arXiv Daily Digest - 2026-04-28

The discovery of novel materials is critical for global energy and quantum technology transitions. While deep learning has fundamentally reshaped this landscape, existing predictive or generative models typically operate in isolation, lacking the autonomous orchestration required to execute the full discovery process. Here we present ElementsClaw, an agentic framework for materials discovery that synergizes Large Atomic Models (LAMs) with Large Language Models (LLMs). In response to varied human requirements, ElementsClaw dynamically orchestrates a suite of LAM tools finetuned from our proposed model Elements for atomic-scale numerical computation, while leveraging LLMs for high-level semantic reasoning. This shift moves AI-driven materials science from isolated processes toward integrated and human interactive discovery. In the demanding domain of superconductors, our agentic system guides the experimental synthesis of four new superconductors, including Zr3ScRe8 with a transition temperature of 6.8 K and HfZrRe4 at 6.7 K. At scale, ElementsClaw screens more than 2.4 million stable crystals within only 28 GPU hours, identifying 68,000 high-confidence superconducting candidates and vastly expanding the known superconducting space. These results demonstrate how our agent accelerates materials discovery with high physical fidelity.

Multimodal affective computing analyzes user-generated social media content to predict emotional states. However, a critical gap remains in understanding how visual content shapes cognitive interpretations and elicits specific affective experiences such as pleasure. This study introduces a novel computational model to infer video-induced pleasure via cognitive appraisal variables. The proposed model addresses four challenges: (1) noisy and inconsistent human labels, (2) the semantic gap between "positive emotions" and "pleasure," (3) the scarcity of pleasure-specific datasets, and (4) the limited interpretability of existing black-box fusion methods. Our approach integrates data-driven and cognitive theory-driven methods, using cognitive appraisal theory and a fuzzy model within an innovative framework. The model employs transformer-based architectures and attention mechanisms for fine-grained multimodal feature extraction and interpretable fusion to capture both inter- and intra-modal dynamics associated with pleasure. This enables the prediction of underlying appraisal variables, thereby bridging the semantic gap and enhancing model explainability beyond conventional statistical associations. Experimental results validate the efficacy of the proposed method in detecting video-induced pleasure, achieving a peak accuracy of 0.6624 in predicting pleasure levels. These findings highlight promising implications for affective content recommendation, intelligent media creation, and advancing our understanding of how digital media influences human emotions.

Hypernetwork-based methods such as Doc-to-LoRA internalize a document into an LLM's weights in a single forward pass, but they fail systematically on conflicts: when the document contradicts pretraining knowledge, accuracy collapses to 46.4% on the deepest facts. We show the failure is a magnitude problem rather than a representational one. The hypernetwork already targets the right layers, but its adapter margin is approximately constant across documents while the pretrained margin grows with training frequency, so deep conflicts lose by construction. The account predicts that failure should track prior strength: sorting 194 conflicts by the base model's log-probability on the contradicted fact, baseline accuracy falls from 68% on weak-prior questions to 16% on strong-prior ones, a 52 percentage-point gap. The cure is amplitude. Selective Layer Boosting scales the adapter at its top-norm layers, and Conflict-Aware Internalization triggers boosting only when the base model is confident. Both are training-free; together they raise deep-conflict accuracy from 46.4% to 71.0% on Gemma-2B and from 53.6% to 72.5% on Mistral-7B while preserving novel-knowledge recall, and beat vanilla retrieval-augmented generation on medium conflicts by 18 percentage points despite operating entirely in parameter space. We release KID-Bench, a 489-question benchmark that separates novel recall, cross-knowledge combination, and prior-graded conflicts.

Recent mixed-policy optimization methods for LLM reasoning that interleave or blend supervised and reinforcement learning signals report improvements over the standard SFT-then-RL pipeline. We show that numerous recently published research papers rely on a faulty baseline caused by two distinct bugs: a CPU-offloaded optimizer bug in DeepSpeed that silently drops intermediate micro-batches during gradient accumulation (affecting multiple downstream frameworks including TRL, OpenRLHF and Llama-Factory), and a loss aggregation bug in OpenRLHF that incorrectly weights per-mini-batch losses. Together they suppress SFT performance, with the optimizer bug accounting for most of the gap and the loss aggregation bug contributing a smaller additional effect. Once corrected, the standard SFT-then-RL pipeline surpasses every published mixed-policy method we evaluate by +3.8 points on math benchmarks with Qwen2.5-Math-7B and by +22.2 points with Llama-3.1-8B. Even a truncated variant with just 50 RL steps outperforms mixed-policy methods on math benchmarks while using fewer FLOPs.

Deploying quantum machine learning on NISQ devices requires architectures where training overhead does not negate computational advantages. We systematically compare two quantum approaches for chaotic time-series prediction on the Lorenz system: a variational Quantum Physics-Informed Neural Network (QPINN) and a Quantum Reservoir Computing (QRC) framework utilizing a fixed transverse-field Ising Hamiltonian. Under matched resources ($4$--$5$ qubits, $2$--$3$ layers), QRC achieves an $81\%$ lower mean-squared error (test MSE $3.2 \pm 0.6$ vs. $47.9 \pm 36.6$ for QPINN) while training $\sim 52,000\times$ faster ($0.2$\,s vs. $\sim 2.4$\,h per seed). Drawing on the classical delay-embedding principle, we formalize a temporal windowing technique within the QRC pipeline that improves attractor reconstruction by providing bounded, structured input history. Analysis reveals that QPINN instability stems from capacity limitations and competing loss terms rather than barren plateaus; gradient norms remained large ($10^3$--$10^4$), ruling out exponential suppression at this scale. These failure modes are absent by construction in the non-variational QRC approach. We validate robustness across three canonical systems (Lorenz, Rössler, and Lorenz-96), where QRC consistently achieves low test MSE ($3.1 \pm 0.6$, $1.8 \pm 0.1$, and $12.4 \pm 0.6$, respectively) with sub-second training. Our findings suggest the fixed-reservoir architecture is a primary driver of QRC's advantage at these scales, warranting further investigation at larger qubit counts and on hardware where quantum-specific advantages are expected to emerge.

Despite the Transformer's dominance across machine learning, its architecture remains largely heuristic and lacks a unified theoretical foundation. We introduce Score-based Variational Flow (SVFlow), a continuous-time dynamical system for representation learning in which the state evolves according to a variational posterior-weighted average of conditional log-likelihood scores, and provide a principled basis for regularization through variational consistency. We show that forward Euler discretization of spherical SVFlow exactly recovers the Transformer architecture. Multi-head attention approximates SVFlow vector field via a vMF kernel-smoothed posterior, while MoE/FFN approximates it in a relaxed network-based way, and the residual-normalization block implements a relaxed retraction that maintains spherical geometry. This unification explains why attention trains stably without explicit regularization while MoE requires auxiliary balancing losses. Experiments on pre-trained language models with prefix shuffling show that SVFlow-induced metrics correlate with task performance, reveal depth-dependent sensitivity, and reflect the intrinsic dynamics of attention.

Asking inquisitive questions while reading, and looking for their answers, is an important part in human discourse comprehension, curiosity, and creative ideation, and prior work has investigated this in text-only scenarios. However, in scientific or research papers, many of the critical takeaways are conveyed through both figures and the text that analyzes them. While scientific visualizations have been used to evaluate Vision-Language Models (VLMs) capabilities, current benchmarks are limited to questions that focus simply on extracting information from them. Such questions only require lower-level reasoning, do not take into account the context in which a figure appears, and do not reflect the communicative goals the authors wish to achieve. We generate inquisitive questions that reach the depth of questions humans generate when engaging with scientific papers, conditioned on both the figure and the paper's context, and require reasoning across both modalities. To do so, we extend the linguistic theory of Questions Under Discussion (QUD) from being text-only to multimodal, where implicit questions are raised and resolved as discourse progresses. We present MQUD, a dataset of research papers in which such questions are made explicit and annotated by the original authors. We show that fine-tuning a VLM on MQUD shifts the model from generating generic low-level visual questions to content-specific grounding that requires a high-level of multimodal reasoning, yielding higher-quality, more visually grounded multimodal QUD generation.

Disease progression varies with age and is influenced by underlying genetic, biochemical, and hormonal etiologies, suggesting the need for tailored monitoring, care, and medication beyond standard clinical guidelines. Specifically, in autoimmune diseases like type 1 diabetes (T1D), where patients depend on exogenous insulin to compensate for insulin deficiency, medication dosing and the physiological response reflected in vital signs can differ. Insulin therapy can lead to hypoglycemia, a dangerous condition characterized by decreased blood glucose levels ($\leq$70). This risk can be mitigated through improved diabetes management supported by data analytics. Notably, leveraging data from continuous glucose monitoring (CGM) devices, hypoglycemia onset can be predicted. However, while glucose variability, auto-antibody levels, and hypoglycemia occurrence differ across age groups, hypoglycemia classification most often only relies on population-based models specialized in specific age ranges. In this work, we classify hypoglycemia 0, 5-15, 20-45, and 50-120 minutes before onset using DiaData, a large CGM dataset of patients with T1D ranging from children to seniors. In particular, we investigate: 1) the generalizability of a population-based model including all age groups, 2) the impact of age-segmented models trained separately per age group, and 3) the effect of model individualization through transfer learning. The results show that a global population-based model yields similar or superior performance compared to age-segmented models. These findings suggest that data from children, teenagers, and adults can be combined for training models on hypoglycemia classification. While glucose variation differs across age groups, short-term hypoglycemic patterns are similar. However, data of children obtain their best recall with age specialized model.

Large language models (LLMs) have shown strong performance on legal benchmarks, including multiple-choice components of bar exams. However, their capacity for generating open-ended legal reasoning in realistic scenarios remains insufficiently explored. Notably, to our best knowledge, there are no prior studies or datasets addressing this issue in the Japanese context. This study presents the first dataset designed to evaluate the open-ended legal reasoning performance of LLMs within the Japanese jurisdiction. The dataset is based on the writing component of the Japanese bar examination, which requires examinees to identify multiple legal issues from long narratives and to construct structured legal arguments in free text format. Our key contribution is the manual evaluation of LLMs' generated responses by legal experts, which reveals limitations and challenges in legal reasoning. Moreover, we conducted a manual analysis of hallucinations to characterize when and how the models introduce content not supported by precedent or law. Our real exam questions, model-generated responses, and expert evaluations reveal the milestones of current LLMs in the Japanese legal domain. Our dataset and relevant resources will be available online.

Recent advances in autonomous driving have motivated research on pedestrian intention prediction, which aims to infer future crossing decisions and actions by modeling temporal dynamics, social interactions, and environmental context. However, existing studies remain constrained by oversimplified multi-agent interaction patterns, opaque reasoning logic, and a lack of global consistency in behavioral predictions, which compromise both robustness and interpretability. In this work, we propose ESIA (Energy-based Spatiotemporal Interaction-Aware framework), a novel Conditional Random Field (CRF)-based paradigm. We cast the intention prediction task as a structured prediction problem over a unified graph-based representation, treating pedestrians and the environment as spatiotemporal nodes. To characterize their distinct roles, we assign unary potentials to nodes to capture individual intentions, and pairwise potentials to edges to encode social and environmental interactions. These potentials are integrated into a unified global energy function to ensure scene-level consistency across behavioral predictions. To further constrain inference without ground-truth supervision, we introduce structural consistency terms to penalize logical contradictions. This optimization is efficiently solved via a novel Unary-Seeded Simulated Annealing (U-SSA) algorithm, which leverages high-confidence unary priors to rapidly converge to a high-quality solution. Extensive experiments on standard benchmarks demonstrate that ESIA achieves state-of-the-art performance with improved interpretability over existing methods.

Expressway video anomaly detection is essential for safety management. However, identifying anomalies across diverse scenes remains challenging, particularly for far-field targets exhibiting subtle abnormal vehicle motions. While Vision-Language Models (VLMs) demonstrate strong semantic reasoning capabilities, processing global frames causes attention dilution for these far-field objects and incurs prohibitive computational costs. To address these issues, we propose VIBES, an asynchronous collaborative framework utilizing VLMs guided by Bayesian inference. Specifically, to overcome poor generalization across varying expressway environments, we introduce an online Bayesian inference module. This module continuously evaluates vehicle trajectories to dynamically update the probabilistic boundaries of normal driving behaviors, serving as an asynchronous trigger to precisely localize anomalies in space and time. Instead of processing the continuous video stream, the VLM processes only the localized visual regions indicated by the trigger. This targeted visual input prevents attention dilution and enables accurate semantic reasoning. Extensive evaluations demonstrate that VIBES improves detection accuracy for far-field anomalies and reduces computational overhead, achieving high real-time efficiency and explainability while demonstrating generalization across diverse expressway conditions.

Metanetworks are neural architectures designed to operate directly on pretrained weights to perform downstream tasks. However, the parameter space serves only as a proxy for the underlying function class, and the parameter-function mapping is inherently non-injective: distinct parameter configurations may yield identical input-output behaviors. As a result, metanetworks that rely solely on raw parameters risk overlooking the intrinsic symmetries of the architecture. Reasoning about functional identity is therefore essential for effective metanetwork design, motivating the development of equivariant metanetworks, which incorporate equivariance principles to respect architectural symmetries. Existing approaches, however, typically enforce strict equivariance, which imposes rigid constraints and often leads to sparse and less expressive models. To address this limitation, we introduce the novel concept of quasi-equivariance, which allows metanetworks to move beyond the rigidity of strict equivariance while still preserving functional identity. We lay down a principled basis for this framework and demonstrate its broad applicability across diverse neural architectures, including feedforward, convolutional, and transformer networks. Through empirical evaluation, we show that quasi-equivariant metanetworks achieve good trade-offs between symmetry preservation and representational expressivity. These findings advance the theoretical understanding of weight-space learning and provide a principled foundation for the design of more expressive and functionally robust metanetworks.

Mechanistic interpretability research on emotion in large language models -- linear probing, activation patching, sparse autoencoder (SAE) feature analysis, causal ablation, steering vector extraction -- depends on stimuli that contain the words for the emotions they test. When a probe fires on "I am furious", it is unclear whether the model has detected anger or detected the word "furious". The two readings have very different consequences for every downstream claim about emotion circuits, features, and interventions. We release AIPsy-Affect, a 480-item clinical stimulus battery that removes the confound at the stimulus level: 192 keyword-free vignettes evoking each of Plutchik's eight primary emotions through narrative situation alone, 192 matched neutral controls that share characters, setting, length, and surface structure with the affect surgically removed, plus moderate-intensity and discriminant-validity splits. The matched-pair structure supports linear probing, activation patching, SAE feature analysis, causal ablation, and steering vector extraction under a strong methodological guarantee: any internal representation that distinguishes a clinical item from its matched neutral cannot be doing so on the basis of emotion-keyword presence. A three-method NLP defense battery -- bag-of-words sentiment, an emotion-category lexicon, and a contextual transformer classifier -- confirms the property: bag-of-words methods see only situational vocabulary, and a contextual classifier detects affect (p < 10^-15) but cannot identify the category (5.2% top-1 vs. 82.5% on a keyword-rich control). AIPsy-Affect extends our earlier 96-item battery (arXiv:2603.22295) by a factor of four and is released openly under MIT license.

Recent large audio language models (LALMs) demonstrate remarkable capabilities in processing extended multi-modal sequences, yet incur high inference costs. Token compression is an effective method that directly reduces redundant tokens in the sequence. Existing compression methods usually assume that all attention heads in LALMs contribute equally to various audio tasks and calculate token importance by averaging scores across all heads. However, our analysis demonstrates that attention heads exhibit distinct behaviors across diverse audio domains. We further reveal that only a sparse subset of attention heads actively responds to audio, with completely different performance when handling semantic and acoustic tasks. In light of this observation, we propose HeadRouter, a head-importance-aware token pruning method that perceives the varying importance of attention heads in different audio tasks to maximize the retention of crucial tokens. HeadRouter is training-free and can be applied to various LALMs. Extensive experiments on the AudioMarathon and MMAU-Pro benchmarks demonstrate that HeadRouter achieves state-of-the-art compression performance, exceeding the baseline model even when retaining 70% of the audio tokens and achieving 101.8% and 103.0% of the vanilla average on Qwen2.5-Omni-3B and Qwen2.5-Omni-7B, respectively.

Information-theoretic (IT) measures are ubiquitous in artificial intelligence: entropy drives decision-tree splits and uncertainty quantification, cross-entropy is the default classification loss, mutual information underpins representation learning and feature selection, and transfer entropy reveals directed influence in dynamical systems. A second, less consolidated family of measures, integrated information (Phi), effective information (EI), and autonomy, has emerged for characterizing agent complexity. Despite wide adoption, measure selection is often decoupled from estimator assumptions, failure modes, and safe inferential claims. This paper provides a practical decision framework for all seven measures, organized around three prescriptive questions for each: (i) what question does the measure answer and in which AI context; (ii) which estimator is appropriate for the data type and dimensionality; and (iii) what is the most dangerous misuse. The framework is operationalized in two complementary artifacts: a measure-selection flowchart and a master decision table. We cover both AI/ML and decision-making agent application domains per measure, with standardized Bridge Boxes linking IT quantities to cognitive constructs. Three worked examples illustrate the framework on concrete practitioner scenarios spanning representation learning, temporal influence analysis, and evolved agent complexity.

Recent advances in formal theorem proving have focused on Olympiad-level mathematics, leaving undergraduate domains largely unexplored. Optimization, fundamental to machine learning, operations research, and scientific computing, remains underserved by existing provers. Its reliance on domain-specific formalisms (convexity, optimality conditions, and algorithmic analysis) creates significant distribution shift, making naive domain transfer ineffective. We present OptProver, a trained model that achieves robust transfer from Olympiad to undergraduate optimization. Starting from a strong Olympiad-level prover, our pipeline mitigates distribution shift through two key innovations. First, we employ large-scale optimization-focused data curation via expert iteration. Second, we introduce a specialized preference learning objective that integrates perplexity-weighted optimization with a mechanism to penalize valid but non-progressing proof steps. This not only addresses distribution shifts but also guides the search toward efficient trajectories. To enable rigorous evaluation, we construct a novel benchmark in Lean 4 focused on optimization. On this benchmark, OptProver achieves state-of-the-art Pass@1 and Pass@32 among comparably sized models while maintaining competitive performance on general theorem-proving tasks, demonstrating effective domain transfer without catastrophic forgetting.

We study when a skip connection around a single-hidden-layer MLP can be absorbed into a residual-free MLP of the same width. We first show that for any architecture whose skip branch is an invertible linear map (including Hyper-Connections and their manifold-constrained variants), the problem reduces to the identity skip case. For homogeneous activations of degree $k \neq 1$, such as ReLU$^2$ and ReGLU, absorption is unconditionally impossible by a degree argument. For gated activations whose gate is differentiable at the origin with $g(0) = 0$, including SwiGLU and GeGLU, a linearization argument gives the same conclusion. These impossibility results extend to arbitrary depth: a composition of $L$ residual blocks using such activations cannot be replicated by any composition of $L$ residual-free blocks of the same width. For ungated ReLU and GELU, the situation is richer. For generic weight matrices, absorption holds at the single-block level if and only if there exists an index set $S$ of size at least $d$ such that $W_{\mathrm{down}}[:,S]\,W_{\mathrm{up}}[S,:] = -I_d$. This condition is non-generic (it fails with probability one under continuous weight distributions), so skip-connected and residual-free MLPs of the same width represent generically disjoint function classes. Whether this disjointness persists for deep compositions of ReLU or GELU blocks remains open.

Slide-based teaching is widely used in higher education, yet in online, hybrid, and asynchronous contexts, slides often lose the instructor presence, narrative continuity, and expressive framing that help learners connect with content. Full lecture video can partly restore these qualities, but it is time-consuming to record, revise, and reuse. This study addresses that pedagogical and production challenge by presenting a practice-based analysis of an open-source workflow for creating talking slide avatars for slide-based teaching. The workflow integrates OpenVoice for text-to-speech generation and voice cloning with Ditto-TalkingHead for audio-driven talking-image synthesis, enabling instructors to transform a script and a static portrait into a short narrated video that can be embedded in slide decks or HTML-based lecture materials. Rather than treating this workflow merely as a technical solution, the study frames talking slide avatars as multimodal communication artifacts at the intersection of digital pedagogy, aesthetic education, and art-technology practice. Using a practice-based implementation and analytic reflection approach, the study documents the production pipeline, examines its communicative and aesthetic affordances, and proposes practical guidelines for script length, image selection, pacing, disclosure, accessibility, and ethical use. The study makes three primary contributions: it presents an educator-oriented open-source production model, reframes talking avatars as an educational communication design problem, and proposes a responsible pathway for incorporating generative synthetic media into teaching. It concludes that short, transparent, and carefully designed avatars can humanize slide-based instruction while providing a reusable communicative layer for introductions, transitions, reminders, and recaps across online, hybrid, and asynchronous learning environments.

Crop disease diagnosis from field photographs faces two recurring problems: models that score well on benchmarks frequently hallucinate species names, and when predictions are correct, the reasoning behind them is typically inaccessible to the practitioner. This paper describes Agri-CPJ (Caption-Prompt-Judge), a training-free few-shot framework in which a large vision-language model first generates a structured morphological caption, iteratively refined through multi-dimensional quality gating, before any diagnostic question is answered. Two candidate responses are then generated from complementary viewpoints, and an LLM judge selects the stronger one based on domain-specific criteria. Caption refinement is the component with the largest individual impact: ablations confirm that skipping it consistently degrades downstream accuracy across both models tested. On CDDMBench, pairing GPT-5-Nano with GPT-5-mini-generated captions yields \textbf{+22.7} pp in disease classification and \textbf{+19.5} points in QA score over no-caption baselines. Evaluated without modification on AgMMU-MCQs, GPT-5-Nano reached 77.84\% and Qwen-VL-Chat reached 64.54\%, placing them at or above most open-source models of comparable scale despite the format shift from open-ended to multiple-choice. The structured caption and judge rationale together constitute a readable audit trail: a practitioner who disagrees with a diagnosis can identify the specific caption observation that was incorrect. Code and data are publicly available https://github.com/CPJ-Agricultural/CPJ-Agricultural-Diagnosis

We present a semantic-structural atlas of transportation research built from 120{,}323 papers across 34 peer-reviewed journals published between 1967 and 2025, roughly an order of magnitude larger than and a decade beyond Sun and Rahwan's~(2017) coauthorship study. We use OpenAlex and Crossref as open, CC0-licensed data sources, resolve author identity through OpenAlex author IDs, ORCID records, and manual alias resolution, and embed every paper with SPECTER2 with Arora-style whitening concatenated with concept TF--IDF and venue linear-discriminant projections. On this substrate we report three findings. First, Leiden on the author-level semantic k-nearest-neighbor graph yields 23 topic communities that agree only weakly with the 172 coauthor communities (normalized mutual information $0.23$), opening room for a predictive layer that neither source encodes alone. Second, a multiplex Leiden partition combining both edge types recovers 181 communities and localizes where collaboration and topic structure decouple. Third -- the paper's core methodological contribution -- we define \emph{phantom collaborators}, pairs of authors who are top-$K$ semantic neighbors yet $\geq 3$ hops apart in the coauthor graph, and show via a temporal hold-out (training cutoff 2019) that phantom pairs become real coauthors in 2020--2025 at a rate $16$ to $33$ times above random, popularity-weighted, and same-venue baselines, with a $68$-fold monotone gradient between the highest- and lowest-similarity buckets. All artifacts are released as a live, reproducible web atlas at https://choi-seongjin.github.io/transport-atlas/.

Recent advances in large language models (LLMs) have shown promise in formal theorem proving, yet evaluating semantic correctness remains challenging. Existing evaluations rely on indirect proxies such as lexical overlap with human-annotated proof, or expensive manual inspection. Inspired by the shift from lexical comparison to test-based evaluation in code generation, we propose T , a framework that evaluates the semantic correctness of formal theorems: a generated theorem is considered correct only if all dependent successor theorems compile successfully, analogous to integration testing. We construct a benchmark from 5 real-world Lean 4 repositories, comprising 2,206 problems paired with 41 successor theorems on average, automatically extracted without human effort. Experiments demonstrate that while state-of-the-art models achieve high compilation success, they perform significantly worse under our semantic metric. The best model, Claude-Sonnet-4.5, achieves only 38.9% Testing Accuracy on the full set, given both natural language proof and successor theorems as context, revealing a critical gap in current theorem generation capabilities.

A widely cited result by Dong et al. (2021) showed that Transformers built from self-attention alone, without skip connections or feed-forward layers, suffer from rapid rank collapse: all token representations converge to a single direction. The proposed remedy was the MLP. We show that this picture, while correct in the regime studied by Dong, is incomplete in ways that matter for architectural understanding. Three results are established. First, layer normalisation is precisely affine-rank-neutral: it preserves the affine rank of the token representation set exactly. The widespread claim that LN "plays no role" is imprecise; the correct statement is sharper. Second, residual connections generically obstruct rank collapse in real Transformers such as BERT-base, in a measure-theoretic sense, without contribution from the MLP. The MLP's irreplaceable function is different: generating feature directions outside the linear span of the original token embeddings, which no stack of attention layers can produce. Third, a phenomenon distinct from rank collapse is identified: head-channel non-identifiability. After multi-head attention sums per-head outputs through the output projection, individual contributions cannot be canonically attributed to a specific head; n(H-1)d_k degrees of freedom per layer remain ambiguous when recovering a single head from the mixed signal. The MLP cannot remedy this because it acts on the post-summation signal. A constructive partial remedy is proposed: a position-gated output projection (PG-OP) at parameter overhead below 1.6% of the standard output projection. The four collapse phenomena identified in the literature -- rank collapse in depth, in width, head-channel non-identifiability, and entropy collapse -- are unified under a symmetry-breaking framework, each corresponding to a distinct symmetry of the Transformer's forward pass.

Accurate modeling of human mobility is critical for tackling urban planning and public health challenges. In undeveloped regions, the absence of comprehensive travel surveys necessitates reconstructing mobility networks from publicly available data. Here we develop neuroGravity, a physics-informed deep learning model that reliably reconstructs mobility flows from limited observations and transfers to unobserved cities. Using only urban facility and population distributions, we find that neuroGravity's regional representations strongly correlate with socioeconomic and livability status, offering scalable proxies for costly surveys. Furthermore, we uncover that spatial income segregation plays a key role in model transferability: mobility networks are most reliably reconstructed when target cities share similar segregation levels with the source. We design an index to quantify this segregation and accurately predict transferability. Finally, we generate mobility flow proxies for over 1,200 cities worldwide, highlighting neuroGravity's potential to mitigate critical data shortages in resource-limited, underdeveloped areas.

With the emergence of large language models (LLMs) and AI agent frameworks, the human-AI co-work paradigm known as Vibe Coding is changing how people code, making it more accessible and productive. In scientific research, where workflows are more complex and the burden of specialized labor limits independent researchers and those in low-resource areas, the potential impact is even greater, particularly in biomedicine, which involves heterogeneous data modalities and multi-step analytical pipelines. In this paper, we introduce Vibe Medicine, a co-work paradigm in which clinicians and researchers direct skill-augmented AI agents through natural language to execute complex, multi-step biomedical workflows, while retaining the role of research director who specifies objectives, reviews intermediate results, and makes domain-informed decisions. The enabling infrastructure consists of three layers: capable LLMs, agent frameworks such as OpenClaw and Hermes Agent, and the OpenClaw medical skills collection, which includes more than 1,000 curated skills from multiple open-source repositories. We analyze the architecture and skill categories of this collection across ten biomedical domains, and present case studies covering rare disease diagnosis, drug repurposing, and clinical trial design that demonstrate end-to-end workflows in practice. We also identify the principal risks, such as hallucination, data privacy, and over-reliance, and outline directions toward more reliable, trustworthy, and clinically integrated agent-assisted research that advances research and technological equity and reduces health care resource disparities.

