks can transition between silent, asynchronous-irregular, and oscillatory states depending on interacting synaptic and temporal time scales, while their joint parameter structure remains incompletely characterized. In this work, we systematically map how postsynaptic decay (τs), conduction delay (d), and plasticity rate (λp) jointly shape oscillatory regimes in recurrent leaky integrate-and-fire networks. By combining Brian2 simulations across the (τs, d, λp) space with a coarse Hopf-reference boundary, we construct regime maps that directly visualize SIL-AI-OSC transitions and corresponding spectral prominence landscapes. The mapped results show that increasing λp expands oscillatory regions toward shorter τs and moderate-to-long delays, while prominence maps identify parameter regions with the strongest rhythmic coherence. Representative control experiments further connect this global landscape to local rhythm-forming mechanisms, showing that STDP freezing weakens rhythmic coherence whereas delay jitter enhances it with minimal change in mean firing rate. As a result, these findings provide a useful reference for operating-point selection, synchrony modulation studies, and future biologically grounded spiking-network modeling within similar balanced-network settings.

Scalar variability -- the finding that representational noise scales proportionally with magnitude, producing a constant coefficient of variation -- is a hallmark of biological magnitude systems. We tested whether transformer language models exhibit this property by analysing the dispersion of hidden-state representations across carrier sentences for 26 numerical magnitudes in three 7-8B parameter models (Llama-3-8B-Instruct, Mistral-7B-Instruct-v0.3, Llama-3-8B-Base; data from Cacioli, 2026). We found the opposite: representational variability decreased with magnitude along the magnitude axis (scaling exponent alpha approx -0.19; 0/16 primary layers with alpha > 0, all three models). The negative sign was consistent in full-dimensional space (alpha approx -0.04) and after sentence-identity correction (alpha approx -0.007). The anti-scalar pattern was 3-5x stronger along the magnitude axis than orthogonal dimensions, and corpus frequency strongly predicted per-magnitude variability (rho = .84). These results demonstrate that distributional learning alone is insufficient to produce scalar variability: transformers reproduce log-compressive magnitude geometry but not the constant-CV noise signature observed in biological systems.

We introduce Mean-Field Game (MFG) epidemiological models, in which immunity either wanes with time in a fully observable way or disappears instantaneously with no direct observation (making a previously recovered individual fully susceptible again without realizing it). Both interpretations create computational challenges for rational noninfected individuals deciding on their contact rates based on their personal current immunity state and the changing epidemiological situation. Both require solving a forward-backward MFG system that includes PDEs (an advection-reaction equation for the immunity-structured population and a Hamilton-Jacobi-Bellman equation for the corresponding value function). We show how this can be done efficiently by solving a two-point boundary value problem for a system of approximating ODEs. We also show how the same approach can be extended to handle an initial uncertainty in the planning horizon.

Amplifying weak molecular signals is essential in both natural and engineered biochemical systems. While most amplification schemes operate out of equilibrium, relying on kinetic barriers and fuel-driven cascades, it is also possible to amplify at thermodynamic equilibrium by shifting the energy landscape upon addition of an analyte. Equilibrium amplification is appealing because, in principle, it can remain indefinitely in the untriggered state. In this work, we establish fundamental structural and thermodynamic limits on equilibrium-based amplification. We first prove that dimerization networks--systems restricted to complexes of at most two monomers--are inherently incapable of equilibrium amplification. This no-go theorem explains the absence of amplification in prior undercomplementary "strand commutation" designs. We then show that allowing trimeric complexes breaks this barrier. We propose an isometric trimer-based amplifier whose output preserves the size of the input, enabling modular composition, and validate it experimentally, achieving an amplification factor close to the expected $2\times$. Finally, we derive universal thermodynamic bounds applicable to any equilibrium network regardless of complex size: the maximum amplification factor scales linearly with the free energy of interaction between the analyte and the amplifier components. For nucleic acid systems, this implies that the analyte length must grow linearly with the desired amplification factor, and that composing modular amplifiers yields diminishing returns for a fixed analyte. Together, these results delineate the structural and energetic boundaries of equilibrium amplification and rigorously justify the necessity of out-of-equilibrium approaches for achieving high gain.

Multimodal deep learning models that fuse whole-slide histopathology images with genomic data have achieved strong discriminative performance for cancer survival prediction, as measured by the concordance index. Yet whether the survival probabilities derived from these models - either directly from native outputs or via standard post-hoc reconstruction - are calibrated remains largely unexamined. We conduct, to our knowledge, the first systematic fold-level 1-calibration audit of multimodal WSI-genomics survival architectures, evaluating native discrete-time survival outputs (Experiment A: 3 models on TCGA-BRCA) and Breslow-reconstructed survival curves from scalar risk scores (Experiment B: 11 architectures across 5 TCGA cancer types). In Experiment A, all three models fail 1-calibration on a majority of folds (12 of 15 fold-level tests reject after Benjamini-Hochberg correction). Across the full 290 fold-level tests, 166 reject the null of correct calibration at the median event time after Benjamini-Hochberg correction (FDR = 0.05). MCAT achieves C-index 0.817 on GBMLGG yet fails 1-calibration on all five folds. Gating-based fusion is associated with better calibration; bilinear and concatenation fusion are not. Post-hoc Platt scaling reduces miscalibration at the evaluated horizon (e.g., MCAT: 5/5 folds failing to 2/5) without affecting discrimination. The concordance index alone is insufficient for evaluating survival models intended for clinical use.

Information flow is central to contemporary accounts of cognition, yet its physical basis in living neural matter remains poorly specified. Here, we develop a multiscale resource-theoretical framework motivated by the \textit{thermocoherent effect}, where heat flow is reciprocally coupled to a delocalized information flow carried by shared coherence and not reducible to local subsystem variables. Extending this line of work in light of recent results on correlation-enabled Mpemba-type thermal relaxation, we argue that the operational relevance of correlations depends less on their taxonomy than on their dynamical accessibility under the underlying interaction geometry. Relational structure encoded in the state of a single composite system -- including quantum entanglement, quantum discord, and classical correlations -- may therefore act as a usable physical resource that remains hidden from local subsystem descriptions. We propose that electrical, chemical, ionic, and thermal transport processes in neural matter may, under suitable microscopic conditions, generate or transduce partially hidden relational resources whose mutual coupling can progressively build larger-scale thermocoherent organization across spatial or spatiotemporal partitions in neural tissue. Ion-channel interfaces, hydrogen-bonded proton networks, aromatic $π$-electron architectures, and phosphate-rich motifs emerge as plausible substrate classes in which such resources may arise, become transiently accessible under environmental coupling, and leave coarse-grained signatures in neural dynamics. The resulting picture is neither a claim of macroscopic quantum cognition nor a reduction of cognition to abstract coding, but a falsifiable framework in which microscopic relational resources can bias transport, relaxation, signaling, and cross-scale neural coordination.

Background music shapes attention, affect, and approach behavior in commercial environments, yet the neural plausibility of AI-generated music for such settings remains poorly characterized. We present an in-silico pilot study that combines Wubble, a generative music system, with TRIBE v2, a publicly released whole-brain encoding model, to estimate cortical response profiles for prompt-conditioned retail music. Five fully instrumental tracks were generated to span low-to-high arousal, sparse-to-dense arrangement, and neutral-to-positive valence prompts, then analyzed with audio-only TRIBE v2 inference on loudness-normalized waveforms. Analysis focused on fsaverage5 cortical predictions summarized over auditory, superior temporal, temporo-parietal, and inferior frontal HCP parcels. The fast bright major-pop condition produced the largest whole-cortex mean activation (0.0402), the strongest prefrontal ROI composite response (0.0704), and the highest parcel means in IFJa (0.1102), IFJp (0.0995), A5 (0.0188), and area 45 (0.0015). Pairwise spatial correlations ranged from 0.787 to 0.974, indicating that prompt variation modulated predicted cortical states rather than yielding a single undifferentiated response profile. Predicted cortical surface maps further revealed visually distinct spatial organization between low-arousal and high-arousal conditions. These results support a cautious claim of cortical neurological plausibility: prompt-conditioned AI music can systematically shift predicted auditory-temporal-prefrontal patterns relevant to salience and valuation. Although the study does not establish subcortical reward engagement or consumer behavior, it provides a reproducible framework for neural pre-screening and pre-optimization of commercial music generation against biologically informed cortical proxies.

Dementia affects over 55 million people worldwide, yet whether distinct domains of physical fitness independently protect against neurodegeneration through shared or divergent biological mechanisms remains unknown. Using the UK Biobank (n = 51,517; 12-year follow-up), we integrated epidemiological, proteomic, and neuroimaging analyses to systematically characterize the multidimensional fitness-dementia relationship. Higher handgrip strength, cardiorespiratory fitness, and pulmonary function were each independently associated with reduced dementia risk (HRs 0.50, 0.62, and 0.73, respectively, for highest vs. lowest tertiles), with stronger associations in women and younger individuals. Plasma proteomic profiling revealed domain-specific molecular signatures--neurofilament light chain predominating for muscular and cardiorespiratory fitness, and inflammatory mediators including GDF15 for pulmonary function--with 22-40 proteins per domain independently predicting dementia, converging on neuroinflammatory and neurovascular pathways. Brain MRI analyses identified hippocampal volume as a significant structural mediator (proportion mediated: 3.7-10.1%), indicating structural preservation as one of multiple mechanistic pathways. Population attributable fraction analyses estimated that suboptimal fitness may account for approximately 26% of dementia cases. These findings reveal that multidimensional physical fitness shapes dementia risk through distinct yet converging neuroinflammatory, neurovascular, and structural brain mechanisms, with implications for life-course prevention.

Stochastic chemical reaction networks (SRNs) in cellular systems are commonly modeled as continuous-time Markov chains (CTMCs) describing the dynamics of molecular copy numbers. The exact evaluation of transient copy number statistics is, however, often hindered by a non-closed hierarchy of moment equations. In this paper, we propose a method for computing theoretically guaranteed upper and lower bounds on transient moments based on the Kolmogorov's backward equation, which provides a dual representation of the CME, the governing equation for the probability distribution of the CTMC. This dual formulation avoids the moment closure problem by shifting the source of infinite dimensionality to the dependence on the initial state. We show that, this dual formulation, combined with the monotonicity of the CTMC generator, leads to a finite-dimensional linear time-invariant system that provides bounds on transient moments. The resulting system enables efficient evaluation of moment bounds across multiple initial conditions by simple inner-product operations without recomputing the bounding system. Further, for certain classes of SRNs, the bounding ODEs admit explicit construction from the reaction model, providing a systematic and constructive framework for computing provable bounds.

Moment dynamics in stochastic chemical kinetics often involve an infinite chain of coupled equations, where lower-order moments depend on higher-order ones, making them analytically intractable. Moment bounding via semidefinite programming provides guaranteed upper and lower bounds on stationary moments. However, this formulation suffers from the rapidly growing size of semidefinite constraints due to the combinatorial growth of moments with the number of molecular species. In this paper, we propose a sparsity-exploiting matrix decomposition method for semidefinite constraints in stationary moment bounding problems to reduce the computational cost of the resulting semidefinite programs. Specifically, we characterize the sparsity structure of moment equations, where each reaction involves only a subset of variables determined by its reactants, and exploit this structure to decompose the semidefinite constraints into smaller ones. We demonstrate that the resulting formulation reduces the computational cost of the optimization problem while providing practically useful bounds.

Rapid advances in singing voice synthesis have increased unauthorized imitation risks, creating an urgent need for better Singing Voice Deepfake (SingFake) Detection, also known as SVDD. Unlike speech, singing contains complex pitch, wide dynamic range, and timbral variations. Conventional 16 kHz-sampled detectors prove inadequate, as they discard vital high-frequency information. This study presents the first systematic analysis of high-resolution (44.1 kHz sampling rate) audio for SVDD. We propose a joint fullband-subband modeling framework: the fullband captures global context, while subband-specific experts isolate fine-grained synthesis artifacts unevenly distributed across the spectrum. Experiments on the WildSVDD dataset demonstrate that high-frequency subbands provide essential complementary cues. Our framework significantly outperforms 16 kHz-sampled models, proving that high-resolution audio and strategic subband integration are critical for robust in-the-wild detection.

We study compressed sensing when the sampling vectors are chosen from the rows of a unitary matrix. In the literature, these sampling vectors are typically chosen randomly; the use of randomness has enabled major empirical and theoretical advances in the field. However, in practice there are often certain crucial sampling vectors, in which case practitioners will depart from the theory and sample such rows deterministically. In this work, we derive an optimized sampling scheme for Bernoulli selectors which naturally combines random and deterministic selection of rows, thus rigorously deciding which rows should be sampled deterministically. This sampling scheme provides measurable improvements in image compressed sensing for both generative and sparse priors when compared to with-replacement and without-replacement sampling schemes, as we show with theoretical results and numerical experiments. Additionally, our theoretical guarantees feature improved sample complexity bounds compared to previous works, and novel denoising guarantees in this setting.

The unscented Kalman filter (UKF) is a commonly used algorithm capable of estimating the states of nonlinear dynamic systems. It carefully chooses a set of sample points, called sigma points that capture the nonlinear system states posterior mean and covariance. The filter is based on the scaled unscented transform, where the scaling parameters impact the spreading of the sigma points, determining the estimated model capturing. In its current form, the UKF employs a single set of scaling parameters shared by all sigma points. Because states in multi-dimensional models often exhibit substantially different behaviors, this imposes a critical limitation: the standard UKF parameters cannot be tuned to extend the spread for one dimension while reducing it for another. To bridge this gap, we propose the multi-scaled UKF to enable spreading differently per state, while maintaining the key properties of the sigma points and UKF. A rigorous mathematical foundation is provided, introducing a novel theoretical approach to multi-scaling. The benefits of this approach are demonstrated through two distinct nonlinear dynamic systems. Consequently, our multi-scaled UKF captures the nonlinear behavior of multi-dimensional states more effectively, leading to improved estimation accuracy.

Digital twins are becoming an important tool for designing, developing, testing, and optimizing next-generation wireless communication systems. Over the past decade, system softwarization has become a reality, and wireless communication systems are no exception. Software-Defined Radios (SDRs), in general, and Universal Software Radio Peripherals (USRPs), in particular, are often used for prototyping and testing advanced wireless systems. Unfortunately, there is currently no end-to-end, software-based, general-purpose testing environment for SDR-based systems: developers often rely on benchtop setups or even small testbeds, but those are costly and cumbersome to build. At the other end of the spectrum, simulations often rely on simplified channel/radio models and typically do not execute full-stack production code, which can increase development effort and reduce fidelity. In this paper, we propose ACHEM (A Channel Emulator), the first software-based, end-to-end wireless channel emulation environment and toolset for communication systems based on SDRs, specifically USRPs. With the proposed emulator and toolkit, any USRP-based system can be fully emulated at the I/Q level in a pure digital environment without requiring specialized hardware (e.g., vehicles, USRPs, FPGAs, or GPUs). The proposed emulator supports multiple transmitters and receivers, MIMO communications, multiple frequencies, heterogeneous sampling rates, real-time node mobility through vehicle emulation, antenna radiation patterns, and various channel models. ACHEM facilitates wireless digital twin development and deployment. ACHEM is validated with several popular open-source USRP-based wireless communication applications, including GNU Radio, srsRAN 4G/5G, and OpenAirInterface.

We aim to make learned point cloud compression deployable for low-latency streaming on mobile systems. While learned point cloud compression has shown strong coding efficiency, practical deployment on mobile platforms remains challenging because neural inference and entropy coding still incur substantial runtime overhead. This issue is critical for immersive 3D communication, where dense geometry must be delivered under tight end-to-end (E2E) latency and compute constraints. In this paper, we present LEAN-3D, a compute-aware point cloud codec for low-latency streaming. LEAN-3D designs a lightweight learned occupancy model at the shallow levels of a sparse occupancy hierarchy, where structural uncertainty is highest, and develops a lightweight deterministic coding scheme for the deep hierarchy tailored to the near-unary regime. We implement the complete encoder/decoder pipeline and evaluate it on an NVIDIA Jetson Orin Nano edge device and a desktop host. In addition, LEAN-3D addresses the decoding failures observed in cross-platform deployment of learned codecs. Such failures arise from numerical inconsistencies in lossless entropy decoding across heterogeneous platforms. Experiments show that LEAN-3D achieves 3-5x latency reduction across datasets, reduces total edge-side energy consumption by up to 5.1x, and delivers lower sustained E2E latency under bandwidth-limited streaming. These results bring learned point cloud compression closer to deployable mobile 3D streaming.

Tensor-valued data arise naturally in multidimensional signal and imaging problems, such as biomedical imaging. When incorporated into generalized linear models (GLMs), naive vectorization can destroy their multi-way structure and lead to high-dimensional, ill-posed estimation. To address this challenge, Low Separation Rank (LSR) decompositions reduce model complexity by imposing low-rank multilinear structure on the coefficient tensor. A representative approach for estimating LSR-based tensor GLMs (LSR-TGLMs) is the Low Separation Rank Tensor Regression (LSRTR) algorithm, which adopts block coordinate descent and enforces orthogonality of the factor matrices through repeated QR-based projections. However, the repeated projection steps can be computationally demanding and slow convergence. Motivated by the need for scalable estimation and classification from such data, we propose LSRTR-M, which incorporates Muon (MomentUm Orthogonalized by Newton-Schulz) updates into the LSRTR framework. Specifically, LSRTR-M preserves the original block coordinate scheme while replacing the projection-based factor updates with Muon steps. Across synthetic linear, logistic, and Poisson LSR-TGLMs, LSRTR-M converges faster in both iteration count and wall-clock time, while achieving lower normalized estimation and prediction errors. On the Vessel MNIST 3D task, it further improves computational efficiency while maintaining competitive classification performance.

In this paper, a simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS)-aided downlink non-orthogonal multiple access (NOMA) Terahertz (THz) wireless system is proposed for indoor and outdoor transmissions. We consider a near-field communication scenario where an access-point (AP) is deployed near a STAR-RIS panel. For links from the STAR-RIS to users, $α-μ$ distribution is adopted for the indoor small-scale fading channels, whereas the outdoor channels are based on Gaussian mixture or mixture of gamma, which follows the recent practical measurement reports. To facilitate performance analysis, we derive exact expressions of a probability density function (PDF) and a cumulative distribution function (CDF) of a weighted sum of $α-μ$ variates. Approximate PDF and CDF expressions of a weighted sum of Gaussian mixture variates are derived as well. Based on these results, closed-form expressions of the outage probability and the ergodic capacity, together with their asymptotic formulas at high signal-to-noise ratio (SNR), are obtained. Moreover, we analyze the capacity of the THz system at the low SNR regime. Impacts of hardware impairments and STAR-RIS protocols (i.e., energy splitting and mode-switching) on the system performance are evaluated. All developed analytical results are validated and demonstrated via numerical simulations.

In this paper, we investigate high-altitude platform station (HAPS) wireless networks for simultaneous non-orthogonal unicast and multicast transmissions. Specifically, stacked intelligent metasurface (SIM)-based wave-domain beamforming is proposed to enable efficient HAPS-to-ground communications. Also, the system performance is investigated from an energy-efficiency (EE) perspective, which is a crucial for HAPS operations. For performance analysis, we derive approximate closed-form expressions for the outage probability over Rician fading channels. For EE optimization, we jointly optimize the transmit power and the SIM phase-shifts for the maximal EE. Two methods are proposed to solve this non-convex optimization problem. The first method develops an efficient alternating optimization (AO) framework based on golden-section search and projected gradient ascent (PGA) for transmit power and phase-shift optimization, respectively. The second method uses unsupervised deep neural network (DNN) that does not require labeling. Performance comparison between the two methods, as well as with other benchmarks schemes are examined. Additionally, the impacts of the number of SIM elements per layers, the number of SIM layers, the maximum transmit power on the EE performance are evaluated. Simulation results are provided to demonstrate the performance of the proposed systems.