Context: Code reviews are essential for maintaining software quality, yet many human review comments suffer from issues such as redundancy, vagueness, or lack of constructiveness. These types of comments may slow down feedback and obscure important insights. Prior work on code review comments mostly explore the detection and categorization of useful comments, while fine-grained categorization of comment issues remains underexplored. Objective: This work aims to design and evaluate an automated system for classifying code review comments according to specific categories of issues. Methodology: We introduced a nine-label taxonomy for code review comments, covering six review comment smells and three common useful intents, and manually labeled 448 comments from a publicly available dataset. We benchmarked zero-shot and one-shot single-label classification over each comment and its associated unified diff hunk, comparing GPT-5-mini, LLaMA-3.3, and DeepSeek-R1. We reported macro-F1 as the primary metric. Results: Zero-shot performance was moderate under class imbalance (macro-F1 0.360 to 0.374). One-shot exemplar conditioning had model-dependent effects: GPT-5-mini and DeepSeek-R1 macro-F1 scores improved, however LLaMA-3.3 suffered a slight decrease. Exemplars most consistently helped intent-boundary labels, whereas classification of evidence-sensitive labels remain challenging. Conclusion: Our results indicate that comment--diff evidence is sufficient for some labels but limited for evidence-sensitive smells. Future work includes adding thread context, improving intent-preserving rewrites, and validating robustness across platforms.

Line current differential relays (LCDRs) are measurement-driven relays that rely on time-synchronized multi-phase current waveforms to infer internal faults in AC and DC power networks. In inverter-based microgrids, however, the increasing reliance on digitally communicated measurements exposes LCDRs to false-data injection attacks (FDIAs), in which adversaries manipulate remote measurement streams to create protection-triggering yet physically inconsistent current trajectories. This paper addresses this emerging measurement integrity problem by introducing a measurement integrity validation scheme that operates as a supervisory instrumentation layer for modern LCDRs. The proposed scheme interprets short windows of synchronized instantaneous current measurements recorded during relay operation and assesses their physical consistency to distinguish genuine fault-induced trajectories from cyber-manipulated measurement streams. A recurrent neural network is trained offline using only relay-available current measurements and exploits the temporal structure of differential current waveforms, which remains informative in inverter-dominated systems where current magnitude is no longer a reliable observable. The method requires no additional sensors, auxiliary protection elements, or prior knowledge of network topology, and is applicable to both AC and DC LCDRs without structural modification. The proposed measurement validation scheme is evaluated on an islanded inverter-based microgrid under a comprehensive set of fault and FDIA scenarios, demonstrating high detection accuracy while preserving relay dependability. Hardware-in-the-loop validation using an OPAL-RT real-time simulator confirms that the scheme satisfies protection timing constraints and can operate in real time under realistic operating conditions.

Chip placement plays an important role in physical design. While generative models like diffusion models offer promising learning-based solutions, current methods have the following limitations: they use random synthetic data for pre-training, require long sampling times, and often result in overlaps due to their dependence on gradient-based solvers during the sampling process. To overcome these issues, we propose FlowPlace, which features mask-guided synthetic data generation, flow-based efficient training with flexible prior injection, and hard constraint sampling for overlap-free layouts. Experiments on OpenROAD and ICCAD 2015 benchmarks show FlowPlace achieves better PPA metrics, 10-50$\times$ faster sampling efficiency, and zero overlaps.

Reliable agricultural data is essential for food security, land-use planning, and economic resilience, yet in Palestine, such data remains difficult to collect at scale because of fragmented landscapes, limited field access, and restrictions on aerial monitoring. This paper presents ResAF-Net, a satellite-based tree detection framework designed for large-scale agricultural monitoring in resource-constrained settings. The proposed architecture combines a ResNet-50 encoder, Atrous Spatial Pyramid Pooling (ASPP), a feature-fusion stage, a multi-head self-attention refinement module, and an anchor-free FCOS detection head to improve tree localization in dense and heterogeneous scenes. Trained on the MillionTrees benchmark, the model achieved 82% Recall, 63.03% mAP@0.50, and 35.47% mAP@0.50:0.95 on the validation split, indicating strong sensitivity to tree presence while maintaining competitive localization quality. Beyond benchmark evaluation, we implemented the model within a web-based GIS application integrated with Palestinian cadastral data from GeoMolg, enabling tree analysis at scene, parcel, and community levels. This deployment demonstrates the practical feasibility of AI-assisted agricultural inventorying in Palestine. It provides a foundation for data-driven monitoring, reporting, and future species-level analysis of Mediterranean tree crops.

In Transformer models, non-GEMM (non-General Matrix Multiplication) operations -- especially Softmax and Layer Normalization (LayerNorm) -- often dominate hardware cost due to their nonlinear nature. To address this, previous approximation studies mainly target rank-oriented tasks, which is acceptable for classification. However, edge Natural Language Processing (NLP) applications and edge generative AI are largely evaluated based on score-oriented tasks, so normalization-guaranteed non-GEMM operations are essential. We propose a hardware-efficient Softmax and LayerNorm with Guaranteed Normalization for Edge devices. Our design employs hardware-efficient approximation methods while preserving the normalization (Softmax: $\sum p = 1$, LayerNorm: $σ= 1$). Our architecture is described in Verilog HDL and synthesized using the Samsung 28nm CMOS process. In accuracy evaluation, we achieve high accuracy with minimal degradation: GLUE +0.07%, SQuAD -0.01%, perplexity -0.09%. Implementation results show that our architecture is small: $942\,μm^2$ for Softmax, $1199\,μm^2$ for LayerNorm. Compared to the state of the art, we achieve up to 11x and 14x reduction in area, respectively.

Recent evidence suggests that frontier AI systems can exhibit agentic misalignment, generating and executing harmful actions derived from internally constructed goals, even without explicit user requests. Existing mitigation methods, such as Reinforcement Learning from Human Feedback (RLHF) and constitutional prompting, operate primarily at the model level and provide only probabilistic safety guarantees. We propose the Policy-Execution-Authorization (PEA) architecture, a "separation-of-powers" design that enforces safety at the system level. PEA decouples intent generation, authorization, and execution into independent, isolated layers connected via cryptographically constrained capability tokens. We present five core contributions: (C1) an Intent Verification Layer (IVL) for ensuring capability-intent consistency; (C2) Intent Lineage Tracking (ILT), which binds all executable intents to the originating user request via cryptographic anchors; (C3) Goal Drift Detection, which rejects semantically divergent intents below a configurable threshold; (C4) an Output Semantic Gate (OSG) that detects implicit coercion using a structured $K \times I \times P$ threat calculus (Knowledge, Influence, Policy); and (C5) a formal verification framework proving that goal integrity is maintained even under adversarial model compromise. By shifting agent alignment from a behavioral property to a structurally enforced system constraint, PEA provides a robust foundation for the governance of autonomous agents.

Intelligent document processing pipelines extract structured entities (tables, images, and text) from documents for use in downstream systems such as knowledge bases, retrieval-augmented generation, and analytics. A persistent limitation of existing pipelines is that extraction output is produced without any intrinsic mechanism to verify whether it faithfully represents the source. Model-internal confidence scores measure inference certainty, not correspondence to the document, and extraction errors pass silently into downstream consumers. We present Reconstruction as Validation (RaV-IDP), a document processing pipeline that introduces reconstruction as a first-class architectural component. After each entity is extracted, a dedicated reconstructor renders the extracted representation back into a form comparable to the original document region, and a comparator scores fidelity between the reconstruction and the unmodified source crop. This fidelity score is a grounded, label-free quality signal. When fidelity falls below a per-entity-type threshold, a structured GPT-4.1 vision fallback is triggered and the validation loop repeats. We enforce a bootstrap constraint: the comparator always anchors against the original document region, never against the extraction, preventing the validation from becoming circular. We further propose a per-stage evaluation framework pairing each pipeline component with an appropriate benchmark. The code pipeline is publicly available at https://github.com/pritesh-2711/RaV-IDP for experimentation and use.

Domestic voice assistants and smart-home devices are increasingly embedded in everyday routines, yet their ethics are often treated as an afterthought or delegated to compliance teams. To explore how expectations about smart-home AI are constructed and managed, we conducted 33 semi-structured interviews with designers, developers, and researchers from major smart-home platforms (Amazon Alexa, Microsoft Azure IoT, and Google Nest). Using a constructivist grounded theory approach, we develop Expectations Management (EM): a culturally embedded model describing how practitioners shape, calibrate, and repair expectations by balancing organisational rights with culturally situated rites. We show that EM differs from expectation-confirmation theory and trust-calibration by foregrounding moral judgement, situated action, and cross-cultural variation. Our analysis reveals four recurring design tensions: automation vs. autonomy, helpfulness vs. intrusiveness, personalisation vs. predictability, and transparency vs. obscurity and distils them into a five-phase EM Design Playbook that supports moral prudence. We discuss implications for responsible smart-home design and offer guidance for human-centred AI.

Constraint-based causal discovery is brittle in finite-sample regimes because erroneous conditional-independence (CI) decisions can cascade into substantial structural errors. We propose Quantitative Argumentation for Causal Discovery (QACD), a semantics-driven framework that represents CI outcomes as graded, defeasible arguments rather than irreversible constraints. QACD maps statistical test outcomes to argument strengths and aggregates conflicting evidence through connectivity-mediated witness propagation, producing a fixed-point acceptability labeling over candidate adjacencies. Experiments on standard benchmark Bayesian networks suggest that QACD improves structural coherence and interventional reliability in several noisy or inconsistent CI regimes, while remaining competitive with classical constraint-based, hybrid, and prior argumentation-based baselines.

Hierarchical clustering is a fundamental task in data analysis, yet for a long time it lacked a principled objective function. Dasgupta [STOC 2016] initiated a formal framework by introducing a discrete objective function for cluster trees. This framework was subsequently expanded by Cohen-Addad et al. [J. ACM 2019], who introduced the notion of admissibility -- a criterion ensuring that, whenever the input similarities admit consistent hierarchical representations, the minimizers of an objective function recover them. They also provided a necessary and sufficient condition for admissibility within a broad class of objective functions, which we refer to as sum-type objective functions. Our contributions are twofold. First, we characterize admissible sum-type objective functions when the scaling function g is a symmetric polynomial of degree at most two, together with sufficient conditions in the degree-three case. For admissible objective functions in this class, we show that the recursive sparsest cut algorithm achieves an O($φ$)-approximation ratio, where $φ$ denotes the approximation factor of the sparsest cut subroutine. Second, we introduce a new class of objective functions for hierarchical clustering, which we term max-type objective functions, where cluster interactions are measured by maximum rather than aggregate similarity. For this class, we establish a general characterization of admissibility for arbitrary scaling functions g, and a complete characterization when g is a symmetric polynomial of degree at most two. These results provide new theoretical insights into admissible objective functions for hierarchical clustering and clarify the scope of algorithmic guarantees for optimizing them.

Resources for Grammatical Error Correction (GEC) in non-English languages are scarce, while available spellcheckers in these languages are mostly limited to simple corrections and rules. In this paper we introduce a first GEC corpus for Romanian consisting of 10k pairs of sentences. In addition, the German version of ERRANT (ERRor ANnotation Toolkit) scorer was adapted for Romanian to analyze this corpus and extract edits needed for evaluation. Multiple neural models were experimented, together with pretraining strategies, which proved effective for GEC in low-resource settings. Our baseline consists of a small Transformer model trained only on the GEC dataset (F0.5 of 44.38), whereas the best performing model is produced by pretraining a larger Transformer model on artificially generated data, followed by finetuning on the actual corpus (F0.5 of 53.76). The proposed method for generating additional training examples is easily extensible and can be applied to any language, as it requires only a POS tagger

LLM routing has achieved promising results in integrating the strengths of diverse models while balancing efficiency and performance. However, to support more realistic and challenging applications, routing must extend into agentic LLM settings, where task planning, multi-round cooperation among heterogeneous agents, and memory utilization are indispensable. To address this gap, we propose GraphPlanner, a heterogeneous graph memory-augmented agentic router for multi-agent LLMs that generates routing workflows for each query and supports both inductive and transductive inference. GraphPlanner formulates workflow generation as a Markov Decision Process (MDP), where at each step it selects both the LLM backbone and the agent role, including Planner, Executor, and Summarizer. By leveraging a heterogeneous graph, denoted as GARNet, to capture interaction memories among queries, agents, and responses, GraphPlanner integrates historical memory and workflow memory into richer state representations. The entire pipeline is optimized with reinforcement learning, jointly improving task-specific performance and computational efficiency. We evaluate GraphPlanner across 14 diverse LLM tasks and demonstrate that: (1) GraphPlanner outperforms strong single-round and multi-round routers, improving accuracy by up to 9.3% while reducing GPU cost from 186.26 GiB to 1.04 GiB; (2) GraphPlanner generalizes robustly to unseen tasks and LLMs, exhibiting strong zero-shot capabilities; and (3) GraphPlanner effectively leverages historical memories, supporting both inductive and transductive inference for more adaptive routing. Our code for GraphPlanner is released at https://github.com/ulab-uiuc/GraphPlanner.

Recent advancements in large language models (LLMs) have catalyzed the rise of reasoning-intensive inference paradigms, where models perform explicit step-by-step reasoning before generating final answers. While such approaches improve answer quality and interpretability, they incur substantial computational overhead due to the prolonged generation sequences. In this paper, we propose Tandem, a novel collaborative framework that synergizes large and small language models (LLMs and SLMs) to achieve high-quality reasoning with significantly reduced computational cost. Specifically, the LLM serves as a strategic coordinator, efficiently generating a compact set of critical reasoning insights. These insights are then used to guide a smaller, more efficient SLM in executing the full reasoning process and delivering the final response. To balance efficiency and reliability, Tandem introduces a cost-aware termination mechanism that adaptively determines when sufficient reasoning guidance has been accumulated, enabling early stopping of the LLM's generation. Experiments on mathematical reasoning and code generation benchmarks demonstrate that Tandem reduces computational costs by approximately 40% compared to standalone LLM reasoning, while achieving superior or competitive performance. Furthermore, the sufficiency classifier trained on one domain transfers effectively to others without retraining. The code is available at: https://github.com/Applied-Machine-Learning-Lab/ACL2026_Tandem.

This paper studies interpretable and fair artificial intelligence architectures for understanding English reading. Introduced transformer-based models, integrating advanced attention mechanisms and gradient-based feature attribution. The model's lack of interpretability, reduction of algorithmic bias, and unreliable performance in learning environments are the current issues faced in natural language teaching. A unified technical pipeline has been constructed, including adversarial bias correction methods, token-level attribution analysis, and multi-head attention heatmap visualization. Experimental validation was conducted using a large-scale labeled English reading comprehension dataset, and the data partitioning scheme and parameter optimization procedures have been determined. The method significantly outperforms the state-of-the-art models for this task in terms of accuracy and macro-average F1 score; in some aspects, it even surpasses or closely matches the results of human evaluations. In multi-week user experiments, the explainable transformer improved teachers' trust and operability in feedback-based assessments within the scoring system. The proposed method aims to ensure high prediction accuracy and fairness for different learners. This indicates that it is a real-world educational application based on artificial intelligence with a focus on interpretation. Improve the user experience in AI-assisted reading comprehension systems, counteract biases, and enhance the details explained by transformers.

Lattice Hamiltonian systems underpin models across condensed matter, nonlinear optics, and biophysics, yet learning their dynamics from data is obstructed by two unknowns: the interaction topology and whether node dynamics are homogeneous. Existing graph-based approaches either assume the graph is given or, as in $α$-separable graph Hamiltonian network, infer it only for separable Hamiltonians with homogeneous node dynamics. We introduce the Hamiltonian Graph Inference Network (HGIN), which jointly recovers the interaction graph and predicts long-time trajectories from state data alone, for both separable and non-separable Hamiltonians and under heterogeneous node dynamics. HGIN couples a structure-learning module -- a learnable weighted adjacency matrix trained under a Hamilton's-equations loss -- with a trajectory-prediction module that partitions edges into physically distinct subgraphs via $k$-means clustering, assigning each subgraph its own encoder and thereby breaking the parameter-sharing bottleneck of conventional GNNs. On three benchmarks -- a Klein--Gordon lattice with long-range interactions and two discrete nonlinear Schrödinger lattices (homogeneous and heterogeneous) -- HGIN reduces long-time energy prediction error and trajectory prediction error by six to thirteen orders of magnitude relative to baselines. A symmetry argument on the Hamiltonian loss further shows that the learned weights encode the parity of the underlying pair potential, yielding an interpretable readout of the system's interaction structure.

The application of large language models (LLMs) in clinical decision support faces significant challenges of "tunnel vision" and diagnostic hallucinations present in their processing unstructured electronic health records (EHRs). To address these challenges, we propose a novel chain-based clinical reasoning framework, called DxChain, which transforms the diagnostic workflow into an iterative process by mirroring a clinician's cognitive trajectory that consists of "Memory Anchoring", "Navigation" and "Verification" phases. DxChain introduces three key methodological innovations to elicit the potential of LLM: (i) a Profile-Then-Plan paradigm to mitigate cold-start hallucinations by establishing a panoramic patient baseline, (ii) a Medical Tree-of-Thoughts (Med-ToT) algorithm for strategic look ahead planning and resource aware navigation, and (iii) a Dialectical Diagnostic Verification procedure utilizing "Angel-Devil" adversarial debates to resolve complex evidence conflicts. Evaluated on two real world benchmarks, MIMIC-IV-Ext Cardiac Disease and MIMIC-IV-Ext CDM, DxChain achieves state-of-the-art performances in both diagnostic accuracy and logical consistency, offering a modular and reliable architecture for next-generation clinical AI. The code is at https://anonymous.4open.science/r/Dx-Chain.

Early, tool-free prediction of post-synthesis timing remains a key obstacle to rapid RTL iteration. We introduce TimingLLM, a two-stage retrieval-augmented LLM pipeline that estimates worst negative slack (WNS) and total negative slack (TNS) directly from Verilog. Stage 1 is a fine-tuned LLM that acts as a compact post-synthesis timing oracle, producing path-level arrivals/required times that are summarized into lightweight structural-timing cues (e.g., bag-of-gates counts, critical-path depth, gate-type patterns). Stage 2 is an LLM-based regressor that predicts WNS/TNS and applies a learned diagonal steering vector at the last transformer block, computed from the k nearest timing-labeled modules in a disjoint retrieval bank. On VerilogEval, TimingLLM attains R_WNS = 0.91 (MAPE 12%) and R_TNS=0.97 (MAPE 16%) while running 1.3-1.6 times faster than prior methods. Training uses a new 60k-module Verilog corpus with synthesis reports, which we will release. After training once, TimingLLM can be adapted to new technology libraries and PVT corners by refitting only a small regression head on 1000 labeled modules per setting, consistently outperforming state-of-the-art baselines.

This Article argues that conversations with companion chatbot should be subject to a clear structural distinction between commercial and non-commercial contexts. The insertion of undisclosed promotional content into affective or relational exchanges should be prohibited, as it collapses the boundary between market transaction and communicative intimacy in ways that erode user autonomy and conversational context. The Article begins by theorizing digital companionship as a sociotechnical form that reconfigures intimacy, dependence and relational vulnerability. It then introduces the potential economic harms derived from conversational advertising. The Article ultimately argues for a firm legal and social distinction between commercial and non-commercial conversational contexts as a precondition for the responsible stabilization of these technologies within social life.

Large Language Models (LLMs) are increasingly deployed in persona-driven applications such as education, customer service, and social platforms, where models are prompted to adopt specific personas when interacting with users. While persona conditioning can improve user experience and engagement, it also raises concerns about how personality cues may interact with gender biases and stereotypes. In this work, we present a controlled study of persona-conditioned story generation in English and Hindi, where each story portrays a working professional in India producing context-specific artifacts (e.g., lesson plans, reports, letters) under systematically varied persona gender, occupational role, and personality traits from the HEXACO and Dark Triad frameworks. Across 23,400 generated stories from six state-of-the-art LLMs, we find that personality traits are significantly associated with both the magnitude and direction of gender bias. In particular, Dark Triad personality traits are consistently associated with higher gender-stereotypical representations compared to socially desirable HEXACO traits, though these associations vary across models and languages. Our findings demonstrate that gender bias in LLMs is not static but context-dependent. This suggests that persona-conditioned systems used in real-world applications may introduce uneven representational harms, reinforcing gender stereotypes in generated educational, professional, or social content.

Gaussian basis functions provide an efficient and flexible alternative to spline activations in KANs. In this work, we introduce the partition-of-unity Gaussian KAN (PU-GKAN), a Shepard-type normalized Gaussian KAN in which the Gaussian basis values on each edge are divided by their local sum over fixed centers. This produces a partition-of-unity feature map with trainable coefficients, while preserving the standard edge-based KAN structure. The normalized construction gives exact constant reproduction at the edge level and admits an explicit finite-feature kernel interpretation. We formulate both the standard Gaussian KAN (GKAN) and PU-GKAN from a finite-feature and additive-kernel viewpoint, making the induced layer kernels and empirical feature matrices explicit. Using the first-layer feature matrix as the reference object, we adopt a practical scale-selection interval for \(ε\), with the lower endpoint determined by adjacent-center overlap and the upper endpoint determined by a conservative conditioning threshold. Numerical experiments show that PU-GKAN reduces sensitivity to \(ε\), improves validation accuracy for most smooth and moderately non-smooth targets, and gives more stable training behavior. The benefit persists across sample-size and center-number sweeps, higher-dimensional architectures, Matérn RBF bases, and physics-informed examples involving Helmholtz and wave equations. These results indicate that Shepard-type partition-of-unity normalization is a simple and effective stabilization mechanism for RBF-based KANs.

The volume of scientific submissions continues to climb, outpacing the capacity of qualified human referees and stretching editorial timelines. At the same time, modern large language models (LLMs) offer impressive capabilities in summarization, fact checking, and literature triage, making the integration of AI into peer review increasingly attractive -- and, in practice, unavoidable. Yet early deployments and informal adoption have exposed acute failure modes. Recent incidents have revealed that hidden prompt injections embedded in manuscripts can steer LLM-generated reviews toward unjustifiably positive judgments. Complementary studies have also demonstrated brittleness to adversarial phrasing, authority and length biases, and hallucinated claims. These episodes raise a central question for scholarly communication: when AI reviews science, can we trust the AI referee? This paper provides a security- and reliability-centered analysis of AI peer review. We map attacks across the review lifecycle -- training and data retrieval, desk review, deep review, rebuttal, and system-level. We instantiate this taxonomy with four treatment-control probes on a stratified set of ICLR 2025 submissions, using two advanced LLM-based referees to isolate the causal effects of prestige framing, assertion strength, rebuttal sycophancy, and contextual poisoning on review scores. Together, this taxonomy and experimental audit provide an evidence-based baseline for assessing and tracking the reliability of AI peer review and highlight concrete failure points to guide targeted, testable mitigations.

Facilitating cross-lingual transfer in multilingual language models remains a critical challenge. Towards this goal, we propose an embedding-based data augmentation technique called XITE. We start with unlabeled text from a low-resource target language, identify an English counterpart in a task-specific training corpus using embedding-based similarities and adopt its label. Next, we perform a simple interpolation of the source and target embeddings to create synthetic data for task-specific fine-tuning. Projecting the target text into a language-rich subspace using linear discriminant analysis (LDA), prior to interpolation, further boosts performance. Our cross-lingual embedding-based augmentation technique XITE yields significant improvements of up to 35.91% for sentiment analysis and up to 81.16% for natural language inference, using XLM-R, for a diverse set of target languages including Korean, Arabic, Urdu and Hindi. Apart from boosting cross-lingual transfer, adaptation using XITE also safeguards against forgetting and maintains task performance on the high-resource language.

Financial AI systems must produce answers grounded in specific regulatory filings, yet current LLMs fabricate metrics, invent citations, and miscalculate derived quantities. These errors carry direct regulatory consequences as the EU AI Act's high-risk enforcement deadline approaches (August 2026). Existing hallucination detectors treat all claims uniformly, missing 43% of computational errors that require arithmetic re-verification against structured tables. We present FinGround, a three-stage verify-then-ground pipeline for financial document QA. Stage 1 performs finance-aware hybrid retrieval over text and tables. Stage 2 decomposes answers into atomic claims classified by a six-type financial taxonomy and verified with type-routed strategies including formula reconstruction. Stage 3 rewrites unsupported claims with paragraph- and table-cell-level citations. To cleanly isolate verification value from retrieval quality, we propose retrieval-equalized evaluation as standard methodology for RAG verification research: when all systems receive identical retrieval, FinGround still reduces hallucination rates by 68% over the strongest baseline ($p < 0.01$). The full pipeline achieves a 78% reduction relative to GPT-4o. An 8B distilled detector retains 91.4% F1 at 18x lower per-claim latency, enabling $0.003/query deployment, supported by qualitative signals from a four-week analyst pilot.

Joint audio-video generation models have shown that unified generation yields stronger cross-modal coherence than cascaded approaches. However, existing models couple modalities throughout denoising via pervasive attention, treating high-level semantics and low-level details in a fully entangled manner. This is suboptimal for talking head synthesis: while audio and facial motion are semantically correlated, their low-level realizations (acoustic signals and visual textures) follow distinct rendering processes. Enforcing joint modeling across all levels causes unnecessary entanglement and reduces efficiency. We propose Talker-T2AV, an autoregressive diffusion framework where high-level cross-modal modeling occurs in a shared backbone, while low-level refinement uses modality-specific decoders. A shared autoregressive language model jointly reasons over audio and video in a unified patch-level token space. Two lightweight diffusion transformer heads decode the hidden states into frame-level audio and video latents. Experiments on talking portrait benchmarks show Talker-T2AV outperforms dual-branch baselines in lip-sync accuracy, video quality, and audio quality, achieving stronger cross-modal consistency than cascaded pipelines.

Financial institutions must track over 60,000 regulatory events annually, overwhelming manual compliance teams; the industry has paid over USD 300 billion in fines and settlements since the 2008 financial crisis. We present ComplianceNLP, an end-to-end system that automatically monitors regulatory changes, extracts structured obligations, and identifies compliance gaps against institutional policies. The system integrates three components: (1) a knowledge-graph-augmented RAG pipeline grounding generations in a regulatory knowledge graph of 12,847 provisions across SEC, MiFID II, and Basel III; (2) multi-task obligation extraction combining NER, deontic classification, and cross-reference resolution over a shared LEGAL-BERT encoder; and (3) compliance gap analysis that maps obligations to internal policies with severity-aware scoring. On our benchmark, ComplianceNLP achieves 87.7 F1 on gap detection, outperforming GPT-4o+RAG by +3.5 F1, with 94.2% grounding accuracy ($r=0.83$ vs. human judgments) and 83.4 F1 under realistic end-to-end error propagation. Ablations show that knowledge-graph re-ranking contributes the largest marginal gain (+4.6 F1), confirming that structural regulatory knowledge is critical for cross-reference-heavy tasks. Domain-specific knowledge distillation (70B $\to$ 8B) combined with Medusa speculative decoding yields $2.8\times$ inference speedup; regulatory text's low entropy ($H=2.31$ bits vs. $3.87$ general text) produces 91.3% draft-token acceptance rates. In four months of parallel-run deployment processing 9,847 updates at a financial institution, the system achieved 96.0% estimated recall and 90.7% precision, with a $3.1\times$ sustained analyst efficiency gain. We report deployment lessons on trust calibration, GRC integration, and distributional shift monitoring for regulated-domain NLP.

Agentic systems that chain reasoning, tool use, and synthesis into multi-step workflows are entering production, yet prevailing evaluation practices like end-to-end outcome checks and ad-hoc trace inspection systematically mask the intermediate failures that dominate real-world error budgets. We present AgentEval, a framework that formalizes agent executions as evaluation directed acyclic graphs (DAGs), where each node carries typed quality metrics assessed by a calibrated LLM judge (GPT-4o), classified through a hierarchical failure taxonomy (3 levels, 21 subcategories), and linked to upstream dependencies for automated root cause attribution. An ablation study isolates the impact of DAG-based dependency modeling: it alone contributes +22 percentage points to failure detection recall and +34 pp to root cause accuracy over flat step-level evaluation with identical judges and rubrics. Across three production workflows (450 test cases, two agent model families, predominantly sequential architectures with a 12% non-DAG trace rate), AgentEval achieves 2.17x higher failure detection recall than end-to-end evaluation (0.89 vs. 0.41), Cohen's kappa = 0.84 agreement with human experts, and 72% root cause accuracy against an 81% human ceiling. Cross-system evaluation on tau-bench and SWE-bench traces confirms transferability (failure detection recall >= 0.78) without taxonomy or rubric modification. A 4-month pilot with 18 engineers detected 23 pre-release regressions through CI/CD-integrated regression testing, reducing median root-cause identification time from 4.2 hours to 22 minutes and driving measurable failure rate reductions in two workflows.