The rapid rise of LLMs over the last few years has promoted growing experimentation with LLM-driven AI tutors. However, the details of implementation, as well as the benefit in a teaching environment, are still in the early days of exploration. This article addresses these issues in the context of implementation of an AI Teaching Assistant (AI-TA) using Retrieval Augmented Generation (RAG) for Trinity College Dublin's Master's Motion Picture Engineering (MPE) course. We provide details of our implementation (including the prompt to the LLM, and code), and highlight how we designed and tuned our RAG pipeline to meet course needs. We describe our survey instrument and report on the impact of the AI-TA through a number of quantitative metrics. The scale of our experiment (43 students, 296 sessions, 1,889 queries over 7 weeks) was sufficient to have confidence in our findings. Unlike previous studies, we experimented with allowing the use of the AI-TA in open-book examinations. Statistical analysis across three exams showed no performance differences regardless of AI-TA access (p > 0.05), demonstrating that thoughtfully designed assessments can maintain academic validity. Student feedback revealed that the AI-TA was beneficial (mean = 4.22/5), while students had mixed feelings about preferring it over human tutoring (mean = 2.78/5).

Reconfigurable antenna technology, such as movable antennas (MAs) and rotatable antennas (RAs), has emerged as a promising solution to enhance the communication performance of wireless systems by exploiting the new degree of freedom (DoF) in antenna reconfiguration. However, existing RA designs mostly consider the array-wise or antenna-wise rotation only, which constrain their great potential in the wide-range radiation pattern control. To overcome this limitation, we propose a new hierarchical rotational six-dimensional MA (HR-6DMA) architecture to improve downlink coverage, which exploits array-wise rotation for global orientation adjustment and individual antenna rotation for fine-grained radiation refinement. Based on this array architecture, we then formulate an optimization problem to maximize the minimum beamforming gain over a target region by jointly optimizing the two-level rotations and transmit beamforming. To solve this non-convex problem, an efficient algorithm is proposed, where the transmit beamforming and per-antenna rotation are optimized via alternating optimization under any feasible array rotation, followed by a low-complexity linear search to determine the optimal array rotation. Last, numerical results show that the proposed HR-6DMA significantly improves the minimum beamforming gain over fixed and single-level rotatable arrays.

The issue of privacy has gained significant attention in recent times. Many real-world applications increasingly require the use of sensitive data, such as in surveillance or tracking and assistance systems. To address these concerns, we propose a framework based on mm-wave radar technology that not only meets privacy requirements but also provides the necessary capabilities for these systems, including reliable current position tracking, sequence tracking, and feedback to the user. While the use of radar technology for surveillance purposes is gaining momentum, there has been no research to date on its application for prayer tracking and assistance systems. Furthermore, there is a lack of comprehensive research that covers all aspects of implementing such a system. Proposed approach offers a versatile solution that can be applied to a broad range of scenarios. Instead of utilizing raw I-Q data, we addressed the challenge of classification based on point cloud information generated by the conventional processing chain of the frequency-modulated continuous wave radar. This information contains corresponding range, reflection amplitude, Doppler and angular values. We have developed and compared different machine-learning classification algorithms to identify the most effective one. Our findings reveal that the convolutional neural network ResNet achieves the best results, with accuracy rates reaching up to 95.4 percent when applied to unknown data. The demonstration video of the developed system can be viewed at the following link: https://youtu.be/PnpGQZWqCr4.

This paper presents RAVEN, a computationally efficient deep learning architecture for FMCW radar perception. The method processes raw ADC data in a chirp-wise streaming manner, preserves MIMO structure through independent receiver state-space encoders, and uses a learnable cross-antenna mixing module to recover compact virtual-array features. It also introduces an early-exit mechanism so the model can make decisions using only a subset of chirps when the latent state has stabilized. Across automotive radar benchmarks, the approach reports strong object detection and BEV free-space segmentation performance while substantially reducing computation and end-to-end latency compared with conventional frame-based radar pipelines.

The integration of cellular communication with Unmanned Aerial Vehicles (UAVs) extends the range of command and control and payload communications of autonomous UAV applications. Accurate modeling of this air-to-ground wireless environment aids UAV mission planning. Models built on and insights obtained from real-life experiments intricately capture the variations in air-to-ground link quality with UAV position, offering more fidelity for simulations and system design than those that rely on generic theoretical models designed for ground scenarios or ray-tracing simulations. In this work, we conduct aerial flights at the Aerial Experimentation and Research Platform for Advanced Wireless (AERPAW) Lake Wheeler testbed to study the variation in key performance indicators (KPIs) of a private 4G/5G cellular base station (BS) with the UAV's altitude, distance from the BS, elevation, and azimuth relative to the BS. Variations in 4G and 5G physical layer KPIs and application layer throughput are logged and analyzed, using two Android smartphones: a Keysight Nemo device, with enhanced KPI access, through a rooted operating system, and a standard smartphone running a custom application that utilizes open-source Android APIs. The observed signal strength measurements are compared to theoretical predictions from free space path loss models that incorporate the BS antenna radiation patterns. Mathematical model parameters for polynomial curve approximations are derived to fit the observed data. Light machine learning approaches, namely random forests, gradient boosting regressors and neural networks, are used to model KPI behaviour as a function of UAV position relative to the BS. The insights and models generated from real-life experiments in this study can serve as valuable tools in the design, simulation and deployment of cellular communication-based UAV systems.

Semantic communications (SCs) aim to transmit only the essential information required to perform given tasks, thereby improving communication efficiency. Deep learning-based joint source-channel coding (deep JSCC) has emerged as a promising approach for SC systems; however, its performance often degrades when the deployment channels differ from the training channel conditions, making robustness a critical requirement. This paper presents a structured overview of recent methodologies for enhancing the robustness of deep JSCC. Specifically, existing approaches are categorized into two classes: robust training approaches and adaptive approaches, with the latter further divided into adaptive semantic feature selection, physical-layer adaptation, and semantic feature adaptation. Finally, we discuss promising directions, including multi-task generalization and explainability in robust SC systems.

Carbon-aware grid optimization relies on accurate locational emission metrics to effectively guide demand-side decarbonization tasks such as spatial load shifting. However, existing metrics are only valid around limited operating regions and unfortunately cannot generalize the emission patterns beyond these regions. When these metrics are used to signal carbon-sensitive resources, they could paradoxically increase system-wide emissions. This work seeks to develop a sensitivity-consistent metric for locational average carbon emissions (LACE-S) using a neural representation approach. To ensure physical validity, the neural model enforces total emission balance through an explicit projection layer while matching marginal emission sensitivities across the entire loading region. Jacobian-based regularization is further introduced to capture the underlying partition of load buses with closely aligned generator responses. Moreover, we present a scalable zonal aggregation strategy, ZACE-S, to reduce the model complexity by mapping nodal inputs to predefined market zones. Numerical tests on the IEEE 30-bus system have verified the performance improvements of LACE-S in matching total emissions and their sensitivities over the non-regularized design. Crucially, while spatial load shifting driven by existing metrics often increases the post-shift emissions, the proposed LACE-S metric has led to a reliable reduction of system-wide emissions, demonstrating its excellent consistency with the global emission patterns.

Uplink performance remains a critical limitation in modern 5G networks, where UEs have to balance limited transmission power against propagation challenges. We conducted extensive measurements in the University of Notre Dame's football stadium, which has a seating capacity of 80,000 spectators, evaluating network behavior under both unloaded (pregame) and severely congested (game day) conditions, with a focus on uplink performance. Analyzing PHY-layer metrics captured via the Rohde & Schwarz QualiPoc, we show that high-frequency TDD bands in the uplink are severely bottlenecked in both the spectral and temporal domains. Despite transmitting near maximum 3GPP power limits, propagation loss inherent to high-frequency bands restricts UEs to low MCS indices and low PRB allocations, even in unloaded networks. This inability to achieve wideband allocation is further compounded by the significantly smaller number of uplink slots compared to downlink slots in TDD frames. Consequently, we observe a severe disparity between uplink and downlink: while high-frequency TDD bands carry the majority of downlink throughput, the network relies heavily on lower-frequency FDD bands for uplink. Additional measurements under favorable propagation conditions around a Verizon COW deployment located in the stadium parking lot also show that this limitation is not solely propagation-driven; rather, the duplexing scheme itself also plays a significant role. Even when TDD bands achieve higher or comparable MCS, FDD bands have a performance edge in the uplink due to the restrictive, downlink-heavy TDD architecture. These findings emphasize the indispensable role of low-frequency FDD spectrum in sustaining uplink capacity, providing insights that will help guide the design of next-generation wireless networks.

Expectation Propagation (EP) is a widely used iterative message-passing algorithm that decomposes a global inference problem into multiple local ones. It approximates marginal distributions as ``beliefs'' using intermediate functions called ``messages''. It has been shown that the stationary points of EP are the same as corresponding constrained Bethe Free Energy (BFE) optimization problem. Therefore, EP is an iterative method of optimizing the constrained BFE. However, the iterative method may fall out of the feasible set of the BFE optimization problem, i.e., the beliefs are not integrable. In most literature, the authors use various methods to keep all the messages integrable. In most Bayesian estimation problems, limiting the messages to be integrable shrinks the actual feasible set. Furthermore, in extreme cases where the factors are not integrable, making the message itself integrable is not enough to have integrable beliefs. In this paper, two EP frameworks are proposed to ensure that EP has integrable beliefs. Both of the methods allows non-integrable messages. We then investigate the signal recovery problem in Generalized Linear Model (GLM) using our proposed methods.

Accurate vehicular localization in Global Navigation Satellite System (GNSS)-denied environments, such as road tunnels, remains a key challenge for cooperative intelligent transport systems (C-ITS). This paper investigates single-anchor positioning by exploiting near-field (NF) propagation and passive radio-reflective structures. We first derive a geometric validity condition for the single-reflector NF (SR-NF) channel model, establishing a bound on the array size under which multipath can be consistently modeled by a single reflector, and linking it to Fresnel-region scaling. Building on this result, we propose JAVELIN, a single-anchor localization framework combining tensor-based NF parameter estimation, adaptive NF/far-field (FF) processing, and recursive Bayesian tracking. The method integrates angle, delay difference, and curvature measurements into a variable-dimension extended Kalman filter with gated nearest-neighbor (NN) association, enabling operation without prior environmental knowledge. Radio-reflective road markings (RRMs) are further introduced to enhance geometric diversity. Simulation results in realistic tunnel scenarios demonstrate accurate and robust localization under different line-of-sight (LoS) conditions, outperforming state-of-the-art single-anchor approaches and benefiting from passive reflector deployment.

Wireless physical neural networks (WPNNs) have emerged as a promising paradigm for performing neural computation directly in the physical layer of wireless systems, offering low latency and high energy efficiency. However, most existing WPNN implementations primarily rely on linear physical transformations, which fundamentally limits their expressiveness. In this work, we propose a relay-assisted WPNN architecture based on activation-integrated stacked intelligent metasurfaces (AI-SIMs), where each passive metasurface layer enabling linear wave manipulation is cascaded with an activation metasurface layer that realizes nonlinear processing in the analog domain. By deliberately structuring multi-hop wireless propagation, the relay amplification matrix and the metasurface phase-shift matrices jointly act as trainable network weights, while hardware-implemented activation functions provide essential nonlinearity. Simulation results demonstrate that the proposed architecture achieves high classification accuracy, and that incorporating hardware-based activation functions significantly improves representational capability and performance compared with purely linear physical implementations.

This paper introduces a novel joint communication and proactive monitoring (JCAM) system that simultaneously monitors multiple untrusted links and serves multiple legitimate users. The system leverages a cell-free massive multiple-input multiple-output (CF-mMIMO) architecture, where one subset of access points (APs) is dedicated to receiving signals from untrusted links, while another subset transmits data to legitimate users and jamming signals into the untrusted links. This dual functionality not only ensures reliable communication for legitimate users but also degrades the performance of untrusted links, thereby enhancing monitoring effectiveness. Closed-form expressions for the spectral efficiency (SE) of legitimate users and the monitoring success probability (MSP) are derived under partial zero-forcing (PZF) precoding/combining schemes with imperfect channel state information. Leveraging these expressions, we develop a simple yet effective AP mode assignment strategy that determines which APs perform downlink transmission and jamming, and which APs are dedicated to receiving signals from untrusted links. The objective is to maximize the MSP while satisfying predefined quality-of-service (QoS) requirements for all legitimate users. Numerical results show that the proposed mode assignment strategy significantly outperforms the benchmark, achieving up to a $32\%$ improvement in monitoring performance, while maintaining low computational complexity. Moreover, our proposed JCAM framework provides nearly a six-fold improvement in the minimum MSP over the co-located massive MIMO baseline.

Emotion is essential in spoken communication, yet most existing frameworks in speech emotion modeling rely on predefined categories or low-dimensional continuous attributes, which offer limited expressive capacity. Recent advances in speech emotion captioning and synthesis have shown that textual descriptions provide a more flexible and interpretable alternative for representing affective characteristics in speech. However, progress in this direction is hindered by the lack of an emotional speech dataset aligned with reliable and fine-grained natural language annotations. To tackle this, we introduce AffectSpeech, a large-scale corpus of human-recorded speech enriched with structured descriptions for fine-grained emotion analysis and generation. Each utterance is characterized across six complementary dimensions, including sentiment polarity, open-vocabulary emotion captions, intensity level, prosodic attributes, prominent segments, and semantic content, enabling multi-granular modeling of vocal expression. To balance annotation quality and scalability, we adopt a human-LLM collaborative annotation pipeline that integrates algorithmic pre-labeling, multi-LLM description generation, and human-in-the-loop verification. Furthermore, these annotations are reformulated into diverse descriptive styles to enhance linguistic diversity and reduce stylistic bias in downstream modeling. Experimental results on speech emotion captioning and synthesis demonstrate that models trained on AffectSpeech consistently achieve superior performance across multiple evaluation settings.

This letter presents a framework for space-to-ground wireless energy transfer (WET) for wirelessly chargeable devices (WCD) located in remote areas or disaster situations. We consider a grid of multi-antenna satellites that charge a WCD within line-of-sight. Closed-form expressions for harvested energy are derived considering maximum ratio transmission (MRT) ensuring that the WCD meets its circuit charging threshold $P_{th}$. Simulations elucidate that milli-joule-level energy can be harvested during satellite grid visibility, with charging efficiency influenced by the number of satellites, their altitude, charging frequency, and grid inclination.

Movable Antenna (MA) technology is emerging as a promising advancement with the potential to significantly enhance the performance of future wireless communication and sensing systems. In this paper, we address two-dimensional (2D) direction of arrival (DOA) estimation via joint shape-position optimization. Specifically, we formulate an optimization problem aimed at minimizing the Cramér-Rao Bound (CRB) based on a 2D DOA estimation model for MA systems. To tackle the highly non-convex nature of this CRB minimization, we investigate the spatial utilization of the movable region (MR) under minimum antenna spacing constraints. By demonstrating that an equilateral triangle yields the minimum overlap area, we strategically design an equilateral triangular MR. This specific geometric configuration enables the exploitation of structural symmetry to simplify the geometric constraints, which effectively reduces the complexity of solving the optimization problem. Subsequently, we derive the optimal MA positions by selecting the candidate locations farthest from the centroid of MR. The results demonstrate that the proposed joint shape-position optimization substantially enhances 2D DOA estimation performance.

We revisit the Gaussian broadcast channel (GBC) and explore the rate region achieved by purely discrete inputs with treating interference as noise (TIN) decoding. Specifically, we introduce a simple scheme based on superposition coding with identically and independently distributed (i.i.d.) inputs drawn from discrete constellations, e.g., pulse amplitude modulations (PAM). Most importantly, we prove that the resulting achievable rate region under TIN decoding is within a constant gap to the capacity region of the GBC, where the gap is independent of all channel parameters. In addition, we show via simulation that the weak user can achieve a higher rate with PAM than with Gaussian signaling in some cases.

Network virtualization and cloudification in Open Radio Access Networks (O-RAN) enable joint orchestration of the processing and fronthaul resources, which are essential for realizing the energy-saving potential of cell-free massive MIMO networks. To harness this potential, we investigate cell-free massive MIMO deployed over an O-RAN architecture with a wireless fronthaul that removes the need for fiber deployment. We first model the end-to-end power consumption under wireless fronthaul. Then, we propose a joint orchestration framework for radio, fronthaul, and processing resources that minimizes end-to-end power consumption while satisfying user-equipment (UE) rate requirements and wireless-fronthaul constraints. Two algorithms are developed: a scenario-sampling/group-Lasso method for centralized precoding and a block-coordinate descent method for distributed precoding. Numerical results show that centralized precoding significantly outperforms distributed precoding. End-to-end resource orchestration provides up to 70% energy-savings compared to cloud-only orchestration and up to 15% compared to radio-only orchestration. Moreover, distributing the same total number of antennas across the coverage area, rather than concentrating them at a few radio units (RUs), substantially reduces network power consumption, demonstrating that cell-free massive MIMO can deliver both high performance and high energy efficiency in future mobile networks.

In high-mobility 6G scenarios, rapidly time-varying channels lead to very short coherence times, which makes conventional pilot-based channel state information (CSI) estimation approaches prone to outdated information or excessive pilot overhead. Therefore, channel prediction becomes essential in such dynamic wireless systems. To address this challenge, large language models (LLMs) are emerging learning frameworks that have recently attracted attention for CSI prediction due to their strong sequence modeling capability and ability to generalize across different environments. This paper proposes an LLM-based framework for channel prediction in high-mobility orthogonal time frequency space (OTFS) communication systems. In this work, we develop a physics-aware LLM-based predictor that learns the temporal evolution of OTFS channel coefficients from historical channel observations while incorporating mobility-related physical descriptors (e.g., maximum Doppler frequency) to achieve accurate prediction of future channel states in rapidly time-varying environments. The effectiveness of the proposed framework is evaluated through extensive simulations under user velocities ranging from 100 to 500 km/h. Numerical results show that the proposed method consistently achieves lower normalized mean square error (NMSE) compared with both classical deep learning predictors and LLM-based predictors without physical channel descriptors. These results demonstrate the advantage of integrating mobility-related channel knowledge with LLM-based sequence modeling for channel prediction in highly dynamic OTFS systems.

This paper presents a measurement driven birth (MDB) model for the generalized labeled multi-Bernoulli (GLMB) filter. The MDB model adaptively generates target births based on measurement data, thereby eliminating the dependence of \textit{a priori} knowledge of target birth distributions. Numerical results are provided to demonstrate the performance.

We extend the algebraic diversity (AD) framework from classical signal processing to quantum measurement theory. The central result -- the Quantum Algebraic Diversity (QAD) Theorem -- establishes that a group-structured positive operator-valued measure (POVM) applied to a single copy of a quantum state produces a group-averaged density matrix estimator that recovers the spectral structure of the true density matrix, analogous to the classical result that a group-averaged outer product recovers covariance eigenstructure from a single observation. We establish a formal Classical-Quantum Duality Map connecting classical covariance estimation to quantum state tomography, and prove an Optimality Inheritance Theorem showing that classical group optimality transfers to quantum settings via the Born map. SIC-POVMs are identified as algebraic diversity with the Heisenberg-Weyl group, and mutually unbiased bases (MUBs) as algebraic diversity with the Clifford group, revealing the hierarchy $\mathrm{HW}(d) \subseteq \mathcal{C}(d) \subseteq S_d$ that mirrors the classical hierarchy $\mathbb{Z}_M \subseteq G_{\min} \subseteq S_M$. The double-commutator eigenvalue theorem provides polynomial-time adaptive POVM selection. A worked qubit example demonstrates that the group-averaged estimator from a single Pauli measurement recovers a full-rank approximation to a mixed qubit state, achieving fidelity 0.91 where standard single-basis tomography produces a rank-1 estimate with fidelity 0.71. Monte Carlo simulations on qudits of dimension $d = 2$ through $d = 13$ (200 random states per dimension) confirm that the Heisenberg-Weyl QAD estimator maintains fidelity above 0.90 across all dimensions from a single measurement outcome, while standard tomography fidelity degrades as $\sim 1/d$, with the improvement ratio scaling linearly with $d$ as predicted by the $O(d)$ copy reduction theorem.

We propose regularized approximate message passing (RAMP), a low-complexity algorithm for discrete signal detection in overloaded multiple-input multiple-output (MIMO) systems where the number of transmit antennas exceeds the number of receive antennas. While the state-of-the-art (SotA) iterative discrete least squares (IDLS) framework achieves near-optimal discrete-aware performance, its iterative matrix inversions impose a prohibitive $\mathcal{O}(M^3)$ complexity. RAMP resolves this by deriving an adaptive, state-dependent scalar denoiser that enforces arbitrary discrete constellation constraints within the approximate message passing (AMP) framework, reducing per-iteration complexity to $\mathcal{O}(NM)$. A robust variant is further proposed by incorporating an $\ell_2$-norm penalty, analogous to a linear minimum mean squared error (LMMSE) estimator, to enhance noise resilience. Simulation results under uncorrelated Rayleigh fading demonstrate that both proposed algorithms closely track their exact IDLS counterparts while avoiding the catastrophic failure of standard AMP in the overloaded regime, achieving steep bit error rate (BER) waterfall curves at a fraction of the computational cost.

The performance of constructive interference precoding (CIP) for multi-user multi-antenna (MU-MIMO) systems is governed by the structure of the constructive interference (CI) regions, yet this is overlooked in conventional constellation design. This work proposes the region-based constellation (RBC) model to lay the foundation for CIP constellation design. An RBC directly defines the mapping between messages and their feasible regions, instead of deriving them from an existing constellation. To provide insight for RBC design, we study the limitations of quadrature-amplitude-modulation (QAM)-based CIP. Analytical results show that the restrictive CI regions of QAM symbols are systematically misaligned with the objective-minimising sign pattern, resulting in a significant gap to the theoretical performance limit. From the perspective of improving sign alignment, two novel RBC schemes with non-convex feasible regions are proposed, namely mirrored-ends QAM (ME-QAM) and real-extended ME-QAM. A low-complexity algorithm is also developed for the resulting mixed-integer quadratic program, achieving a complexity comparable to QAM-based CIP. Simulation results with constellation sizes $\{16,64\}$ demonstrate up to $4$~dB signal-to-noise-ratio gain of the proposed schemes over QAM-based CIP. The proposed RBC model is also applicable to other systems with non-bijective modulation, representing a promising direction for future research.

With the growing demand for large-scale and high-quality data in edge intelligence systems, mobile robots are increasingly deployed to collect data proactively, particularly in complex environments. However, existing robot-assisted data collection methods face significant challenges in achieving reliable and efficient performance, especially in non-line-of-sight (NLoS) environments. This paper proposes a communication-and-learning dual-driven (CLD) autonomous navigation scheme that incorporates region-aware propagation characteristics and a non-point-mass robot representation. This scheme enables simultaneous optimization of navigation, communication, and learning performance. An efficient algorithm based on majorization-minimization (MM) is proposed to solve the non-convex and non-smooth CLD problem. Simulation results demonstrate that the proposed scheme achieves superior performance in collision-avoidance navigation, data collection, and model training compared to benchmark methods. It is also shown that CLD can adapt to different scenarios by flexibly adjusting the weight factor among navigation, communication and learning objectives.

This paper proposes a method for estimating a surface that contains a given set of points from noisy measurements. More precisely, by assuming that the surface is described by the zero set of a function in the span of a given set of features and a parametric description of the distribution of the noise, a computationally efficient method is described that estimates both the surface and the noise distribution parameters. In the provided examples, polynomial and sinusoidal basis functions were used. However, any chosen basis that satisfies the outlined conditions mentioned in the paper can be approximated as a combination of trigonometric, exponential, and/or polynomial terms, making the presented approach highly generalizable. The proposed algorithm exhibits linear computational complexity in the number of samples. Our approach requires no hyperparameter tuning or data preprocessing and effectively handles data in dimensions beyond 2D and 3D. The theoretical results demonstrating the convergence of the proposed algorithm have been provided. To highlight the performance of the proposed method, comprehensive numerical results are conducted, evaluating our method against state-of-the-art algorithms, including Poisson Reconstruction and the Neural Network-based Encoder-X, on 2D and 3D shapes. The results demonstrate the superiority of our method under the same conditions.

Fast Radio Bursts (FRBs) carry fossil information about non-Gaussianity generated during inflation. This primordial signal is most accessible on the largest scales, where the scale-dependent bias correction $\propto f_\mathrm{NL}\,H_0^2/k^2$ dominates, but where systematic effects are also strongest. A central challenge is the degeneracy between the intergalactic-medium electron bias $b_e$ and the primordial non-Gaussianity (PNG) signal, which can degrade $σ(f_\mathrm{NL})$ by orders of magnitude when $b_e$ is marginalised. We show this degeneracy can be broken internally by exploiting the cross-power spectrum $C_\ell^{DΔt}$ between the FRB dispersion measure (DM) field and Shapiro timing delays along multiple interferometric sightlines. The DM field traces the biased electron density, while the Shapiro timing signal probes the Newtonian gravitational potential independently of astrophysical bias. Their cross-correlation is directly proportional to $b_e$, independently of the matter power spectrum, providing a self-calibration of the electron bias. We derive $C_\ell^{DΔt}$ analytically in the Limber approximation and find a correlation coefficient $|ρ(\ell)|\approx 0.51$--$0.79$ across $\ell = 2$--$100$. A joint Fisher matrix analysis over $\{f_\mathrm{NL},\,b_e^0,\,z_\mathrm{fb}\}$ shows that including the cross-spectrum reduces $σ(b_e^0)$ by a factor of $2.1$--$5.1$ relative to a DM-only analysis. After full marginalisation, the joint analysis recovers $σ(f_\mathrm{NL})$ within a factor of $1.0$--$1.9$ of the fixed-bias benchmark, compared with $1.7$--$3.3$ degradation without the cross-spectrum. For a shallow survey with a 500\,AU baseline and $10^4$ FRBs, the joint constraint achieves $σ(f_\mathrm{NL})\approx 790$, within 4\% of the fixed-bias result and a factor $3.3$ better than the marginalised DM-only case.

Considering $Z_4$ symmetry in Type I seesaw scenario, one could obtain mass-squared differences of light neutrinos, mixings and $CP$ violating phase within $3 σ$ confidence level based on neutrino oscillation data. This is possible with only three independent complex parameters for allowed Yukawa couplings and one real mass parameter for heavy right handed neutrino fields around electroweak scale. After considering only three more real parameters as coming from small soft-symmetry breaking terms, the lightest right handed neutrino could be considered as dark matter candidate via freeze-in mechanism and the other two heavier right handed neutrinos through their decays, could generate the baryonic asymmetry of the universe naturally via resonant leptogenesis.

We extend a recently developed Big Bang Nucleosynthesis (BBN) code, {\tt PRyMordial}, to constrain a broad class of Grand Unified Theories to which BBN is sensitive, since these lead to varying fundamental couplings. A previously developed self-consistent perturbative analysis of the effects of these variations has been implemented in {\tt PRyMordial}, leading to robust constraints of the value of the fine-structure constant, $α$, at the BBN epoch using current observations of Helium-4 and Deuterium abundances. We explored two different viable scenarios, relying on alternative assumptions on the gravitational sector: the variation of the gravitational coupling can be implemented by varying either particle masses, or Newton's gravitational constant. For the variation of masses, we obtained at $68\%$ confidence level a constraint on the relative variation of $α$, between the BBN epoch and the present-day laboratory value, of $Δα/α=2\pm51$ ppm (parts per million), while for the variation of Newton's constant the analogous constraint is $Δα/α=2\pm22$ ppm. We also show that, given these constraints, these models do not provide a solution to the cosmological Lithium problem.

Joint analyses of high-resolution CMB temperature maps with galaxy surveys provide a unique way to reconstruct the radial velocity field of the underlying matter distribution via the kinematic Sunyaev-Zeldovich (kSZ) effect. Using data from the Atacama Cosmology Telescope (ACT) DR6 and the Dark Energy Spectroscopic Instrument (DESI) DR2, we present radial velocity reconstructions for luminous red galaxies (LRGs), emission-line galaxies (ELGs), and quasars (QSOs). Leveraging the spectroscopic data, we are able to reliably model the foreground contamination and report a negligible impact on our main observables. We detect the velocity-galaxy cross-correlation at $17.0σ$ for LRGs, and for the first time, at $8.3σ$ for ELGs and $6.8σ$ for QSOs. We further report the first detection of the velocity-velocity correlation using LRGs at $3.1σ$, as well as the highest cumulative detection of the kSZ effect to date at $20.8 σ$. Similarly to previous results, we find a lower amplitude of the kSZ signal compared to our fiducial halo model prediction and electron profile assuming a Battaglia profile. Combining these new observables, we obtain constraints on local-type primordial non-Gaussianity (PNG): $f_{\rm NL}^{\rm loc} = 15.9_{-34.4}^{+34.6}$ at 68\% confidence, which represents the tightest constraint to date derived from the velocity field. The measurements presented here already exhibit lower noise on a per-mode basis than the galaxy auto-correlation on the largest scales, $k<0.004~\rm{Mpc^{-1}}$, highlighting the key role these observables will play in the context of future CMB experiments such as the Simons Observatory.

Tidal disruption events (TDEs) can be observed when stars get too close to supermassive black holes and are torn apart and accreted. The delay time distribution of TDEs, or rate of TDEs as a function of time since a burst of star formation, can be used to determine what mechanisms influence the TDE rate. We compile a catalog of 41 TDE host galaxies with optical spectra, model the stellar populations with Bagpipes, and retrieve the age of the most recent burst of star formation to construct the delay time distribution of TDEs. TDEs occur more frequently in post-starburst galaxies than in other types of galaxies, though the mechanism causing this rate enhancement is unknown. We find that the TDE rate increases with post-burst age to reach a peak at ~1 Gyr relative to a control sample. We compare the observational TDE delay time distribution to theoretical models, which propose overdense stellar nuclei, radial anisotropies in stellar orbits, supermassive black hole binaries, and AGN disks as potential mechanisms that may enhance the TDE rate in post-starburst galaxies. Most models predict a TDE rate that declines with post-burst age, in contrast to our observational results, though some models are still feasible at certain ages (e.g., the black hole binary model matches at old burst ages and the stellar overdensity model matches at intermediate burst ages).

Using the hydrodynamical (fluid) approximation, we present a self-consistent theoretical framework that couples the origin, evolution and decay of spiral structures to the secular dynamical evolution of their host galactic disks. Our approach highlights non-resonant spiral excitation through azimuthal forcing that leverages mild, pervasive gradients in the disk's mass and angular momentum distributions, structural features we term cavernae. These cavernae are weaker but more widespread than the sharp features behind groove mode excitation and commonplace in exponential disks. We discuss how non-resonant features combine with other responses -- resonant dressing, steady waves, groove modes -- to produce a global, evolving spiral nexum that transports angular momentum and reshapes the disk. Using expressions for torques, angular momentum transport and heating, we demonstrate that global spirals are intrinsically self-limiting; the angular momentum changes and heating they generate quenches their own growth, dictating a finite lifetime for any single spiral episode. A succession of transient episodes, each with properties adjusted to the changed disk conditions, lays the pathway to long-lived spiral activity. This behavior suggests that the character of secular evolution shifts over time. We find that the short-lived, high-multiplicity (high-m) spirals that dominate in dynamically cold disks induce widespread, impulse-like non-resonant heating, yet with a low ratio of heating to radial migration. As the disk warms, high-m features are suppressed, leading to steadier, lower-m spirals that heat progressively more efficiently near resonances. In this light, the dynamical coldness of disk galaxies today requires a past dominated by high-m transient perturbations, whereas warmer, more compact systems reflect an advanced stage of evolution regulated by transient, low-m spirals.

Galactic cosmic rays (CRs) are a fundamental non-thermal component of the interstellar medium (ISM). Understanding the transport of super-high-energy particles is essential for interpreting observations of Galactic PeVatrons. Classical diffusion models assuming a homogeneous and isothermal medium oversimplify the multiphase ISM. We utilize high-resolution 3D MHD simulations to self-consistently generate a multiphase ISM, comprising the warm (WNM), unstable (UNM), and cold neutral medium (CNM), and investigate 1.5-15 PeV particle transport using a test-particle approach. We find that thermal phase transitions induce steep magnetic field strength gradients at phase boundaries, creating localized magnetic fluctuations that act as efficient sites for adiabatic mirror reflections and non-adiabatic pitch-angle scattering, strongly enhancing cross-field transport at these interfaces. However, because phase boundaries occupy only a small volume fraction and particles spend most of their trajectory in the weakly scattering WNM and UNM, the global pitch-angle scattering coefficient in the multiphase ISM is smaller than in an equivalent isothermal medium. This locally strong scattering nevertheless drives both parallel and perpendicular spatial diffusion coefficients to $\sim10^{30} {\rm cm^2 s^{-1}$ at 1.5~PeV, with the perpendicular component exceeding its isothermal counterpart ($\sim 10^{28}{\rm cm^2 s^{-1}$) by two orders of magnitude. Using a phase--phase diffusion matrix decomposition, we show that global CR transport is governed by the volume-filling, trans-Alfvénic WNM and UNM, where particles stream along stochastically wandering field lines. Cross-phase displacement correlations are universally positive, indicating cooperative transport between thermal phases. In contrast, the super-Alfvénic CNM acts as an efficient confinement that substantially suppresses local diffusion.

Using DESI SV3 spectroscopic group centrals together with deep HSC photometric data, we measure the conditional luminosity functions (CLFs) of central and satellite galaxies, separately for red and blue populations, in dark matter halos spanning $M_h\sim10^{12}- 10^{15}M_{\odot}$ and $0<z<1$. The depth of the HSC imaging enables CLFs to be measured to unprecedentedly faint limits, reaching $M_r \approx -15$ at $0.2 \leqslant z < 0.5$ and $M_r \approx -17$ at $0.5 \leqslant z < 1.0$. We find a remarkably weak evolution in the CLF of satellite galaxies over $0<z<1$. The Blue satellite CLFs are well described by a single Schechter function across all halo masses and redshifts, with a nearly constant slope of $-1.25\lesssim α\lesssim -1.2$. In contrast, red satellite CLFs exhibit a pronounced and ubiquitous faint-end upturn in all halo mass and redshift bins, with little evolution in the faint-end slope ($-1.8\lesssim α_f\lesssim -1.7$). These results indicate that the low-mass end of the red sequence in clusters/groups was already established by $z\sim1$. Both satellite characteristic magnitudes and central galaxy luminosities fade with time. Red central galaxies are consistent with passive evolution, whereas the luminosity evolution of blue centrals is dominated by ongoing star formation. Satellite galaxies evolve more rapidly than predicted by simple stellar population models, highlighting the importance of environmental effects. The quenched fraction of satellite galaxies as a function of stellar mass exhibits a universal minimum at $M_{\ast} \sim 10^9M_{\odot}$, independent of halo mass and redshift. We discuss possible interpretations of these results and their implications for galaxy formation and evolution.

Cygnus X-3 has recently been established as a variable ultra-high-energy(UHE) gamma-ray source with photons detected up to 3.7~PeV. The temporal correlation between its PeV activity and GeV flares, together with the possible orbital modulation, suggests that the emission is produced within or close to the binary system. In this work, we test whether the contemporaneous GeV emission zone can also host the acceleration of the parent protons responsible for the multi-PeV photons. We jointly model the contemporaneous \textit{Fermi}-LAT spectrum and orbital light curve with a one-zone leptonic scenario dominated by anisotropic external inverse-Compton scattering. The fit places the GeV emission region at $H\sim2.8\times10^{11}\,$cm and constrains the magnetic field--size product to $BH\lesssim10^{13.3}\,$G\,cm at the $3σ$ level. This implies a maximum proton energy of only $\sim0.3$~PeV from the Hillas criterion, far below that required by the observed PeV emission. We therefore conclude that the GeV zone cannot be the main PeV acceleration site. Instead, the PeV emission should originate from a more compact inner region, and the jet magnetic field must dissipate rapidly between the PeV and GeV emitting zones.

Power-law (PL) distribution functions (DF) are prevalent in highly diverse systems. The systems range in size from nanometer to mega light years, in complexity from dust grains to living organisms, and characterize the distribution of various events in nature and in various human activities. To gain some insight on why PL DF are so prevalent, we explore the conditions leading to the formation of a PL DF in a simple system of colliding hard sphere. We follow the time evolution of the energy DF through direct Monte Carlo simulations. In statistical equilibrium, the DF evolves into the Maxwell-Boltzmann (MB) DF. A transition to a PL DF occurs when: 1. The system is initially far from equilibrium. For example, a mix of light and heavy particles with the same velocity. 2. The system dynamics is scale-free, which holds in the intermediate asymptotic regime, far from the initial and the final equilibrium states. The scale-free dynamics leads to a DF which evolves in a self-similar form. 3. The system is open with a scale-free boundary condition. For example, a constant injection of particles far from equilibrium. The DF PL index is set by the time dependence of the self-similar DF and by the boundary condition. The PL index is independent of the self-similar DF form. Conditions 1-3 are common in a great variety of systems, which may explain why PL DF are so prevalent in nature.

We perform a comparative phase space analysis of the light mass Galileon model and standard Quintessence in the context of late--time cosmic acceleration. Focusing on a spatially flat FLRW background, we consider a cubic Galileon interaction supplemented by a scalar potential and examine three representative choices of the potential: a generalized cosh potential, a simple cosh potential, and a linear potential. By introducing suitable dimensionless variables, the cosmological field equations are reformulated as an autonomous dynamical system, allowing a systematic investigation of the stationary points and their stability properties. For the light mass Galileon scenario, we find that although the phase space admits scalar field dominated solutions, all critical points are of saddle type for the potentials considered. In particular, no stable late-time accelerating attractor emerges, even in the presence of de-Sitter like configurations. In contrast, the Quintessence limit admits stable de-Sitter attractors for cosh potentials, providing a viable description of the observed late--time acceleration. Our results highlight a key qualitative distinction between Galileon and Quintessence cosmologies and indicate that, within the light mass Galileon framework, the higher-order Galileon interactions may be required to realize a stable accelerating Universe.

Compact, steep-spectrum radio sources are key tracers of exotic astrophysical objects such as pulsars and high-redshift radio galaxies. All-sky radio surveys at different frequencies, like the TIFR-GMRT Sky Survey (TGSS) and the NRAO VLA Sky Survey (NVSS), have been usually exploited to identify such tracers. The more recent imaging survey, Rapid ASKAP Continuum Survey (RACS), with higher angular resolution and better sensitivity offers an avenue for a far better identification and characterization of compact, steep-spectrum sources. In this work, using publicly available RACS images at 887 MHz and 1.4 GHz, we present an image-domain characterization of 171 compact source candidates between declinations -40 degrees and +41 degrees, that were detected and appeared compact at 147 MHz in TGSS but not detected at 1.4 GHz in NVSS. Our detailed characterization resulted in the identification of 66 compact sources, 87 non-compact, diffuse or resolved sources, and 18 sources that are not detected in either of the RACS or NVSS images, implying spectral indices steeper than -2.0. Out of the 66 compact sources, 34 have spectral indices steeper than -1.5. We demonstrate that a large fraction of the sources in our sample were earlier not detected and resulted in incorrect spectral index limits due to poor imaging quality of NVSS in the Galactic plane. We present the spectral indices and morphological classification of all the sources in our sample and discuss their usefulness in identifying and studying interesting sources such as radio pulsars, high-redshift radio galaxies, and other extragalactic sources.