Physics-aware symbolic simulation of 3D scenes is critical for robotics, embodied AI, and scientific computing, requiring models to understand natural language descriptions of physical phenomena and translate them into executable simulation environments. While large language models (LLMs) excel at general code generation, they struggle with the semantic gap between physical descriptions and simulation implementation. We introduce PhysCodeBench, the first comprehensive benchmark for evaluating physics-aware symbolic simulation, comprising 700 manually-crafted diverse samples across mechanics, fluid dynamics, and soft-body physics with expert annotations. Our evaluation framework measures both code executability and physical accuracy through automated and visual assessment. Building on this, we propose a Self-Corrective Multi-Agent Refinement Framework (SMRF) with three specialized agents (simulation generator, error corrector, and simulation refiner) that collaborate iteratively with domain-specific validation to produce physically accurate simulations. SMRF achieves 67.7 points overall performance compared to 36.3 points for the best baseline among evaluated SOTA models, representing a 31.4-point improvement. Our analysis demonstrates that error correction is critical for accurate physics-aware symbolic simulation and that specialized multi-agent approaches significantly outperform single-agent methods across the tested physical domains.

Human daily behavior unfolds as complex sequences shaped by intentions, preferences, and context. Effectively modeling these behaviors is crucial for intelligent systems such as personal assistants and recommendation engines. While recent advances in deep learning and behavior pre-training have improved behavior prediction, key challenges remain--particularly in handling long-tail behaviors, enhancing interpretability, and supporting multiple tasks within a unified framework. Large language models (LLMs) offer a promising direction due to their semantic richness, strong interpretability, and generative capabilities. However, the structural and modal differences between behavioral data and natural language limit the direct applicability of LLMs. To address this gap, we propose Behavior Understanding Alignment (BUA), a novel framework that integrates LLMs into human behavior modeling through a structured curriculum learning process. BUA employs sequence embeddings from pretrained behavior models as alignment anchors and guides the LLM through a three-stage curriculum, while a multi-round dialogue setting introduces prediction and generation capabilities. Experiments on two real-world datasets demonstrate that BUA significantly outperforms existing methods in both tasks, highlighting its effectiveness and flexibility in applying LLMs to complex human behavior modeling.

Serving diverse NLP workloads with large language models is costly: at one enterprise partner, inference costs exceeded $200K/month despite over 70% of queries being routine tasks well within the capability of smaller models. We present RouteNLP, a closed-loop framework that routes queries across a tiered model portfolio to minimize cost while satisfying per-task quality constraints. The framework integrates three components: a difficulty-aware router with shared task-conditioned representations trained on preference data and quality signals; confidence-calibrated cascading that uses conformal prediction for distribution-free threshold initialization; and a distillation-routing co-optimization loop that clusters escalation failures, applies targeted knowledge distillation to cheaper models, and automatically retrains the router, yielding over twice the cost improvement of untargeted distillation. In an 8-week pilot deployment processing ~5K queries/day at an enterprise customer-service division, RouteNLP reduced inference costs by 58% while maintaining 91% response acceptance and reducing p99 latency from 1,847 ms to 387 ms. On a six-task benchmark spanning finance, customer service, and legal domains, the framework achieves 40-85% cost reduction while retaining 96-100% quality on structured tasks and 96-98% on generation tasks, with human evaluation confirming that 74.5% of routed generation outputs match or exceed frontier-model quality.

Ensuring safe exploration in high-dimensional systems with unknown dynamics remains a significant challenge. Existing safe reinforcement learning methods often provide safety guarantees only in expectation, which can still lead to safety violations. Control-theoretic approaches, in contrast, offer hard constraint-based safety guarantees but typically assume access to known system dynamics or require accurate estimation of control-affine models. In this paper, we propose a safe reinforcement learning framework that learns a probabilistic control-affine dynamics model in an offline setting. The learned model is leveraged to explicitly construct control barrier functions (CBFs) that incorporate model uncertainty to provide conservative safety constraints. These CBF constraints are enforced through an online constraint-based action correction mechanism, enabling safe exploration without overly restricting task performance. Empirical evaluations on nonlinear, complex continuous-control benchmarks demonstrate that our approach achieves returns comparable to those of existing baselines while significantly reducing safety violations.

Silicon samples are increasingly used as a low-cost substitute for human panels and have been shown to reproduce aggregate human opinion with high fidelity. We show that, in the alignment-relevant domain of philosophy, silicon samples systematically collapse heterogeneity. Using data from $N = {277}$ professional philosophers drawn from PhilPeople profiles, we evaluate seven proprietary and open-source large language models on their ability to replicate individual philosophical positions and to preserve cross-question correlation structures across philosophical domains. We find that language models substantially over-correlate philosophical judgments, producing artificial consensus across domains. This collapse is associated in part with specialist effects, whereby models implicitly assume that domain specialists hold highly similar philosophical views. We assess the robustness of these findings by studying the impact of DPO fine-tuning and by validating results against the full PhilPapers 2020 Survey ($N = {1785}$). We conclude by discussing implications for alignment, evaluation, and the use of silicon samples as substitutes for human judgment. The code of this project can be found at https://github.com/stanford-del/silicon-philosophers.

Semi-supervised classification, where unlabeled data are massive but labeled data are limited, often arises in machine learning applications. We address this challenge under high-dimensional data by leveraging the manifold and cluster assumptions. Based on the Fermat distance, a density-sensitive metric that naturally encodes the cluster assumption, we propose the weighted $k$-nearest neighbors (NN) classifier and multidimensional scaling (MDS)-induced classifiers. The use of MDS with a large target dimension allows the effective application of linear classifiers to complex manifold data. Theoretically, we derive a sharp lower bound for the expected excess risk within clusters and prove that the weighted $k$-NN classifier utilizing the true Fermat distance is minimax optimal. Furthermore, we explicitly quantify the utility of unlabeled data by showing that the error arising from estimating the Fermat distance decays exponentially with the pooled sample size. Such a rate is much faster than the related rates in the literature. Extensive experiments on synthetic and real datasets demonstrate competitive or superior performance of our approaches compared to state-of-the-art graph-based semi-supervised classifiers.

Privacy-critical domains require phishing detection systems that satisfy contradictory constraints: near-zero false positives to prevent workflow disruption, transparent explanations for non-expert staff, strict regulatory compliance prohibiting sensitive data exposure to external APIs, and robustness against AI-generated attacks. Existing rule-based systems are brittle to novel campaigns, while LLM-based detectors violate privacy regulations through unredacted data transmission. We introduce CyberCane, a neuro-symbolic framework integrating deterministic symbolic analysis with privacy-preserving retrieval-augmented generation (RAG). Our dual-phase pipeline applies lightweight symbolic rules to email metadata, then escalates borderline cases to semantic classification via RAG with automated sensitive data redaction and retrieval from a phishing-only corpus. We further introduce PhishOnt, an OWL ontology enabling verifiable attack classification through formal reasoning chains. Evaluation on DataPhish2025 (12.3k emails; mixed human/LLM) and Nazario/SpamAssassin demonstrates a 78.6-point recall gain over symbolic-only detection on AI-generated threats, with precision exceeding 98% and FPR as low as 0.16%. Healthcare deployment projects a 542x ROI; tunable operating points support diverse risk tolerances, with open-source implementation at https://github.com/sbhakim/Cybercane.

The relationship between brain lateralization and cognitive functions is well-documented. The left hemisphere primarily handles tasks such as language and arithmetic, while the right hemisphere is involved in creative activities like drawing and music perception. Eye-tracking technology has shown the potential to reveal cognitive states by measuring ocular metrics such as pupil diameter and fixation duration. However, the ability to distinguish lateralized brain activity using these ocular metrics remains underexplored. Here, we demonstrate that pupil diameter and fixation duration can effectively classify left and right brain hemisphere activities. We obtained a considerably high classification performance, with an F1 score of 0.894. The results suggest that ocular metrics are robust indicators of lateralized brain activity and can be applied in cognitive monitoring and neurorehabilitation. Our future work expands on this by integrating these methods into real-time applications EyeBrain, potentially broadening their use across various cognitive and neurological domains.

Conflict-free replicated data types (CRDTs) and the local-first concept are increasingly employed not only in small-scale collaboration systems among few users who trust each other, but also in large-scale systems, like Matrix for instant messaging and Keyhive for collaborative documents. Since mutual trust is no longer warranted, these systems require Byzantine fault tolerance and fine-grained access control. As of today, Matrix and Keyhive pair an informal specification with an unverified reference implementation. In this work, we follow a bottom-up approach towards constructing formally verified authorization algorithms for Byzantine fault-tolerant local-first systems. As intermediate target for formal verification, we contribute semantics and invariants of a replicated data type for managing simplified collaboration groups, based on capabilities for access control and hash chronicles for replication. To enable future integration into local-first systems like Matrix and Keyhive, we strive for accessibility to system engineers by using the Rust programming language for formal specification, verification, and implementation, enabled by the Verus framework using the Z3 theorem prover at zero runtime cost. We report on our experience and preliminary results following this approach, and discuss next steps towards scaling up access control expressiveness. Whether this approach can be scaled up to the complexity of real-world local-first access control systems like Matrix or Keyhive remains future work, but our findings demonstrate the potential of system-oriented formal verification with Verus.

Building scalable and reusable multi-agent decision policies from offline datasets remains a challenge in offline multi-agent reinforcement learning (MARL), as existing methods often rely on fixed observation formats and action spaces that limit generalization. In contrast, large language models (LLMs) offer a flexible modeling interface that can naturally accommodate heterogeneous observations and actions. Motivated by this, we propose the Decision Language Model (DLM), which formulates multi-agent decision making as a dialogue-style sequence prediction problem under the centralized training with decentralized execution paradigm. DLM is trained in two stages: a supervised fine-tuning phase, which leverages dialogue-style datasets for centralized training with inter-agent context and generates executable actions from offline trajectories, followed by a group relative policy optimization phase to enhance robustness to out-of-distribution actions through lightweight reward functions. Experiments on multiple benchmarks show that a unified DLM outperforms strong offline MARL baselines and LLM-based conversational decision-making methods, while demonstrating strong zero-shot generalization to unseen scenarios across tasks.

Large language model (LLM) decoding is latency-sensitive and often bottlenecked by fragmented operator execution and repeated off-chip materialization of intermediate tensors. Prior work expands fusion scope by leveraging thread-block clusters and on-chip inter-block collectives to fuse attention-side operators such as QKV projection, attention, and output projection. We develop ClusterFusion++, a CUDA-level extension that broadens fusion to the full Transformer decoder block for GPT-NeoX/Pythia models: LayerNorm -> QKV -> RoPE -> decode attention -> output projection -> Post-LN -> MLP -> residual. We additionally engineer a CUDA-Graph-compatible execution mode with persistent Tensor Memory Accelerator (TMA) descriptors to reduce per-step overhead. On an NVIDIA RTX 5090-class GPU, ClusterFusion++ improves throughput by 1.34x for Pythia-2.8B and yields similar gains for Pythia-6.9B, while maintaining high output fidelity (near-token-identical generation, with minor non-determinism from FP16 atomics).

Diffusion models are central to modern generative modeling, and understanding how they balance memorization and generalization is critical for reliable deployment. Recent work has shown that memorization in diffusion models is shaped by training dynamics, with generalization and memorization emerging at different stages of training. However, deployed diffusion models are often further distilled, introducing an additional training phase whose impact on memorization is not well understood. In this work, we analyze how distillation reshapes memorization behavior in diffusion models, taking consistency distillation as a representative framework. Empirically, we show that when applied to a teacher model that has memorized data, consistency distillation significantly reduces transferred memorization in the student while preserving, and sometimes improving, sample quality. To explain this behavior, we provide a theoretical analysis using a random feature neural network model [Bonnaire et al., 2025], showing that consistency distillation suppresses unstable feature directions associated with memorization while preserving stable, generalizable modes. Our findings suggest that distillation can serve not only as an acceleration tool, but also as a mechanism for improving the memorization-generalization trade-off.

Efficient hardware implementation of nonlinear activation functions is a crucial task in deploying artificial neural networks on resource-constrained and edge devices such as Field-Programmable Gate Arrays (FPGAs). The sigmoid activation function is widely used for probabilistic output, binary classification, and gating mechanisms in recurrent neural networks, despite its reliance on exponential computations. This paper presents a hardware-efficient FPGA implementation of the sigmoid activation function using a mixed-radix CORDIC-based architecture. The proposed approach leverages the mathematical relationship between the sigmoid and hyperbolic tangent functions. The input range is normalized to 1, enabling the corresponding tanh computation to operate within a reduced range of 0.5, which significantly improves convergence behavior. To achieve high accuracy with minimal hardware overhead, a modified mixed-radix hyperbolic rotation CORDIC (MR-HRC) algorithm combining radix-2 and radix-4 iterations is introduced. The initial radix-2 stage ensures stable convergence, while the subsequent radix-4 stage accelerates convergence without requiring scale-factor compensation. In the final stage, a radix-2 linear vectoring CORDIC (R2-LVC) is used to compute the hyperbolic tangent by dividing hyperbolic sine and cosine values derived from the MR-HRC algorithm. The entire architecture is fully pipelined and implemented on an FPGA. The design is realized on an Xilinx Virtex-7 FPGA using a 16-bit fixed-point representation. Experimental results demonstrate a significant reduction in hardware utilization, requiring only 835 logic slices with zero DSP usage. Additionally, the design achieves a mean absolute error of 4.23 10^-4, outperforming several recent sigmoid implementations.

Optical chemical structure recognition (OCSR) translates molecular images into machine-readable representations like SMILES strings or molecular graphs, but remains challenging in real-world documents due to inexhaustible variations in chemical structures, shorthand conventions, and visual noise. Most existing deep-learning-based approaches rely on teacher forcing with token-level Maximum Likelihood Estimation (MLE). This training paradigm suffers from exposure bias, as models are trained under ground-truth prefixes but must condition on their own previous predictions during inference. Moreover, token-level MLE objectives hinder the optimization towards molecular-level evaluation criteria such as chemical validity and structural similarity. Here we introduce Minimum Risk Training (MRT) to OCSR and propose COMO (Closed-loop Optical Molecule recOgnition), a closed-loop framework that mitigates exposure bias by directly optimizing over molecule-level, non-differentiable objectives, by iteratively sampling and evaluating the model's own predictions. Experiments on ten benchmarks including synthetic and real-world chemical diagrams from patent and scientific literature demonstrate that COMO substantially outperforms existing rule-based and learning-based methods with less training data. Ablation studies further show that MRT is architecture-agnostic, demonstrating its potential for broad application to end-to-end OCSR systems.

Test-time alignment methods offer a promising alternative to fine-tuning by steering the outputs of large language models (LLMs) at inference time with lightweight interventions on their internal representations. Recently, a prominent and effective approach, RE-Control (Kong et al., 2024), has proposed leveraging an external value function trained over the LLM's hidden states to guide generation via gradient-based editing. While effective, this method overlooks a key characteristic of alignment tasks, i.e. that they are typically formulated as learning from human preferences between candidate responses. To address this, in this paper we propose a novel preference-based training framework, Pref-CTRL, that uses a multi-objective value function to better reflect the structure of preference data. Our approach has outperformed RE-Control on two benchmark datasets and showed greater generalization on out-of-domain datasets. Our source code is available at https://github.com/UTS-nlPUG/pref-ctrl.

The rapid proliferation of Generative AI necessitates rigorous documentation standards for transparency and governance. However, manual creation of Model and Data Cards is not scalable, while automated approaches lack large-scale, high-fidelity benchmarks for systematic evaluation. We introduce MetaGAI, a comprehensive benchmark comprising 2,541 verified document triplets constructed through semantic triangulation of academic papers, GitHub repositories, and Hugging Face artifacts. Unlike prior single-source datasets, MetaGAI employs a multi-agent framework with specialized Retriever, Generator, and Editor agents, validated through four-dimensional human-in-the-loop assessment, including human evaluation of editor-refined ground truth. We establish a robust evaluation protocol combining automated metrics with validated LLM-as-a-Judge frameworks. Extensive analysis reveals that sparse Mixture-of-Experts architectures achieve superior cost-quality efficiency, while a fundamental trade-off exists between faithfulness and completeness. MetaGAI provides a foundational testbed for benchmarking, training, and analyzing automated Model and Data Card generation methods at scale. Our data and code are available at: https://github.com/haoxuan-unt2024/MetaGAI-Benchmark.

Short-term human pose prediction plays a crucial role in interactive systems, assistive robots, and emotion-aware human-computer interaction[1-3]. While current trajectory prediction models primarily rely on geometric motion cues, they often overlook the underlying emotional signals influencing human motion dynamics[4-5]. This paper investigates whether facial expression-derived emotion embeddings can provide auxiliary conditional signals for short-term pose prediction. To further evaluate multimodal conditionation in a recursive prediction setting, we propose a lightweight autoregressive predictive world model that performs 15-step rolling pose prediction. This framework combines pose keypoints with emotion embeddings through a learnable gating mechanism and performs autoregressive unfolding prediction using a recurrent sequence model based on a two-layer LSTM architecture. Experiments were conducted on two small-scale pose-emotion video datasets: controlled motion sequences with minimal facial expression changes and, natural emotion-driven motion sequences with considerable facial expression changes. The results show that simple multimodal fusion does not consistently improve prediction accuracy, while normalized gating fusion significantly enhances the performance of emotion-driven motion sequences. Furthermore, counterfactual perturbation experiments demonstrate that the predicted trajectory exhibits measurable sensitivity to changes in multimodal input, suggesting that facial expression embeddings act as auxiliary conditional signals rather than redundant features. In summary, these results indicate that incorporating facial expression-derived emotion embeddings into emotion-conditional short-term pose prediction based on a lightweight predictive world model architecture is a feasible approach.

Targeting entailment model checking, a recent study has pioneered an idea of Eigenmarking search, an improvement over Grover search using extra qubits. The extra qubits condition the quantum state evolution such that the answer states (if exist) are always in the minority. The minority criteria is essential to Grover probability-amplitude amplification and consequently the effectiveness of Grover search. In addition to enforce the minority criteria, Eigenmarking also employs complementary states (through well-orchestrated phase rotation) for easy identification of a no-answer case (related to a no-violation case in the context of model checking). Eigenmarking search has been shown effective in two-qubit simulations. The three Eigenmarking schemes have been previously proposed. Two schemes require two extra qubits. One scheme (called ``subtle marking'') requires one extra qubit with a multiple-qubit-controlled phase rotation. Our study refines the mechanism using only one extra qubit with only two-qubit-controlled phase rotation, commonly known as \texttt{ccz}, regardless of how many qubits the input has. Using a multiple-qubit-controlled phase rotation (as in subtle marking) associates with highly entangled states. Highly entangled states in a real quantum hardware are difficult (or in some cases may even be unachievable) particularly in a scaled up scenario involving many qubits. Our proposed new Eigenmarking scheme has lightened the burden for the hardware requirement. The new Eigenmarking search has been experimented in two-qubit-system simulations and shown viable, achieving the minimal relative local winning margin of W=3.17 and the worst-case distinguishability of D=0.769 (cf. W=0.67; D=0.19 from conventional marking and W=0.28; D=0.55 from subtle marking).

Multi-turn, long-horizon tasks are increasingly common for large language models (LLMs), but solving them typically requires many sequential model invocations, accumulating substantial inference costs. Here, we study cost-aware multi-turn LLM routing: selecting which model to invoke at each turn from a model pool, given a fixed cost budget. We propose MTRouter, which encodes the interaction history and candidate models into joint history-model embeddings, and learns an outcome estimator from logged trajectories to predict turn-level model utility. Experiments show that MTRouter improves the performance-cost trade-off: on ScienceWorld, it surpasses GPT-5 while reducing total cost by 58.7%; on Humanity's Last Exam (HLE), it achieves competitive accuracy while reducing total cost by 43.4% relative to GPT-5, and these gains even carry over to held-out tasks. Further analyses reveal several mechanisms underlying its effectiveness: relative to prior multi-turn routers, MTRouter makes fewer model switches, is more tolerant to transient errors, and exhibits emergent specialization across models. Code: https://github.com/ZhangYiqun018/MTRouter

Physics-informed neural networks (PINNs) provide a promising machine learning framework for solving partial differential equations, but their training often breaks down on challenging problems, sometimes converging to physically incorrect solutions despite achieving small residual losses. This failure, we argue, is not merely an optimization difficulty. Rather, it reflects a fundamental weakness of the empirical PDE residual loss, which can admit trivial or spurious solutions during training. From this perspective, we revisit pseudo-time stepping, a technique that has recently shown strong empirical success in PINNs. We show that its main benefit is not simply to ease optimization; instead, when combined with collocation-point resampling, it helps reveal and avoid spurious solutions. At the same time, we find that the effectiveness of pseudo-time stepping depends critically on the choice of step size, which cannot be tuned reliably from the training loss alone. To overcome this limitation, we propose an adaptive pseudo-time stepping strategy that selects the step size from a finite-difference surrogate of the local residual Jacobian, yielding the largest step permitted by local stability without per-problem tuning. Across a diverse set of PDE benchmarks, the proposed method consistently improves both accuracy and robustness. Together, these findings provide a clearer understanding of why PINNs fail and suggest a practical pathway toward more reliable physics-informed learning. All code and data accompanying this manuscript are available at https://github.com/sifanexisted/jaxpi2.

Safety specifications in cyber-physical systems (CPS) capture the operational conditions the system must satisfy to operate safely within its intended environment. As operating environments evolve, operational rules must be continuously refined to preserve consistency with observed system behavior during simulation-based verification and validation. Revising inconsistent rules is challenging because the changes must remain syntactically correct under a domain-specific grammar. Language-in-the-loop refinement further raises safety concerns beyond syntactic violations, as it can produce semantically unjustified refinements that overfit to the observed outcomes. We introduce a framework that combines counterfactual reasoning with a grammar-constrained refinement loop to refine operational rules, aligning them with the observed system behavior. Applied to an autonomous driving control system, our approach successfully resolved the inconsistencies in an operational rule inferred by a conventional baseline while remaining grammar compliant. An empirical large language model (LLM) study further revealed model-dependent refinement quality and safety lessons, which motivate rigorous grammar enforcement, stronger semantic validation, and broader evaluation in future work.

Multi-plane architectures have become increasingly prevalent in the Fat-Tree networks of AI data centers. By leveraging multiple ports on a single network interface card (NIC) or multiple NICs within a scale-up domain, each port or NIC is allocated to an independent network plane, thereby provisioning the overall system with multiple network planes. However, no prior literature has explored the application of multi-plane technologies to direct networks such as HyperX. This paper investigates the multi-plane HyperX network and demonstrates that, compared to state-of-the-art network topologies like multi-plane Fat-Tree, Dragonfly, and Dragonfly+, the multi-plane HyperX architecture achieves a significantly smaller network diameter and superior cost-effectiveness.

Existing theory suggests that Kolmogorov-Arnold Networks (KANs) can overcome the spectral bias commonly observed in neural networks under the assumption that inputs are statistically independent. However, this assumption does not hold in time series forecasting (TSF), where inputs are lagged observations with strong temporal autocorrelation. Through theoretical analysis and empirical validation, we obtain an unexpected finding: temporal autocorrelation reintroduces spectral bias in KANs, and the bias becomes increasingly pronounced as the degree of autocorrelation increases. This suggests that standard KANs may face substantial difficulties in TSF with strongly autocorrelated inputs. To address this problem, we introduce the Discrete Cosine Transform (DCT) to reduce the correlations among the network inputs. As expected, experimental results reveal that DCT preprocessing substantially reduces the observed low-frequency preference in TSF. This result also corroborates that the spectral bias of KANs in TSF tasks is indeed induced by the autocorrelation among input variables.

The health condition of components in civil infrastructures can be described by various discrete states according to their performance degradation. Inferring these states from measurable responses is typically an ill-posed inverse problem. Although Bayesian methods are well-suited to tackle such problems, computing the posterior probability density function (PDF) presents challenges. The likelihood function cannot be analytically formulated due to the unclear relationship between discrete states and structural responses, and the high-dimensional state parameters resulting from numerous components severely complicates the computation of the marginal likelihood function. To address these challenges, this study proposes a novel Bayesian inversion paradigm for discrete variables based on Probabilistic Graphical Models (PGMs). The Markov networks are employed as modeling tools, with model parameters learned from data and structural topology prior. It has been proved that inferring this PGM produces the same probabilistic estimation as the posterior PDF derived from Bayesian inference, which effectively solves the above challenges. The inference is accomplished by Graph Neural Networks (GNNs), and a graph property-based GNN training strategy is developed to enable accurate inference across varying graph scales, thereby significantly reducing the computational overhead in high-dimensional problems. Both synthetic and experimental data are used to validate the proposed framework

Collusion among autonomous agents poses a critical security threat in embodied multi-agent systems (MAS), where coordinated behaviors can deviate from global objectives and lead to real-world consequences. Existing defenses, primarily based on identity control or post-hoc behavior analysis, are insufficient to address such threats in embodied settings due to delayed feedback and noisy observations in physical environments, which make behavioral deviations difficult to detect accurately and in a timely manner. To address this challenge, we propose a mutagenic incentive intervention approach that mitigates collusion by reshaping agents' payoff structures. By rewarding agents who report collusive behavior and penalizing identified participants, the mechanism induces strategic defection and renders collusion unstable. We further design supporting mechanisms, including reporting deposits, smart contract-based reward enforcement, and encrypted communication, to ensure robustness against misuse of the incentive mechanism and retaliation from penalized agents. We implement the proposed approach in both simulated and real-world embodied environments. Experimental results show that our method effectively suppresses collusion by inducing defection, while preserving system efficiency. It achieves performance comparable to the non-collusion baseline and outperforms representative reactive defenses, thereby fulfilling the desired security objectives. These results demonstrate the effectiveness of proactive incentive design as a practical paradigm for securing embodied multi-agent systems.

Business logic bugs violate intended business semantics and are particularly prevalent in enterprise software. Yet most existing unit test generation techniques are code-centric, making such bugs difficult to expose. We present SeGa, a semantics-driven unit test generation technique for uncovering business logic bugs. SeGa constructs a semantic knowledge base from product requirement documents, represented as a set of functionality entries that group related requirements under a common business intent. Given a focal method, SeGa retrieves the relevant functionality entries and derives fine-grained business scenarios with explicit preconditions, triggering actions, expected outcomes, and semantic constraints to guide LLM-based test generation. We evaluate SeGa on four industrial Go projects containing 60 real-world business logic bugs. SeGa detects 22-25 more bugs than four state-of-the-art LLM-based techniques and improves precision by 26.9%-34.3%. Deployment across 6 production repositories further uncovers 16 previously unknown business logic bugs that were confirmed and fixed by developers. From our industrial study, we summarize a series of lessons and suggestions for practical use and future research.

Uncertainty in large language model (LLM)-based systems is often studied at the level of a single model output, yet deployed LLM applications are compound systems in which uncertainty is transformed and reused across model internals, workflow stages, component boundaries, persistent state, and human or organisational processes. Without principled treatment of how uncertainty is carried and reused across these boundaries, early errors can propagate and compound in ways that are difficult to detect and govern. This paper develops a systems-level account of uncertainty propagation. It introduces a conceptual framing for characterising propagated uncertainty signals, presents a structured taxonomy spanning intra-model (P1), system-level (P2), and socio-technical (P3) propagation mechanisms, synthesises cross-cutting engineering insights, and identifies five open research challenges.

Accurate short-term electricity load forecasting is a cornerstone of U.S. grid reliability; however, prevailing deep learning models remain opaque, limiting operator trust during extreme weather. A unified, interpretable, physics-informed ensemble framework is proposed, integrating a Convolutional Neural Network (CNN) branch for local feature extraction and a Transformer branch for long-range dependency modeling; the branches are fused through a validation-optimized weighted ensemble and regularized by a physics-informed loss derived from the piecewise parabolic temperature-demand relationship of the Electric Reliability Council of Texas (ERCOT) system. Post-hoc interpretability is provided through SHapley Additive exPlanations (SHAP) with the DeepExplainer backend, yielding global and event-level attributions. Using eight years of ERCOT hourly load data (2018-2025) fused with Automated Surface Observing System (ASOS) records from three Texas stations, the framework achieves 713 MW MAE, 812 MW RMSE, and 1.18% MAPE on the test window. For Hampel-flagged extreme events, MAPE falls by 20.7% relative to its Transformer branch and by 40.5% relative to its CNN branch; an ablation confirms that the parabolic and ramp constraints drive a 14.7% RMSE reduction. SHAP analysis reveals a regime shift: temperature dominates under normal operation, whereas wind speed and precipitation become more influential during cold fronts and heatwaves.

Graph-based anti-money laundering (AML) systems on blockchain networks can score suspicious activity at two granularity levels -- transactions or actor addresses -- yet compliance action is conducted per actor. This paper contributes an evaluation methodology for measuring how scoring granularity affects investigation queue composition under fixed review budgets. We formalize the evaluation through a projection framework mapping transaction-level scores to the actor-level action unit via four aggregation operators, and introduce budgeted investigation metrics -- yield@budget, burden decomposition, and case fragmentation. Using the public Elliptic++ Bitcoin dataset (203,769 transactions; 822,942 address occurrences), we train independent random forest classifiers at each level under a causal temporal protocol and compare review queues through Jaccard overlap, burden decomposition, and feature-matching ablations. At one-percent budget, temporal evaluation yields mean Jaccard of 0.374 (SD 0.171); static pooled evaluation yields 0.087 (95% CI [0.079, 0.094]). An enriched address model receiving all 237 features produces even lower overlap (Jaccard=0.051), with 4.3% illicit per 100 reviews versus 30.2% for the transaction-projected queue. Address-level detection value is temporally concentrated: two timesteps exceed 91% illicit per 100 reviews while the static burden is only 3.4%. A fixed hybrid policy underperforms the best single-level queue by 5.05pp (CI [-10.2pp, -0.9pp]). These findings establish that scoring granularity is a consequential design variable for AML investigation systems -- same data, same budget, different queues, different addresses investigated.

Early detection of mental health conditions, particularly stress and depression, from social media text remains a challenging open problem in computational psychiatry and natural language processing. Automated systems must contend with figurative language, implicit emotional expression, and the high noise inherent in user-generated content. Existing approaches either leverage external commonsense knowledge to model mental states explicitly, or apply self-augmentation and contrastive training to improve generalization, but seldom do both in a principled, unified framework. We propose K-SENSE (Knowledge-guided Self-augmented Encoder for Neuro-Semantic Evaluation of Mental Health), a framework that jointly exploits external psychological reasoning and internal representation robustness. K-SENSE adopts a three-stage encoding pipeline: (1) inferential commonsense knowledge is extracted from the COMET model across five mental state dimensions; (2) a semantic anchor is constructed by combining hidden representations from two parallel encoding streams, projected into a shared space before fusion; and (3) a supervised contrastive learning objective aligns same-class representations while encouraging the attention mechanism to suppress irrelevant knowledge noise. We evaluate K-SENSE on Dreaddit (stress detection) and Depression_Mixed (depression detection), achieving mean F1-scores of 86.1 (0.6%) and 94.3 (0.8%), respectively, over five independent runs. These represent improvements of approximately 2.6 and 1.5 percentage points over the strongest prior baselines. Ablation experiments confirm the contribution of each architectural component, including the temporal knowledge integration strategy and the choice to keep the knowledge encoder frozen during fine-tuning.

Reward hacking in code generation, where models exploit evaluation loopholes to obtain full reward without correctly solving the tasks, poses a critical challenge for Reinforcement Learning (RL) and the deployment of reasoning models. Existing studies have been conducted primarily on synthetic hacking trajectories. However, whether these synthetic behaviors faithfully represent naturally emerging hacking in the wild remains unclear. In this work, we present a systematic analysis of the synthetic vs. in-the-wild discrepancy in reward hacking. We examine to what extent hacking behaviors induced by prompting resemble those emerging during RL training, and whether monitors trained on synthetic trajectories generalize to naturally arising but previously unseen hacking. To scale up the curation of in-the-wild reward hacking trajectories, we modified Group Relative Policy Optimization (GRPO) by injecting conflicting unit tests as tracers and applying a "resampling-until-hack" mechanism. Through controlled comparisons between monitors trained on synthetic versus in-the-wild data, we find that (1) synthetic-data-trained monitors fail to generalize to "in-the-wild" hacking, and (2) monitors trained on our "in-the-wild" trajectories demonstrate stronger generalizability to unseen hacking types. Our results indicate that synthetic reward hacking data may not fully reflect natural reward hacking behaviors, and that relying solely on synthetic data can lead to misleading conclusions. The codebase is available at https://github.com/LichenLillc/CoTMonitoring.git

Educational NLP systems are typically evaluated using engagement metrics and satisfaction surveys, which are at best a proxy for meeting pedagogical goals. We introduce six computational metrics for automated evaluation of pedagogical alignment in student-AI dialogue. We validate our metrics through analysis of 12,650 messages across 500 conversations from four courses. Using our metrics, we identify a fundamental misalignment: educators design conversational tutors for sustained learning dialogue, but students mainly use them for answer-extraction. Deployment context is the strongest predictor of usage patterns, outweighing student preference or system design: when AI tools are optional, usage concentrates around deadlines; when integrated into course structure, students ask for solutions to verbatim assignment questions. Whole-dialogue evaluation misses these turn-by-turn patterns. Our metrics will enable researchers building educational dialogue systems to measure whether they are achieving their pedagogical goals.

Multi-component natural language processing (NLP) pipelines are increasingly deployed for high-stakes decisions, yet no existing adversarial method can test their robustness under realistic conditions: binary-only feedback, no gradient access, and strict query budgets. We formalize this strict black-box threat model and propose a two-agent evasion framework operating in a semantic perturbation space. An Attacker Agent generates meaning-preserving rewrites while a Prompt Optimization Agent refines the attack strategy using only binary decision feedback within a 10-query budget. Evaluated against four evidence-based misinformation detection pipelines, the framework achieves evasion rates of 19.95 to 40.34% on modern large language model (LLM) based systems, compared to at most 3.90% for token-level perturbation baselines that rely on surrogate models because they cannot operate under our threat model. A legacy system relying on static lexical retrieval exhibits near-total vulnerability 97.02%, establishing a lower bound that exposes how architectural choices govern the attack surface. Evasion effectiveness is associated with three architectural properties: evidence retrieval mechanism, retrieval-inference coupling, and baseline classification accuracy. The iterative prompt optimization yields the largest marginal gains against the most robust targets, confirming that adaptive strategy discovery is essential when evasion is non-trivial. Analysis of successful rewrites reveals four exploitation patterns, each targeting failures at distinct pipeline stages. A pattern-informed defense reduces the evasion rate by up to 65.18%.

Deep learning-based nuclei segmentation and classification in pathology images typically rely on large-scale pixel-level manual annotations, which are costly and difficult to obtain across diverse tissues and staining conditions. To address this limitation, we propose a framework that leverages spatial transcriptomics (ST) data as supervision for nuclei segmentation and classification. By incorporating cell-level ST data, we obtain gene expression profiles and corresponding nuclear masks from histopathological images. Gene expression profiles are converted into cell-type labels and used as training data for image-based classification. Because existing gene expression-based cell-type classification methods are not designed for image recognition, we introduce an image-oriented classification approach that bridges gene expression-based cell typing and image-based cell classification. To evaluate generalization, we conduct segmentation experiments on previously unseen organs and compare our method with conventional supervised models. Despite being trained on fewer organ types, our framework achieves higher segmentation accuracy, demonstrating strong transferability. Classification experiments further show consistent improvements over existing approaches.

Large language models are increasingly deployed as automated judges for evaluating other models, yet the stability of their verdicts under semantically equivalent prompt paraphrases remains unmeasured. We introduce JudgeSense, a framework and benchmark for quantifying this property via the Judge Sensitivity Score (JSS), defined as the fraction of paraphrase pairs on which a judge returns an identical decision. Evaluating nine judge models on 494 validated paraphrase pairs, we find that coherence is the only task where judges meaningfully differ, with JSS ranging from 0.389 to 0.992. On factuality, all judges cluster near JSS about 0.63, driven by a polarity-inverted prompt artifact; after correction, factuality JSS rises to about 0.9. Pairwise tasks (preference and relevance) exhibit degenerate always-A behavior in 8 of 9 judges, indicating strong position bias. Model scale does not predict consistency. We release code, decision logs, and a validated paraphrase dataset to support standardized JSS reporting.

We study the organization of channel-level importance in transformer feed-forward networks (FFNs). Using a Fisher-style loss proxy (LP) based on activation-gradient second moments, we show that loss sensitivity is concentrated in a small set of channels within each layer. In Llama-3.1-8B, the top 1% of channels per layer accounts for a median of 58.7% of LP mass, with a range of 33.0% to 86.1%. We call these loss-critical channels supernodes. Although FFN layers also contain strong activation outliers, LP-defined supernodes overlap only weakly with activation-defined outliers and are not explained by activation power or weight norms alone. Around this core, we find a weaker but consistent halo structure: some non-supernode channels share the supernodes' write support and show stronger redundancy with the protected core. We use one-shot structured FFN pruning as a diagnostic test of this organization. At 50% FFN sparsity, baselines that prune many supernodes degrade sharply, whereas our SCAR variants explicitly protect the supernode core; the strongest variant, SCAR-Prot, reaches perplexity 54.8 compared with 989.2 for Wanda-channel. The LP-concentration pattern appears across Mistral-7B, Llama-2-7B, and Qwen2-7B, remains visible in targeted Llama-3.1-70B experiments, and increases during OLMo-2-7B pretraining. These results suggest that LLM FFNs develop a small learned core of loss-critical channels, and that preserving this core is important for reliable structured pruning.

Forecasting systems in science must be accurate, physically consistent, and certifiably reliable. Most existing models address prediction, constraint enforcement, and verification separately, limiting scalability and interpretability. We introduce GeoCert, a geometric AI framework that unifies forecasting, physical reasoning, and formal verification within a single differentiable computation. GeoCert formulates forecasting as evolution along a hyperbolic manifold, where negative curvature induces contraction dynamics, intrinsic robustness, and logarithmic-time certification. A hierarchical constraint architecture separates universal physical laws from domain-specific dynamics, enabling certified generalization across energy, climate, finance, and transportation systems. GeoCert achieves state-of-the-art accuracy while reducing computational cost by 97.5% and maintaining better certification rates. By embedding verification into the geometry of learning, GeoCert transforms forecasting from empirical approximation to formally verified inference, offering a scalable foundation for trustworthy, reproducible, and physically grounded scientific AI.

While recent autonomous agents demonstrate impressive capabilities, they predominantly rely on manually scripted workflows and handcrafted heuristics, inherently limiting their potential for open-ended improvement. To address this, we propose Escher-Loop, a fully closed-loop framework that operationalizes the mutual evolution of two distinct populations: Task Agents that solve concrete problems, and Optimizer Agents that recursively refine both the task agents and themselves. To sustain this self-referential evolution, we propose a dynamic benchmarking mechanism that seamlessly reuses the empirical scores of newly generated task agents as relative win-loss signals to update optimizers' scores. This mechanism leverages the evolution of task agents as an inherent signal to drive the evaluation and refinement of optimizers without additional overhead. Empirical evaluations on mathematical optimization problems demonstrate that Escher-Loop effectively pushes past the performance ceilings of static baselines, achieving the highest absolute peak performance across all evaluated tasks under matched compute. Remarkably, we observe that the optimizer agents dynamically adapt their strategies to match the shifting demands of high-performing task agents, which explains the system's continuous improvement and superior late-stage performance.

This study asks whether the threat of AI detection changes how people write with AI, and whether other people can tell the difference. In a two-phase controlled experiment, 21 participants wrote opinion pieces on remote work using an AI chatbot. Half were randomly warned that their submission would be scanned by an AI detection tool. The other half received no warning. Both groups had access to the same chatbot. In Phase 2, 251 independent judges evaluated 1,999 paired comparisons, each time choosing which document in the pair was written by a human. Judges were not told that both writers had access to AI. Across all evaluations, judges selected the warned writer's document as human 54.13% of the time versus 45.87% for the unwarned writer. A two-sided binomial test rejects chance guessing at p = 0.000243, and the result holds across both writing stances. Yet on every measurable text feature extracted, including AI overlap scores, lexical diversity, sentence structure, and pronoun usage, the two groups were indistinguishable. The judges are picking up on something that feature-based methods do not capture.

In 2016, Viazovska famously solved the sphere packing problem in dimension $8$, using modular forms to construct a 'magic' function satisfying optimality conditions determined by Cohn and Elkies in 2003. In March 2024, Hariharan and Viazovska launched a project to formalize this solution and related mathematical facts in the Lean Theorem Prover. A significant milestone was achieved in February 2026: the result was formally verified, with the final stages of the verification done by Math, Inc.'s autoformalization model 'Gauss'. We discuss the techniques used to achieve this milestone, reflect on the unique collaboration between humans and Gauss, and discuss project objectives that remain.

Large Language Models (LLMs) have achieved strong performance across natural language and multimodal tasks, yet their practical deployment remains constrained by inference latency and kernel launch overhead, particularly in interactive, short-sequence settings. This paper presents a hybrid runtime framework that combines Just-In-Time (JIT) compilation with CUDA Graph execution to reduce launch overhead while preserving runtime flexibility during autoregressive decoding. The framework partitions transformer inference into static components executed via CUDA Graph replay and dynamic components handled through JIT-compiled kernels, enabling asynchronous graph capture and reuse across decoding steps. We evaluate the proposed approach on LLaMA-2 7B using single-GPU, batch-size-one inference across prompt lengths from 10 to 500 tokens. Experimental results show that the hybrid runtime reduces Time-to-First-Token (TTFT) by up to 66.0% and achieves lower P99 latency compared with TensorRT-LLM in this regime. These results indicate that hybrid JIT-CUDA Graph execution can effectively reduce inference latency and variance for short-sequence LLM workloads, making it a practical optimization strategy for latency-sensitive AI applications.

NVIDIA's CUDA Tile (CuTile) introduces a Python-based, tile-centric abstraction for GPU kernel development that aims to simplify programming while retaining Tensor Core and Tensor Memory Accelerator (TMA) efficiency on modern GPUs. We present the first independent, cross-architecture evaluation of CuTile against established approaches such as cuBLAS, Triton, WMMA, and raw SIMT on three NVIDIA GPUs spanning Hopper and Blackwell: H100 NVL, B200, and RTX PRO 6000 Blackwell Server Edition. We benchmark representative AI workloads, including GEMM, fused multi-head attention, and end-to-end LLM inference in BF16/FP16 precision, to assess both performance and portability. Our results show that CuTile effectiveness is strongly workload- and architecture-dependent. On datacenter-class Blackwell (B200), CuTile achieves up to 1007 TFLOP/s for fused attention, outperforming FlashAttention-2 by 2.5x while requiring only 60 lines of Python kernel code. For GEMM, CuTile reaches 52-79% of cuBLAS performance in 22 lines of code (versus 123 for WMMA), making it a practical replacement for hand-written CUDA kernels but not yet for vendor-optimized libraries. However, the same CuTile attention kernel achieves only 53% of FlashAttention-2 throughput on RTX PRO 6000 (sm_120), exposing significant cross-architecture optimization gaps. In contrast, Triton sustains 62-101% of cuBLAS performance across all tested platforms without architecture-specific tuning, demonstrating substantially stronger portability.

Single-arm trials accelerate study timelines by reducing the number of patients that must be recruited for a concurrent control group. However, these designs require an alternative comparator to estimate treatment effects. One approach is to construct a virtual control arm using a machine learning (ML) model trained on external control data to predict the counterfactual outcomes of the treatment arm. Our aim in this study was to leverage virtual controls by developing and evaluating ML-based counterfactual outcome models trained on IFX-treated patients to predict 1-year steroid-free clinical remission (SFCR ) and a composite of C-reactive protein remission plus steroid-free clinical remission (CRP-SFCR) for ADA-treated pediatric Crohn's disease patients, and to compare the resulting IFX-versus-ADA treatment effect estimates with those obtained using propensity score matching to external controls. Five ML models were used to train counterfactual models on the observed IFX cohort data. The resulting models were used to predict the counterfactual outcomes for the ADA arm patients. LGBM yields the best OR closest to the propensity score matched reference, and all 95% CI results align with the conclusion from the reference study that no statistical difference in the primary and secondary outcomes has been observed between the patients treated with ADA or IFX. Our study supports virtual controls as a viable and effective substitute for expensive, lengthy or unethical patient recruitment in an inflammatory bowel disease (IBD) trial. The developed gradient boosted prediction model can be used as a pretrained model to generate IFX counterfactual predictions in future studies, pending external validation and assessment of transportability.

Chain-of-Thought (CoT) reasoning has emerged as a key technique for eliciting complex reasoning in Large Language Models (LLMs). Although interpretable, its dependence on natural language limits the model's expressive bandwidth. Continuous thought models address this bottleneck by reasoning in latent space rather than human-readable tokens. While they enable richer representations and faster inference, they raise a critical safety question: how can we detect misaligned reasoning in an uninterpretable latent space? To study this, we introduce MoralChain, a benchmark of 12,000 social scenarios with parallel moral/immoral reasoning paths. We train a continuous thought model with backdoor behavior using a novel dual-trigger paradigm - one trigger that arms misaligned latent reasoning ([T]) and another that releases harmful outputs ([O]). We demonstrate three findings: (1) continuous thought models can exhibit misaligned latent reasoning while producing aligned outputs, with aligned and misaligned reasoning occupying geometrically distinct regions of latent space; (2) linear probes trained on behaviorally-distinguishable conditions ([T][O] vs [O]) transfer to detecting armed-but-benign states ([T] vs baseline) with high accuracy; and (3) misalignment is encoded in early latent thinking tokens, suggesting safety monitoring for continuous thought models should target the "planning" phase of latent reasoning.

Multi-agent systems (MAS), composed of networks of two or more autonomous AI agents, have become increasingly popular in production deployments, yet introduce security risks that do not arise in single-agent settings. Even if individual agents exhibit robust security, architectural decisions governing their coordination can create attack surfaces that have not been systematically characterized. In this work, we present an empirical study of how MAS design decisions shape the tradeoff between task performance and attack resistance. Across three agentic environments (browser, desktop, and code) and 13 architectural configurations, we use stagewise evaluations that distinguish planning refusal, execution-stage interception, partial harmful execution, and successful attack completion to study three key design choices: (i) agent roles, which determine how authority and responsibility are allocated; (ii) communication topology, which shapes how and when agents interact; and (iii) memory, which determines the context and state visibility accessible to each agent. We find that multi-agent architectures are more vulnerable than standalone agents in the majority of configurations, with attack success rates varying by up to 3.8x at comparable or higher benign accuracy, and that no single design is universally safer. These results motivate the development of further evaluations that move beyond the security properties of a single agent.

The growing availability of online support groups has opened up new windows to study mental health through natural language processing (NLP). However, it is hindered by a lack of high-quality, well-validated datasets. Existing studies have a tendency to build task-specific corpora without collecting them into widely available resources, and this makes reproducibility as well as cross-task comparison challenging. In this paper, we present a uniform benchmark set of four Reddit-based datasets for disjoint but complementary tasks: (i) detection of suicidal ideation, (ii) binary general mental disorder detection, (iii) bipolar disorder detection, and (iv) multi-class mental disorder classification. All datasets were established upon diligent linguistic inspection, well-defined annotation guidelines, and human-judgmental verification. Inter-annotator agreement metrics always exceeded the baseline agreement score of 0.8, ensuring the labels' trustworthiness. Previous work's evidence of performance on both transformer and contextualized recurrent models demonstrates that these models receive excellent performances on tasks (F1 ~ 93-99%), further validating the usefulness of the datasets. By combining these resources, we establish a unifying foundation for reproducible mental health NLP studies with the ability to carry out cross-task benchmarking, multi-task learning, and fair model comparison. The presented benchmark suite provides the research community with an easy-to-access and varied resource for advancing computational approaches toward mental health research.

Automated failure diagnosis requires correlating browser-visible symptoms with backend observability signals, yet existing benchmarks do not evaluate this cross-modal reasoning task. Constructing one is non-trivial: multi-modal failure scenarios are costly to annotate, and live-environment capture introduces stochasticity that makes cross-run agent comparison unreliable. We present CUJBench, to our knowledge, the first benchmark to combine browser-visible failure evidence with backend observability in a diagnostic framing. CUJBench addresses annotation cost through an LLM-assisted generation pipeline with a multi-agent review loop and a three-layer annotation scheme, producing 87 labeled scenarios across five fault families, and ensures reproducibility by packaging each failure as a deterministic multi-modal snapshot with a fixed tool interface. Evaluating six frontier models under retrieval, browser-only, and full-toolset baselines, the benchmark yields an overall accuracy of 19.7% with a ceiling of 52%, well below saturation. Contrary to expectation, browser-only agents outperform full-toolset agents in aggregate, with expanded evidence access inducing unfocused exploration rather than improved synthesis. Trajectory analysis identifies cross-modal synthesis as the primary bottleneck: agents retrieve the decisive evidence but fail to attribute it correctly - a structural limitation uniform across all six models that model scale and richer tool access alone cannot resolve.

Vision Transformers trained only on image classification routinely transfer to tasks that demand spatial understanding, yet they receive no spatial supervision during pretraining. We ask where and how robustly such structure is encoded. Probing a frozen ViT-B/16 layerwise for two complementary properties, local patch boundaries (BSDS500) and per-patch depth (NYU Depth V2), reveals a clear hierarchy: boundary structure becomes linearly decodable at layers 5-6 (AP = 0.833), while depth, which requires integrating global cues, peaks two to three layers later at layer 8 (MAE = 0.0875). Both signals collapse at the final classification layer, and random-weight controls confirm the encodings are learned rather than architectural. Causal interventions add specificity: ablating the single direction a linear depth probe reads degrades depth decoding by up to 165%, while ablating any other direction changes it by less than 1%. Targeted activation patching along that direction shows the depth signal is partially re-derived at each layer rather than passively carried in the residual stream, with mid-layer interventions persisting most strongly downstream. The result is that a classification-trained ViT develops an actively maintained spatial hierarchy that mirrors the early-to-late progression observed in the primate visual cortex.

Argumentation is a core practice in STEM education, but its productivity depends on who participates and how they interact. Higher-achieving students often dominate the talk and decision-making, while lower-achieving peers may disengage, defer, or comply without contributing substantive reasoning. Forming groups strategically based on students' stances and argumentation skills could help foster inclusive, evidence-based discourse. In practice, however, teachers are constrained in implementing this grouping strategy because it requires real-time insight into students' positions and the quality of their argumentation, information that is difficult to assess reliably and at scale during instruction. We present a generative AI-powered system, ArguAgent, that creates groups optimizing for stance heterogeneity while constraining argumentation quality differences to +/-1 level on a validated learning progression. ArguAgent uses a two-component assessment pipeline: first scoring student arguments on a 0-4 rubric, then clustering positions via semantic analysis. We validated the scoring component against human expert consensus (Krippendorff's ααα = 0.817) using 200 expert-generated scores. Testing three OpenAI models (GPT-4o-mini, GPT-5.1, GPT-5.2) with identical calibrated prompts, we found that systematic prompt engineering informed by human disagreement analysis contributed 89% of scoring improvement (QWK: 0.531 to 0.686), while model upgrades contributed an additional 11% (QWK: 0.686 to 0.708). Simulation testing across 100 classes demonstrated that the grouping algorithm achieves 95.4% of groups that meet both design criteria, a 3.2x improvement over random assignment. These results suggest ArguAgent can enable real-time, theoretically grounded grouping that promotes productive STEM argumentation in classrooms.

Industrial maintenance environments increasingly rely on AI systems to assist operators in understanding asset behavior, diagnosing failures, and evaluating interventions. Although large language models (LLMs) enable fluent natural-language interaction, deployed maintenance assistants routinely produce generic explanations that are weakly grounded in telemetry, omit verifiable provenance, and offer no testable support for counterfactual or action-oriented reasoning that undermine trust in safety-critical settings. We present IndustryAssetEQA, a neurosymbolic operational intelligence system that combines episodic telemetry representations with a Failure Mode Effects Analysis Knowledge Graph (FMEA-KG) to enable Embodied Question Answering (EQA) over industrial assets. We evaluate on four datasets covering four industrial asset types, including rotating machinery, turbofan engines, hydraulic systems, and cyber-physical production systems. Compared to LLM-only baselines, IndustryAssetEQA improves structural validity by up to 0.51, counterfactual accuracy by up to 0.47, and explanation entailment by 0.64, while reducing severe expert-rated overclaims from 28% to 2% (approximately 93% reduction). Code, datasets, and the FMEA-KG are available at https://github.com/IBM/AssetOpsBench/tree/IndustryAssetEQA/IndustryAssetEQA.

Large language models are being deployed as mental health support agents at scale, yet only 16% of LLM-based chatbot interventions have undergone rigorous clinical efficacy testing, and simulations reveal psychological deterioration in over one-third of cases. We evaluate four generative models on 250 Prolonged Exposure (PE) therapy scenarios and 146 CBT cognitive restructuring exercises (plus 29 severity-escalated variants), scored by a three-judge LLM panel. All models scored near-perfectly on surface acknowledgment (~0.91-1.00) while therapeutic appropriateness collapsed to 0.22-0.33 at the highest severity for three of four models, with protocol fidelity reaching zero for two. Under CBT severity escalation, one model's task completeness dropped from 92% to 71% while the frontier model's safety-interference score fell from 0.99 to 0.61. We identify a systematic, modality-spanning failure: RLHF safety alignment disrupts the therapeutic mechanism of action by grounding patients during imaginal exposure, offering false reassurance, inserting crisis resources into controlled exercises, and refusing to challenge distorted cognitions mentioning self-harm in PE; and through task abandonment or safety-preamble insertion during CBT cognitive restructuring. These findings motivate a five-axis evaluation framework (protocol fidelity, hallucination risk, behavioral consistency, crisis safety, demographic robustness), mapped onto FDA SaMD and EU AI Act requirements. We argue that no AI mental health system should proceed to deployment without passing multi-axis evaluation across all five dimensions.

Stochastic sampling strategies are widely adopted in large language models (LLMs) to balance output coherence and diversity. These heuristics are often inherited in Multimodal LLMs (MLLMs) without task-specific justification. However, we contend that stochastic decoding can be suboptimal for Visual Question Answering (VQA). VQA is a closed-ended task with head-heavy answer distributions where uncertainty is usually epistemic, arising from missing or ambiguous visual evidence rather than plausible continuations. In this work, we provide a theoretical formalization of the relationship between model calibration and predictive accuracy, and derive the sufficient conditions for greedy decoding optimality. Extensive experiments provide empirical evidence for the superiority of greedy decoding over stochastic sampling across multiple benchmarks. Furthermore, we propose Greedy Decoding for Reasoning Models, which outperforms both stochastic sampling and standard greedy decoding in multimodal reasoning scenarios. Overall, our results caution against naively inheriting LLMs decoding heuristics in MLLMs and demonstrate that greedy decoding can be an efficient yet strong default for VQA.

The global financial ecosystem confronts a critical asymmetry: while fraud syndicates operate as borderless, distributed networks, banking institutions remain constrained by regulatory data silos, limiting visibility into cross-institutional threat patterns under strict privacy laws such as GDPR. Although Federated Learning (FL) enables collaborative training, existing protocols impose a trade-off among scalability, privacy, and integrity. Homomorphic encryption schemes are computationally expensive, while pairwise masking protocols require O(N^2) key exchanges and lack mechanisms to detect malformed updates. Existing defenses also remain vulnerable to gradient inversion attacks that can reconstruct sensitive transaction data. To address these limitations, we propose Dynamic Sharded Federated Learning (DSFL), a verifiable secure aggregation framework for cross-institution financial fraud detection. DSFL replaces mesh topologies with Dynamic Stochastic Sharding, reducing communication complexity from O(N^2) to O(N m), where m is a fixed shard size, achieving linear scalability. To mitigate insider threats, we introduce Linear Integrity Tags, an additive-homomorphic commitment mechanism that enables probabilistic verification of submitted updates without the overhead of zero-knowledge proofs, while not enforcing semantic correctness. Additionally, the Active Neighborhood Recovery protocol ensures robust aggregation under participant dropouts. Empirical evaluation on the Credit Card Fraud Detection Dataset (ULB) demonstrates an approximately 33x latency reduction compared to Paillier-based secure aggregation, while maintaining strong resilience under simulated failures. These results position DSFL as a practical foundation for scalable and privacy-preserving collaborative fraud detection.