We infer recoil (kick) velocities for all binary black hole merger events reported up to the GWTC--4 catalog, together with candidate intermediate-mass black hole events. We obtain informative kick constraints for GW231028\_153006 ($839^{+1018}_{-681}\,\mathrm{km\,s^{-1}}$) and GW231123\_135430 ($974^{+944}_{-760}\,\mathrm{km\,s^{-1}}$). Additionally, we compute recoil velocities for recently reported events from the ongoing fourth observing run: GW241011\_233834, GW241110\_124123, and GW250114\_082203, obtaining $v_{\rm kick} = 974^{+555}_{-466}\,\mathrm{km\,s^{-1}}$, $394^{+582}_{-207}\,\mathrm{km\,s^{-1}}$, and $115^{+301}_{-95}\,\mathrm{km\,s^{-1}}$, respectively. The remnant of GW241011\_233834 is therefore inferred to have one of the largest recoil velocities among currently known events. We find that present recoil kick constraints are driven primarily by measurements of the mass ratio and spin magnitudes, while the contribution from spin orientation angles remains subdominant in most cases. We estimate typical retention probabilities of the remnant black holes in GWTC catalogs to be $\sim 1$--$5\%$ for globular clusters, $\sim 15$--$30\%$ for nuclear star clusters, $\sim 5$--$40\%$ for dwarf galaxies, and $\sim 70$--$100\%$ for elliptical galaxies. We further show that, even for remnants retained in globular clusters, recoil-induced spatial displacements from the cluster core are often significant, which can substantially suppress the chances of hierarchical mergers. We find that the probability for a GWTC merger remnant to participate in hierarchical mergers is $\sim 0.1$--$1\%$ in globular clusters and $\sim 1$--$15\%$ in nuclear star clusters.

Supercritical gas filaments in molecular clouds host the dense cores in which new stars form. However, the mechanisms governing their formation and subsequent gas accretion remain poorly understood. In this study, we conduct a statistical analysis of a large sample of sub-parsec supercritical filaments using H13COp J=1-0 data from the ALMA Three-millimeter Observations of Massive Star-forming regions (ATOMS) Survey. We identified velocity-coherent filaments in position-position-velocity (PPV) space and systematically examined velocity gradients both along and perpendicular to their skeletons. Our analysis uncovers a remarkable result: at scales of ~ 0.1-1 pc, the local velocity gradients within these supercritical filaments show no preferred alignment with the filament skeletons and exhibit no correlation with the local gravitational field. This random orientation suggests the presence of chaotic gas motions deep inside these dense structures. These findings may indicate that turbulence-rather than gravity-dominates gas dynamics and structural evolution at small scales, even in regions on the verge of star formation, challenging the paradigm of gravity-dominated structure formation within molecular clouds. This scenario should be further tested by more state-of-the-art simulations. This study offers key observational insights into the roles of turbulence and gravity in establishing the initial conditions for star formation.

We study oscillon/I-ball solutions in a real scalar version of the Friedberg-Lee-Sirlin (FLS) model. Using the two-timing analysis, we derive the conditions for oscillon solutions and explore multi-field oscillon configurations. In these configurations, the two fields form co-located oscillons that oscillate with frequencies set by their respective masses. These multi-field oscillons can be viewed as a bound state of two oscillons due to attractive interactions between the fields. We confirm these analytical predictions through numerical lattice calculations. This work extends the standard picture of single-field oscillons and may be relevant for cosmological scenarios involving multiple interacting real scalar fields.

We present a model-independent test of anisotropy in the cosmic distance duality relation (CDDR), $D_L=(1+z)^2 D_A$, using the Pantheon+ type Ia supernova sample and baryon acoustic oscillation (BAO) data. The angular diameter distance is reconstructed via Gaussian Processes, enabling an estimate of $η(z)=D_L/[D_A(1+z)^2]$ without assuming a background cosmology. We also allow for a possible isotropic evolution, parameterized as $η(z)=1+η_1 z$, and find a redshift-dependent deviation whose significance depends on the assumed supernova calibration. Anisotropy is modeled through a dipole modulation and constrained using a full covariance-based likelihood. To assess statistical significance, we construct null realizations that preserve both the redshift distribution and the survey selection function. We find that the observed dipole amplitude is consistent with isotropic expectations and lies below the levels induced by statistical fluctuations and survey geometry. We obtain a robust 95\% upper bound $A_{95}=0.025$, stable across different supernova calibration choices. We find no evidence for intrinsic anisotropy in the CDDR. Our results highlight the importance of accounting for survey selection effects in anisotropy searches and provide a viable framework for testing directional deviations in cosmological relations.

The Hubble tension is shaped not only by shifts between early- and late-time parameter estimates, but also by the stiffness of the constraints that define them. In this work, we analyze this geometric structure in the wCDM model by separating the discrepancy into two components: a parameter displacement and a directional Fisher curvature. Within the local Gaussian approximation, the quadratic tension along a given direction factorizes into the squared shift and the combined directional curvature contributed by the datasets. Applying this framework to Planck, DESI DR2, and SH0ES, we show that extending \LambdaCDM to wCDM primarily reshapes the Fisher geometry of the CMB constraint rather than opening a genuinely new route to concordance. Allowing the dark-energy equation-of-state parameter w to vary suppresses the leading Planck Fisher eigenvalue to only \sim 2.7 % of its \LambdaCDM value, while producing only a modest rotation of the dominant acoustic-scale eigenmode. The net effect is a strong softening of the effective acoustic rigidity. At the same time, high-precision late-time data, especially from DESI DR2, inject substantial curvature along the expansion-rate direction. This added stiffness acts as a geometric wall, closing off phantom-like escape routes and sharply limiting tension relief within the extended parameter space. Our results indicate that changes in the inferred H_0 tension under model extension are best understood as a reconfiguration of the constraint manifold rather than as evidence for new physical agreement. The shift-curvature decomposition thus offers a simple, fast, and physically transparent way to diagnose cosmological tensions.

Using $Λ$CDM and Padé-(2,1) cosmography, we study directional variations in the Hubble constant, $H_0$, using galaxy cluster and Type Ia Supernovae (from Pantheon Plus) by the hemisphere decomposition method. Since there is a degeneracy between $H_0$ and absolute magnitude $M_B$ for Supernovae, Cepheid host calibration is usually required to constrain $H_0$. Hence, in this work in order to complement the Cepheid host calibration in Supernovae, we also use calibrations based on galaxy cluster scaling relations. We find that there is a $\lesssim 1σ$ difference in $H_0$ variations when using galaxy clusters as calibrators compared to Cepheids highlighting that the variations in $H_0$ are robust across different calibration methods. Across all combinations of models and data sets used, we obtain a consistent deviation $\sim 2σ$ from isotropy. In nearly all cases, we notice that the maximum $ΔH_0$ aligns with the CMB dipole direction.

Gaia two-parameter (G2P) stars have cumulative errors in parallax and proper motion so great that only their mean positions were reported in DR3. One potential cause of these high errors is another star as indicated by two intensity peaks in the scans. Speckle interferometry astrometric measurements of 25 G2P stars with high multi-peak percentages were obtained with the 1.5m telescope at Mt. Wilson Observatory. Of the 25 observed G2P stars, seven had no reported Gaia companions within 5.0". We found nearby companions for all seven. The 18 other G2P stars had known Gaia companions within 2.0". Of these, 13 had separations that agreed closely with the speckle measurements but with some discrepancy in position angles, three stars did not agree in either separation or position angle and no companion was detected for the remaining two. Although some of these issues may be resolved in DR4 or DR5, others may be inherent limitations of Gaia capabilities that speckle interferometry observations may be able to fill.

We examine the impact of input neodymium (Nd) atomic data on the light curves and spectra of kilonovae, probing the sensitivity of kilonova observables to the atomic physics of this important lanthanide element. We use the SuperNu Monte Carlo radiative transfer code, simulating a simple semi-analytic 1D kilonova with a pure Nd atmosphere, fixing the radiative transfer method while using input atomic data generated by three different codes: the LANL suite of atomic physics codes, HULLAC, and Autostructure. We see that the choice of atomic data significantly shapes the resulting light curves and spectra. Peak bolometric luminosities differ by a ratio of nearly 1.5 between HULLAC/Autostructure and LANL data sets. Moreover, we observe significant near- to mid-IR differences in the structure of the spectra. We specifically attribute these differences to the choice of atomic data for neutral Nd I. Many of the results here have been adapted from a presentation at "Radiative Transfer and Atomic Physics of Kilonovae" in Stockholm, 2023. We additionally present a LANL data set with energies calibrated to available values in the NIST Atomic Spectra Database, and demonstrate that this calibration also significantly affects IR spectral structure at late time. The substantial differences in kilonova observables that arise from tuning the atomic data of just one lanthanide element highlight the special attention that must be paid to atomic physics uncertainties when modeling kilonovae, from AT2017gfo to beyond.

Bayesian inference of nanohertz gravitational-wave background models in pulsar timing array analyses often relies on Gaussian-process interpolators to avoid repeated, computationally expensive strain-spectrum calculations. However, Gaussian-process training becomes a bottleneck for large training sets. We test whether probabilistic neural networks can replace Gaussian processes in this role for both a self-interacting dark matter model and a phenomenological environmental model. We find that neural networks recover consistent posteriors while significantly reducing both training and Markov chain Monte Carlo runtime, with the largest gains for the more computationally demanding model.

Most barred galaxies exhibit symmetric structures. However, recent studies have shown that a subset of barred galaxies exhibit lopsided morphologies. To quantify their occurrence and investigate their physical origins, we analyze barred galaxies in the IllustrisTNG TNG50 simulation. We select 519 clearly barred galaxies in their stellar mass maps out of 770 barred galaxies from the TNG50-1 catalog. We classify the bar morphologies into four subgroups - `Lopsided', `Perturbed', `Symmetric', and `Indeterminate' - and perform a comparative analysis of their physical properties. We find that galaxies hosting asymmetric bars (`Lopsided' and `Perturbed') tend to have higher gas densities around the bar region, enhanced star formation activity, and more recent bar-formation epochs than galaxies with symmetric bars. However, the factor that most consistently distinguishes the four subgroups is the stellar mass distribution of the host galaxy, and there appears to be no physical correlation with bar size. These findings suggest that asymmetric bars form preferentially in less massive galaxies and may evolve into symmetric bars over time through secular processes. However, this conclusion should be considered with caution, as the fraction of asymmetric bars in the TNG50 simulation is systematically higher than that observed in the local universe.

We present a search for strong H$β$+[O III] emitters at $z=9.4-11.3$ in the HUDF using ultra-deep JWST/MIRI F560W imaging from the MIDIS survey. Three galaxies are identified via pronounced F560W flux excesses, consistent with strong rest-frame optical line emission. SED modelling yields rest-frame H$β$+[O III] equivalent widths of $\sim 600-1300$AA (median $\simeq 1260$AA), placing these sources among the most extreme known at these epochs. Combining these with a literature sample of 16 spectroscopically confirmed galaxies at $z\geq 9$, we find a median ${\rm EW}^{\rm Hβ+[O III]}_{\rm rest}\simeq 1300$AA, similar to values at $z\sim6-9$. We find no evidence for either a strong increase or decline in EW beyond $z\sim9$. A tentative trend of higher EW with increasing UV luminosity is observed, while no statistically significant anti-correlation with stellar mass is found. We place a first constraint on the H$β$+[O III] luminosity function at $z\simeq9-11$ ($Φ\sim10^{-3.4}\,{\rm Mpc^{-3}\,dex^{-1}}$ at $\log( L_{\rm Hβ+[OIII]}/{\rm erg\,s^{-1}})=42.5$), consistent with a decline relative to $z\sim7-8$. The MIDIS sources have $\log(ξ_{\rm ion}/{\rm Hz\,erg^{-1}})=25.1-25.4$. We find significant correlations between $ξ_{\rm ion}$ and EW and $β$, but not with UV luminosity, consistent with trends at lower redshift. These results suggest that the physical conditions governing nebular emission and ionising efficiency are already in place at $z\sim9-11$, extending trends established at $z\sim6-9$.

We present fully kinetic simulations of driven 2D turbulence in a relativistic plasma, designed for the first time to induce a fast magnetosonic cascade. As the driving strength increases, turbulence transitions from a weak wave-dominated regime to strong shock-driven dynamics. Using spatiotemporal Fourier analysis, we identify fast modes, finding that the weak turbulence regime exhibits spectral properties that are in excellent agreement with theoretical expectations. Our results are relevant for the modeling of turbulence in high-energy astrophysical plasmas.

We report the detection of weak high-velocity Na$\;$I absorption at V$_{\odot}$ = $-61.0\pm0.2$ km s$^{-1}$ in the spectrum of the stellar remnant at the center of the Galactic supernova remnant of 1181 AD. This velocity is not unlike that seen in old, more evolved SN remnants, but is much less than the remnant's $\simeq10^{3}$ km s$^{-1}$ expanding optical nebula. We briefly discuss its possible nature and origin.

We present a theoretical framework demonstrating a deterministic initialization mechanism for Warm Inflation via classical conformal boundary conditions. A persistent challenge in dissipative inflationary models is the "cold start" paradox: initializing the requisite thermal bath to generate the dissipative friction that subsequently sustains radiation production. Postulating an idealized, asymptotically scale invariant pre-inflationary phase, we mathematically prove that a conformal Weyl mapping to the emergent metric furnishes a finite, analytically derived initial radiation density. Implementing a spontaneous conformal symmetry-breaking ansatz, an emergent inflaton field is subjected to this inherited thermal bath. We analytically derive the initial kinematics of this framework, demonstrating that for strict sub-Planckian temperatures, the universe naturally initializes in the weak dissipative regime (Q << 1). The initial Hubble friction provided by the boundary radiation enables a smooth, deterministic kinematic handoff to the warm slow-roll steady-state attractor. As a mathematical proof-of-concept, this mechanism provides a fully realized framework to bypass the bootstrap problem of warm inflation.

We investigate the origin of the broad exponential wings observed in a significant fraction of the Halpha profiles of JWST-discovered little red dots (LRDs) and little blue dots (LBDs). Recent studies have shown that exponential broad-line profiles are not a prerogative of LRDs, are often also present in LBDs, and need not imply that electron scattering is the dominant broadening mechanism in every source. Motivated by our unification picture in which LRDs are the dust-reddened, high-inclination counterparts of compact blue broad-line AGNs, we model the broad Balmer emission with a virialized, radially stratified broad-line region (BLR). In this framework, the observed profile is the luminosity-weighted superposition of clouds spanning a range of radii and therefore a range of characteristic virial velocities. We show that such a stratified BLR can reproduce the extended exponential-like wings observed in three representative LRDs, without requiring electron scattering to be the primary origin of the broad wings. Our results support a picture in which the broad wings and the line cores encode different physics: the wings arise primarily from virial BLR stratification, whereas the cores retain additional imprints of absorption and radiative transfer in dense gas. The successful fits further suggest that the cloud radial distribution peaks near the dust sublimation radius, while the exponential wings are shaped by the line-emitting inner BLR shells where the higher virial velocities produce the high-velocity tails. This offers a simple physical explanation for the exponential wings of little dots, without invoking exotic new components or scenarios.

We present a Gaussian-process (GP) emulator for the monopole of the redshift-space halo power spectrum in $Λ$CDM cosmologies with massive neutrinos. The emulator is trained on 1000 COLA simulations distributed in a Latin-hypercube design over the six-dimensional cosmological parameter space $\{Ω_m h^2,Ω_b h^2,Ω_νh^2,σ_8,h,n_s\}$, with outputs at 11 snapshots spanning $0.5 \le z \le 2.0$. From redshift-space halo catalogues we measure shot-noise-subtracted monopole spectra over $0.01 \le k \le 0.50\,h\,\mathrm{Mpc}^{-1}$. We also generate 1000 fixed-cosmology realizations to estimate the covariance matrix and to construct synthetic data vectors for likelihood tests. On held-out cosmologies, the emulator reproduces the simulated spectra to typically better than $2\%$ across the scales and redshifts considered. Combined with its GP-based estimate of interpolation uncertainty, this speed and accuracy make the emulator well suited to repeated likelihood evaluations in Markov Chain Monte Carlo analyses. The resulting framework provides an efficient route toward neutrino-mass inference from DESI-motivated redshift-space clustering measurements.

X-ray variability provides a powerful probe of the innermost regions of active galactic nuclei (AGNs), offering valuable insights into the accretion process and the structure of the corona. Previous studies have established a correlation between the X-ray variability timescale, black hole mass, and luminosity, forming the AGN X-ray variability plane. A possible link between the X-ray spectral photon index and X-ray variability was noted in early studies but has rarely been incorporated into subsequent analyses of the variability plane. Moreover, the limited sample sizes in earlier works have limited the robustness and universality of the X-ray variability plane. In this work, we compile a sample of 112 AGNs with 399 exposures from the 4XMM-DR14 catalog and constrain the correlations between X-ray variability timescale, black hole mass, luminosity, and photon index using the recently developed fitting method, BADDAT {(Baseline-Aware Dependence fitting for DAmping Timescales)}, which enables a robust exploration of an extended parameter space. Our analysis confirms the dependence of the rest-frame variability timescale ($τ_{\rm rest}$) on black hole mass ($M_{\rm BH}$) and further incorporates the photon index ($Γ$) into the variability plane, yielding a best-fit relation of $\log (τ_{\rm rest}/{\rm s}) = 1.22\log (M_{\rm BH}/M_\odot) - 0.24Γ- 3.53$, which is strongly favored over the model with $M_{\rm BH}$ alone. In contrast, the inclusion of luminosity does not produce a comparable improvement. The correlation with $Γ$ likely reflects the effects of Comptonization and the geometry of the corona.

We derive explicit, algebraic expressions for the steady-state number density of cosmic ray electrons as a function of position and energy using Green's function of the diffusion equation with energy losses for an axisymmetric distributions of the particle sources in the galactocentric radius $r$ and distance to the mid-plane $z$. The solution is obtained for a Gaussian distribution of the particle sources in $r$ and $z$ but we show that it can be used for an arbitrary spatial distribution of the sources. The accuracy of our results is about 10% or better in a wide ranges of $r$ and $z$ and particle energies. These solutions can be used in the interpretation of radio astronomical observations of galaxies, particularly in the studies of the radio luminosities for large galaxy samples, and represent a physically justifiable and efficient alternative to the assumption of the energy equipartition between cosmic rays and interstellar magnetic fields.

Recent cosmological observations, including the latest Dark Energy Spectroscopic Instrument (DESI) data releases DR1 and DR2, have renewed interest in the possibility that dark energy may exhibit dynamical behavior rather than being a strict cosmological constant. In this work, we perform a fully model-independent reconstruction of the quintessence scalar field potential using Gaussian Process regression and current Hubble measurements. Instead of assuming a specific functional form for the scalar field potential, we reconstruct the quintessence potential and the corresponding kinetic energy directly from observational data. Our analysis is based on Hubble parameter measurements obtained from cosmic chronometers and the latest high-precision DESI DR2 baryon acoustic oscillation (BAO) data, together with Type Ia supernova data from the Pantheon+ compilation. Gaussian Processes provide a nonparametric and model-independent framework that allows the data to guide the reconstruction. We employ two covariance functions, namely the squared exponential and the Matern ($ν= 9/2$) kernels, in order to assess the sensitivity of the reconstruction to the kernel choice. We further explore the impact of background cosmological assumptions by considering different priors on the matter density and spatial curvature. Finally, we compare the reconstructed scalar field potential with two theoretically motivated benchmark models: a power law potential and an exponential potential. We find that both models remain consistent with the reconstructed potential within the inferred confidence intervals.

We introduce a numerical optimization method to extract the X-ray reverberation response functions from the multi-band light curves of the active galactic nuclei. This approach does not require prior assumptions about the accretion disc and corona geometry, provided that the light curves result from the superposition of direct and singly-convolved signals, consistently across all bands. By reformulating the light curve equations into the matrix form, the optimal response matrix is derived by minimizing the squared difference between the observed and reconstructed light curves using a gradient-based optimization algorithm. We demonstrate that the method can robustly accommodate up to two convolution processes, such as the reverberation and the propagation, simultaneously. When tested on the synthesized light curves, the method demonstrates robustness of the solutions to variations in the relative contributions of each light curve component as the recovered response kernel remains acceptably close to the ground truth, as evaluated by both the response geometry and the reconstructed light curves. The method's tolerance to random noise was also assessed. With appropriate denoising, the response kernel can be reliably recovered when the signal-to-noise ratio is at least $100$. We show, as a proof of concept, that the proposed method is geometrical-model independent and has the potential to offer a flexible complement to traditional approaches.

Temporal offsets between Gamma-Ray Bursts (GRBs) and high-energy neutrinos provide a useful probe of propagation effects in extreme astrophysical environments. We investigate whether such offsets can be generated by photon propagation through dense axion clouds gravitationally bound to magnetars. Working within the Euler-Heisenberg effective theory extended by the axion sector, we derive the modified photon dispersion relations in the presence of a strong magnetic background and an oscillating axion field. We show that axion-photon mixing turns the magnetized vacuum into an anisotropic birefringent medium, leading to geometry-dependent deviations from luminal propagation and kinematic time delays that reach $Δt_{\perp}\simeq1.33\times10^{-12}\,\mathrm{s}$ for orthogonal propagation. Although this effect is many orders of magnitude larger than the delays expected in diffuse astrophysical backgrounds, it remains far too small to account for the macroscopic offsets discussed in current multimessenger candidates. We further show that the same birefringent medium constrains the survival of the intrinsic linear polarization of prompt GRB emission, yielding the environmental bound $g_{aγγ}\lesssim6.02\times10^{-14}\,\mathrm{GeV}^{-1}$ for benchmark magnetar-scale parameters and axion masses near $m_a\sim10^{-4}\,\mathrm{eV}$. Magnetar-hosted axion clouds thus emerge as complementary environments in which dispersive transport and polarimetric observables jointly probe axion electrodynamics.

We assess whether intracluster light (ICL) can serve as an observational discriminator of dark matter physics. The self-interacting dark matter (SIDM) model has gained increasing attention as a possible resolution to small-scale discrepancies between collisionless cold dark matter (CDM) simulations and observations, predicting distinct tidal interaction histories within galaxy clusters. We analyze Cluster-EAGLE zoom-in galaxy clusters re-simulated from identical initial conditions in both CDM and SIDM frameworks. The morphological similarity between dark matter and multiple baryonic tracers -- gas, all stars, galaxies, and the combined brightest cluster galaxy plus ICL (BCG+ICL) -- is quantified using the Weighted Overlap Coefficient, a contour-overlap statistic. We find that dark matter is traced most accurately by BCG+ICL, followed by gas, all stars, and galaxies. The BCG+ICL component remains a robust tracer even at high redshift, while gas initially traces dark matter poorly but improves over time, eventually approaching the performance of BCG+ICL. Notably, in the SIDM case the gas distribution more closely resembles dark matter than in CDM. This reflects the underlying physics: in CDM, collisionless dark matter behaves similarly to the collisionless BCG+ICL, whereas in SIDM, self-interactions introduce an effective collisionality, making dark matter evolve more like the gas component. We also find that dwarf and satellite galaxies are more sensitive to the underlying dark matter model, despite their poorer overall tracing performance. Our results demonstrate the potential of ICL as a novel observational probe of dark matter physics and provide a first step toward using diffuse cluster light to constrain the nature of dark matter.

The inflaton equation of motion including one loop radiative corrections from spectator fields is obtained. We consider a massless scalar conformally coupled to gravity and a massless fermion Yukawa coupled to the inflaton as models for spectators that \emph{do not feature} gravitational particle production, their production during slow roll is solely a consequence of their coupling to the inflaton. The one-loop self energy and the fully renormalized equation of motion of the inflaton are obtained and solved explicitly for an inflaton potential $m^2\varphi^2/2$. The solution features Sudakov-type logarithmic secular terms, which are resumed via the dynamical renormalization group and compared to the solutions with a phenomenological friction term. During $N_e$ e-folds of slow roll inflation the inflaton evolves as $\varphi^{(0)}_{Isr}(t)\,e^{\frac{m^2Γ}{9H^3}\,N_e(t)}$ for the phenomenological friction term $Γ$ and $\varphi^{(0)}_{Isr}(t)\,e^{ΥN^2_e}$ with $Υ= -\frac{λ^2}{24π^2 H^2} ; \frac{y^2_R}{12π^2}$ for the radiative corrections from bosonic and fermionic spectators respectively where $\varphi^{(0)}_{Isr}(t)$ is the slow roll solution in absence of interactions, showing that a phenomenological friction term is not reliable. A generalization of the optical theorem to a finite time domain and cosmological expansion is introduced to obtain the distribution function $f(k,t)$ and total number of spectators produced \emph{during slow roll}. $f(k,t)$ is peaked at superhorizon scales and the total number of particles grows $\propto e^{3N_e}$. A non-perturbative mean field theory is introduced to describe the self-consistent evolution of the inflaton coupled to spectators, its linearized version reproduces the self-energy, the inflaton equation of motion and the results on particle production.

Flaring activity from the jets of active galactic nuclei has been studied for several decades, closely related to the loading and evolution of the jet. In this work, we focus on the sub-hundred parsec jet region, well traced by non-thermal radio and gamma-ray emission. Only in recent years have light curves capturing the decade-long behavior of such sources become available for a large ensemble of objects. While previous studies have focused on a direct correlation or few-month lag between gamma-ray and radio activity, recent neutrino-bright blazars observed by the IceCube Neutrino Observatory present multi-year delays between initial gamma-ray activity and subsequent radio flares. In this work, we search for similar-timescale correlations between Fermi-LAT gamma-ray data and RATAN-600 radio data from ~100 blazars. We consider two gamma-ray bands, 100 MeV-1 GeV and 1 GeV-500 GeV, as well as the integral band, to compare correlations between potential opaque and unabsorbed regions of the jet. Gaussian process modeling is used for smooth light curve function prediction. We also analyze morphological AGN core data from the MOJAVE survey, forming a sub-selection to better illustrate potential dependence on location. In the broader selection, several sources exhibit delayed flares on the order of 1-3 years. In the stacked analysis, we find the highest correlation for a radio delay on the order of 180 days. The stacked correlation resulting from the MOJAVE sub-selection corresponds to a slightly smaller time lag. Delayed radio flares or extended radio emission appear to be notable features within the general blazar population.

Gauss--Bonnet inflation typically affects the dynamics over an extended portion of the trajectory, making it difficult to isolate a controlled imprint at CMB scales. We consider a trajectory-selective coupling \(μ(φ,X)\) that gates the Gauss--Bonnet sector in phase space, enabling the higher-curvature contribution to be localized within a finite e-fold window while remaining negligible elsewhere. We identify stable inflationary solutions consistent with this localization and enforce standard ghost and gradient stability conditions for both scalar and tensor perturbations. For these viable backgrounds we compute pivot-scale observables and examine their dependence on the overall Gauss--Bonnet strength and on the kinetic gating. The framework offers a controlled route for realizing localized higher-curvature effects with predictable consequences for CMB-scale measurements.

Reconstructing the assembly history of the Milky Way requires precise constraints on the dynamical age of its merger remnants -- the time elapsed since a progenitor satellite was disrupted by the Galactic tidal force. We present a new framework to derive this dynamical age for disrupted stellar systems by extending the Fourier analysis of the orbital frequency distribution proposed by Gomez and Helmi. To overcome the smearing of frequency-space structures caused by observational noise, we introduce the Greedy Optimistic Clustering algorithm. This method allows for an optimistic exploration of the density contrasts in the orbital frequency space by taking into account the observational uncertainty in the data, effectively sharpening the signal required for age estimation. By applying this method to the Helmi stream, we derive a dynamical age of $6.8 \pm 0.8$ Gyr. Our derived accretion epoch is consistent with the observed kinematic properties of the Helmi stream. In particular, the marked asymmetry in the vertical velocity distribution -- where approximately two-thirds of the stars have negative $v_z$ in the solar neighborhood -- supports a relatively recent arrival. This suggests that the progenitor of the Helmi stream was accreted during an epoch of Galactic growth distinct from the much earlier Gaia-Sausage-Enceladus merger ($\sim 10$ Gyr ago). We validate our methodology using error-added mock simulations, demonstrating the reliability of our approach. Our results establish the Greedy Optimistic Clustering framework as a powerful chronometric tool for reconstructing the hierarchical assembly of the Milky Way using current and future high-precision astrometric datasets.

We present constraints on the Hubble constant ($H_0$) derived from the observed dispersion measure (DM) distribution of unlocalized fast radio bursts (FRBs). While localized FRBs with redshift measurements have been used to investigate the Hubble tension, their sample remains limited. Here we demonstrate that unlocalized FRBs -- which are far more numerous -- can independently constrain $H_0$ without requiring redshift information, as cosmic expansion imprints itself on their DM distribution. Analyzing a selected sample of 2124 unlocalized FRBs from the CHIME Catalog II, we obtain $H_0 = 73.8^{+14.0}_{-12.3}~\mathrm{km\,s^{-1}\,Mpc^{-1}}$ at the $1σ$ confidence level, corresponding to an uncertainty of about 18%. Breaking the degeneracy between $H_0$ and the characteristic cutoff energy $E_*$ of the FRB isotropic energy distribution would reduce this uncertainty to 9%. This work constitutes the first $H_0$ measurement derived solely from the DM distribution of unlocalized FRBs, highlighting their potential as a new cosmological probe. Future joint analyses with localized FRBs promise even tighter constraints.

We present the discovery of a low-mass companion located at $ρ$ $\sim$ 0\farcs{}85 ($r_{\rm proj} \approx 62~au$) from the early-type 1.2 Gyr-old star HIP 53005 using direct imaging data from the Subaru and Keck Telescopes and astrometry from the Hipparcos-Gaia Catalog of Accelerations. The companion, HIP 53005 C, is a component of a multiple system also including a $\approx$ 12\farcs{}4-separation M dwarf companion inducing a negligible proper motion acceleration. HIP~53005 C's position on color-magnitude diagrams, the fit of its spectral energy distribution to atmosphere models, and its location on an empirical mass-magnitude diagram all suggest that it lies at the M/L transition and near the hydrogen-burning limit ($\sim80~M_{\rm Jup}$). However, our orbital fitting combining direct-imaging relative astrometry with proper motion acceleration favors a much higher dynamical mass of $\sim185\ M_{\rm Jup}$. An additional unseen, more closely-orbiting companion below the detection limit (at $ρ\lesssim0\farcs2$)) may explain this discrepancy. Alternatively, HIP~53005C could be a low-mass binary like Gliese~229Bab, making this system an intriguing laboratory for studying multiple star formation.

The initial distance to PEARLSDG estimated from the Tip of the Red Giant Branch suggested it was an exotic isolated quiescent dwarf galaxy. We combine recent and archival Hectospec spectroscopy to place it at $z = 0.02843 \pm 0.00012$ ($D \approx 124$\,Mpc) within a galaxy group, revising the distance from 30\,Mpc to $\sim$124\,Mpc. We then carry out {\sc Prospector} SED fitting using parametric and non-parametric star-formation histories sampled with \texttt{dynesty}, \texttt{nautilus}, and \texttt{emcee}, recovering metallicity $\log(Z/Z_\odot) = -0.44^{+0.35}_{-0.06}$, stellar mass $\log_{10}(M_*/M_\odot) = 9.25^{+0.02}_{-3.73}$, and dust attenuation $\hatτ_V = 0.67^{+0.02}_{-0.05}$. The updated metallicity places PEARLSDG squarely on the standard mass--metallicity relation, resolving its former outlier status, with its quenched star-formation history consistent with environmental quenching in a group setting.

We study the performance of the flat $Λ$CDM model and the dynamical dark energy parameterizations $w_0$CDM and $w_0w_a$CDM, in which the dark energy (DE) equation of state is either constant ($w=w_0$) or redshift-dependent [$w(z)=w_0+w_a z/(1+z)$], without and with a varying CMB lensing consistency parameter $A_L$, using combinations of Planck PR4 CMB data (PR4 and lensing), and a compilation of non-CMB data composed of baryon acoustic oscillation (BAO) data that do not include DESI BAO data, Pantheon+ type Ia supernova observations, Hubble parameter measurements $H(z)$, and growth rate $fσ_8$ data. We also compare results from earlier Planck PR3 data with those obtained using PR4 data in order to assess the stability of cosmological constraints. For the largest data combinations, PR3/PR4+lensing+non-CMB, the cosmological parameters inferred from PR3 and PR4 data are consistent, almost all differing by $1σ$ or less. For the $Λ$CDM$+A_L$ model, we have $A_L=1.087 \pm 0.035$ for PR3 and $A_L=1.053 \pm 0.034$ ($1.6σ$ above unity) for PR4, which indicates that the CMB lensing anomaly is reduced when PR4 data are used. For the $w_0 w_a$CDM parameterization, we find $w_0 = -0.863\pm0.060$ (quintessence-like) and $w_0+w_a=-1.37^{+0.19}_{-0.17}$ (phantom-like), suggesting that the current observations favor dynamical DE over a cosmological constant at about $1.8σ$. For the $w_0w_a$CDM$+A_L$ parameterization, we find $w_0=-0.877\pm 0.060$ and $w_0 + w_a =-1.29_{-0.17}^{+0.20}$, corresponding to a preference for dynamical DE over a cosmological constant of about $1.5σ$ and with $A_L = 1.042 \pm 0.037$ exceeding unity at $1.1σ$. These results indicate that while the PR4 data mildly favor a time-evolving DE, part of this preference may be associated with possible residual excess smoothing present in the Planck PR4 CMB anisotropy spectra (abridged).

The extended main sequence (eMS) and extended main sequence turnoff (eMSTO) phenomena have been observed in some young and intermediate-age star clusters in the Milky Way and in the Magellanic Clouds. In this study, we conduct a survey of 53 galactic open clusters (OCs) to investigate the roles of stellar rotation, differential extinction, and cluster properties in the emergence of eMS and eMSTO. The projected rotational velocities are taken from the Gaia ESO spectroscopic survey and the Gaia DR3 line-broadening velocities. Stellar members of each OC are identified using the ML-MOC algorithm with Gaia DR3 astrometry. We divide clusters into four classes based on the color-rotation distribution, extinction, and MSTO morphology and report 14 clusters ($A_{\rm v} \lesssim 0.15$ mag, Class I) that exhibit split MS with fast and slow rotators populating the redder and bluer parts of MSTO. For the remaining clusters, differential extinction hampers the color-rotation distinction and also inflates MSTO width and therefore introduces a systematic offset in the MSTO-age relation. We also quantify the fraction of slow rotators among MSTO stars, finding a median value of $f_{\rm slow\, rot}^{v \sin i<100} \approx 0.41$ and the fraction reaching the spin-down limit, $f_{\rm slow\, rot}^{v \sin i<30}$, is $ \approx 0.08$. We find no statistically significant correlation between $f_{\rm slow\, rot}$ and either the binary fraction or cluster age.

Probing dark matter distribution on sub-galactic scales is essential for testing the Cold Dark Matter ($Λ$CDM) paradigm. Strong gravitational lensing, as one of the most powerful approach by far, provides a direct, purely gravitational probe of these substructures. However, extracting cosmological constraints is severely bottlenecked by the mass-sheet degeneracy (MSD) and the unscalable nature of manual and neural-network modeling. Here, we introduce LensAgent, a pioneering training-free, large language model (LLM)-driven agentic framework for the autonomous physical inference of mass distributions. Operating as an autonomous scientific agent, LensAgent couples high-level logical reasoning with deterministic physical modeling tools, demonstarting successful reconstruction of mass distribution in SLACS Grade A strong lensing systems. This self-evolving architecture enables the robust extraction of sub-galactic substructures at scale, unlocking the cosmological potential of upcoming wide-field surveys such as the Rubin Observatory (LSST) and Euclid.

Scalar-induced gravitational waves (SIGWs) provide a powerful probe of inflationary dynamics on scales far smaller than those accessible to the cosmic microwave background and large-scale structure. In scenarios with a transient ultra-slow-roll (USR) phase, the curvature power spectrum can be strongly enhanced on small scales, potentially generating an observable stochastic GW background. In this regime, scalar dynamics during inflation can become nonlinear, challenging the validity of standard perturbative predictions. Existing semi-analytical calculations of SIGWs rely on the linear evolution of inflation fluctuations. In this work, we compute SIGWs from USR inflation using lattice simulations. We evolve the inflaton field fully nonlinearly during inflation and extract the curvature perturbation nonperturbatively, then simulate its post-reheating horizon re-entry by evolving the Newtonian potential linearly while retaining the full non-Gaussian structure of the initial conditions for the primordial fluctuations in the tensor source. For moderate non-Gaussianity, the semi-analytical prediction captures the correct order of magnitude of the GW signal but receives important corrections. When inflationary non-Gaussianities are large, it can fail dramatically in both amplitude and spectral shape, independently of the overall size of the tensor power spectrum. Our results show that reliable predictions of SIGWs in such scenarios require nonperturbative control of the inflationary scalar dynamics. The code used for this work is available at https://github.com/caravangelo/inflation-easy.git.

Rest-frame ultraviolet (UV) spectra of Little Red Dots (LRDs) often show Ly$α$ emission. Along with broad Balmer emission, LRDs are expected to produce broad Ly$α$ emission. However, the large column density of neutral gas invoked to explain the Balmer break should significantly redshift and further broaden the Ly$α$ line, making it challenging to detect without sensitive, moderate-resolution spectra. We present ultra-deep (29 hours) G140M JWST/NIRSpec observations covering the rest-UV of two LRDs in Abell2744 from the SPURS Cycle 4 Large Program. One of our targets is Abell2744-QSO1, a gravitationally-lensed LRD at $z=7.04$ with faint UV emission (M$_{\rm UV}=-16.9$), and the other source (UNCOVER-2476) is newly-confirmed at $z=4.02$ with a very bright UV continuum (M$_{\rm UV}=-19.6$). We find that Abell2744-QSO1 has a broad Ly$α$ profile, along with narrow CIV, FeII$\lambda1786$, and OI$\lambda1302$ emission. The Ly$α$ profile suggests an origin similar to the broad H$α$, but the line is considerably less redshifted than expected from existing dense gas models. We show that the line profile can be explained if the dense neutral gas is clumpy, allowing Ly$α$ to escape by scattering off of the clump surfaces. We find that UNCOVER-2476 has narrow [NeIV] emission, indicating either a hard radiation field or shocks. We confirm two close neighbors with Ly$α$ emission around Abell2744-QSO1, indicating it traces a dense environment that may have ionized its surrounding IGM. We suggest that LRDs may preferentially trace bubbles carved by their dense environments, contributing to the prevalence of Ly$α$ in the population.

The diversity of galaxy morphologies and their relations with galaxy and halo properties is fundamental to understanding galaxy formation. Cosmological simulations of representative volumes can help disentangle the origin of observed correlations, but most suffer from two main limitations that affect morphologies: an over-pressurised interstellar medium and spurious interactions between stellar and dark matter particles. We present an overview of galaxy morphologies in the COLIBRE simulations, which address these limitations and reproduce many observed galaxy scaling relations. To quantify galaxy morphology, we use four (strongly-correlated) theory-space metrics, three kinematic and one spatial. We explore how different choices and limitations affect these indicators, including luminosity- versus mass-weighting, aperture size and shot noise. Overall, we find good convergence in present-day morphologies across two orders of magnitude in mass resolution. COLIBRE predicts that kinematic morphology correlates strongly with stellar mass and colour, and that galaxies with stellar masses of $\approx(1-2)\times 10^{10}\,\mathrm{M}_{\odot}$ tend to be the most rotationally-dominated. At fixed stellar mass, the morphology of central galaxies correlates weakly with the properties of their host halo. Morphology correlates more strongly with internal galaxy properties, with more disky galaxies being more gas-rich, having higher star formation rates and exhibiting younger and more extended stellar populations. Other properties, like the mass of the most massive black hole, the fraction of stars that are accreted and stellar metallicity, also correlate with morphology, but with correlation strengths sensitive to the stellar mass of the galaxy and whether it is a central or satellite.