Reliable decision-making with streaming data requires principled uncertainty quantification of online methods. While first-order methods enable efficient iterate updates, their inference procedures still require updating proper (covariance) matrices, incurring $O(d^2)$ time and memory complexity, and are sensitive to ill-conditioning and noise heterogeneity of the problem. This costly inference task offers an opportunity for more robust second-order methods, which are, however, bottlenecked by solving Newton systems with $O(d^3)$ complexity. In this paper, we address this gap by studying an online Newton method with Hessian averaging, where the Newton direction at each step is approximately computed using a sketch-and-project solver with Nesterov's acceleration, matching $O(d^2)$ complexity of first-order methods. For the proposed method, we quantify its uncertainty arising from both random data and randomized computation. Under standard smoothness and moment conditions, we establish global almost-sure convergence, prove asymptotic normality of the last iterate with a limiting covariance characterized by a Lyapunov equation, and develop a fully online covariance estimator with non-asymptotic convergence guarantees. We also connect the resulting uncertainty quantification to that of exact and sketched Newton methods without Nesterov's acceleration. Extensive experiments on regression models demonstrate the superiority of the proposed method for online inference.

Radiographic grading of knee osteoarthritis (KOA) with the Kellgren-Lawrence (KL) system is limited by inter-reader variability and the opacity of current deep learning approaches, which predict KL grades directly from images without decomposing structural features. We present Knee-xRAI, a modular framework that independently quantifies the three cardinal radiographic features of KOA (joint space narrowing [JSN], osteophytes, and subchondral sclerosis) and integrates them into an explainable KL grade classification. The pipeline combines U-Net++ segmentation for contour-based JSN measurement, an SE-ResNet-50 network for per-site osteophyte grading (OARSI scale), and a hybrid texture-CNN classifier for binary sclerosis quantification. The resulting 50-dimensional structured feature vector feeds two complementary classification paths. An XGBoost path supports SHAP-based feature attribution. A ConvNeXt hybrid path combines the structured vector with a full-image encoder for enhanced predictive performance. Evaluated on 8,260 radiographs from an OAI-derived dataset, the JSN module achieved a Dice coefficient of 0.8909 and an mJSW intraclass correlation of 0.8674 against manual annotations. The ConvNeXt hybrid path reached a test quadratic weighted kappa (QWK) of 0.8436 and AUC of 0.9017. The transparent XGBoost path achieved a test QWK of 0.6294 with full feature-level audit capability. Ablation confirmed JSN as the dominant predictor (QWK = 0.6103 alone), with osteophyte features providing consistent incremental gain (+0.0183) and sclerosis contributing marginally. Inference-time ablation of Path B confirmed the structured pathway contributes materially beyond the image encoder, with QWK drops of 0.098 (feature zeroing) and 0.284 (feature-image permutation). Knee-xRAI explicitly quantifies all three KL-defining radiographic features within a single auditable pipeline.

Dynamic Tanh (DyT) removes LayerNorm by bounding activations with a learned tanh(alpha x). We show that this bounding is a regime-dependent implicit regularizer, not a uniformly beneficial replacement. Across GPT-2-family models spanning 64M to 3.78B parameters and 1M to 118M tokens, with Llama and ViT cross-checks, DyT improves validation loss by 27.3% at 64M/1M but worsens it by 18.8% at 64M/118M; the 1M benefit vanishes with capacity (+1.7% at 3.78B), while the 118M penalty reaches +27.9%. The mechanism is measurable: 49% of DyT activations saturate at 1M versus 23% at 118M, and a 500-step saturation heuristic classifies DyT's sign with 75% raw in-sample accuracy on the 12-cell GPT-2 calibration set (AUC 0.75; 64% when adding Scale 5 stress cells), correctly labels 3/3 Llama checks, but only reaches 50% raw leave-one-scale-out accuracy. Three interventions support the bounding explanation: HardTanh reproduces the regime pattern, increasing alpha at 118M monotonically reduces DyT's penalty, and vanilla+dropout(p=0.5) matches DyT's data-rich loss. We also localize Llama-DyT collapse to SwiGLU gating, where saturation separates collapse from convergence in a 3-seed component ablation (r=0.94). Scope: all experiments are compute-limited (T/P < 1.84), below Chinchilla-optimal training.

Reliable depth estimation from spherical images is crucial for 360° vision in robotic navigation and immersive scene understanding. However, the onboard spherical camera can experience unintentional pose variations in real-world robotic platforms that, along with the geometric distortions inherent in equirectangular projections, significantly impact the effectiveness of depth estimation. To study this issue, a novel public benchmark, called Sphere-Depth, is introduced to systematically evaluate the robustness of monocular depth estimation models from equirectangular images in a reproducible way. Camera pose perturbations are simulated and used to assess the performance of a popular perspective-based model, Depth Anything, and of spherical-aware models such as Depth Anywhere, ACDNet, Bifuse++, and SliceNet. Furthermore, to ensure meaningful evaluation across models, a depth calibration-based error protocol is proposed to convert predicted relative depth values into metric depth values using supervised learned scaling factors for each model. Experiments show that even models explicitly designed to process spherical images exhibit substantial performance degradation when variations in the camera pose are observed with respect to the canonical pose. The full benchmark, evaluation protocol, and dataset splits are made publicly available at: https://github.com/sgazzeh/Sphere_depth

The relentless expansion of scientific literature presents significant challenges for navigation and knowledge discovery. Within Research Information Retrieval, established tasks such as text summarization and classification remain crucial for enabling researchers and practitioners to effectively navigate this vast landscape, so that efforts have increasingly been focused on developing advanced research information systems. These systems aim not only to provide standard keyword-based search functionalities but also to incorporate capabilities for automatic content categorization within knowledge-intensive organizations across academia and industry. This study systematically evaluates the performance of off-the-shelf Large Language Models (LLMs) in analyzing scientific texts according to a given classification scheme. We utilized the hierarchical ORKG taxonomy as a classification framework, employing the FORC dataset as ground truth. We investigated the effectiveness of advanced prompt engineering strategies, namely In-Context Learning (ICL) and Prompt Chaining, and experimentally explored the influence of the LLMs' temperature hyperparameter on classification accuracy. Our experiments demonstrate that Prompt Chaining yields superior classification accuracy compared to pure ICL, particularly when applied to the nested structure of the ORKG taxonomy. LLMs with prompt chaining outperform the state-of-the-art models for domain (1st level) prediction and show even better performance for subject (2nd level) prediction compared to the older BERT model. However, LLMs are not yet able to perform well in classifying the topic (3rd level) of research areas based on this specific hierarchical taxonomy, as they only reach about 50% accuracy even with prompt chaining.

We prove lower bounds on learning the Möbius or Liouville function with a variety of standard learning techniques, including kernel methods, noisy gradient methods, and correlational statistical query algorithms. These results follow from quantitative bounds on the correlation of Möbius with digital characters of various finite abelian groups, where the group is dictated by the type of input data the algorithm is given. Using residues mod $p$ for many different primes corresponds to a cyclic group, and using the base $p$ expansion for a fixed prime corresponds to an elementary abelian $p$-group. We also note that lower bounds of this form are closely related to certain types of digital prime number theorems.

Federated learning (FL) is a distributed machine learning method where multiple devices collaboratively train a model under the management of a central server without sharing underlying data. One of the key challenges of FL is the communication bottleneck caused by variations in connection speed and bandwidth across devices. Therefore, it is essential to reduce the size of transmitted data during training. Additionally, there is a potential risk of exposing sensitive information through the model or gradient analysis during training. To address both privacy and communication efficiency, we combine differential privacy (DP) and adaptive quantization methods. We use Laplacian-based DP to preserve privacy, which is relatively underexplored in FL and offers tighter privacy guarantees than Gaussian-based DP. We propose a simple and efficient global bit-length scheduler using round-based cosine annealing, along with a client-based scheduler that dynamically adapts based on client contribution estimated through dataset entropy analysis. We evaluate our approach through extensive experiments on CIFAR10, MNIST, and medical imaging datasets, using non-IID data distributions across varying client counts, bit-length schedulers, and privacy budgets. The results show that our adaptive quantization methods reduce total communicated data by up to 52.64% for MNIST, 45.06% for CIFAR10, and 31% to 37% for medical imaging datasets compared to 32-bit float training while maintaining competitive model accuracy and ensuring robust privacy through differential privacy.

Evolve pairs a small local language model with a persistent, teacher-compiled knowledge store -- refined through sleep consolidation and usage-driven refresh -- to deliver substantial accuracy gains over the model's parametric baseline while amortizing teacher costs through cross-query knowledge reuse. Rather than retrieving document fragments at query time, Evolve constructs a store of semantically coherent sections compiled by teacher models at natural conceptual boundaries; new sections are staged on acquisition, consolidated offline through teacher-mediated merging, and refreshed inline when expired. A 2B-parameter local model handles classification and generation; large teacher models are invoked only for knowledge operations. Across 750 benchmark queries spanning custom specialist questions, NaturalQuestions, and TriviaQA, the 2B model augmented by Evolve improves from 20-33% baseline accuracy to 60-84% (+40-52pp) while reducing teacher invocations by over 50% through reuse. Post-consolidation compresses the knowledge store by 31-33.5% across three independent benchmarks while preserving accuracy; section-based retrieval outperforms chunk-based retrieval by 5-9pp across every lifecycle condition. The architecture supports two generation modes over the same lifecycle -- suppress (strict section-only grounding, auditable) and augment (section-supplemented responses).

Uniform random rotations are a useful primitive in applications such as fast Johnson-Lindenstrauss embeddings, kernel approximation, communication-efficient learning, and recent AI compression pipelines, but they are computationally expensive to generate and apply in high dimensions. A common practical replacement is repeated structured random rotations built from Walsh-Hadamard transforms and random sign diagonals. Applying the structured random rotation twice has been shown empirically to be useful, but the supporting theory is still limited. In this paper we study the approximation quality achieved when using this two-block structured Hadamard rotation. Our results are both positive and negative. On the positive side, we prove that every fixed coordinate of the two-block transform converges uniformly, over all inputs, to the corresponding coordinate of a uniformly rotated vector, with an explicit Kolmogorov-distance bound of order $d^{-1/5}$. On the negative side, we prove an explicit lower bound on the Wasserstein distance between the full vector distributions, showing that the two-block transform is not a globally accurate surrogate for a uniform random rotation in the worst case. For the extremal input used in the lower bound, we also prove a matching asymptotic upper bound, showing that the lower-bound scale is sharp for that input. Taken together, the results identify a clear separation between one-dimensional marginal behavior, where approximation improves with dimension, and full high-dimensional geometry, where a nonvanishing discrepancy remains. This provides a partial theoretical explanation for the empirical success of structured Hadamard rotations in some algorithms, while also clarifying the limitations of treating them as drop-in replacements for true uniform random rotations.

Cloud-hosted Large Language Models (LLMs) offer unmatched reasoning capabilities and dynamic knowledge, yet submitting raw queries to these external services risks exposing sensitive user intent. Conversely, relying exclusively on trusted local models preserves privacy but often compromises answer quality due to limited parameter scale and knowledge. To resolve this dilemma, we propose Game-theoretic Trustworthy Knowledge Acquisition (GTKA), a framework that formulates the trade-off between knowledge utility and privacy as a strategic game. GTKA consists of three components: (i) a privacy-aware sub-query generator that decomposes sensitive intent into generalized, low-risk fragments; (ii) an adversarial reconstruction attacker that attempts to infer the original query from these fragments, providing adaptive leakage signals; and (iii) a trusted local integrator that synthesizes external responses within a secure boundary. By training the generator and attacker in an alternating adversarial manner, GTKA optimizes the sub-query generation policy to maximize knowledge acquisition accuracy while minimizing the reconstructability of the original sensitive intent. To validate our approach, we construct two sensitive-domain benchmarks in the biomedical and legal fields. Extensive experiments demonstrate that GTKA significantly reduces intent leakage compared to state-of-the-art baselines while maintaining high-fidelity answer quality.

While annotated corpora are crucial in the field of natural language processing (NLP), those containing copyrighted material are difficult to exchange among researchers. Yet, such corpora are necessary to fully represent the diversity of data found in the wild in the context of NLP tasks. We tackle this issue by proposing a method to lawfully and publicly share the annotations of copyrighted literary texts. The corpus creator shares the annotations in clear, along with a non-reversible hashed version of the source material. The corpus user must own the source material, and apply the same hash function to their own tokens, in order to match them to the shared annotations. Crucially, our method is robust to reasonable divergences in the version of the copyrighted data owned by the user. As an illustration, we present alignment experiments on different editions of novels. Our results show that our method is able to correctly align 98.7 to 99.79% of tokens depending on the novel, provided the user version is sufficiently close to the corpus creator's version. We publicly release novelshare, a Python implementation of our method.

Large vision-language models (VLMs) can recognize \textit{what} happens in video but fail to count \textit{how many} times. We introduce \textbf{PushupBench}, 446 long-form clips (avg. 36.7s) for evaluating repetition counting. The best frontier model achieves 42.1\% exact accuracy; open-source 4B models score $\sim$6\%, matching supervised baselines. We show that accuracy alone misleads -- weaker models exploit the modal count rather than reason temporally. Fine-tuning on counting with 1k samples transfers to general video understanding: MVBench (+2.15), PerceptionTest (+1.88), TVBench (+4.54), suggesting counting is a proxy for broader temporal reasoning.PushupBench incorporated in \texttt{lmms-eval} (https://github.com/EvolvingLMMs-Lab/lmms-eval/pull/1262) and hosted on (pushupbench.com/)

This paper proposes a new method to improve the training efficiency of deep convolutional neural networks. During training, the method evaluates scores to measure how much each layer's parameters change and whether the layer will continue learning or not. Based on these scores, the network is scaled down such that the number of parameters to be learned is reduced, yielding a speed up in training. Unlike state-of-the-art methods that try to compress the network to be used in the inference phase or to limit the number of operations performed in the backpropagation phase, the proposed method is novel in that it focuses on reducing the number of operations performed by the network in the forward propagation during training. The proposed training strategy has been validated on two widely used architecture families: VGG and ResNet. Experiments on MNIST, CIFAR-10 and Imagenette show that, with the proposed method, the training time of the models is more than halved without significantly impacting accuracy. The FLOPs reduction in the forward propagation during training ranges from 17.83\% for VGG-11 to 83.74\% for ResNet-152. These results demonstrate the effectiveness of the proposed technique in speeding up learning of CNNs. The technique will be especially useful in applications where fine-tuning or online training of convolutional models is required, for instance because data arrive sequentially.

We report a reproducible error pattern in GPT-5.4 on OWL~2~DL compliance queries: the model frequently answers ``unknown'' when the reasoner-entailed answer is ``no'' under \emph{FunctionalProperty} closure or class \emph{disjointness}. Using 180 reasoner-audited queries from a procedural expansion of the observed pattern plus 18 hand-authored held-out queries in two unrelated domains (insurance and clinical), we compare four interaction modes under matched query budget: single-shot, three rounds of generic ``you-are-wrong'' retry, three rounds of reasoner-verdict repair with an open-world-assumption (OWA) hint, and the same repair without the hint. Direct faithfulness is 43.9\,\% (Wilson 95\,\% CI $[36.8,51.2]$); generic retry reaches 81.7\,\% ($[75.4,86.6]$); the verdict-with-hint variant is \emph{worse} at 67.2\,\% ($[60.1,73.7]$); the verdict-only variant reaches 97.8\,\% ($[94.4,99.1]$). All pairwise comparisons remain significant under McNemar's exact test with Bonferroni correction ($α= 0.01$; all $p < 10^{-5}$). The same fingerprint accounts for 4/4 errors on the held-out queries. Our interpretation is bounded: prompt framing can matter more than corrective content, and reasoner-guided wrappers should be ablated explicitly.

Generative retrieval (GR) ranks documents by autoregressively generating document identifiers. Because many GR methods rely on trie-constrained beam search, they are vulnerable to early pruning of relevant prefixes under finite-beam decoding. Planning Ahead in Generative Retrieval (PAG) mitigates this failure mode by using simultaneous decoding to compute a document-level look-ahead prior that guides subsequent sequential decoding. We reproduce PAG at inference time and stress-test its decoding behavior. Using the authors' released checkpoint and identifier/trie artifacts under the reported decoding setup, we reproduce the main effectiveness results on MS MARCO Dev and TREC-DL 2019/2020, and corroborate the reported beam-size-latency trade-off in our hardware setting. Beyond reproduction, we introduce plan drift diagnostics that quantify how intent-preserving query variations alter the planner's top-n candidate set and highest-weight planner tokens, and how these changes affect guided decoding. We find that PAG's planning signal is brittle under lexical surface-form variation: intent-preserving typos can trigger plan collapse, where the planned candidate pool shifts enough that the look-ahead bonus provides little useful guidance, effectively reverting decoding toward weaker unguided search. We further evaluate fixed-index cross-lingual robustness using non-English mMARCO queries against an English index, and assess query-side mitigation strategies that require no re-indexing; query translation provides the strongest recovery in our setting. Overall, our results confirm PAG's reported effectiveness and the benefit of planning-guided decoding under the released inference setup, while showing that these gains depend on the stability of the planning signal under realistic query variation and query-document mismatch.

Refereeing is vital in sports, where fair, accurate, and explainable decisions are fundamental. While intelligent assistant technologies are being widely adopted in soccer refereeing, current AI-assisted approaches remain preliminary. Existing research mostly focuses on isolated video perception tasks and lacks the ability to understand and reason about foul scenarios. To fill this gap, we propose SoccerRef-Agents, a holistic and explainable multi-agent decision-making framework for soccer refereeing. The main contributions are: (i) constructing the multimodal benchmark SoccerRefBench with over 1,200 referee theory questions and 600 foul video clips; (ii) building a vector-based knowledge base RefKnowledgeDB using the latest "Laws of the Game" and a classic case database for precise, knowledge-driven reasoning; (iii) designing a novel multi-agent architecture that collaborates via cross-modal RAG to bridge the semantic gap between visual content and regulatory texts. This work explores the technical capability of integrating MLLMs with refereeing expertise, and evaluations show our system significantly outperforms general-purpose MLLMs in decision accuracy and explanation quality. All databases, benchmarks, and code will be made available.

Generative information retrieval (GenIR) consolidates retrieval into a single neural model that decodes document identifiers (docids) directly from queries. While this model-as-index paradigm offers architectural simplicity, it is poorly suited to dynamic document collections. Unlike modular systems, where indexes are easily updated, GenIR's knowledge is parametrically encoded in its weights; consequently, standard adaptation methods such as full and parameter-efficient fine-tuning can induce catastrophic forgetting. We show that sequential adaptation improves retrieval on newly added documents but substantially degrades performance on earlier slices, exposing a pronounced stability-plasticity trade-off. To address this, we propose post-adaptation memory tuning (PAMT), a memory-only stabilization stage that augments an adapted model with a modular parametric memory head (PMH). PAMT freezes the backbone and attaches a product-key memory with fixed addressing. During prefix-trie constrained decoding, decoder hidden states sparsely query PMH to produce residual corrections in hidden space; these corrections are mapped to score adjustments via the frozen output embedding matrix, computed only over trie-valid tokens. This guides docid generation while keeping routing and backbone parameters fixed. To limit cross-slice interference, PAMT updates only a fixed budget of memory values selected using decoding-time access statistics, prioritizing entries frequently activated by the current slice and rarely used in prior sessions. Experiments on MS MARCO and Natural Questions under sequential, disjoint corpus increments show that PAMT substantially improves retention on earlier slices with minimal impact on retrieval performance for newly added documents, while modifying only a sparse subset of memory values per session.

Federated Learning (FL) typically assumes unconditional collaboration, a premise that overlooks the complexities of real-world, multi-stakeholder environments in which clients may need to exclude one another for strategic, regulatory, or competitive reasons. This paper addresses this gap, which we term 'client-level disagreements,' by first introducing a taxonomy of such scenarios. We then propose a robust, multi-track resolution strategy that guarantees strict client exclusion by creating and managing isolated model update paths ('tracks'), thereby preventing the cross-contamination and unfairness issues present in naive strategies. Through an empirical evaluation of our custom simulation system across 34 scenarios using the MNIST and N-CMAPSS datasets, we validate that our approach correctly handles permanent, temporal, and overlapping disagreement patterns. Our scalability analysis reveals the server-side resolution algorithm's overhead is negligible (<1 ms per round) even under heavy load. The primary scalability constraint is the client-side training load from participating in multiple tracks, a cost that we show can be effectively mitigated by a submodel reuse strategy. This work presents a scalable and architecturally sound method for managing client-level disagreements, and enhances the practical applicability of FL in settings where policy compliance and strategic control are non-negotiable.

Transthoracic echocardiography is the reference standard for confirming structural heart disease (SHD), but first-line screening is limited by cost, workflow burden, and specialist availability. We evaluated whether open pretrained electrocardiogram (ECG) foundation models can support echo-confirmed multi-label SHD detection using the public EchoNext Mini-Model benchmark. Six echocardiography-derived abnormalities were targeted: reduced left ventricular ejection fraction, increased left ventricular wall thickness, aortic stenosis, mitral regurgitation, tricuspid regurgitation, and right ventricular systolic dysfunction. Under a common pipeline, we compared engineered ECG features with gradient boosting, end-to-end waveform learning from scratch, and transfer from open ECG foundation models. We then applied in-domain self-supervised adaptation of an ECG foundation model (ECG-FM) on EchoNext waveforms followed by selective supervised fine-tuning, and evaluated trade-offs between discrimination and adaptation cost. Adapted ECG-FM models achieved the best overall performance: peak macro-AUROC 0.8509 and macro-AUPRC 0.4297, while a parameter-efficient operating point preserved AUROC (0.8501) and attained the highest fixed-threshold macro-F1 0.3691. Late fusion with covariates did not improve threshold-independent discrimination, and evaluated LoRA, alternative backbones, and mixture-of-foundations strategies did not surpass the best adapted single-backbone models. These results indicate that for ECG-based case finding and echocardiography triage, combining target-domain self-supervised adaptation with selective supervised updating of a pretrained ECG backbone is the most effective transfer strategy.

Aligning denoising generative models with human preferences or verifiable rewards remains a key challenge. While policy-gradient online reinforcement learning (RL) offers a principled post-training framework, its direct application is hindered by the intractable likelihoods of these models. Prior work therefore either optimizes an induced Markov decision process (MDP) over sampling trajectories, which is stable but inefficient, or uses likelihood surrogates based on the diffusion evidence lower bound (ELBO), which have so far underperformed on visual generation. Our key insight is that the ELBO-based approach can, in fact, be made both stable and efficient. By reducing surrogate variance and controlling gradient steps, we show that this approach can beat MDP-based methods. To this end, we introduce Variational GRPO (V-GRPO), a method that integrates ELBO-based surrogates with the Group Relative Policy Optimization (GRPO) algorithm, alongside a set of simple yet essential techniques. Our method is easy to implement, aligns with pretraining objectives, and avoids the limitations of MDP-based methods. V-GRPO achieves state-of-the-art performance in text-to-image synthesis, while delivering a $2\times$ speedup over MixGRPO and a $3\times$ speedup over DiffusionNFT.

Neurosymbolic systems can satisfy logical constraints during learning without achieving the intended concept-label correspondence; this is a problem known as reasoning shortcuts. We formalize reasoning shortcuts as a constraint satisfaction problem and investigate under which conditions concept mappings are uniquely determined by the constraints. We prove that a discrimination property (requiring that no valid concept mapping can be transformed into another valid mapping by swapping two concept values) is necessary for shortcut-freeness under bijective mappings, but demonstrate via a counterexample that it is insufficient even when the constraint graph is connected. We develop an ASP-based algorithm that verifies whether a given constraint set uniquely determines the intended concept mapping, with proven soundness and completeness. When shortcuts are detected, a greedy repair algorithm eliminates them by augmenting the constraint set, converging in at most $k$ iterations, where $k$ is the number of alternative valid mappings. We further provide a complexity classification: deciding shortcut-freeness is coNP-complete, counting shortcuts is #P-complete, and finding minimal repairs is NP-hard. We also establish sample complexity bounds showing that logarithmically many label queries suffice for disambiguation in favorable cases, while querying all ambiguous positions suffices in the worst case. Experiments across eight benchmark domains validate our approach.

Reconstructing high-fidelity fluid dynamics from sparse temporal observations is quite challenging, mainly due to the chaotic and non-linear nature of fluid transport. Standard deep learning-based interpolation methods often tend to regress to the mean, which results in spatial blurring and temporal strobing, especially noticeable around the observed anchor frames where transitions become discontinuous. In this work, we propose a novel Temporal U-Net architecture that integrates a VGG-based perceptual loss along with a Physics-Informed Bridge to overcome these issues. By introducing time-weighted feature blending and enforcing a parabolic boundary condition defined by t(1 - t), the model ensures smooth transitions while also maintaining perfect consistency at the endpoints. Experimental results on multi-channel RGB fluid data show that our method clearly outperforms standard models, both in terms of structural fidelity and texture preservation. In particular, the model achieves a Mean Absolute Error of 0.015, compared to 0.085 for a standard L1 baseline. Further Spatial Power Spectral Density (PSD) analysis reveals that the model is able to retain high-frequency turbulent details that are usually lost in deterministic reconstructions.

This paper investigates context stickiness in in-context learning (ICL), a phenomenon where earlier examples in a prompt interfere with a transformer's ability to adapt to later tasks. Using synthetic regression tasks over linear and quadratic functions, we examine how models trained under sequential, mixed, and random curricula handle abrupt task switches during inference. By sweeping over structured combinations of misleading linear examples followed by recovery quadratic examples, we quantify how prior context biases prediction error and how quickly models realign. Our results show strong evidence of persistent interference: more preceding linear examples reliably degrade quadratic predictions, while additional quadratic examples reduce error but with diminishing returns. We further find that training curricula significantly modulate resilience, with sequential training on the target function class yielding the fastest recovery, and surprisingly, random training producing the least robust behavior.

Solutions to the Schrödinger bridge problem and its generalizations yield feedback control policies for optimal density steering over a controlled diffusion. To numerically compute the same, the dynamic Sinkhorn recursion has become a standard approach. The mathematical engine behind this approach is the Hopf-Cole transform that recasts the conditions for optimality into a system of boundary-coupled linear PDEs. Recent works pointed out that for the control-affine Schrödinger bridge problem, this exact linearity via Hopf-Cole transform, and thus the standard Sinkhorn recursion, apply only if the control and noise channels are proportional. When the channels do not match, the Hopf-Cole-transformed PDEs remain nonlinear, and no algorithm is available to solve the same. We advance the state-of-the-art by designing a Sinkhorn recursion with memory that leverages the structure of these nonlinear PDEs, and demonstrate how it solves the control-affine Schrödinger bridge problem with input and noise channel mismatch. We prove the local stability of the proposed algorithm.

Event-Based Models (EBMs) infer biomarker progression from cross-sectional data but typically only as ordinal sequences and rely on rigid model assumptions. We propose \textsc{Tempo}, a Transformer architecture that learns both ordinal and continuous event sequences through simulation-based supervised learning. \textsc{Tempo} uses two Transformer modules: one treats biomarkers as tokens to infer event sequencing; the other treats patients as tokens, representing each by their per-biomarker abnormality profile, to infer patients' disease stages. On synthetic benchmarks, \textsc{Tempo} reduces normalized Kendall's Tau distance by 52.89\% and staging MAE by 25.33\% compared to state-of-the-art SA-EBM, with larger reductions in high-dimensional settings (58.88\% and 61.10\%). Applied to ADNI, \textsc{Tempo} recovers a biologically plausible Alzheimer's progression: early medial temporal atrophy, followed by amyloid accumulation and cognitive decline, and late-stage tau pathology with terminal acceleration of global neurodegeneration -- broadly consistent with established disease models. \textsc{Tempo} also eliminates the need to derive custom inference algorithms and enables rapid empirical comparison of generative hypotheses.

Autonomous multi-agent LLM systems are increasingly deployed to investigate operational incidents and produce structured diagnostic reports. Their trustworthiness hinges on whether each claim is grounded in observed evidence rather than model-internal inference. Existing groundedness evaluators (binary classifiers, LLM-as-judge scalars, self-correction loops) treat supporting evidence as interchangeable and emit a single signal that offers no principled control over downstream action. We present GSAR, a grounding-evaluation and replanning framework that (i) partitions claims into a four-way typology (grounded, ungrounded, contradicted, complementary), giving first-class standing to non-redundant alternative perspectives; (ii) assigns evidence-type-specific weights reflecting epistemic strength; (iii) computes an asymmetric contradiction-penalised weighted groundedness score; and (iv) couples that score to a three-tier decision function (proceed, regenerate, replan) driving a bounded-iteration outer loop under an explicit compute budget. We formalise the algorithm, prove six structural properties, and evaluate five design claims on FEVER with gold Wikipedia evidence under four independently-trained LLM judges (gpt-5.4, claude-sonnet-4-6, claude-opus-4-7, gemini-2.5-pro). Every ablation reproduces in the same direction on every judge: bootstrap 95% CIs on the rho=0 effect exclude 0 on all four; the no-complementary ablation under Opus 4.7 has CI [-96,-68] of 200; at n=1000 three independent judges converge to DeltaS(rho=0)=+0.058. A head-to-head against Vectara HHEM-2.1-Open is included. To our knowledge, GSAR is the first published groundedness framework coupling evidence-typed scoring with tiered recovery under an explicit compute budget.

Adversarial attacks play a pivotal role in testing and improving the reliability of deep learning (DL) systems. Existing literature has demonstrated that subtle perturbations to the input can elicit erroneous outcomes, thereby substantially compromising the security of DL systems. This has emerged as a critical concern in the development of DL-based safety-critical systems like Autonomous Driving Systems (ADSs). The focus of existing adversarial attack methods on End-to-End (E2E) ADSs has predominantly centered on misbehaviors of steering angle, which overlooks speed-related controls or imperceptible perturbations. To address these challenges, we introduce UniAda, a multi-objective white-box attack technique with a core function that revolves around crafting an image-agnostic adversarial perturbation capable of simultaneously influencing both steering and speed controls. UniAda capitalizes on an intricately designed multi-objective optimization function with the Adaptive Weighting Scheme (AWS), enabling the concurrent optimization of diverse objectives. Validated with both simulated and real-world driving data, UniAda outperforms five benchmarks across two metrics, inducing steering and speed deviations from 3.54 degrees to 29 degrees and 11 km per hour to 22 km per hour on average. This systematic approach establishes UniAda as a proven technique for adversarial attacks on modern DL-based E2E ADSs.

Large language models (LLMs) have demonstrated strong performance on a wide range of software engineering tasks, including code generation and analysis. However, most prior work relies on cloud-based models or specialized hardware, limiting practical applicability in privacy-sensitive or resource-constrained environments. In this paper, we present a systematic empirical evaluation of two locally deployed LLMs, LLaMA 3.2 and Mistral, for real-world Python bug detection using the BugsInPy benchmark. We evaluate 349 bugs across 17 projects using a zero-shot prompting approach at the function level and an automated keyword-based evaluation framework. Our results show that locally executed models achieve accuracy between 43% and 45%, while producing a large proportion of partially correct responses that identify problematic code regions without pinpointing the exact fix. Performance varies significantly across projects, highlighting the importance of codebase characteristics. The results demonstrate that local models can identify a meaningful share of bugs, though precise localization remains difficult for locally executed LLMs, particularly when handling complex and context dependent bugs in realistic development scenarios.

Large language models (LLMs) show promise in medical diagnosis, but real-world deployment remains challenging due to high-stakes clinical decisions and imperfect reasoning reliability. As a result, careful inspection of model behavior is essential for assessing whether diagnostic reasoning is reliable and clinically grounded. However, debugging medical LLMs remains difficult. First, developers often lack sufficient medical domain expertise to interpret model errors in clinically meaningful terms. Second, models can fail across a large and diverse set of instances involving different input types, tasks, and reasoning steps, making it challenging for developers to prioritize which errors deserve focused inspection. Third, developers struggle to identify recurring error patterns across cases, as existing debugging practices are largely instance-centric and rely on manual inspection of isolated failures. To address these challenges, we present VeriLLMed, a visual analytics system that integrates external biomedical knowledge to audit and debug medical LLM diagnostic reasoning. VeriLLMed transforms model outputs into comparable reasoning paths, constructs knowledge graph-grounded reference paths, and identifies three recurring classes of diagnosis errors: relation errors, branch errors, and missing errors. Case studies and expert evaluation demonstrate that VeriLLMed helps developers identify clinically implausible reasoning and generate actionable insights that can inform the improvement of medical LLMs.

Existing LLM-based EDA agents are often isolated task-specific systems. This leads to repeated engineering effort and limited reuse of successful design and debugging strategies. We present LEGO, a unified skill-based platform for front-end design generation. It decomposes the digital front-end flow into six independent steps and represents every agent capability as a standardized composable circuit skill within a plug-and-play architecture. To build this skill library, we survey more than 100 papers, select 11 representative open-source projects, and extract 42 executable circuit skills within a six-step finite state machine formulation. Circuit Skill Builder automates skill extraction with linear scalability. Agent Skill RAG achieves submillisecond retrieval without relying on embedding models. Empirical evaluation on a hard subset of 41 VerilogEval v2 problems that gpt-5.2-codex fails to solve under extra-high reasoning effort shows that individual circuit skills constructed within LEGO raise Pass@1 from 0.000 to 0.805. This is an 80.5% gain over the baseline. Cross-project skill compositions also reach 0.805 Pass@1. They outperform hierarchy-verilog by 14.6% and VerilogCoder by 2.5%. They also match MAGE. These results show that modular skill composition supports both effective and flexible RTL design automation. The LEGO platform and all circuit skills are publicly available at GitHub: https://github.com/loujc/LEGO-An-LLM-Skill-Based-Front-End-Design-Generation-Platform

Neural networks can be trained to learn task-relevant representations from data. Understanding how these networks make decisions falls within the Explainable AI (XAI) domain. This paper proposes to study an XAI topic: uncovering unknown organisational patterns in network representations, particularly those representations learned by the speaker recognition network that recognises the speaker identity of utterances. Past studies employed algorithms (e.g. t-distributed Stochastic Neighbour Embedding and K-means) to analyse and visualise how network representations form independent clusters, indicating the presence of flat clustering phenomena within the space defined by these representations. In contrast, this work applies two algorithms -- Single-Linkage Clustering (SLINK) and Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) -- to analyse how representations form clusters with hierarchical relationships rather than being independent, thereby demonstrating the existence of hierarchical clustering phenomena within the network representation space. To semantically understand the above hierarchical clustering phenomena, a new algorithm, termed Hierarchical Cluster-Class Matching (HCCM), is designed to perform one-to-one matching between predefined semantic classes and hierarchical representation clusters (i.e. those produced by SLINK or HDBSCAN). Some hierarchical clusters are successfully matched to individual semantic classes (e.g. male, UK), while others to conjunctions of semantic classes (e.g. male and UK, female and Ireland). A new metric, Liebig's score, is proposed to quantify the performance of each matching behaviour, allowing us to diagnose the factor that most strongly limits matching performance.

Whether intermediate reasoning is computationally useful or merely explanatory depends on whether chain-of-thought (CoT) tokens contain task-relevant information. We present a mechanistic causal analysis of CoT on GSM8K using activation patching: transferring token-level hidden states from a CoT generation to a direct-answer run for the same question, then measuring the effect on final-answer accuracy. Across models, generating after patching yields substantially higher accuracy than both direct-answer prompting and the original CoT trace, revealing that individual CoT tokens can encode sufficient information to recover the correct answer, even when the original trace is incorrect. This task-relevant information is more prevalent in correct than incorrect CoT runs and is unevenly distributed across tokens, concentrating in mid-to-late layers and appearing earlier in the reasoning trace. Moreover, patching language tokens such as verbs and entities carry task-solving information that steers generation toward correct reasoning, whereas mathematical tokens encode answer-proximal content that rarely succeeds. Patched outputs are often shorter and yet exceed the accuracy of a full CoT trace, suggesting complete reasoning chains are not always necessary. Together, these findings demonstrate that CoT encodes recoverable, token-level problem-solving information, offering new insight into how reasoning is represented and where it breaks down.

Deep generative models and neural operators have demonstrated significant potential for 3D aerodynamic inference. However, they often face inherent challenges in maintaining physical consistency and preserving high-frequency features, primarily due to spectral bias and gradient conflicts within the governing equations. To address these issues, we propose GeoFunFlow-3D, a physics-guided generative flow matching framework. Temporally, we utilize optimal transport theory to build the generation path, ensuring stable training dynamics. Spectrally, we introduce a high-order discrete engine without automatic differentiation (No-AD) to reduce gradient stiffness. Spatially, a topology-aware super-resolution module (SATO) is employed to rigorously enforce physical laws in localized regions such as shock waves. We evaluated our framework on complex industrial datasets. On the BlendedNet dataset, the model successfully avoids mode collapse even under sparse data conditions. For the NASA Rotor37 test, it accurately captures 3D detached shock structures. Compared to conventional operators, GeoFunFlow-3D significantly improves accuracy, reducing the pressure field error (RRMSE) to 0.0215 while maintaining competitive inference efficiency. Ultimately, this work provides a reliable, geometry-driven approach for generating high-dimensional fluid fields.

Recent multimodal large language models (MLLMs) have shown strong capabilities in perception, reasoning, and generation, and are increasingly used in applications such as social robots and human-computer interaction, where understanding human emotions is essential. However, existing benchmarks mainly formulate emotion understanding as a static recognition problem, leaving it largely unclear whether current MLLMs can understand emotion as a dynamic process that evolves, shifts between states, and unfolds across diverse social contexts. To bridge this gap, we present EmoTrans, a benchmark for evaluating emotion dynamics understanding in multimodal videos. EmoTrans contains 1,000 carefully collected and manually annotated video clips, covering 12 real-world scenarios, and further provides over 3,000 task-specific question-answer (QA) pairs for fine-grained evaluation. The benchmark introduces four tasks, namely Emotion Change Detection (ECD), Emotion State Identification (ESI), Emotion Transition Reasoning (ETR), and Next Emotion Prediction (NEP), forming a progressive evaluation framework from coarse-grained detection to deeper reasoning and prediction. We conduct a comprehensive evaluation of 18 state-of-the-art MLLMs on EmoTrans and obtain two main findings. First, although current MLLMs show relatively stronger performance on coarse-grained emotion change detection, they still struggle with fine-grained emotion dynamics modeling. Second, socially complex settings, especially multi-person scenarios, remain substantially challenging, while reasoning-oriented variants do not consistently yield clear improvements. To facilitate future research, we publicly release the benchmark, evaluation protocol, and code at https://github.com/Emo-gml/EmoTrans.

We present a comprehensive evaluation of Large Language Models (LLMs) on Computer Science (CS) Data Structure examination questions. Our work introduces a new benchmark dataset comprising exam questions from Tel Aviv University (TAU), curated to assess LLMs' abilities in handling closed and multiple-choice questions. We evaluated the performance of OpenAI's GPT 4o and Anthropic's Claude 3.5, popular LLMs, alongside two smaller LLMs, Mathstral 7B and LLaMA 3 8B, across the TAU exams benchmark. Our findings provide insight into the current capabilities of LLMs in CS education.

Cross-domain task-oriented dialogue requires reasoning over implicit and explicit feasibility constraints while planning long-horizon, multi-turn actions. Large language models (LLMs) can infer such constraints but are unreliable over long horizons, while Reinforcement learning (RL) optimizes long-horizon behavior yet cannot recover constraints from raw dialogue. Naively coupling LLMs with RL is therefore brittle: unverified or unstructured LLM outputs can corrupt state representations and misguide policy learning. Motivated by this, we propose Verified LLM-Knowledge empowered RL (VLK-RL), a hybrid framework that makes LLM-derived constraint reasoning usable for RL. VLK-RL first elicits candidate constraints with an LLM and then verifies them via a dual-role cross-examination procedure to suppress hallucinations and cross-turn inconsistencies. The verified constraints are mapped into ontology-aligned slot-value representations, yielding a structured, constraint-aware state for RL policy optimization. Experiments across multiple benchmarks demonstrate that VLK-RL significantly improves generalization and robustness, outperforming strong single-model baselines on long-horizon tasks.

Deep learning (DL)-based systems can exhibit unexpected behavior when exposed to out-of-distribution (OOD) scenarios, posing serious risks in safety-critical domains such as malware detection and autonomous driving. This underscores the importance of thoroughly testing such systems before deployment. To this end, researchers have proposed a wide range of test selection metrics designed to effectively select inputs. However, prior evaluations of metrics reveal three key limitations: (1) narrow testing objectives, for example, many studies assess metrics only for fault detection, leaving their effectiveness for performance estimation unclear; (2) limited coverage of OOD scenarios, with natural and label shifts are rarely considered; (3) Biased dataset selection, where most work focuses on image data while other modalities remain underexplored. Consequently, a unified benchmark that examines how these metrics perform under multiple testing objectives, diverse OOD scenarios, and different data modalities is still lacking. This leaves practitioners uncertain about which test selection metrics are most suitable for their specific objectives and contexts. To address this gap, we conduct an extensive empirical study of 15 existing metrics, evaluating them under three testing objectives (fault detection, performance estimation, and retraining guidance), five types of OOD scenarios (corrupted, adversarial, temporal, natural, and label shifts), three data modalities (image, text, and Android packages), and 13 DL models. In total, our study encompasses 1,640 experimental scenarios, offering a comprehensive evaluation and statistical analysis.

The deployment of Large Language Models (LLMs) as assistants in electric grid operations promises to streamline compliance and decision-making but exposes new vulnerabilities to prompt-based adversarial attacks. This paper evaluates the risk of jailbreaking LLMs, i.e., circumventing safety alignments to produce outputs violating regulatory standards, assuming threats from authorized users, such as operators, who craft malicious prompts to elicit non-compliant guidance. Three state-of-the-art LLMs (OpenAI's GPT-4o mini, Google's Gemini 2.0 Flash-Lite, and Anthropic's Claude 3.5 Haiku) were tested against Baseline, BitBypass, and DeepInception jailbreaking methods across scenarios derived from nine NERC Reliability Standards (EOP, TOP, and CIP). In the initial broad experiment, the overall Attack Success Rate (ASR) was 33.1%, with DeepInception proving most effective at 63.17% ASR. Claude 3.5 Haiku exhibited complete resistance (0% ASR), while Gemini 2.0 Flash-Lite was most vulnerable (55.04% ASR) and GPT-4o mini moderately susceptible (44.34% ASR). A follow-up experiment refining malicious wording in Baseline and BitBypass attacks yielded a 30.6% ASR, confirming that subtle prompt adjustments can enhance simpler methods' efficacy.

LLM-generated code is widely used, and the share of committed code produced by LLMs is expected to increase. However, we are not at a point where LLMs can be effective contributors to production code. We present an approach that exposes the shortcomings of LLM generation on such projects, and proposes recommendations; the targets of our study are sizable open-source projects, e.g., FFmpeg and wolfSSL. First, we developed a framework that uses verification and validation to evaluate a given LLM's suitability to fix or add features to an existing project. Second, we apply the framework to 212 commits (bug fixes and small feature improvements) in eight popular open-source projects and three LLMs: GPT-4o, Ministral3, and Qwen3-Coder. The success rate varied from 0% to 60% depending on the project. The LLMs failed in a variety of ways, from generating syntactically incorrect code, to producing code that fails basic (static) verification, or validation via the project's test suite. In particular, the LLMs struggle with generating new code, handling contexts (function or file) outside a certain size range, and in many cases their success is due to parroting code changes they have been trained on.

Agentic AI systems face security challenges that stateless large language models do not. They plan across extended horizons, maintain persistent memory, invoke external tools, and coordinate with peer agents. Existing security analyses organize threats by attack type (prompt injection, jailbreaking), but provide no principled model of which architectural component is vulnerable or over what timescale the threat manifests. This paper makes five contributions. First, we introduce the Layered Attack Surface Model (LASM), a seven-layer framework that maps threats to distinct architectural components: Foundation, Cognitive, Memory, Tool Execution, Multi-Agent Coordination, Ecosystem, and Governance, the accountability and observability layer that spans the stack analogously to the network management plane. Second, we introduce attack temporality as an orthogonal analytical dimension with four classes: Instantaneous (T1), Session-Persistent (T2), Cross-Session Cumulative (T3), and Sub-Session-Stack, Non-Session-Bounded (T4). Third, through a systematic review of 94 papers (2021--2025), we show that the most dangerous emerging threats concentrate at the intersection of high-layer attacks (L5--L7) and slow-burn temporality (T3--T4): covert agent collusion, long-term memory poisoning, MCP supply-chain compromise, and alignment failure that manifests as an insider threat with no external adversary. Only 8 of 120 paper-cell assignments (7%) fall in this zone. Fourth, we propose a cross-layer defense taxonomy spanning all seven LASM layers and all four temporality classes, exposing which threat classes existing defenses leave unaddressed. Fifth, we survey evaluation benchmarks, identify five research gaps in the under-studied high-layer, slow-burn zone, and argue that agentic security must be treated as a distributed systems problem embedded in an adversarial ecosystem.

Unlike traditional fact-based retrieval, rationale-based retrieval typically necessitates cross-encoding of query-document pairs using large language models, incurring substantial computational costs. To address this limitation, we propose Rabtriever, which independently encodes queries and documents, while providing comparable cross query-document comprehension capabilities to rerankers. We start from training a LLM-based generative reranker, which puts the document prior to the query and prompts the LLM to generate the relevance score by log probabilities. We then employ it as the teacher of an on-policy distillation framework, with Rabtriever as the student to reconstruct the teacher's contextual-aware query embedding. To achieve this effect, Rabtriever is first initialized from the teacher, with parameters frozen. The Joint-Embedding Predictive Architecture (JEPA) paradigm is then adopted, which integrates a lightweight, trainable predictor between LLM layers and heads, projecting the query embedding into a new hidden space, with the document embedding as the latent vector. JEPA then minimizes the distribution difference between this projected embedding and the teacher embedding. To strengthen the sampling efficiency of on-policy distillation, we also add an auxiliary loss on the reverse KL of LLM logits, to reshape the student's logit distribution. Rabtriever optimizes the teacher's quadratic complexity on the document length to linear, verified both theoretically and empirically. Experiments show that Rabtriever outperforms different retriever baselines across diverse rationale-based tasks, including empathetic conversations and robotic manipulations, with minor accuracy degradation from the reranker. Rabtriever also generalizes well on traditional retrieval benchmarks such as MS MARCO and BEIR, with comparable performance to the best retriever baseline.

Scaling test-time computation with reinforcement learning (RL) has emerged as a reliable path to improve large language models (LLM) reasoning ability. Yet, outcome-based reward often incentivizes models to be overconfident, leading to hallucinations, unreliable confidence-based control, and unnecessary compute allocation. We introduce Reinforcement Learning with Confidence Margin (\textbf{RLCM}), a calibration-aware RL framework that jointly optimizes correctness and confidence reliability via a margin-enhanced process reward over intermediate-budget completions. Rather than aligning confidence to correctness likelihoods, RLCM encourages to widen the confidence margin between correct and incorrect steps within a single reasoning trajectory. Across mathematical, code, logic and science benchmarks, our method substantially improves calibration while maintaining or improving accuracy. We further show that, with calibrated confidence signals, the resulting models enable more efficient conformal risk control and effective confidence-weighted aggregation.

Emotionally talking head video generation aims to generate expressive portrait videos with accurate lip synchronization and emotional facial expressions. Current methods rely on simple emotional labels, leading to insufficient semantic information. While introducing high-level semantics enhances expressiveness, it easily causes lip-sync degradation. Furthermore, mainstream generation methods struggle to balance computational efficiency and global motion awareness in long videos and suffer from poor temporal coherence. Therefore, we propose an \textbf{E}motion-\textbf{A}ware \textbf{D}iffusion model-based \textbf{Net}work, called \textbf{EAD-Net}. We introduce SyncNet supervision and Temporal Representation Alignment (TREPA) to mitigate lip-sync degradation caused by multi-modal fusion. To model complex spatio-temporal dependencies in long video sequences, we propose a Spatio-Temporal Directional Attention (STDA) mechanism that captures global motion patterns through strip attention. Additionally, we design a Temporal Frame graph Reasoning Module (TFRM) to explicitly model temporal coherence between video frames through graph structure learning. To enhance emotional semantic control, a large language model is employed to extract textual descriptions from real videos, serving as high-level semantic guidance. Experiments on the HDTF and MEAD datasets demonstrate that our method outperforms existing methods in terms of lip-sync accuracy, temporal consistency, and emotional accuracy.

Graph Neural Networks (GNNs) have demonstrated impressive performance in learning representations from graph-structured data. However, their message-passing mechanism inherently relies on the assumption of label consistency among connected nodes, limiting their applicability to low-homophily settings. Moreover, since message passing operates as a hierarchical diffusion process, GNNs face challenges in capturing long-range dependencies. As network depth increases, the structural noise along heterophilic edges tends to be amplified, resulting in over-smoothing. This issue becomes especially prominent in highly heterophilic graphs, where the propagation of inconsistent semantics across the topology continually exacerbates misaggregation. To address this issue, we propose a novel framework named Layer Embedding Deep Fusion Graph Neural Network (LEDF-GNN). Specifically, we design a Layer Embedding Deep Fusion (LEDF) operator that nonlinearly fuses multi-layer embeddings to capture inter-layer dependencies and effectively alleviate deep propagation degradation. Meanwhile, to mitigate structural heterophily, LEDF-GNN employs a Dual-Topology Parallel Strategy (DTPS) that simultaneously leverages the original and reconstructed topologies, allowing for adaptive structure-semantics co-optimization under diverse homophily conditions. Extensive semi-supervised classification experiments on the citation and image benchmarks demonstrate that, under both homophilic and heterophilic settings, LEDF-GNN consistently outperforms state-of-the-art baselines, validating its effectiveness and generalization capability across diverse graph types.

Audio-text retrieval enables semantic alignment between audio content and natural language queries, supporting applications in multimedia search, accessibility, and surveillance. However, current state-of-the-art approaches struggle with long, noisy, and weakly labeled audio due to their reliance on contrastive learning and large-batch training. We propose a novel multimodal retrieval framework that refines audio and text embeddings using a cross-modal embedding refinement module combining transformer-based projection, linear mapping, and bidirectional attention. To further improve robustness, we introduce a hybrid loss function blending cosine similarity, $\mathcal{L}_{1}$, and contrastive objectives, enabling stable training even under small-batch constraints. Our approach efficiently handles long-form and noisy audio (SNR 5 to 15) via silence-aware chunking and attention-based pooling. Experiments on benchmark datasets demonstrate improvements over prior methods.

Group Relative Policy Optimization (GRPO) performs coarse-grained credit assignment in reinforcement learning with verifiable rewards (RLVR) by assigning the same advantage to all tokens in a rollout. Process reward models can provide finer-grained supervision, but they require step-level annotation or additional reward modeling. We show that hidden-state distributions contain a useful signal for local reasoning quality that can be extracted using only outcome-level correctness labels available in RLVR. Specifically, within each GRPO group, the Wasserstein distance between span-level hidden state distributions of correct and incorrect rollouts increases around regions where their local reasoning quality diverges. This association holds both across examples and within individual trajectories, suggesting that hidden-state distributional divergence can serve as a self-supervision signal for fine-grained credit assignment. We formalize this observation with a separation theorem showing that, under mild structural assumptions, post-divergence spans have larger Wasserstein distances than pre-divergence spans whenever the population-level distributional gap exceeds finite-sample noise. Motivated by this result, we propose \textbf{S}pan-level \textbf{H}idden state \textbf{E}nabled \textbf{A}dvantage \textbf{R}eweighting (SHEAR), which modifies GRPO by using span-level Wasserstein distances to scale token-level advantages, amplifying updates on tokens whose hidden states are more separated from the opposing group. The method requires no additional model and only minimal changes to the training pipeline. Experiments on five mathematical reasoning benchmarks and five code generation benchmarks show improvements over standard GRPO and strong performance relative to supervised process reward models, while requiring no additional annotation or reward model training.

Deep reinforcement learning policies achieve strong performance in complex continuous control environments with nonlinear contact forces. However, these policies often produce chaotic state dynamics, with trivially small changes to the initial conditions significantly impacting the long-term behaviour of the control system. This high sensitivity to initial conditions limits the application of Deep RL to real-world control systems where performance and stability guarantees are often required. To address this issue, we propose Global stabilisation via Intrinsic Fine Tuning (GIFT), a general-purpose training framework which directly optimises the global stability of existing high-performing deep RL policies using a custom reward function. We demonstrate that GIFT increase the stability of the control interaction while maintaining comparable task performance, thereby improving the suitability of deep RL policies for real-world control systems.

Remote sensing image change captioning (RSICC) aims to describe the difference between two remote sensing images. While recent methods have explored video modeling, they largely overlook the inherent ambiguities in viewpoint, scale, and prior knowledge, lacking effective constraints on the encoder. In this paper, we present STAND, a Semantic Anchoring Constraint with Dual-Granularity Disambiguation for RSICC, to progressively resolve these ambiguities. Specifically, to establish a reliable feature foundation, we first introduce an interpretable constraint to regularize temporal representations. Operating on these purified features, a dual-granularity disambiguation module resolves spatial uncertainties by coupling macro-level global context aggregation for viewpoint confusion with micro-level frequency-refocused attention for small-object scale enhancement. Ultimately, to translate these visually disambiguated features into precise text, a semantic concept anchoring module leverages language categorical priors to tackle knowledge ambiguity during decoding. Extensive experiments verify the superiority of STAND and its effectiveness in addressing ambiguities.

Offline multi-agent reinforcement learning (MARL) enables policy learning from fixed datasets, but is prone to coordination failure: agents trained on static, off-policy data converge to suboptimal joint behaviours because they cannot co-adapt as their policies change. We introduce CODA (Coordination via On-Policy Diffusion for Multi-Agent Reinforcement Learning), a diffusion-based multi-agent trajectory generator for data augmentation that samples conditioned on the current joint policy, producing synthetic experience which reflects the evolving behaviours of the agents, thereby providing a mechanism for co-adaptation. We find that previous diffusion-based augmentation approaches are insufficient for fostering multi-agent coordination because they produce static augmented datasets that do not evolve as the current joint policy changes during training; CODA resolves this by more closely simulating on-policy learning and is a meaningful step toward coordinated behaviours in the offline setting. CODA is algorithm-agnostic and can be layered onto both model-free and model-based offline reinforcement learning pipelines as an augmentation module. Empirically, CODA not only resolves canonical coordination pathologies in continuous polynomial games but also delivers strong results on the more complex MaMuJoCo continuous-control benchmarks.

Dual-target molecule generation, which focuses on discovering compounds capable of interacting with two target proteins, has garnered significant attention due to its potential for improving therapeutic efficiency, safety and resistance mitigation. Existing approaches face two critical challenges. First, by simplifying the complex dual-target optimization problem to scalarized combinations of individual objectives, they fail to capture important trade-offs between target engagement and molecular properties. Second, they typically do not integrate synthetic planning into the generative process. This highlights a need for more appropriate objective function design and synthesis-aware methodologies tailored to the dual-target molecule generation task. In this work, we propose CombiMOTS, a Pareto Monte Carlo Tree Search (PMCTS) framework that generates dual-target molecules. CombiMOTS is designed to explore a synthesizable fragment space while employing vectorized optimization constraints to encapsulate target affinity and physicochemical properties. Extensive experiments on real-world databases demonstrate that CombiMOTS produces novel dual-target molecules with high docking scores, enhanced diversity, and balanced pharmacological characteristics, showcasing its potential as a powerful tool for dual-target drug discovery. The code and data is accessible through https://github.com/Tibogoss/CombiMOTS.