We aim to confirm the compact nature and constrain the radio spectra of candidate persistent radio sources (PRSs) associated with repeating fast radio bursts (FRBs). We performed European VLBI Network (EVN) observations at 5 and 8 GHz targeting two candidates identified in a recent VLA survey. We measured flux densities and upper limits at milliarcsecond resolution and combined them with published VLBI data at lower frequencies to derive spectral constraints. We detect a compact source associated with FRB 20190417A at 5 GHz with a flux density of $150\pm45$ uJy, while no detection is obtained at 8 GHz. The source is unresolved and has a brightness temperature $T_{\rm b} \gtrsim 10^{6-7}$ K, confirming its non-thermal nature. Combining our measurement with VLBI data at 1.4 GHz, we derive a spectral index $α= -0.19 \pm 0.29$, consistent with a nearly flat spectrum. This makes FRB 20190417A only the second PRS with a spectral index constrained using VLBI data. The inferred luminosity places the source on the proposed $L_ν$-|RM| relation. Including this source yields a scatter of $σ_Δ= 0.65$, corresponding to $\hatα|ε| = 1.5 \pm 0.7$, consistent with forward shocks in the free-expansion phase or young pulsar wind nebulae. For the candidate PRS associated with FRB 20181030A, we report upper limits of 80 uJy at 5 GHz and 150 uJy at 8 GHz, corresponding to $L_{5\,\mathrm{GHz}} \lesssim 3.8 \times 10^{25}\ {\rm erg\ s^{-1}\ Hz^{-1}}$, and implying a steep spectral index ($α\lesssim -1.2$) if the VLA emission arises from a compact component. Our results highlight the importance of VLBI in isolating compact emission from FRB engines and provide one of the few spectral constraints for PRSs at milliarcsecond resolution. The consistency of FRB 20190417A with the $L_ν$-|RM| relation supports a nebular origin for the persistent emission.

Population inference in gravitational-wave astronomy allows us to connect individual detections to the astrophysics of compact objects and their environments. Current approaches employed for population inference with LIGO-Virgo-KAGRA data approximate evaluation of the hierarchical population likelihood via post-processing of individual-event posteriors. However, the case of the Laser Interferometer Space Antenna (LISA) will be more complex for two main reasons: the transdimensional "global fit" approach to LISA data analysis which models all signals and noise simultaneously, and the presence of both individually-resolved signals and the unresolved stochastic ``Galactic foreground" arising from the Galactic binary population, which induces a circular dependence between the resolved and unresolved systems and our ability to detect the former. These challenges are not without opportunity; LISA's data will contain every mHz compact binary in the Milky Way -- either individually or within the Galactic foreground -- with great potential for Galactic and stellar astrophysics. We therefore propose an alternative approach: direct evaluation of the full hierarchical population likelihood within the LISA global fit. We develop a statistical formalism for joint inference of individually-resolved gravitational-wave sources, an unresolved stochastic foreground, and a shared, underlying astrophysical population, present PELARGIR, a prototype GPU-accelerated population inference module for the LISA global fit, demonstrate the formalism and PELARGIR via a toy model analysis, and lay out a roadmap towards an astrophysically-motivated LISA global fit with embedded population inference. While we apply the formalism here to the population of LISA Galactic binaries, it is applicable across the gravitational-wave spectrum with use cases in pulsar timing and next-generation terrestrial observatories.

Simulations predict that bars in galaxies should slow down over time. This is often attributed to the exchange of angular momentum between the bar and other regions of the galaxy, such as the outer disc and dark matter halo, which implies that galaxies with a more massive halo or disc should be able to slow down the bar more efficiently. However, observational evidence for this process has been limited. In this work, we provide observational support for the slowing down of bars as predicted by simulations. We combine bar kinematics measurements obtained with the Tremaine-Weinberg method and host galaxy mass estimates derived from Jeans anisotropic models for a sample of 30 galaxies from the MaNGA survey. We find a statistically significant anti-correlation (>4sigma) between the bar pattern speed and both the stellar and total dynamical mass, which suggests that the slowest bars reside in the most massive galaxies. However, while the slope of the best-fit line between the pattern speed and dark matter mass is negative, it is not statistically significant (2.43sigma). We also find that bars with lower pattern speeds have more extended NFW dark matter profiles with lower central densities. Additionally, we find statistically significant correlations (>3sigma) between the corotation radius and the stellar mass, dark matter mass, and total dynamical mass. Finally, we find no significant correlations that involve the dark matter fraction or R, likely due to the inherent challenges associated with measuring these specific parameters accurately.

PSR J1928+1815 is a 10.55 ms millisecond pulsar in a 3.6 hr orbit with a massive ($1.0$-$1.6\,M_{\odot}$) companion that produces extended radio eclipses. The companion, proposed to be a stripped helium star, is undetected in optical and infrared surveys. We present deep near-infrared imaging using Keck/NIRC2 with laser guide star adaptive optics. No source is detected at the pulsar position down to a $5σ$ limit of $K_s \approx 21.3$. Using stripped-star atmosphere models and conservative extinction estimates, we show that any plausible helium star companion would have been detected, ruling out this interpretation. A massive white dwarf (WD) companion remains consistent with the non-detection. We consider two possible origins for the eclipses: (1) absorption in a wind driven by a young, hot WD, and (2) material ablated from the WD by the pulsar. The former can naturally arise following Case BB mass transfer, which produces $\sim 1.2\,M_\odot$ WDs capable of sustaining winds of $\dot{M} \gtrsim 10^{-12}$-$10^{-13}\,M_\odot\,{\rm yr}^{-1}$ for $\sim 10^4$-$10^5$ yr, sufficient to obscure the pulsar at GHz frequencies. The latter requires efficient coupling of the pulsar's spin-down luminosity to the companion to drive the needed mass loss, which may be difficult to achieve. If the eclipse is powered by a WD wind, the system is likely observed in a short-lived phase; alternatively, if the companion is an older WD, the origin of the eclipsing material remains unclear. The apparent uniqueness of PSR J1928+1815 is consistent with a short detectability lifetime, though formation rate estimates remain uncertain.

We investigate whether Big Bang nucleosynthesis (BBN) remains compatible with the Covarying Coupling Constants plus Tired Light (CCC+TL) cosmology. In this framework, only quantities with explicit length dimensionality covary through a universal scaling function $f \left( z \right)$, while dimensionless constants and dimensionless ratios remain invariant. At the redshifts $z$ relevant to BBN, $f \left( z \right )$ approaches a constant plateau $f_{\text{max}} \left( z \right) \simeq 3$, and the tired-light contribution is negligible, so the early-time dynamics reduce to a global rescaling of dimensioned quantities. In particular, the Hubble expansion rate $H$ at fixed temperature $T$ satisfies $H_{\text{CTL}} \left( T \right) = f^{-1}_{\text{max}} H_{Λ\text{CDM}}\left( T\right)$, implying a longer cooling time $Δt$ between weak freeze-out and the onset of nucleosynthesis by the same factor (CCC+TL labeled as $\textit{CTL}$). We find that BBN predictions are preserved provided the relevant interaction rates $Γ$ and decay rates governing the neutron lifetime $τ_n$ share the same plateau scaling as $H$, so that governing combinations such as $Γ\text{/}H$ and $\text{exp} \left( -Δt \text{/} τ_n \right)$ remain invariant. Implementing these plateau rescalings in the Kawano/NUC123 network (via a single control parameter $\texttt{fctl} \equiv f_{\text{max}}$) yields identical light-element abundances for $\texttt{fctl}= 1$ ($Λ$CDM) and $\texttt{fctl} = 3\left( \text{CCC+TL} \right)$ to within $10^{-3} - 10^{-4}$ level, consistent with numerical rounding. We also illustrate that adopting the lower late-time CCC+TL baryon density from the Pantheon+ data fit can reduce the ${}^7\text{Li}$ discrepancy but simultaneously increases D/H, implying that BBN alone does not select between the late-time baryon-density inferences considered here.

High-resolution X-ray spectroscopy provides a powerful probe of the interstellar medium (ISM), giving direct access to the composition and physical state of dust grains and atomic species in dense environments. We present a study of the gas and dust along the line of sight to the bright low-mass X-ray binary GX 340+0, which samples higher-density gases in the ISM. Using deep Chandra/HETG spectra, we characterize X-ray absorption fine structure from dust, gas absorption lines, and the optical depths of the Si, S, and Fe K-edges. By fitting these three edges simultaneously, we reduce degeneracies in the dust composition and find that amorphous olivine dominates the fractional contribution among the dust columns ($\sim$65%), followed by metallic iron ($\sim$19%), iron sulfides (pyrrhotite and troilite; $\sim$10%), and fayalite ($\sim$5%), with the remaining species contributing only a few percent in total. From the inferred stoichiometry, we estimate that $\sim$74% of Fe is associated with silicates, $\sim$8% with sulfides, and $\sim$18% with metallic iron, suggesting that Fe is predominantly incorporated in iron rich silicate grains along this sightline. We detect S ii absorption and infer a sulfur dust fraction of $\sim$35%. We also detect absorption structure near the Ca and Ar K edges, highlighting the need for improved atomic photoabsorption data. The Chandra/HETG spectral resolution remains essential to disentangle dust and gas contributions at the Si and S K edges, providing a benchmark for dust characterization in high-density regions in the ISM.

Contemporary all-sky surveys have observed thousands of extragalactic transients in the nearby universe, and upcoming surveys will discover exponentially more at higher redshifts. With these large samples, population-level analysis of the photometric behavior of different transient classes is now possible, allowing for photometric classification and physical parameter inference from relatively sparse individual light curves. To enable such studies, we introduce Gaussian process Optimized Photometric Regression of Extragalactic Archival Ultraviolet-infrared eXplosions, a.k.a GOPREAUX--a Python package for Gaussian Process Regression of multi-wavelength transient photometry. Our modeling is unique in that it interpolates transient emission across phase and wavelength in a non-parametric, data-driven way. This allows for predictions of light curves and spectra at higher redshifts, where the rest-frame ultraviolet (UV) emission is redshifted into the observer-frame optical or infrared. To this end, we aggregate a sample of almost 1,300 transients observed in the UV and optical with the Neil Gehrels Swift Telescope, complemented with additional optical and infrared coverage from surveys such as ZTF and open-source data releases. Our sample includes 275 Type II SNe, 172 stripped-envelope SNe, 72 superluminous SNe, and 58 tidal disruption events, among other classes. Our code and reduced photometry--comprising over 146,000 photometric observations--are available as open-source software and data products. Here we discuss our sample criteria, data reduction and modeling methodologies, the multi-wavelength light curves and spectral templates produced by our models, and the future directions in photometric classification and physical parameter inference this code and data repository enables.

It has been suggested that "Little Red Dots" (LRDs) might be accreting black holes enshrouded by dense gas in a nearly closed geometry, which completely covers the central black hole, leading to an atmosphere-like structure known as the "black-hole star" ($\rm BH^\star$). We test this scenario by analysing new JWST spectroscopy in the far ultraviolet (FUV, rest-frame) of the prototypical LRD Abell2744-QSO1, at $z=7.04$. We found the presence of broad Ly$α$ emission with an FWHM of $\sim 1000$ km/s, and detections of OI, CIV, and/or FeII emission lines. The NIRCam imaging and NIRSpec slit images indicate that the low-velocity component ($v\lesssim 200$ km/s) of Ly$α$ is likely spatially extended, but the high-velocity component ($v\gtrsim 200$ km/s) of Ly$α$ remains unresolved. Based on the multi-component kinematics and flux of Ly$α$ relative to Balmer lines, we conclude that the observed line profile is unlikely to be broadened by subsequent resonant scattering through the interstellar medium. This suggests that the high-velocity component of Ly$α$ originates in the broad-line region, although resonant scattering in the dense gas likely makes Ly$α$ broader than H$α$ as observed. The nebular features of this LRD indicate that there is at least one relatively optically thin direction where Ly$α$ can escape from the broad-line region (BLR). We also found indications that photons from the BLR are powering fluorescence of FeII and OI on a larger physical scale. The FUV features thus challenge the fully-covered geometry interpretation and suggest that there are "holes" in the $\rm BH^\star$, or the absorbing medium is simply clumpy.

We perform fully relativistic GRMHD simulations of magnetized, self-gravitating black hole-disk (BHD) systems in which the black hole spin is misaligned with the disk angular momentum. Massive disks (disk to BH mass ratios of $16-28\%$) around rapidly rotating black holes ($χ\lesssim 0.97$) develop a nonaxisymmetric instability for tilt angles from $0^\circ$ to $180^\circ$. Magnetic stresses damp, but do not completely suppress, the nonaxisymmetric instability, and corresponding gravitational wave (GW) emission, in aligned systems, while they enhance it in antialigned BHDs: MRI-driven turbulence enhances angular momentum transport and accelerates nonlinear instability evolution in misaligned configurations. All models launch magnetically driven jets consistent with the Blandford-Znajek (BZ) mechanism, with collimation depending on spin orientation. The GWs reflect strong nonaxisymmetric structure from a persistent $m=1$ mode. The coupling between fast MRI and the slower nonaxisymmetric instability growth governs the outcome, with tilt controlling how MRI modifies the global mode. These simulations provide the first self-consistent GRMHD treatment of tilted, self-gravitating BHD systems and support their role as multimessenger sources.

This chapter provides an almost comprehensive guide to the Imaging X-ray Polarimetry Explorer (IXPE) data analysis. The chapter briefly introduces the IXPE spacecraft and the instrument onboard; subsequently, the strategies adopted to extract polarimetric information and to optimize the response are given. Moreover, the data formats and processing steps to avoid potential systematic errors and to achieve the best results from the IXPE data are reported. Both the model-independent and the model-dependent analyses are summarized, as the instrument response functions and their different available flavors. The idea behind this chapter is to collect suggestions and answers to questions that are typically raised by users, aiming to enable researchers to maximize the scientific return from IXPE observations.

Galaxies evolve in tandem with their environments -- mergers and gas inflows drive galaxy growth while galactic outflows launched by supernovae may seed the galactic environment with gas, metals, and energy, fueling star-formation far from the main bodies of galaxies. The formation histories of young stars in the stellar halos of nearby galaxies can help understand this interplay. We thus present the most detailed map to date of young stars in the stellar halo of M82, a starburst galaxy in the M81 Group that hosts a prototypical outflow, using Hubble Space Telescope (HST) and Subaru Hyper-Suprime Cam observations. We find widespread extraplanar populations of stars with ages $\lesssim630$ Myr, with clear detections of stars up to $\sim5$ kpc to the south in unique arc-like stellar features (Southern Arcs) and in a new stellar trail up to $\sim20$ kpc to the east (M82's Tail), originating from the Southern Arcs. We estimate a total halo star formation of $\sim4\times10^6\,M_\odot$ in the last $630$ Myr. Overall, the star formation history (SFH) of the M82 Tail is correlated with periods of heightened star cluster formation in the M82 disk, which suggests the influence of the starburst outflow. Further, the fraction of young stars decreases as we move away from M82 to the east. We forward a picture where the M82 Tail formed from ram pressure stripped gas arising from M82's westward motion, triggered by shocks from the outflow.

At Stage-III sensitivities, cosmic shear $B$ modes unambiguously indicate systematic contamination and are often used to inform data selection and scale cuts for cosmological inference. We validate $B$ modes for the Ultraviolet Near-Infrared Optical Northern Survey (UNIONS)-3500 (2894 deg$^2$, $n_\mathrm{eff} \approx 5.0$ arcmin$^{-2}$) using three $E$/$B$-separable statistics: pure-mode correlation functions $ξ_\pm^{\mathrm{B}}(θ)$, Complete Orthogonal Sets of $E$/$B$-mode Integrals (COSEBI) $B$-mode amplitudes $B_n$, and harmonic-space power spectra $C_\ell^{BB}$. For each statistic, we compute probability-to-exceed (PTE) values over a two-dimensional grid of scale-cut boundaries; our adopted cuts lie in broad stable regions of acceptable PTE. $B$-mode detections and PTE failures on initial catalog versions led us to investigate galaxy size cuts and stellar halo masking. After cuts, all three statistics pass the null test (minimum PTE $= 0.18$). Before scale cuts, we measure an oscillatory COSEBI $B$-mode pattern consistent with repeating additive shear bias, a detector-level effect seen across multiple Stage-III surveys including CFHTLenS, which used the same MegaCam camera; scale cuts that exclude the charge-coupled device (CCD) angular scale suppress it. Although these statistics probe the same two-point shear field, scale cuts in one do not map exactly onto cuts in another, because their respective filter functions weight angular scales differently. The most conservative validation therefore requires scale and sample selections that pass null tests across all frameworks simultaneously, an approach that applies directly to Stage-IV surveys where systematic errors dominate.

We revisit the derivation of the Rhoades-Ruffini bound on the upper limit for the maximum mass of neutron stars and find that the assumption made there for the onset of an ultimately stiff phase of high-density matter is not stringent. Relaxing this assumption and allowing for an onset of stiff non-nucleonic matter under neutron star constraints at the saturation density or below boost the upper limit of the theoretically possible maximum mass to $4~M_\odot$ or higher, in the mass-gap region between neutron stars and stellar-mass black holes. We provide a fit formula for the dependence of this upper limit on the speed of sound and the onset density of the deconfinement transition.

The first Euclid Quick Data Release (Q1) provides extensive imaging and spectroscopic data for hundreds of millions of photometric objects across several deep fields. Accurate classifications and photometric redshifts (photo-z) for these sources are crucial to maximizing the value of these data. In this work, we perform source classification and photo-z estimation for the Euclid Deep Field North (EDF-N) around the North Ecliptic Pole, using a deep learning framework (DeepDISC) that learns and infers using 9-band images simultaneously. We train three dedicated models for (1) source detection and classification, (2) galaxy photo-z, and (3) quasar photo-z. The Euclid Q1 input source catalog, and classifications and spectroscopic redshifts (spec-z) from the Dark Energy Spectroscopic Instrument Data Release 1 are adopted as our training data. DeepDISC source detection achieves overall completeness of ~93% and purity of ~80% if using the Euclid source catalog as the ground truth. Using a JWST source catalog within EDF-N as the reference, we estimate a true purity of ~ 90% for DeepDISC sources. About 99.2%, 99.0%, and 84.8% of stars, galaxies, and quasars, respectively, are correctly recovered with their spectroscopic classifications. The DeepDISC photo-zs show good agreement with spectroscopic redshifts, for both galaxies and quasars. Comparisons with other Euclid Q1 products demonstrate that DeepDISC provides comparable or improved performance in source detection/deblending, classification and photo-z, especially for quasars. These results demonstrate the potential of pixel-level deep learning approaches for large-scale sky surveys such as Euclid and Roman, which will continue to improve with better training labels. We release the full DeepDISC source catalog (~13 million objects) for EDF-N with classifications and photo-zs, including photo-z probability distributions.

The redshift drift provides a kinematic test of the cosmic expansion history through the slow time variation of the redshift of comoving sources. Motivated by the expected Sandage-Loeb measurements from future facilities, we investigate the drift within a cosmographic framework, modeling the Hubble rate through both a second-order Taylor expansion and a $(2,1)$ Padé approximant. We constrain the cosmographic parameters $(H_0,q_0,j_0)$ by combining Pantheon+ and SH0ES type Ia supernovae with gamma-ray bursts and then examine the impact of adding baryon acoustic oscillation measurements from the second DESI data release. The resulting constraints are used to construct a mock Sandage-Loeb catalog, after which the analyses are repeated including the simulated drift data. In this way, we assess the internal consistency of the reconstructed background rather than perform an independent forecast. Accordingly, we find that, for the SNeIa+GRB analysis, the Taylor reconstruction is compatible at the $1σ$ level with the $ω_0ω_1$CDM scenario, whereas the Padé parameterization improves the agreement of $q_0$ with the $Λ$CDM paradigm. Once DESI BAO data are included, the agreement with the reference background models weakens to the $2σ$ level. The addition of the mock Sandage-Loeb sample mainly tightens the bounds on $q_0$ and $j_0$, with moderate shifts in the central values. We finally compare the reconstructed redshift drift with the corresponding behavior predicted by the $Λ$CDM and $ω_0ω_1$CDM scenarios.