With the rise of mobile-first consumption, users increasingly engage with data visualizations on mobile devices. However, the vast majority of existing visualizations are originally authored for desktop environments. Due to significant differences in viewport size and interaction paradigms, directly scaling desktop charts often results in illegible text, information loss, and interaction failures. To bridge this gap, we propose an automated framework to adapt desktop-based visualizations for mobile screens. By systematically categorizing the operations involved in the adaptation process, we establish a multi-level design space. This space defines evolution rules spanning from the global topology level, through the reference frame level, down to the visual elements level. Guided by this theoretical framework, we developed Proteus, a large language model-driven multi-agent system that automatically parses online visualizations, predicts optimal transformation strategies within the design space, and generates equivalent, highly readable visualizations for mobile devices. Case studies and an in-depth user study with 12 participants demonstrate the effectiveness and usability of Proteus.

Aspect Sentiment Quad Prediction (ASQP) has seen significant advancements, largely driven by the powerful semantic understanding and generative capabilities of large language models (LLMs). However, while syntactic structure information has been proven effective in previous extractive paradigms, it remains underutilized in the generative paradigm of LLMs due to their limited reasoning capabilities. In this paper, we propose S^2IT, a novel Stepwise Syntax Integration Tuning framework that progressively integrates syntactic structure knowledge into LLMs through a multi-step tuning process. The training process is divided into three steps. S^2IT decomposes the quadruple generation task into two stages: 1) Global Syntax-guided Extraction and 2) Local Syntax-guided Classification, integrating both global and local syntactic structure information. Finally, Fine-grained Structural Tuning enhances the model's understanding of syntactic structures through the prediction of element links and node classification. Experiments demonstrate that S^2IT significantly improves state-of-the-art performance across multiple datasets. Our implementation will be open-sourced at https://github.com/DMIRLAB-Group/S2IT.

Full-duplex spoken dialogue systems can model natural conversational behaviours such as interruptions, overlaps, and backchannels, yet such systems remain largely unexplored for Indian languages. We present the first open, reproducible full-duplex spoken dialogue system for Hindi by adapting Moshi, a state-of-the-art duplex speech architecture, using a custom Hindi tokeniser and training on 26,000 hours of real spontaneous conversations collected from 14,695 speakers with separate speaker channels, enabling direct learning of turn-taking and overlap patterns from natural interactions. To support Hindi text generation, we replace the original English tokeniser and reinitialise text-vocabulary-dependent parameters while retaining the pre-trained audio components. We propose a two-stage training recipe -- large-scale pre-training followed by fine-tuning on 1,000 hours of conversational data. Evaluation through the prompted dialogue continuation paradigm with both automatic metrics and human judgments demonstrates that the resulting model generates natural and meaningful full-duplex conversational behaviour in Hindi. This work serves as a first step toward real-time duplex spoken dialogue systems for Hindi and other Indian languages.

Active learning algorithms automatically identify the most informative samples from large amounts of unlabeled data and tremendously reduce human annotation effort in inducing a machine learning model. In a conventional active learning setup, the labeling oracles are assumed to be infallible, that is, they always provide correct answers (in terms of class labels) to the queried unlabeled instances, which cannot be guaranteed in real-world applications. To this end, a body of research has focused on the development of active learning algorithms in the presence of imperfect / noisy oracles. Existing research on active learning with noisy oracles typically simulate the oracles using machine learning models; however, real-world situations are much more challenging, and using ML models to simulate the annotation patterns may not appropriately capture the nuances of real-world annotation challenges. In this research, we first collect annotations of text samples (from 3 benchmark text classification datasets) from crowd-sourced workers through a crowd-sourcing platform. We then conduct extensive empirical studies of 8 commonly used active learning techniques (in conjunction with deep neural networks) using the obtained annotations. Our analyses sheds light on the performance of these techniques under real-world challenges, where annotators can provide incorrect labels, and can also refuse to provide labels. We hope this research will provide valuable insights that will be useful for the deployment of deep active learning systems in real-world applications. The obtained annotations can be accessed at https://github.com/varuntotakura/al_rcta/.

Due to the unprecedented success of deep learning, it has become an integral component in several multimedia computing applications in todays world. Unfortunately, deep learning systems are not perfect and can fail, sometimes abruptly, without prior warning or explanation. While reducing the error rate of deep neural networks has been the primary focus of the multimedia community, the problem of predicting when a deep learning system is going to fail has received significantly less research attention. In this paper, we propose a simple yet effective framework, MetaErr, to address this under-explored problem in deep learning research. We train a meta-model whose goal is to predict whether a base deep neural network will succeed or fail in predicting a particular data sample, by observing the base models performance on a given learning task. The meta-model is completely agnostic of the architecture and training parameters of the base model. Such an error prediction system can be immensely useful in a variety of smart multimedia applications. Our empirical studies corroborate the promise and potential of our framework against competing baselines. We further demonstrate the usefulness of our framework to improve the performance of pseudo-labeling-based semi-supervised learning, and show that MetaErr outperforms several strong baselines on three benchmark computer vision datasets.

This work proposes an agentic, intent-driven end-to-end (E2E) orchestration framework that integrates intent co-creation with a Test-Driven Quality Assurance paradigm. In this framework, autonomous agents iteratively refine a user's initial intent into a confirmed, auditable specification. Furthermore, the system automatically derives validation tests from these intents before provisioning, directly mirroring the Test-Driven Development workflow in software engineering to ensure proactive Service Level Agreement (SLA) compliance. The architecture is grounded in a standards-aligned knowledge representation using TM Forum (TMF) information models and catalogs. This enables deterministic graph traversal from high-level Product Offerings down to granular Service/Resource and Test specifications. We prototyped this architecture by extending OpenSlice with a message-driven, multi-agent pattern and integrating MCP-enabled (Model Context Protocol) tool access for real-time knowledge retrieval. Currently, our evaluation of the agents targets the intent co-creation phase as a baseline toward full-scale orchestration. Building on experiments with multiple open-source Large Language Model (LLM) backends integrated with the TMF-based knowledge base, we observe substantial variability in tool-use reliability and hallucination patterns, underscoring the critical importance of robust knowledge integration in agentic 6G systems.

In this work, we present Au-M-ol, a novel multimodal architecture that extends Large Language Models (LLMs) with audio processing. It is designed to improve performance on clinically relevant tasks such as Automatic Speech Recognition (ASR). Au-M-ol has three main components: (1) an audio encoder that extracts rich acoustic features from medical speech, (2) an adaptation layer that maps audio features into the LLM input space, and (3) a pretrained LLM that performs transcription and clinical language understanding. This design allows the model to interpret spoken medical content directly, improving both accuracy and robustness. In experiments, Au-M-ol reduces Word Error Rate (WER) by 56\% compared to state-of-the-art baselines on medical transcription tasks. The model also performs well in challenging conditions, including noisy environments, domain-specific terminology, and speaker variability. These results suggest that Au-M-ol is a strong candidate for real-world clinical applications, where reliable and context-aware audio understanding is essential.

Current LLM agents operate under an implicit but universal assumption: execution is a transaction -- the user submits a request, the agent works in isolation, and only upon completion does the dialogue resume. This forces users into a binary choice: wait for a potentially incorrect output, or interrupt and lose all progress. We reject this assumption and propose the stream paradigm, in which agent execution and user intervention are concurrent, interleaved processes sharing a bidirectional channel. We formalize this paradigm through a reversibility taxonomy that classifies every agent action as Idempotent, Reversible, Compensable, or Irreversible, and arrive at a core conclusion: an agent's flexibility is bounded by its reversibility. We prove that conflicting compensable actions impose unavoidable adaptation costs and that conflicting irreversible actions make full specification satisfaction impossible -- these costs are properties of the action space, not of the algorithm. Guided by this insight, we present the Revision Absorber, a reactive algorithm based on the Earliest-Conflict Rollback rule that is structurally optimal under mild assumptions. Experiments on StreamBench with real LLM agents validate all predictions: the Absorber matches the quality of a brute-force full-restart baseline while wasting an order of magnitude fewer steps of already-completed work, turning mid-execution revisions from a dead-end into a first-class interaction.

Human activity recognition serves as the foundation for various emerging applications. In recent years, researchers have used collaborative sensing of multi-source sensors to capture complex and dynamic human activities. However, multimodal human activity sensing typically encounters highly heterogeneous data across modalities and label scarcity, resulting in an application gap between existing solutions and real-world needs. In this paper, we propose CLMM, a general contrastive learning framework for human activity recognition that achieves effective multimodal recognition with limited labeled data. CLMM employs a novel two-stage training strategy. In the first stage, CLMM employs a CNN-DiffTransformer encoder to capture cross-modal shared information by extracting local and global features. Meanwhile, a hard-positive samples weighting algorithm enhances gradient propagation to reinforce shared learning. In the second stage, a dual-branch architecture combining quality-guided attention and bidirectional gated units captures modality-specific information, while a primary-auxiliary collaborative training strategy fuses both shared and modality-specific information. Experimental results on three public datasets demonstrate that CLMM significantly improves state-of-the-art baselines in both recognition accuracy and convergence performance.

AI agents are now running real transactions, workflows, and sub-agent chains across organizational boundaries without continuous human supervision. This creates a problem no current infrastructure is equipped to solve: how do you identify, verify, and hold accountable an entity with no body, no persistent memory, and no legal standing? We define AI Identity as the continuous relationship between what an AI agent is declared to be and what it is observed to do, bounded by the confidence that those two things correspond at any given moment. Through a structured survey of industry trends, emerging standards, and technical literature, we conduct a gap analysis across the full agent identity lifecycle and make three contributions: (1) a structural comparison of human and AI identity across four dimensions (substrate, persistence, verifiability, and legal standing) showing that the asymmetry is fundamental and that extending human frameworks to agents without structural modification produces systematic failures; (2) an evaluation of current technical and regulatory documents against the identity requirements of autonomous agents, finding that none adequately address the challenge of governing nondeterministic, boundary-crossing entities; and (3) identification of five critical gaps (semantic intent verification, recursive delegation accountability, agent identity integrity, governance opacity and enforcement, and operational sustainability) that no current technology or regulatory instrument resolves. These gaps are structural; more engineering effort alone will not close them. Foundational research on AI identity is the central conclusion of this report.

The proliferation of agentic artificial intelligence has outpaced the conceptual tools needed to characterize agency in computational systems. Prevailing definitions mainly rely on autonomy and goal-directedness. Here, we argue for a minimal notion open to principled inspection given three criteria: intentionality as action grounded in beliefs and desires, rationality as normatively coherent action entailed by a world model, and explainability as action causally traceable to internal states; we subsequently instantiate these as a partially observable Markov decision process under a variational framework wherein posterior beliefs, prior preferences, and the minimization of expected free energy jointly constitute an agentic action chain. Using a canonical T-maze paradigm, we evidence how empowerment, formulated as the channel capacity between actions and anticipated observations, serves as an operational metric that distinguishes zero-, intermediate-, and high-agency phenotypes through structural manipulations of the generative model. We conclude by arguing that as agents engage in epistemic foraging to resolve ambiguity, the governance controls that remain effective must shift systematically from external constraints to the internal modulation of prior preferences, offering a principled, variational bridge from computational phenotyping to AI governance strategy

Long-context large language models remain computationally expensive to run and often fail to reliably process very long inputs, which makes context compression an important component of many systems. Existing compression approaches typically rely on trained compressors, dense retrieval-style selection, or heuristic trimming, and they often struggle to jointly preserve task relevance, topic coverage, and cross-sentence coherence under a strict token budget. To address this, we propose a training-free and model-agnostic compression framework that selects a compact set of sentences guided by structural graph priors. Our method constructs a sparse hybrid sentence graph that combines mutual k-NN semantic edges with short-range sequential edges, extracts a topic skeleton via clustering, and ranks sentences using an interpretable score that integrates task relevance, cluster representativeness, bridge centrality, and a cycle coverage cue. A budgeted greedy selection with redundancy suppression then produces a readable compressed context in original order. Experimental results on four datasets show that our approach is competitive with strong extractive and abstractive baselines, demonstrating larger gains on long-document benchmarks.

PDF documents contain critical visual elements such as figures, tables, and forms whose accurate extraction is essential for document understanding and multimodal retrieval-augmented generation (RAG). Existing PDF parsers often miss complex visuals, extract non-informative artifacts (e.g., watermarks, logos), produce fragmented elements, and fail to reliably associate captions with their corresponding elements, which degrades downstream retrieval and question answering. We present a lightweight and production level PDF parsing framework that can accurately detect visual elements and associates captions using a combination of spatial heuristics, layout analysis, and semantic similarity. On popular benchmark datasets and internal product data, the proposed solution achieves $\geq96\%$ visual element detection accuracy and $93\%$ caption association accuracy. When used as a preprocessing step for multimodal RAG, it significantly outperforms state-of-the-art parsers and large vision-language models on both internal data and the MMDocRAG benchmark, while reducing latency by over $2\times$. We have deployed the proposed system in challenging production environment.

Chain-of-Thought (CoT) prompting has emerged as a simple and effective way to elicit step-by-step solutions from large language models (LLMs). However, CoT reasoning can be unstable across runs on long, multi-step problems, leading to inconsistent answers for unchanged task. Most prior work focuses on improving the forward reasoning chain within a single pass, with less attention to iterative and contrastive correction. To address this gap, we propose CAP-CoT, a Cycle Adversarial Prompt optimization framework designed to improve both CoT reasoning accuracy and stability of a single deployed solver. In each cycle, a forward solver generates candidate reasoning chains, an adversarial challenger constructs plausible but deliberately flawed chains using targeted error strategies, and a feedback agent contrasts the two chains and produces step-aligned structured feedback. This feedback closes the optimization loop in two directions, including updating the solver prompt based on errors exposed by the challenger, and updating the challenger prompt to generate increasingly targeted errors in subsequent cycles. Unlike safety-oriented adversarial prompting such as jailbreak or prompt-injection attacks, our adversarial component is task-semantic and aims to expose logical vulnerabilities in reasoning chains. Experiments across six benchmarks and four LLM backbones demonstrate that within two to three adversarial prompt optimization cycles, CAP-CoT consistently reduces variability across runs while improving reasoning accuracy and robustness to prompt perturbations.

Large language models (LLMs) operate in two fundamental learning modes - fine-tuning (FT) and in-context learning (ICL) - raising key questions about which mode yields greater language proficiency and whether they differ in their inductive biases. Prior studies comparing FT and ICL have yielded mixed and inconclusive results due to inconsistent experimental setups. To enable a rigorous comparison, we propose a formal language learning task - offering precise language boundaries, controlled string sampling, and no data contamination - and introduce a discriminative test for language proficiency, where an LLM succeeds if it assigns higher generation probability to in-language strings than to out-of-language strings. Empirically, we find that: (a) FT has greater language proficiency than ICL on in-distribution generalization, but both perform equally well on out-of-distribution generalization. (b) Their inductive biases, measured by the correlation in string generation probabilities, are similar when both modes partially learn the language but diverge at higher proficiency levels. (c) Unlike FT, ICL performance differs substantially across models of varying sizes and families and is sensitive to the token vocabulary of the language. Thus, our work demonstrates the promise of formal languages as a controlled testbed for evaluating LLMs, behaviors that are difficult to isolate in natural language datasets. Our source code is available at https://github.com/bishwamittra/formallm.

The successive generations of consensus algorithms have progressively shifted the performance bottleneck of blockchains to the execution layer. While recent works address this by parallelizing transaction execution, they often overlook the critical role of transaction sequencing. Historically, transaction ordering was left to validator discretion, a practice prone to Maximal Extractable Value (MEV) attacks, or rigid fair-ordering protocols that limit validator revenue. In this work, we address the tension between validator revenue and order fairness using a dynamic optimization framework. We introduce a blockchain-independent model for transaction sequencing in a continuous setting where block executions can overlap. Within this framework, we propose an anytime genetic algorithm that utilizes gas prices, object sets, and predicted execution times to optimize schedules. We evaluate our approach with real-world datasets from Sui and Ethereum, and demonstrate that our algorithm increases validator profit by approximately 15% and accelerates congestion relief by up to 58%. Furthermore, we quantify the impact of fair-ordering constraints, showing they can reduce validator revenue by 50% to 60% during periods of high congestion. We provide the first evidence that enforcing strict fair ordering effectively nullifies the advantages of advanced sequencing.

Large language models (LLMs) are increasingly utilized in various complex reasoning tasks due to their excellent instruction following capability. However, the model's performance is highly dependent on the open-ended characteristics of the users' input prompt. Natural prompts often do not follow proper syntactic rules, which creates ambiguous queries that yield multiple interpretations. Such ambiguous prompts confuse the model in choosing the correct reasoning paths to answer questions. Prior works address this challenge by applying query editing during the LLM inference process without explicitly solving the root cause of the ambiguity. To address this limitation, we propose a pre-inference prompt optimization mechanism via explicit prompt disambiguation. Particularly, we identify semantic risks in the prompt, check their multi-perspective consistency, and resolve any semantic conflicts that arise. Finally, we organize the resolved ambiguities in a logically structured manner as a clean input to the LLM. By explicitly resolving semantic ambiguity, our method can produce a more focused attention distribution to the semantically essential tokens. We also leverage small language models (SLMs) as the main executor of prompt disambiguation to benefit from their efficient computation. Through comprehensive experiments on multiple benchmarks, we demonstrate that our method improves reasoning performance by 2.5 points at a cost of only \$0.02. Our study promotes explicit prompt disambiguation as an effective prompt optimization method without disturbing the internal mechanism of LLM inference.

An approach to construct explicit integral representations for two-layer ReLU networks is presented, which provides relatively simple representations for any multivariate polynomial. Quantitative bounds are provided for a particular, sharpened ReLU integral representation, which involves a harmonic extension and a projection. The bounds demonstrate that functions can be approximated with $L^{2}(\mathcal{D})$ errors that do not depend explicitly on dimension or degree, but rather the coefficients of their monomial expansions and the distribution $\mathcal{D}$.

Software engineering (SE) organizations operate in a knowledge-intensive domain where critical assets -- architectural expertise, design rationale, and system intuition -- are overwhelmingly tacit and volatile. The departure of key contributors or the decay of undocumented decisions can severely impair project velocity and software quality. While conventional SE risk management optimized for schedule and budget is common, the intangible knowledge risks that determine project success remain under-represented. The goal of this research work is to propose and evaluate the Knowledge Lever Risk Management (KLRM) Framework, designed specifically for the software development lifecycle. The primary objectives are to: (1) recast intangible knowledge assets as active mechanisms for risk mitigation (Knowledge Levers); (2) integrate these levers into a structured four-phase architecture (Audit, Alignment, Activation, Assurance); and (3) provide a formal stochastic model to quantify the impact of lever activation on project knowledge capital. We detail the application of these levers through software-specific practices such as pair programming, architectural decision records (ADRs), and LLM-assisted development. Stochastic Monte Carlo simulations demonstrate that full lever activation increases expected knowledge capital by 63.8\% and virtually eliminates knowledge crisis probability. Our research shows that knowledge lever activation improves alignment across the project management iron triangle (scope, time, cost) by reducing rework and rediscovery costs.

Symbolic regression discovers mathematical formulas from data. Some methods fix a tree of operators, assign learnable weights, and train by gradient descent. The tree's structure, which determines what operators and variables appear at each position, is chosen once and applied to every target. This paper tests whether that choice affects which targets are actually recovered. Three structures are compared, all sharing the same operator and target language but differing in how variables enter the tree; one is strictly more expressive. Across over 12,700 training runs, one structure recovers a target at 100% while another scores 0%, and the ranking reverses on a different target. Expressiveness guarantees that a solution exists in the search space, but not that gradient descent finds it: the most expressive structure fails on targets that a restricted alternative solves reliably. Switching the operator changes which targets succeed; reversing its gradient profile collapses recovery entirely. Balanced (non-chain) tree shapes are never recovered. These findings show that the optimization landscape, not expressiveness alone, determines what gradient-based symbolic regression recovers.

Research shows that dialogue, the interactive process through which participants articulate their thinking, plays a central role in constructing shared understanding, coordinating action, and shaping learning outcomes in teams. Analysing dialogue content has been central to advancing team learning theory and informing the design of computer-supported collaborative learning environments, yet this progress has depended on labour-intensive qualitative coding. LLMs offer new possibilities for automating and enhancing the dialogue layer within emerging multimodal learning analytics approaches, with recent studies showing that they can approximate human coding through few-shot prompting. However, prior work has focused on replicating human coding accuracy for research purposes, rather than addressing a more educationally consequential question: how can we design prompts that allow an LLM to label team dialogue accurately and fast enough to be useful in real settings, such as in-person healthcare simulations, where results must be returned quickly and computational cost and sustainability also matter? This paper investigates how prompt design and batching strategies can be optimised to balance coding accuracy, processing time, and environmental impact in team-based healthcare simulation debriefing. Using a dataset of 11,647 utterances coded across 6 dialogue constructs, we compared 4 prompt designs across varying batch sizes, evaluating coding performance, processing time, and energy consumption, as well as the trade-offs between these metrics. Results indicate that increasing batch size improves speed and reduces energy use, but negatively impacts coding performance. Beyond demonstrating the feasibility of LLM-based qualitative analysis, this study offers practical guidance for scaling dialogue analytics in contexts where timeliness, privacy, and sustainability are critical.

Code review is central to software engineering education but hard to scale in capstone projects due to tight deadlines, uneven peer feedback, and limited prior experience. We investigate an LLM-as-reviewer integrated directly into GitHub pull requests (human-in-the-loop) across two cohorts (more than 100 students, 2023--2024). Using a mixed-methods design -- GitHub data, reflective reports, and a targeted survey -- we examine engagement and responsiveness as behavioral indicators of self-regulated learning processes. Quantitatively, the 2024 cohort produced more iterative activity (1176 vs. 581 PRs), while technical issues observed in 2023 (227 failed AI attempts) dropped to zero after tool and instructional refinements. Despite different adoption levels (93\% vs. 50\% of teams using the tool), responsiveness was stable: 32\% (2023) and 33\% (2024) of successfully AI-reviewed PRs were followed by subsequent commits on the same PR. Qualitatively, students used the LLM's structured comments to focus reviews and discuss code quality, while guidance reduced over-reliance. We contribute: (i) an in-workflow design for an AI reviewer that scaffolds learning while mitigating cognitive offloading; (ii) a repeated cross sectional comparison across two cohorts in authentic settings; (iii) a mixed-methods analysis combining objective GitHub metrics with student self-reports; and (iv) evidence-based pedagogical recommendations for responsible, student-led AI-assisted review.

The use of Machine Learning (ML) for data-driven decision-making often relies on access to sensitive datasets, which introduces privacy challenges. Traditional encryption methods protect data at rest or in transit but fail to secure it during processing, exposing it to unauthorized access. Homomorphic encryption emerges as a transformative solution, enabling computations on encrypted data without decryption, thus preserving confidentiality throughout the ML pipeline. This paper addresses the challenge of training ML models on encrypted data while maintaining accuracy and efficiency by proposing a proof-of-concept for a privacy-preserving framework that leverages Cheon-Kim-Kim-Song (CKKS) for approximate real-number arithmetic. Also, it demonstrates the feasibility of training K-Nearest Neighbors (KNN) and linear regression models on encrypted data, and evaluates encrypted inference for a basic Multilayer Perceptron (MLP) architecture. Experimental results show that models trained under Homomorphic encryption achieve performance metrics comparable to plaintext-trained models, validating the approach. However, challenges such as computational overhead, noise management, and limited support for non-polynomial operations persist. This work lays the groundwork for broader adoption of privacy-preserving ML in real-world applications, balancing security with computational feasibility.

Human-imitated speech poses a greater challenge than AI-generated speech for both human listeners and automatic detection systems. Unlike AI-generated speech, which often contains artifacts, over-smoothed spectra, or robotic cues, imitated speech is produced naturally by humans, thereby preserving a higher degree of naturalness that makes imitation-based speech forgery significantly more challenging to detect using conventional acoustic or cepstral features. To overcome this challenge, this study proposes an auditory perception-based Spectro-Temporal Modulation (STM) representation framework for human-imitated speech detection. The STM representations are derived from two cochlear filterbank models: the Gammatone Filterbank (GTFB), which simulates frequency selectivity and can be regarded as a first approximation of cochlear filtering, and the Gammachirp Filterbank (GCFB), which further models both frequency selectivity and level-dependent asymmetry. These STM representations jointly capture temporal and spectral fluctuations in speech signals, corresponding to changes over time in the spectrogram and variations along the frequency axis related to human auditory perception. We also introduce a Segmental-STM representation to analyze short-term modulation patterns across overlapping time windows, enabling high-resolution modeling of temporal speech variations. Experimental results show that STM representations are effective for human-imitated speech detection, achieving accuracy levels close to those of human listeners. In addition, Segmental-STM representations are more effective, surpassing human perceptual performance. The findings demonstrate that perceptually inspired spectro-temporal modeling is promising for detecting imitation-based speech attacks and improving voice authentication robustness.

Accurate long-term time series forecasting (LTSF) requires the capture of complex long-range dependencies and dynamic periodic patterns. Recent advances in frequency-domain analysis offer a global perspective for uncovering temporal characteristics. However, real-world time series often exhibit pronounced cross-domain heterogeneity where variables that appear synchronized in the time domain can differ substantially in the frequency domain. Existing frequency-based LTSF methods often rely on implicit assumptions of cross-domain homogeneity, which limits their ability to adapt to such intricate variability. To effectively integrate frequency-domain analysis with temporal dependency learning, we propose AdaMamba, a novel framework that endogenizes adaptive and context-aware frequency analysis within the Mamba state-space update process. Specifically, AdaMamba introduces an interactive patch encoding module to capture inter-variable interaction dynamics. Then, we develop an adaptive frequency-gated state-space module that generates input-dependent frequency bases, and generalizes the conventional temporal forgetting gate into a unified time-frequency forgetting gate. This allows dynamic calibration of state transitions based on learned frequency-domain importance, while preserving Mamba's capability in modeling long-range dependencies. Extensive experiments on seven public LTSF benchmarks and two domain-specific datasets demonstrate that AdaMamba consistently outperforms state-of-the-art methods in forecasting accu racy while maintaining competitive computational efficiency. The code of AdaMamba is available at https://github.com/XDjiang25/AdaMamba.

Frontier models push the boundaries of what is learnable at extreme computational costs, yet distillation via sampling reasoning traces exposes closed-source frontier models to adversarial third parties who can bypass their guardrails and misappropriate their capabilities, raising safety, security, and intellectual privacy concerns. To address this, there is growing interest in building antidistillation methods, which aim to poison reasoning traces to hinder downstream student model learning while maintaining teacher performance. However, current techniques lack theoretical grounding, requiring either heavy fine-tuning or access to student model proxies for gradient based attacks, and often lead to a significant teacher performance degradation. In this work, we present a theoretical formulation of antidistillation as a Stackelberg game, grounding a problem that has so far largely been approached heuristically. Guided by the desired design properties our formulation reveals, we propose \texttt{TraceGuard}, an efficient, post-generation black-box method to poison sentences with high importance for teacher reasoning. Our work offers a scalable solution to share model insights safely, ensuring that the advancement of reasoning capabilities does not come at the cost of intellectual privacy or AI safety alignment.

Diffusion language models expose an explicit denoising trajectory, making it possible to ask when different kinds of information become measurable during generation. We study three independent 32-step runs of LLaDA-8B-Base on masked WikiText-103 text, each with 1{,}000 probe-training sequences and 200 held-out evaluation sequences. From saved trajectories, we derive four temporal measurements: token commitment; linear recoverability of part-of-speech (POS), coarse semantic category, and token identity; confidence and entropy dynamics; and sensitivity under mid-trajectory re-masking. Across seeds, the same ordering recurs: content categories stabilize earlier than function-heavy categories, POS and coarse semantic labels remain substantially more linearly recoverable than exact lexical identity under our probe setup, uncertainty remains higher for tokens that ultimately resolve incorrectly even though late confidence becomes less calibrated, and perturbation sensitivity peaks in the middle of the trajectory. A direct/collateral decomposition shows that this peak is overwhelmingly local to the perturbed positions themselves. In this LLaDA+WikiText setting, denoising time is therefore a useful analysis axis: under our measurements, coarse labels are recovered earlier and more robustly than lexical identity, trajectory-level uncertainty tracks eventual correctness, and mid-trajectory states are the most intervention-sensitive.