The potential association between gravitational waves (GWs) and fast radio bursts (FRBs) offers a unique multi-messenger probe for cosmology. In this paper, we develop a redshift-independent framework to constrain cosmological parameters using the luminosity distance - dispersion measure relation, accounting for realistic astrophysical uncertainties. We perform a comprehensive comparative analysis across different GWs detector sensitivities and modeling assumptions. Specifically, we investigate the performance of the current LIGO-Virgo (LV) network (at $z < 0.2$) versus the future Cosmic Explorer (CE). Our study further evaluates the impact of different dispersion measure (DM) distributions -- specifically the corrected Macquart's PDF (Zhuge+2025) and the log-normal distribution -- and explores the influence of including or excluding host galaxy DM contributions. Using realistic simulated observations, we find that while the current LV network lacks the precision to provide meaningful constraints, CE will enable high-precision cosmology. Even without spectroscopic redshifts, CE observations can effectively break parameter degeneracies and robustly constrain both cosmology and host galaxy parameters. These results highlight the necessity of next-generation detectors.

We investigate how isochrones computed with different input physics and initial conditions affect the age dating of the open cluster NGC 6866, and compare the results with asteroseismic ages derived from Kepler photometry. Using Gaia DR3 data, we identified 180 cluster members with a clustering algorithm. We then developed an isochrone-cloud fitting method that accounts for a range of free parameters in the input physics. Variable stars were subsequently identified among the cluster members. For 19 g-mode pulsators, we carried out modelling with a dedicated grid of rotating stellar models, constrained by spectroscopic and photometric parameters, the asymptotic gravity-mode period spacing, and the near-core rotation rate. We considered two cases: modelling each pulsator individually and modelling them under the assumption of a common cluster age. PARSEC and MIST isochrones yield discrepant ages of 690 and 467 Myr, respectively. The isochrone-cloud fit indicates an initial critical rotation distribution peaking at 0.6, about a factor of two higher than inferred from asteroseismology. The seismic modelling shows agreement between seismic and isochronal masses, but substantial differences in the derived ages due to differences in internal mixing. When the g-mode pulsators are modelled with a shared cluster age, we obtain 759 Myr, consistent with the PARSEC isochronal age. We conclude that age dating of open clusters is sensitive to the adopted input physics and initial conditions, highlighting the need for better calibrated stellar evolutionary models.

We investigate the co-evolution of the stellar and dark matter mass of satellite galaxies using the COLIBRE cosmological hydrodynamical simulations with subhaloes resolved by the history-based HBT-HERONS subhalo finder. We identify a universal tidal track connecting stellar mass loss to subhalo mass loss characterized by two distinct phases, which can be well described by the two-parameter model. The initial phase consists primarily of dark matter stripping, whereas stellar stripping becomes significant only after the subhalo bound mass fraction drops below a critical value ($\sim 0.057$). We find a bimodal mass loss rate distribution of subhaloes. In satellites with modest mass loss rates, the stellar mass is largely frozen. By contrast, the galaxy quickly becomes unresolved, along with the dark matter component for the extreme-mass-loss population, naturally explaining the lack of ``orphan'' galaxies in previous hydrodynamical simulations. Our model also predicts the formation condition for dark-matter-deficient galaxies (DMDGs), whose abundance peaks at $m_{*}\sim 10^{9.5}\,\rm{M}_{\odot}$. The abundance of DMDGs can be very sensitive to numerical effects, with COLIBRE resolving a much larger DMDG population than previous hydrodynamical simulations. We also estimate the influence of artificial disruption on the satellite stellar mass function, which can amount to 20 (50) per cent at $m_* \sim 10^{9} (10^{8}) \, \rm M_\odot$, given a baryonic mass resolution of $\sim 10^{6}\,\rm{M}_{\odot}$.

Dark-matter haloes do not form in isolation but within the surrounding cosmic web. By the time a halo begins to collapse, its larger-scale environment has typically collapsed along two axes, forming filaments that channel anisotropic infall toward the halo. In this work, we derive from first principles the characteristic Lagrangian scale ratio at which such an anisotropic tidal field most strongly influences halo formation. Specifically, we identify the inflection point of the conditional probability that the tidal field, smoothed on a scale Rsd, undergoes two-dimensional compression, given the presence of a density peak of rarity nu on a smaller scale Rpk. For a standard LambdaCDM cosmology, we find (Rsd/Rpk)infl = 2.2 + (nu-2.5) for Rpk corresponding to a tophat filter of 8Mpc/h. This result implies that the anisotropic tidal influence on a collapsing halo typically extends to 2-3 times the size of its Lagrangian patch. Recast as a function of formation redshift z, the characteristic filament scale around 2.5 sigma peaks can be approximated by Rsd(z) = 31 /(2+(1+z)**2)Mpc/h. We provide practical scaling laws for selecting dynamically relevant smoothing scales in large-scale surveys and for setting initial patch sizes in high-resolution zoom simulations.

We aim to understand the progenitor channels of CCSNe via a statistical study of the ages of their environments. We compiled a large and minimally biased sample of 129 CCSNe discovered by untargeted wide-field transient surveys and with archival VLT/MUSE integral-field-unit spectroscopy. We measured the local Hα luminosity within a 300-pc aperture centered on the SN explosion site as an empirical proxy for the environmental age. We find that the environments of Type II(P), IIb and Ib SNe do not show a significant age difference while Type Ic SNe are located in systematically younger environments than the other types (i.e. II $\approx$ IIb $\approx$ Ib > Ic). This is inconsistent with some previous reports of monotonically younger CCSNe environments with increasing envelope stripping (II > IIb > Ib > Ic). Our result suggests that Type Ic SNe have much younger and more massive progenitors than the other CCSN types and they likely originate from a distinct progenitor channel. The distinction between Types II(P), IIb and Ib SNe is insensitive to progenitor mass and mainly due to the different binary separation; in contrast, Type Ic SNe predominantly require much higher-mass progenitors accompanied by close companions with large mass ratios and/or much stronger stellar wind that depends sensitively on progenitor mass.

Ionization is a major driver of both physical and chemical evolution in protostellar systems. Recent observations reveal substantial chemical processing in protoplanetary disks by the time the surrounding envelope has cleared. Thus, physical conditions during the preceeding phase, when an infalling envelope of material is still present, are crucial for determining the extent of chemical processing at early stages. We used observations of H13CO+ and C18O from the Northern Extended Millimeter Array (NOEMA) and IRAM 30m telescope to constrain the ionization rate in the envelopes of three Class 0 protostars: NGC-1333 IRAS4A, L1448-C, and L1157. We find ionization rates in the range zeta = 1e-16 - 1e-13 s$^{-1}$ , several orders of magnitude above the ionization rate of zeta = 6e-17 s$^{-1}$ in the diffuse interstellar medium. This supports the idea that ionization driven chemistry is more efficient at earlier stages (< 1e5 years) of protostellar evolution.

We investigate the synthetic model of globular cluster (GC) systems of 17 compact massive galaxies (CMGs) from the Illustris TNG100 simulation to explore their connection with massive relic galaxies, systems that have undergone little structural evolution across cosmic time. The co-evolution of the GC systems and their host galaxies is based on a GC formation and evolution model that assigns clusters to stellar particles according to age and local conditions, providing positional, kinematic, and chemical information for individual GCs. By combining stellar assembly histories, effective radius evolution, and GC properties such as in-situ vs. ex-situ origin, metallicity, and spatial distribution, we identify consistent signatures of early formation and late-time accretion. We find that the GC mass fraction traces the host assembly history more robustly than the GC number fraction, as massive clusters better preserve the imprint of the early accretion history. Three CMGs from TNG100 emerge as strong massive relic analogs, exhibiting high in-situ GC fractions, narrow metallicity distributions, and compact spatial distributions. A tight correlation between the host stripped fraction and the extent of the ex-situ GC population further reveals the possibility to consider GC spatial profiles as a signature to identify tidal stripping processes. These results indicate that the combined analysis of GC populations and host stellar assembly offers a robust diagnostic for identifying massive relic galaxies and constraining their evolutionary histories.

We discuss flat Friedmann-Lemaitre-Robertson-Walker (FLRW) metric-affine cosmology where the metric and connection as well as the matter energy-momentum and hypermomentum all obey the symmetry of spatial homogeneity and isotropy. In particular, we outline a scenario where a dark dust fluid carries spin hypermomentum which makes its effective equation of state dynamical and might relate to the DESI DR2 data.

We conducted a homogeneous chemical analysis of pre-main sequence stars with effective temperatures ranging from $\sim$ 3000 K to $\sim$ 5500 K in eight nearby star-forming regions (SFRs): Chamaeleon I, $η$ Chamaeleonis, Lupus, Orion OB1a, Orion OB1b, $σ$Orionis, Taurus, and Corona-Australis. Our study aims to: 1) derive the lithium abundance (A(Li)) and highlight the impact of veiling correction on both A(Li) and age determination; 2) perform the iron (Fe) and barium (Ba) abundance analysis in regions with scarce previous measurements; 3) investigate the possible Ba enhancement. The analyzed data were obtained as part of the PENELLOPE Large Program using the ESPRESSO, UVES, and X-Shooter instruments. We measured the equivalent width of the lithium line (EWLi) at $λ$ = 6707.8 Angstrom, from which A(Li) is derived using the curves of growth method. The Fe and Ba abundances have been measured through spectral synthesis analysis. Using the EAGLES code, we derived an upper limit on the age of the eight SFRs. Our findings underscore the necessity of veiling corrections on EWLi, which can shift A(Li) and age estimates by up to $\sim$ 0.7 dex and $\sim$ 20 Myr, respectively. Accounting for veiling, the A(Li) distributions peak in a range between 3.3 and 3.8 dex for most clusters, and the upper age limit is approximately 5 Myr for all SFRs. We successfully measured the mean iron and barium abundances in Lupus, Taurus, Cha I, and $η$ Cha, showing slightly sub-solar iron abundance, and a clear Ba overabundance, with [Ba/H] values reaching up to 0.75 dex.

The supernova remnant (SNR) G25.1-2.3 was identified in the radio band during the Sino-German $λ$6 cm survey of the Galactic plane. We present a detailed investigation of the optical, HI, and CO emission towards the G25.1-2.3 to better understand its characteristics and environment. In this study, optical spectroscopic data of the remnant and its environment have been analysed for the first time, providing new insights into their emission properties. The Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST) and 1.5-m Russian-Turkish Telescope (RTT150) data show variations across the observed regions, with [SII]/H$α$ ranging from 0.16 to 0.83. We identified shock-heated gas in the northern and southern regions and several photoionized regions around the SNR based on their [SII]/H$α$ ratios derived from spectra. The [SII]$λ$6716/$λ$6731 ratio observed in the northern region suggests electron densities ($n_{\rm e}$) ranging from 120 to 1030 cm$^{-3}$, whereas the southern regions show higher values, between 490 and 4500 cm$^{-3}$. The variations in the observed H$α$/H$β$ ratios indicate significant differences in extinction across the regions. H$α$ images obtained using the 1-m Turkish Telescope (T100) reveal optical emission in the northern and southern, characterized by filamentary and diffuse structures. We newly found a hole-like distribution of HI, whose spatial extent is roughly consistent with the diameter of the SNR. Based on radio data, we examine the evolutionary stage of G25.1-2.3 using the surface brightness-diameter ($Σ-D$) relation and the equipartition method.

In this letter, we report a novel discovery of unique characteristics for the tidal dwarf galaxy (candidate) Holmberg IX via a multiwavelength investigation. New observations are taken for deeply mapping Hα emission and combined with archival/published data for comprehensively probing dust, gas, and stellar populations in this galaxy. We find in Holmberg IX a dearth of dust incompatible with its rich gas and metal; globally young stellar populations with prominent FUV but deficient and marginal Hα emissions, distinct from other tidal dwarf galaxies ever known. By assuming normal IMF, Holmberg IX is suggested to be born ~ 130 Myr ago from a bursty star formation event which then rapidly ceased, with very few stars formed in the past ~ 80 Myr that demarcates a lower age limit for the galactic mainbody; current star formation occurs only in outskirts, bringing a conundrum about the reason for the recent quenching in such a gas-rich environment. Contradicting general expectation for tidal dwarf galaxies hosting continuous star formation, the present quiescence implies Holmberg IX currently staying in a rarely-seen transient period. Without star formation continuing, Holmberg IX is likely transforming into a dwarf spheroidal galaxy, or oppositely into a(n) (ultra-)diffuse system which will probably dissolve in the end. Instead, if Holmberg IX possesses peculiar IMF and hosts low-mass, weak-Hα star formation, it is able to maintain long-term survival in its current status. On whichever evolutionary pathway in reality, Holmberg IX appears as a special case updating conventional understandings of tidal dwarf galaxies and hinting potential existence of similar analogs in the universe.

The early dark energy resolution of Hubble tension seems to be suggesting a scale-invariant Harrison-Zeldovich spectrum of primordial scalar perturbation, i.e. $n_s=1$ ($|n_s-1|\sim {\cal O}(0.001)$) for $H_0\sim 73$km/s/Mpc. In this work, we propose a possibility to acquire $n_s=1$ in single field slow-roll models of inflation. In our consideration, the potential of inflaton during inflation still preserve the shape of well-known single field inflation models in deep slow-roll region, but inflation ends suddenly due to a large step of inflaton potential. In particular, we investigate the implication of our scheme for chaotic inflation and Starobinski inflation, and show how they can be compatible with the observation for $n_s=1$.

Investigating the properties of the first stars in the universe is essential, yet it remains an open question. One way to explore these stars is by examining their effects on the surrounding gas during the epoch of reionization. In this study, we investigate whether the 21-cm global signal can constrain the typical mass and star formation efficiency of first-generation stars. We perform semi-numerical simulations that include the escape fraction of ionizing photons, which depends on stellar and halo masses, as well as the heating structure surrounding a halo that hosts the first star, determined by radiation hydrodynamics (RHD) simulations. By applying Fisher analysis, while accounting for foreground emissions, we demonstrate that future observations with instruments such as the Radio Experiment for the Analysis of Cosmic Hydrogen (REACH) could provide meaningful constraints on these properties.

We review selected aspects of the interior physics of compact stars, focusing on the microscopic and macroscopic manifestations of spin, magnetic fields, and nucleonic superfluidity and superconductivity. Spin statistics of fermions allows quantum degeneracy pressure to determine the stability and global properties of neutron stars, whose structure depends sensitively on the strong interactions among baryons in dense matter. Using a generic meta-modeling framework based on an expansion of the nuclear energy density around the isospin-symmetric and saturation-density limits, we highlight how various lesser-known terms in this expansion affect compact-star observables and review multimessenger constraints on mass, radius, and moment of inertia. The influence of magnetic fields on dense matter is examined, showing that substantial effects in their structure require extremely strong fields, whereas lower fields are sufficient to affect their superfluid phases. At the mesoscopic scale, the coexistence of superfluid and superconducting components features vortex and flux-tube lattices, with pinning and mutual friction processes playing central roles in neutron-star rotational dynamics. We discuss unresolved issues concerning vortex structure, flux-tube configurations, and the origin of pulsar glitches and post-glitch relaxation. We also briefly address the possible emergence of deconfined quark phases in compact-star cores, including their color-superconducting properties, as well as the associated vortex structures and magnetic-field responses in such phases.

The point-particle approximation is foundational to modelling clustering of matter in the universe, but is fundamentally inconsistent within General Relativity due to associated spacetime singularities. This bottleneck has historically restricted the study of matter clustering to linear scales. We resolve this by utilising the recent observation that a matter horizon precedes the formation of caustics in expanding spacetimes. This allows for the isolation of singularities via spacetime surgery. By glueing distinct spacetime sheets related by a discrete transformation across the shared boundary, we derive a covariant backreaction term that contributes to the effective energy-momentum tensor. We demonstrate that the spacetime backreaction contribution modifies local particle trajectories, naturally producing flat galaxy rotation curves in the outskirts without the need for dark matter particles.

Unlocking the full physical information encoded in low-resolution spectra poses a significant challenge for astronomical survey analysis. Such a task demands modeling spectra and optimizing astrophysical parameters in high-dimensional space, as a consequence of line blending. Here we present PhDLspec -- a deep learning framework embedded with physical priors for stellar spectra modeling and analysis. By imposing differential spectra derived from ab initio stellar atmospheric model calculation on a transformer framework, PhDLspec can rigorously and precisely model stellar spectra by simultaneously taking into account more than 30 physical parameters, at a computational speed hundreds of times faster than ab initio model calculation. With such a flexible stellar modeling approach, PhDLspec can effectively derive ~30 stellar labels from a low-resolution spectrum using affordable optimization techniques. Application to LAMOST spectra (R~1800) yields stellar elemental abundances in good agreement with high-resolution spectroscopic surveys, following essential calibrations to correct systematic biases in elemental abundance estimates using wide binaries and reference high-resolution datasets. We provide a catalog of 25 elemental abundances for 116,611 subgiant stars with precise age estimates. The successful application of PhDLspec to LAMOST spectra for high-dimensional parameter determination sheds light on similar challenges faced by other surveys and disciplines.

The Euclid Q1 Strong Lensing Discovery Engine identified 497 galaxy-scale lens candidates across 63 deg^2, yet none received spectroscopic characterization. We present the first spectroscopic redshifts for this sample from NISP SIR slitless spectroscopy (1.25-1.85 um, R~450) at zero additional telescope cost. Of 579 published Q1 lenses, 473 fall within SIR coverage. We detect emission lines in 461 systems and measure secure source redshifts (>=3 lines) for 419, deflector redshifts (>=2 absorption features) for 199, and complete (z_src, z_def) pairs for 178 systems (148 with dual-grism confirmation). Source redshifts span 0.70 < z_src < 2.88 (median 1.59); deflector redshifts span 0.24 < z_def < 2.47 (median 1.06). This is the largest single-campaign spectroscopic lens characterization to date, exceeding SLACS (85), BELLS (25), SL2S (~56), and AGEL (139) individually, with no dedicated follow-up time required. We present a quality-tiered catalog (148 gold-complete, 188 gold-source, 108 silver, 17 bronze) for mass modeling and lens statistics. Extrapolating to Euclid's 14,500 deg^2 wide survey implies ~100,000 spectroscopic lens redshifts -- three orders of magnitude beyond existing samples.

Low eccentricity, short orbital period pulsar companions may provide a probe to study novel dense and stable exoplanet internal compositions due to the potentially significant orbital evolution they experience caused by strong gravitational tides. We model the tidal characteristics such as apsidal motion constants, orbital precession, and tidal deformability for a variety of equations of state to be compared with values recovered via pulsar timing for a sample of four systems: PSR J1719-1438b, PSR J0636+5128b, PSR J2322+2650b, and PSR J1807-2459A b. With this method, we hope to place stringent limits on the chemical and structural composition of these objects. Through limiting the internal composition of pulsar companions, we aim to elucidate their unique history and formation.

By combining multi-band data from Gaia DR3, MWISP CO, and LAMOST DR11 LSR/MSR, we investigate the co-evolution of stars and their parent molecular cloud in a snake-like stellar structure, named Snake III. Based on 5-D phase-space selection, we identified 5683 member stars (median age 7.6 Myr) across approximately $300 \times 500 \times 175$ pc$^3$ volume, along with 12 embedded open clusters. Then we use BEEP distances combined with $^{12}$CO velocities to clearly identify the molecular clouds associated with the stellar complex in spatial and kinematics. The molecular cloud density increases with Galactic longitude, with older open clusters forming in cavities near higher-density regions (except ASCC 125), while young field stars currently form preferentially in present-day high-density environments, indicating that cloud density regulates the star-formation sequence. $^{12}$CO excitation temperature, centroid velocity, velocity dispersion and H$α$ emission reveal that early feedback first compresses cloud edges to trigger new stars, then sweeps and disperses the parent clouds. The extremely young cluster (ASCC 125, 4.4 Myr) lies near the densest region yet is surrounded by a shell with bidirectional density-velocity perturbations, consistent with a delayed-triggering scenario under the combined influence of UBC 178 stellar-wind feedback and a suspected supernova blast. Our results naturally demonstrate that snake-like stellar structures are filamentary relics of hierarchical star formation within giant molecular clouds. They provide direct observational evidence that cloud density and early feedback jointly modulate the progression of star formation, offering a clear and young laboratory for studying star-cloud co-evolution.

T Corona Borealis (T CrB) is a symbiotic recurrent nova with an $\simeq 80$ yr recurrence interval, the eruptions of which occur on top of a $\simeq 15$ yr long high-brightness state. We show that the high-brightness state is best explained as the response of a high-viscosity ($α=3$) accretion disk to a unique event in which the mass transfer rate from the donor star increases by a factor $\simeq 100$, from $\dot{M}\mathrm{(quies)}= 2 \times 10^{-9} M_\odot$ yr$^{-1}$ up to $\dot{M}\mathrm{(out)}= 1.9 \times 10^{-7} M_\odot$ yr$^{-1}$; it can not be a thermal-viscous disk instability outburst neither a steady nuclear burning event. The constraint that the matter accreted onto the white dwarf in between eruptions equals the envelope mass $M_{ig}$ needed to trigger nova eruptions at the observed recurrence interval requires a white dwarf mass of $M_1= 1.29 M_\odot$, a donor star mass of $M_2= 0.7 M_\odot$, and an inclination of $i= 57.3^o$. As the high-brightness state responds for 95% of $M_{ig}$, the nova eruptions of T CrB are induced by accretion events. Without the 15 yr long enhanced mass transfer events, its nova recurrence interval would be significantly longer, $\simeq 5500$ yr. T CrB exhibits a conspicuous decrease in brightness during the 1-2 yr prior to the nova event. We argue that this pre-eruption dip occurs during the convection phase that precedes the nova eruption and is best explained by the slow, accelerated expansion of the accreted envelope (and inner disk radius) at an average velocity of $v_\mathrm{exp}= 0.02$ km s$^{-1}$ over a 2 yr timescale, likely as a consequence of excess heat being increasingly deposited at the accreted layer by thermonuclear reactions before the nova eruption stage.

The interstellar scintillation observed in radio pulsars arises from interference between electromagnetic waves scattered by electron density fluctuations in the turbulent interstellar plasma, providing a critical tool for probing the small-scale structure of the ionized interstellar medium and the pulsar system itself. The primary aim of this work is to study long-term scintillation variations for a bright and nearby pulsar, PSR J0814$+$7429, carried out from 2013 September to 2023 September with the LOw-Frequency ARray (LOFAR) High Band Antennae in the frequency range of 120 - 170 MHz. We derive the basic scintillation parameters, scintillation bandwidth ($Δν_{\rm d}$) and scintillation timescale ($Δτ_{\rm d}$), from the two-dimensional (2D) auto-covariance function of the dynamic spectra that are a 2D matrix of pulse intensity as a function of time and frequency. We present the long-term monitoring of $Δν_{\rm d}$ and $Δτ_{\rm d}$ for PSR J0814$+7429$, which shows a strong annual variation in the time series of the $Δτ_{\rm d}$. From our modeling of the annual variations of scintillation velocities, the scattering screen is anisotropic and located at $0.23\pm0.02$ kpc from the Earth, likely corresponding to the boundary of the Local Bubble.

With recent data from the \emph{James Webb Space Telescope} (JWST), it is possible to calculate the mass of the supermassive black holes at the centre of galaxies, and the stellar mass of the host galaxies at $z \gtrsim 5$. In this work, we apply the method of extreme-value statistics to calculate the distributions of extreme black hole and stellar mass for the redshift range $3.5 \lesssim z \lesssim 8.5$. We sample these distributions to obtain a prediction for the black hole to stellar mass ratio of $\sim0.24$ over this redshift range, with the median in each bin varying in the range $0.18-0.35$. Our predictions are consistent with the highest observed values of the ratio from JWST observations of high-redshift galaxies.

In the framework of the $Λ$CDM model, galaxies evolve within dark matter (DM) haloes, where baryonic processes modify the inner structure of the DM distribution. In particular, baryon condensation and feedback can alter the inner density profiles of haloes, motivating studies of their central regions. The aim of this work is to investigate the inner slope of the DM density profiles of galaxies in the TNG50 simulation, its relation to galaxy properties, its evolution with redshift, and the impact of baryonic processes by comparing galaxies to a corresponding dark matter-only (DMO) realisation. Spherically averaged DM density profiles are constructed for galaxies in TNG50 and the DMO run. The inner slope is quantified using an Inner Linear Fit (ILF), defined as a power-law fit to the central region of the density profiles and motivated by the asymptotic behaviour of generalized NFW models. Subhaloes are matched between simulations and tracked across $z=0$, $0.2$, $0.7$, and $1$. The inner DM structure of galaxies in TNG50 shows that high-stellar-mass systems ($M_\star \gtrsim 10^{11}$ M$_\odot$) exhibit shallow inner slopes irrespective of being centrals or satellites, while lower-mass galaxies ($M_\star \lesssim 10^{9}$ M$_\odot$) show a broader diversity of profiles. At fixed stellar mass, low-mass satellites tend to be more cuspy, with the steepest slopes found in redder systems with lower $V_{\max}$ in more massive host haloes. We find a clear cosmic evolution, from shallower slopes at $z \sim 1$ to steeper profiles towards low redshift in both hydrodynamical and DMO runs, with hydrodynamical galaxies steeper. Finally, we verify that the population exhibiting the steepest slopes remains qualitatively robust to variations in the adopted fitting range, as extending the fit to larger radii$-$thereby excluding the innermost regions$-$generally leads to even steeper inferred slopes.

The South Pole Telescope (SPT) collaboration has recently embarked upon a campaign to monitor the brightness of a sample of active galactic nuclei (AGN), both in real time and in archival SPT data. The original design of the SPT was optimized for observations of the cosmic microwave background (CMB) at arc-minute and larger angular scales, and it has been used for this purpose for nearly twenty years, using three generations of CMB cameras. Recently it has been recognized that data from CMB experiments have the potential to be used for AGN monitoring. In this paper, we present the first public release of data from a full sample of SPT-monitored AGN, comprising 158 AGN light curves and associated data from the SPTpol camera, which was operational from 2012-2016. These light curves were created using observations from the SPTpol 500 deg$^{2}$ survey, in which the instrument was used to scan a 500 deg$^2$ patch of the sky several times per day with detectors sensitive to radiation in bands centered at 90 and 150 GHz. We provide a comprehensive description of the observations, the data processing methods, and the resulting light curve catalog. As an example of analyses that these data enable, we searched for a correlation between variability and spectral index, and we looked for ``bluer-when-brighter'' trends in the sample. Our analysis finds $> 10 σ$ correlation between fractional intrinsic variance and mean spectral index in the sample, but no significant evidence for bluer-when-brighter trends. The datasets from this study can be accessed through the SPT Treasury Record of AGN With Historical Activity and Time-Series or STRAWHAT catalog. This initial data release includes SPTpol light curves at 90 and 150 GHz, focusing on total intensity. In later updates, SPTpol polarization data and new observations from the SPT-3G instrument at 90, 150, and 220 GHz will be included.

We used the FourStar near-IR camera on Magellan-Baade to obtain high S/N H-Band imaging of 66 galaxies with radial velocities of 2000 < V < 5000 km/s. Our goal was to use the superior distance measurements of surface-brightness-fluctuations (SBF) to derive ``peculiar velocities'' to test claims that the CMB dipole anisotropy, equivalent to $\approx$600 km/s with respect to the Local Group, arises from a 'local' overdensity in the galaxy/dark-matter distribution -- the Great Attractor. SBF's ability to measure distances with 5% accuracy confirms a strong flow over a steradian of the sky peaking at Vpec $\sim$ 1000 km/s and converging to zero at D $\approx$70 Mpc from the Local Group. The modest spatial extent of this flow $R_V$ $\sim$ 5000 km/s is consistent with the original Great Attractor model (a diameter D $\sim$ 140 Mpc), as well as the magnitude and direction of the CMB dipole anisotropy, and the power spectrum of CMB fluctuations -- the latter two arguably the most secure measurements in astrophysics. In contrast, our results are at-odds with reports of comparable amplitude 'bulk flows' on scales of hundreds of Mpc that themselves may be inconsistent with the expected fluctuations in the CMB for a $Λ$CDM universe. We contend that only distance-estimators as accurate as SBF are able settle the question of whether the CMB dipole arises from the gravitational influence of large-scale structure within, or without $\sim$100 Mpc of the Local Group.

We study DESI DR1 galaxies to quantify colour dependence on cosmic web environment for three tracers spanning complementary regimes: BGS ($0.15\le z<0.55$), LRG ($0.6\le z<0.9$), and ELG ($0.6\le z<1.6$). Web environments are reconstructed with the tidal-tensor (T-Web) formalism on a $256^3$ grid in an $800\,Mpc$ cube and classified into voids, sheets, filaments, and knots. Sheets and filaments dominate volume ($\sim 45$--$48\%$ and $\sim 37$--$40\%$), voids $\sim 6$--$16\%$ knots $\sim 4$--$6\%$. A mass-dependent Otsu method separates red and blue populations. The BGS red fraction evolves non-monotonically: at $z\approx0.20$, voids ($13.89\pm5.76\%$), sheets ($6.13\pm1.27\%$), filaments ($9.24\pm1.66\%$), knots ($6.12\pm3.42\%$); at $z\approx0.30$, values range from $0.63\pm0.44\%$ to $2.01\pm0.99\%$; at $z\approx0.50$, from $17.93\pm0.44\%$ to $19.63\pm1.08\%$; environmental differences are small. LRGs show environment-dependent quenching: at $z\approx0.66$, knots ($65.90\pm0.45\%$), voids ($62.40\pm1.81\%$), filaments ($60.21\pm0.48\%$), sheets ($58.37\pm3.15\%$); by $z\approx0.88$, these converge to $\sim 68$--$70\%$. ELGs exhibit strong redshift evolution: filaments drop from $55.18\pm0.31\%$ at $z\approx0.65$ to $33.22\pm0.21\%$ at $z\approx0.95$; voids and sheets show similar declines, with weak and non-monotonic. High-mass selection increases red fractions but preserves trends. Relative red and blue fractions (RRF/RBF) show filaments and sheets host the largest shares of both red and blue galaxies; knots contribute less despite elevated red fractions. The $(g-r)$ colour distributions reveal an enhanced red component in knots and bluer colours in voids, with the clearest bimodality at low redshift. Overall, stellar mass drives the primary quenching trend, while environment provides a systematic secondary modulation, strongest in dense knots and at lower stellar masses.

In this work, we derive the general solutions for a cylindrically symmetric space-time filled with a cosmological perfect fluid obeying $p=γρ$ ($0\leq γ\leq 1$), where $γ=1$ represents a stiff or Zeldovich fluid. Using Marder's metric with coefficients depending on $t$ and $r$, we obtain explicit solutions of the gravitational field equations for the three cases $δ= 1, 0, -1$, corresponding to exponential, power-law, and trigonometric behaviors of the metric functions. The resulting space-times exhibit anisotropic evolution, nontrivial expansion and shear, and curvature singularities, with energy density and pressure profiles determined by the integration constants. These solutions provide a comprehensive framework for modeling cylindrically symmetric cosmologies, offering insights into early-universe dynamics and anisotropic gravitational phenomena. The versatility of the solutions also opens avenues for extensions to higher-dimensional or modified gravity scenarios, making them a valuable tool for both theoretical and phenomenological studies in general relativity.

We carry out a test of the fundamental Etherington relation (cosmic distance duality relation) which relates the luminosity distance $D_{\rm L}$ and angular diameter distance $D_{\rm A}$ in metric theories of gravity. We use the latest measurements of the angular diameter distance as a function of redshift from the Dark Energy Spectroscopic Instrument Data Release 2 (DESI-DR2) and the luminosity distance from a variety of compilations of Supernovae of Type Ia (SNeIa). Our results indicate that these measurements are statistically consistent with the Etherington relation. In addition to providing a confirmation of the underlying assumptions of the Etherington relation, i.e., the metric nature of gravity, Lorentz invariance and photon number conservation, our results are also a stringent test of any residual systematic effects. We interpret the absence of evidence of any deviation from this relation to constrain the evolution of the absolute magnitude of SNeIa to $dM/dz = 0.07 \pm 0.07$ over and above the systematics that are already accounted for in the SNeIa analyses. We discuss how the Etherington relation can be used to constrain systematic parameters in the analyses of dynamical dark energy using geometric probes, to make it more robust against systematic effects.

Spontaneous breaking of discrete symmetries play non-trivial role in many well-motivated particle physics models. However, it leads to a network of cosmologically unwanted domain walls (DWs) which can be made unstable by introducing a bias term in the scalar potential. In this letter, we provide a novel origin of such bias terms at finite temperature due to radiative corrections from a Dirac fermion with large asymmetry $\sim \mathcal{O}(0.1)$ in its number density. In addition to getting a new viable region of parameter space for collapsing DWs not explored previously and resulting gravitational waves (GWs) accessible at future experiments, the viability of the scenario crucially depends on the temperature of asymmetry generation too. This provides a unique way of probing both the amount of asymmetry and the corresponding temperature via future observations of GWs from collapsing DWs. The large asymmetry in the Dirac fermion can also have interesting implications for the observed baryon asymmetry as well as dark matter and large neutrino asymmetry.

Multiply imaged, gravitationally lensed supernovae are rare but powerful tools for providing independent measurements on cosmological parameters. Supernova (SN) 2025wny ("SN Winny") is the first gravitationally-lensed Type I superluminous supernova and the first lensed supernova in a galaxy-scale system that is suitable for time-delay cosmography studies. In this work, we present high-resolution $K_p$-band adaptive optics imaging of SN Winny obtained with the near-infrared camera (NIRC2) on the W. M. Keck II telescope. With exquisite image quality (FWHM$\approx0.^{\prime\prime}065$) we determine and make use of the precise astrometric positions of the five multiple images as constraints for our lens mass models. With lenstronomy and Glee, we parameterize the total mass of the system with a singular isothermal ellipsoid, a singular isothermal sphere, and external shear. The two independent models are in excellent agreement and reproduce the observed image positions with sub-milli-arcsecond residuals. The inferred projected total masses enclosed within the Einstein radii of the primary and secondary lens galaxies are M$_1$ = 4.44$^{+0.06}_{-0.05}\times10^{11} M_\odot$ and M$_2$ = 0.96$^{+0.02}_{-0.02}\times10^{11} M_\odot$, respectively. Likewise, the inferred effective velocity dispersion of the primary lens is $σ_{1} = $ 277.4$^{+0.9}_{-0.7}$ km/s, consistent with the independent spectroscopic measurement made by DESI of $σ_{\star,1} = $ 298$\,\pm\,37$ km/s. Our modeling results are also consistent with previous results for the same system with data from the Large Binocular Telescope (LBT), using the same lens modeling codes. We also corroborate their finding that the SN multiple image A has an anomalous excess of flux by a factor of ~2-3 beyond what our smooth mass models predict.

Stars are natural sources of feebly interacting particles, including putative particles with mass $m_χ$ and electric charge $qe$. The emission of such millicharged particles (MCPs) causes an energy loss which can alter stellar evolution. While MCP production rates have been computed for different plasma parameters, they have yet to be derived for the conditions relevant to late stages of stellar evolution, in which the temperature can reach values $T\simeq 10-100\,\rm keV$ while the plasma frequency is $ω_{\rm pl}\ll T$. In this paper, we compute the MCP energy-loss rates relevant for pre-supernova objects, finding three different regimes in which the dominant processes are respectively plasmon decay ($m_χ< ω_{\rm pl}/2$), Compton-like scattering ($m_χ> ω_{\rm pl}/2$, $T\lesssim 0.5\,\rm MeV$), and electron-positron annihilation. We obtain semi-analytical fits for the energy-loss rates suitable for implementation in stellar evolution codes.

We investigate the impact of millicharged particles (MCPs) on massive stars undergoing pulsational pair-instability supernovae and on the location of the lower edge of the black hole mass gap. We find that energy losses due to MCP emission weaken the pulsations, allowing the star to retain more mass and thereby shifting the lower edge of the mass gap to higher black hole masses. The mass gap is sensitive to a region of MCP parameter space with masses $35\,{\rm keV}\lesssim m_χ\lesssim 200\,{\rm keV}$ and charges $10^{-10}\lesssim q \lesssim 10^{-9}$, which remains unconstrained by existing astrophysical probes. If confirmed, recent gravitational wave observations placing the lower edge of the mass gap near $45\,{\rm M}_\odot$ would translate directly into bounds on this parameter space.

Globular clusters (GCs) display much lower binary fractions than found among main-sequence stars in the solar neighborhood. The physical cause of this difference is debatable: does it reflect different star formation outcomes at low metallicity and/or high density, the dynamical processing of primordial binaries over cluster lifetimes, or a combination of the two? Starting from the assumption that the initial binary distribution in GCs is the same as the binary distribution observed in the solar neighborhood, we show with straightforward analytical calculations that the dynamical dissolution of "soft" primordial binaries can fully explain the main-sequence binary fractions in present-day GCs. We validate our estimates against a detailed N-body simulation with the Cluster Monte Carlo code. Adopting the view that the observed binary fraction in a given cluster constrains the location of the hard/soft boundary at birth, we infer that surviving Milky Way GCs had a similar distribution of birth radii to young massive clusters in the local universe. Our findings underscore the crucial role of stellar black holes (through "black hole burning") in sculpting GC binary populations and reinforce the need for realistic initial conditions in theoretical modeling of GC dynamics.

Radio relics are diffuse, extended synchrotron sources located at the outskirts of merging galaxy clusters. Their origin has been linked with shock waves injected into the intracluster medium, but the acceleration mechanism at the shock front is still under debate. Some clusters, like MACS J1752.0+4440, host a double relic system, with two relics found on opposite sides with respect to the cluster center. To investigate the acceleration mechanism that generates radio relics, we study the morphological and spectral properties of the double relic system in MACS J1752. We present new wideband radio continuum observations made with uGMRT and JVLA, and LOFAR data. We perform a detailed, high-resolution spectral analysis of the double relic system in MACS J1752, observing and characterizing substructures, particularly for the brighter relic. We find a double-peaked surface brightness and spectral index profile for the NE relic and identify a "bright bar" substructure. Moreover, we observed surprisingly flat integrated spectral indices for both relics, at $α_{\mathrm{int}}^{\mathrm{NE}} = -0.91 \pm 0.06$ and $α_{\mathrm{int}}^{\mathrm{SW}} = -0.83 \pm 0.05$. We study the spatial variation of the spectral index, observing a coherent trend with the observed substructures. We estimate an injection Mach number of $\mathcal{M}_{\mathrm{NE}} = 3.1^{+0.1}_{-0.1}$ and $\mathcal{M}_{\mathrm{SW}} = 3.2^{+0.1}_{-0.1}$. By performing a spectral curvature analysis for both relics, generating color-color plots and a spectral curvature maps, we observe two "concave" spectra represented by positive spectral curvature, in contrast with particle population ageing models. The observed properties of the NE relic are not consistent with a simple scenario with a single shock front. Multiple shock surfaces, re-acceleration, and projection effects likely play a role in shaping the morphology of the relic.