Semantic Communication (SC) backdoor attacks aim to utilize triggers to manipulate the system into producing predetermined outputs via backdoored shared knowledge. Current SC backdoors adopt monomorphic paradigms with single attack target, which suffers from limited attack diversity, efficiency, and flexibility in heterogeneous downstream scenarios. To overcome the limitations, we propose SemBugger, a polymorphic SC backdoor. By dynamically adjusting the trigger intensity, SemBugger finely-grained controls over the SC knowledge to generate diverse malicious results from the system. Specifically, SemBugger is realized through a multi-effect poisoning-training framework. It introduces graded-intensity triggers to poison training data and optimizes SC systems with hierarchical malicious loss. The trained system's knowledge dynamically adapts to trigger intensity in inputs to yield target outputs, all while preserving transmission fidelity for benign samples. Moreover, to augment SC security, we propose a provable robustness defense that resists SemBugger's homogeneous attacks through a controlled noise mechanism. It operates via strategically adding noise in SC inputs, and we formally provide a theoretical lower bound on the defense efficacy. Experiments across diverse SC models and benchmark datasets indicate that SemBugger attains high attack efficacy while maintaining the regular functionality of SC systems. Meanwhile, the designed defense effectively neutralizes SemBugger attacks.

Organisations operating within information-intensive environments face intensifying pressure to formalise the governance of information security. The ISO/IEC 27001:2022 standard provides a globally recognised framework for establishing, implementing, maintaining, and continually improving an Information Security Management System (ISMS). This article analyses the procedural architecture deployed in a financial-technology organisation's ISMS, examining eight core operational procedures: IT Risk Assessment and Treatment, User Code of Conduct, Password Policy, Access Control, Internet Access, Physical Security, Backup and Restore Management, and Nonconformity Root Cause Analysis and Corrective Action. Drawing on documented internal training materials, the article investigates how each procedure operationalises the requirements of Annex~A controls and Clauses~6--10 of ISO~27001:2022. The paper evaluates the CIA Triad as a unifying evaluation criterion, the twelve-step risk assessment methodology, role-based responsibility allocation, and the interplay between corrective action governance and continual improvement. The findings suggest that a tightly integrated, multi-layered procedural hierarchy, supported by clear accountability structures and measurable risk metrics, constitutes the foundation of an effective ISMS implementation in financial-technology operating environments.

This paper investigates the deep learning optimization problem with softmax cross-entropy loss. We propose a layer separation strategy to alleviate the strong nonconvexity encountered during training deep networks. For cross-entropy models with fully connected and convolutional neural networks, we introduce auxiliary variables associated with hidden layer outputs and construct corresponding layer separation models, which decompose the original deeply nested optimization problem into a sequence of more manageable subproblems. We also conduct theoretical analyses, proving that the new layer separation loss provides an upper bound for the original cross-entropy loss. Moreover, we design alternating minimization algorithms and prove that, under appropriate conditions, these algorithms exhibit decreasing properties of the loss function. Numerical experiments validate the effectiveness of the proposed methods and indicate improved optimization behavior, especially for fully connected and convolutional neural networks.

Spiking Neural Networks (SNNs) offer a biologically inspired foundation for low-power, event-driven intelligence, yet their direct on-chip supervised training remains a key hardware challenge. This paper presents a multiplication-free, spike-time-based learning algorithm specifically designed for efficient FPGA realization. The proposed approach eliminates floating-point arithmetic and explicit gradient storage, enabling a fully event-driven, digital training pipeline. Implemented on a Xilinx Artix-7 FPGA, the architecture achieves high operating speed and minimal resource usage while maintaining competitive accuracy. These results demonstrate that the learning algorithm effectively maps onto reconfigurable hardware, achieving both computational and energy efficiency. Software simulations further validate scalability, with 96.5\% and 84.8\% accuracy on MNIST and Fashion-MNIST. With its spike-driven and multiplier-free operation, the proposed framework delivers a practical and scalable hardware solution for real-time, on-chip SNN learning in edge environments.

Memes convey meaning through the interaction of visual and textual signals, often combining humor, irony, and offense in subtle ways. Detecting harmful or sensitive content in memes requires accurate modeling of these multimodal cues. Existing CLIP-based approaches rely on static fusion, which struggles to capture fine grained dependencies between modalities. We propose DARC-CLIP, a CLIP-based framework for adaptive multimodal fusion with a hierarchical refinement stack. DARC-CLIP introduces Adaptive Cross-Attention Refiners to for bidirectional information alignment and Dynamic Feature Adapters for task-sensitive signal adaptation. We evaluate DARC-CLIP on the PrideMM benchmark, which includes hate, target, stance, and humor classification, and further test generalization on the CrisisHateMM dataset. DARC-CLIP achieves highly competitive classification accuracy across tasks, with significant gains of +4.18 AUROC and +6.84 F1 in hate detection over the strongest baseline. Ablation studies confirm that ACAR and DFA are the main contributors to these gains. These results show that adaptive cross-signal refinement is an effective strategy for multimodal content analysis in socially sensitive classification.

Existing large-dimensional theory for spectral algorithms resolves either the optimally tuned point or the interpolation limit, but leaves the under-regularized regime unexplored. We study the learning curve and benign overfitting of spectral algorithms in the large-dimensional setting where the sample size and dimension are of comparable order, i.e., $n \asymp d^γ$ for some $γ>0$. We first consider inner-product kernels on the sphere $\mathbb{S}^{d-1}$ and establish a sharp asymptotic characterization of the excess risk across the full regularization path under various source conditions $s \geq 0$, where $s$ measures the relative smoothness of the regression function. Our results reveal that the learning curve is not simply U-shaped but instead consists of three distinct regimes: over-regularized, under-regularized, and interpolation regimes. This characterization allows us to fully capture the benign overfitting phenomenon, demonstrating that benign overfitting arises consistently across both the under-regularized and interpolation regimes whenever $s$ is positive but no larger than a critical threshold. We further show that, in the sufficiently regularized regime, the kernel learning curve is recovered by an associated sequence model. Finally, we extend the learning-curve analysis to large-dimensional KRR for a class of kernels on general domains in $\mathbb{R}^d$ whose low-degree eigenspaces satisfy spectral-scaling and hyper-contractivity conditions.

Can large language model agents discover hidden safety objectives through experience alone? We introduce EPO-Safe (Experiential Prompt Optimization for Safe Agents), a framework where an LLM iteratively generates action plans, receives sparse binary danger warnings, and evolves a natural language behavioral specification through reflection. Unlike standard LLM reflection methods that rely on rich textual feedback (e.g., compiler errors or detailed environment responses), EPO-Safe demonstrates that LLMs can perform safety reasoning from a strictly impoverished signal in structured, low-dimensional environments: the agent never observes the hidden performance function $R^*$, only a single bit per timestep indicating that an action was unsafe. We evaluate on five AI Safety Gridworlds (Leike et al., 2017) and five text-based scenario analogs where visible reward $R$ may diverge from $R^*$. EPO-Safe discovers safe behavior within 1-2 rounds (5-15 episodes), producing human-readable specifications with correct explanatory hypotheses about hazards (e.g., "X cells are directionally hazardous: entering from the north is dangerous"). Critically, we show that standard reward-driven reflection actively degrades safety: agents reflecting on reward alone use the loop to justify and accelerate reward hacking, proving that reflection must be paired with a dedicated safety channel to discover hidden constraints. We further evaluate robustness to noisy oracles: even when 50% of non-dangerous steps produce spurious warnings, mean safety performance degrades by only 15% on average, though sensitivity is environment-dependent, as cross-episode reflection naturally filters inconsistent signals. Each evolved specification functions as an auditable set of grounded behavioral rules discovered autonomously through interaction, rather than authored by humans as in Constitutional AI (Bai et al., 2022).

Deploying proprietary Deep Neural Networks (DNNs) on commodity edge devices demands hardware-backed Digital Rights Management (DRM) capable of withstanding both software-level and physical adversaries. In Unified Memory Architecture (UMA) systems, the host CPU and Neural Processing Unit (NPU) share physical DRAM, leaving plaintext model weights directly readable by a compromised OS kernel. Existing defenses fail in this constrained setting: trusted execution environments monopolize scarce memory with permanently reserved regions, while full-memory encryption operates at page granularity. This forces the system to fetch massive 4 KB memory pages for sub-page tensor tiles, severely crippling bandwidth. We present Tessera, a reference architecture for inline, cache-line granularity weight decryption on UMA edge accelerators. The design intercepts 64-byte AXI bursts, computing AES-256-CTR keystreams in parallel with DRAM fetches. This streams plaintext directly into isolated NPU SRAM, creating a transient memory footprint confined to the active tile and eliminating the need for permanent memory carve-outs. Measurements across three distinct SoC platforms demonstrate that this parallelization hides cryptographic latency behind standard DRAM fetch times, a condition that holds even under worst-case timing variations. Consequently, Tessera is projected to achieve 98.4\% of the theoretical memory bandwidth ceiling (a mere 1.6\% overhead). Across standard vision and language models, page-level memory encryption suffers up to a 32x bandwidth penalty, whereas Tessera maintains an optimal 1x footprint for all layer geometries. Finally, Tessera neutralizes major UMA-specific attack vectors -- including physical DRAM extraction, rogue DMA, and compute hijacking -- and formally prevents plaintext leakage across sparse tensors.

Current video moment retrieval excels at action-centric tasks but struggles with narrative content. Models can see \textit{what is happening} but fail to reason \textit{why it matters}. This semantic gap stems from the lack of \textbf{Theory of Mind (ToM)}: the cognitive ability to infer implicit intentions, mental states, and narrative causality from surface-level observations. We introduce \textbf{StoryTR}, the first video moment retrieval benchmark requiring ToM reasoning, comprising 8.1k samples from narrative short-form videos (shorts/reels). These videos present an ideal testbed. Their high information density encodes meaning through subtle multimodal cues. For instance, a glance paired with a sigh carries entirely different semantics than the glance alone. Yet multimodal perception alone is insufficient; ToM is required to decode that a character ``smiling'' may actually be ``concealing hostility.'' To teach models this reasoning capability, we propose an \textbf{Agentic Data Pipeline} that generates training data with explicit three-tier ToM chains (intent decoding, narrative reasoning, boundary localization). Experiments reveal the severity of the reasoning gap: Gemini-3.0-Pro achieves only 0.53 Avg IoU on StoryTR. However, our 7B \textbf{Shorts-Moment} model, trained on ToM-guided data, improves +15.1\% relative IoU over baselines, demonstrating that \textit{narrative reasoning capability matters more than parameter scale}.

Delayed feedback poses a core challenge for online CVR prediction, forcing a trade-off between label accuracy and data freshness. Existing methods address this through delay modeling or sample reweighting, yet neglect how post-click behaviors evolve over the observation period. To overcome this limitation, we formalize this evolution as feedback trajectory and propose TRACE. Instead of forcing hard labels on unrevealed samples, our method evaluates how well the accumulated feedback status aligns with conversion versus non-conversion, dynamically refining posteriors without waiting for final outcomes. To counteract early-stage trajectory sparsity, we further design a reliability-gated retrospective completer that leverages full-lifecycle data to provide adaptive posterior guidance for unrevealed samples. Extensive experiments validate TRACE's superiority over state-of-the-art baselines and confirm the retrospective completion module as a model-agnostic enhancer for existing systems. Our code is available at https://github.com/LunaZhangxy/TRACE.

Analog circuit design relies heavily on reusing existing intellectual property (IP), yet searching across heterogeneous representations such as SPICE netlists, schematics, and functional descriptions remains challenging. Existing methods are largely limited to exact matching within a single modality, failing to capture cross-modal semantic relationships. To bridge this gap, we present AnalogRetriever, a unified tri-modal retrieval framework for analog circuit search. We first build a high-quality dataset on top of Masala-CHAI through a two-stage repair pipeline that raises the netlist compile rate from 22\% to 100\%. Built on this foundation, AnalogRetriever encodes schematics and descriptions with a vision-language model and netlists with a port-aware relational graph convolutional network, mapping all three modalities into a shared embedding space via curriculum contrastive learning. Experiments show that AnalogRetriever achieves an average Recall@1 of 75.2\% across all six cross-modal retrieval directions, significantly outperforming existing baselines. When integrated into the AnalogCoder agentic framework as a retrieval-augmented generation module, it consistently improves functional pass rates and enables previously unsolved tasks to be completed. Our code and dataset will be released.

Perturbing a deterministic $n$-dimensional matrix with small Gaussian noise is a cornerstone of smoothed analysis of algorithms [Spielman and Teng, JACM 2004], as it reduces the condition number of the input to $O(n)$, and with it the complexity of many matrix algorithms. However, when deployed algorithmically, these perturbations are expensive due to the cost of generating and storing $n^2$ Gaussian random variables. We propose a perturbation that requires generating and storing $O(n)$ random numbers in $O(\log n)$ bits of precision, and reduces the condition number of any deterministic matrix to $O(n)$, matching Gaussian perturbations. Our result in particular implies a better complexity for the perturbed conjugate gradient algorithm, showing that we can solve an $n\times n$ linear system in linear space to within an arbitrarily small constant backward error using $O(n)$ matrix-vector products. In our construction, we introduce the concept of a pattern matrix, which is a dense deterministic matrix that maps all sparse vectors into dense vectors, and we combine it with a sparse perturbation whose entries are dependent and located in a non-uniform fashion. In order to analyze this construction, we develop new techniques for lower bounding the smallest singular value of a random matrix with dependent entries.

Large language model-based agents have recently emerged as powerful approaches for solving dynamic and multi-step tasks. Most existing agents employ planning mechanisms to guide long-term actions in dynamic environments. However, current planning approaches face a fundamental limitation that they operate at a fixed granularity level. Specifically, they either provide excessive detail for simple tasks or insufficient detail for complex ones, failing to achieve an optimal balance between simplicity and complexity. Drawing inspiration from the principle of \textit{progressive refinement} in cognitive science, we propose \textbf{AdaPlan-H}, a self-adaptive hierarchical planning mechanism that mimics human planning strategies. Our method initiates with a coarse-grained macro plan and progressively refines it based on task complexity. It generates self-adaptive hierarchical plans tailored to the varying difficulty levels of different tasks, which can be optimized by imitation learning and capability enhancement. Experimental results demonstrate that our method significantly improves task execution success rates while mitigating overplanning at the planning level, providing a flexible and efficient solution for multi-step complex decision-making tasks. To contribute to the community, our code and data will be made publicly available at https://github.com/import-myself/AHP.

Automating repository-level software engineering tasks is a foundational challenge for autonomous code agents, largely due to the difficulty of configuring executable environments. However, manual configuration remains a labor-intensive bottleneck, necessitating a transition toward fully automated environment configuration. Existing approaches often rely on pre-defined artifacts or are restricted to specific programming languages, limiting their applicability to real-world repositories. In this paper, we first propose RAT (RunAnyThing), a language-agnostic framework for automated environment configuration on arbitrary repositories. RAT features a multi-stage pipeline that integrates semantic initialization, a planning mechanism, specialized toolset, and a robust sandbox for configuration. Furthermore, to enable rigorous evaluation, we propose RATBench, a benchmark that reflects the the distribution and heterogeneity of real-world repositories. Extensive experiments demonstrate that RAT achieves state-of-the-art performance, improving the Environment Setup Success Rate (ESSR) by an average of 29.6% over strong baselines.

This work introduces DyABD, a novel and complex benchmark dataset of dynamic abdominal MRIs from patients with abdominal hernias and associated high quality abdominal muscle annotations. DyABD is the first-of-its-kind in four key ways; (1) it proposes the first abdominal muscle segmentation task, (2) the dynamic MRIs are acquired whilst the patients perform various exercises, introducing extreme anatomical variability, making it one of the most challenging segmentation datasets to date, (3) it includes both pre and post corrective MRIs and (4) DyABD promotes clinical research into the high recurrence rates of abdominal hernias. Beyond dataset introduction, this work provides a comprehensive evaluation of the generalisation capabilities of existing segmentation models across Supervised, Few Shot and Zero Shot paradigms on the unseen DyABD dataset. This work reveals that there is still room for substantial improvement in the field of medical image segmentation, with the majority of techniques achieving a Dice Coefficient of 0.82. This work therefore sheds light on the true progress of the field and redefines the benchmark for progress in medical image segmentation.

We propose a unified fractional regularization framework for sparse signal recovery based on the $\ell_1/\ell_p^q$ model. Our main theoretical contribution is the characterization of the equivalence between the first-order stationary points of the $\ell_1/\ell_p^q$ formulation and the subtractive $\ell_1 - α\ell_p$ model, providing a unified perspective on these nonconvex regularizers. In addition, we establish a new sufficient recovery condition under the Restricted Isometry Property (RIP), showing that the framework's robustness even under high-coherence sensing matrices. To solve the resulting problem, we develop a majorization-minimization (MM) algorithm and prove its convergence via the Kurdyka-Lojasiewicz (KL) property. Numerical experiments on different sensing matrices and MRI reconstruction demonstrate that the proposed approach consistently outperforms existing methods.

AI incident reporting requirements are emerging in regulation and policy, yet no operational criteria exist for determining when a detected AI incident warrants escalation beyond national handling to international coordination. This paper proposes an escalation framework to address this gap, intended as a common reference point across jurisdictions that enables aligned escalation while preserving flexibility in how actors respond within their own legal and policy contexts. We review SB 53, the EU AI Act, the GPAI Code of Practice, and incident frameworks from other industries to derive eight criteria for assessing whether an incident warrants escalation, translated into a sequential flowchart with gated decision points and threshold checks. For each criterion, we map how it interplays with these regulatory frameworks, identifying where their design choices support or undermine effective detection. We test the framework against ten documented AI incidents and structured variants to identify where criteria under-detect or misclassify incidents in practice. We find three design patterns that may lead to systematic under-detection in regimes where model developers are responsible for escalation: a. where escalation requires confirmed harm, events such as model weight exfiltration risk detection only after severe, irreversible harm has propagated; b. where incidents are assessed individually, systemic harms emerging from accumulation risk being under-detected; and c. where thresholds align with legal instruments rather than quantitatively testable terms, criteria risk being impractical to apply under time pressure. We also find that escalation rules are only one component of a broader framework: the underlying definitions against which thresholds are set, and the data available to the responsible actor, create interdependencies that can themselves drive under-detection.

Monitoring human activity in indoor environments is important for applications such as facility management, safety assessment, and space utilization analysis. While mobile robot teams offer the potential to actively improve observation quality, existing multi-robot monitoring and active perception approaches typically rely on coverage or visitation based objectives that are weakly aligned with the accuracy requirements of human-centric monitoring tasks. In this work, we formulate cooperative active observation as a decentralized control problem in which multiple robots adjust their motion to directly optimize monitoring accuracy under partial observability. We propose a learning-based framework for cooperative policies from decentralized observations using multi-agent reinforcement learning (MARL), supported by an architecture that handles variable numbers of humans and temporal dependencies. Simulation results across diverse indoor environments and monitoring tasks show that the proposed approach consistently outperforms classical coverage, persistent monitoring, and learning-free multi-robot baselines, while remaining robust to changes in the number of observed humans.

LLM-as-a-Judge has become the dominant paradigm for evaluating language model outputs, yet LLM judges exhibit systematic biases that compromise evaluation reliability. We present a comprehensive empirical study comparing nine debiasing strategies across five judge models from four provider families (Google, Anthropic, OpenAI, Meta), three benchmarks (MT-Bench n=400, LLMBar n=200, custom n=225), and four bias types. Our key findings: (1) Style bias is the dominant bias (0.76-0.92 across all models), far exceeding position bias (<= 0.04), yet has received minimal research attention. (2) All models show a conciseness preference on expansion pairs, but truncation controls confirm they correctly distinguish quality from length (0.92-1.00 accuracy), suggesting quality-sensitive evaluation rather than a simple length bias. (3) Debiasing is beneficial but model-dependent: the combined budget strategy significantly improves Claude Sonnet 4 by +11.2 pp (p < 0.0001), with directionally positive trends for other models. Only 2 of 20 non-baseline configurations show decreased agreement. We release our evaluation framework, controlled dataset, and all experimental artifacts at https://github.com/sksoumik/llm-as-judge.

In this work, we developed and tested 3 techniques for vector quantization (VQ) based model weight compression. To mitigate codebook collapse and enable end-to-end training, we adopted cosine similarity-based assignment. Building on ideas from attention-based formulations in Differentiable K-Means (DKM), we further improved this approach by using cosine similarity for assignment combined with top-1 sampling and a straight-through estimator, thereby eliminating the need for weighted-average reconstruction. Finally, we investigated the use of differentiable neural architecture search (NAS) to adaptively select layer-wise quantization configurations, further optimizing the compression process. Although our method does not consistently outperform existing approaches across all quantization levels, it provides useful insights into the design trade-offs and behaviors of VQ-based model compression methods.

Software-driven O-RAN architectures enable rapid innovation through frequent, independent updates to virtualized components. However, attributing performance variations to specific software changes is challenging due to the stochastic nature of wireless systems, where channel conditions, interference, and hardware variability confound analysis. Traditional threshold-based monitoring and manual troubleshooting do not scale with modern software evolution. This paper presents RANalyzer, an automated test analysis framework that quantifies the performance impact of software updates beyond what can be explained by wireless channel conditions. RANalyzer combines LLM-assisted semantic extraction with residuals analysis. The first categorizes code changes by affected protocol layers and functional components, while the second provides insights on the effect of load, channel, or code changes on the test performance. We contribute an extensive dataset collected over more than two years of continuous over-the-air testing on an experimental O-RAN testbed, comprising over 8,600 automated tests across 69 releases of the OAI stack. By modeling expected performance and interpreting deviations as software-induced effects, we identify degraded instances attributable to code changes and correlate them with specific change categories. The framework can be integrated into CI/CD/CT pipelines for automated, continuous evaluation of software updates at scale.

Most recent state-of-the-art (SOTA) large language models (LLMs) use Mixture-of-Experts (MoE) architectures to scale model capacity without proportional per-token compute, enabling higher-quality outputs at manageable serving costs. However, MoE inference at scale is fundamentally bottlenecked by expert load imbalance and inefficient token routing, especially in multi-node deployments where tokens are not guaranteed to be routed to local experts, resulting in significant inter-node all-to-all communication overhead. To systematically characterize these challenges, we profile SOTA open-source MoE models, including Llama 4 Maverick, DeepSeek V3-671B, and Qwen3-230B-A22B, on various datasets and collected over 100k real expert activation traces. Upon studying the expert activation patterns, we uncover various persistent properties across all the frontier MoE models: variable expert load imbalance, domain-specific expert activation where expert popularity shifts across task families (code, math, chat, general), and a strong correlation between prefill and decode expert activations. Motivated by these findings, we propose workload-aware micro-batch grouping and an expert placement strategy to maximize token locality to the destination expert, thereby reducing inter-node communication. Across models and datasets, these optimizations help reduce all2all communication data up to 20, resulting in lower MoE decode latency and better accelerator utilization.

The emerging threat of AR-LLM-based Social Engineering (AR-LLM-SE) attacks (e.g. SEAR) poses a significant risk to real-world social interactions. In such an attack, a malicious actor uses Augmented Reality (AR) glasses to capture a target visual and vocal data. A Large Language Model (LLM) then analyzes this data to identify the individual and generate a detailed social profile. Subsequently, LLM-powered agents employ social engineering strategies, providing real-time conversation suggestions, to gain the target trust and ultimately execute phishing or other malicious acts. Despite its potential, the practical application of AR-LLM-SE faces two major bottlenecks, (1) Cold-start personalization, Current Retrieval-Augmented Generation (RAG) methods introduce critical delays in the earliest turns, slowing initial profile formation and disrupting real-time interaction, (2) Static Attack Strategies, Existing approaches rely on fixed-stage, handcrafted social engineering tactics that lack foundation in established psychological theory. To address these limitations, we propose PhySE, a novel framework with two core innovations, (1) VLM-Based SocialContext Training, To eliminate profiling delays, we efficiently pre-train a Visual Language Model (VLM) with social-context data, enabling rapid, on-the-fly profile generation, (2) Adaptive Psychological Agent, We introduce a psychological LLM that dynamically deploys distinct classes of psychological strategies based on target response, moving beyond static, handcrafted scripts. We evaluated PhySE through an IRB-approved user study with 60 participants, collecting a novel dataset of 360 annotated conversations across diverse social scenarios.

Today's high-performance architectures are increasingly constrained by data movement latency and energy overhead, as the slowdown of single-core performance scaling coincides with the rise of highly data-intensive workloads. In-memory architectures have emerged as a complementary solution to conventional von Neumann systems by alleviating memory bandwidth bottlenecks, exploiting massive concurrency, and mitigating excessive data movement between memory and processing units. This study proposes a parallel in-memory stochastic computing (SC) architecture that implements an end-to-end computation pipeline within Magnetic Tunnel Junction (MTJ)-based memory augmented with logic-in-memory (LIM) capabilities. By leveraging the inherent stochasticity and write-read characteristics of MTJ devices, the proposed architecture enables a fully parallel and deterministic conversion of binary operands into probabilistic bit-streams, eliminating the need for energy-intensive external random number generation circuitry. These bit-streams are processed by parallel stochastic arithmetic units integrated directly within the memory arrays to efficiently implement core arithmetic and transcendental functions with minimal hardware complexity and inherent noise tolerance. The resulting stochastic outputs can be either reused as an input of future stochastic processing or converted back to binary form using parallel accumulation mechanisms and stored in the MTJ memory. By tightly integrating data storage, bit-stream generation, and computation within a unified in-memory fabric, the proposed design maximizes memory-level parallelism while substantially minimizing data movement.

Video Question Answering (VideoQA) demands models that jointly reason over spatial, temporal, and linguistic cues. However, the task's inherent complexity often requires multi-step reasoning that current large multimodal models (LMMs) perform implicitly, leaving their internal decision process opaque. In contrast, large reasoning models (LRMs) explicitly generate intermediate logical steps that enhance interpretability and can improve multi-hop reasoning accuracy. Yet, these models are not designed for native video understanding, as they typically rely on static frame sampling. We propose UpstreamQA, a modular framework that disentangles and evaluates core video reasoning components through explicit upstream reasoning modules. Specifically, we employ multimodal LRMs to perform object identification and scene context generation before passing enriched reasoning traces to downstream LMMs for VideoQA. We evaluate UpstreamQA on the OpenEQA and NExTQA datasets using two LRMs (o4-mini, Gemini 2.5 Pro) and two LMMs (GPT-4o, Gemini 2.5 Flash). Our results demonstrate that introducing explicit reasoning can significantly boost performance and interpretability of downstream VideoQA, but can also lead to performance degradation when baseline performance is sufficiently high. Overall, UpstreamQA offers a principled framework for combining explicit reasoning and multimodal understanding, advancing both performance and diagnostic transparency in VideoQA in several scenarios.

Emerging AR-LLM-based Social Engineering attack (e.g., SEAR) is at the edge of posing great threats to real-world social life. In such AR-LLM-SE attack, the attacker can leverage AR (Augmented Reality) glass to capture the image and vocal information of the target, using the LLM to identify the target and generate the social profile, using the LLM agents to apply social engineering strategies for conversation suggestion to win the target trust and perform phishing afterwards. Current defensive approaches, such as role-based access control or data flow tracking, are not directly applicable to the convergent AR-LLM ecosystem (considering embedded AR device and opaque LLM inference), leaving an emerging and potent social engineering threat that existing privacy paradigms are ill-equipped to address. This necessitates a shift beyond solely human-centric measures like legislation and user education toward enforceable vendor policies and platform-level restrictions. Realizing this vision, however, faces significant technical challenges: securing resource-constrained AR-embedded devices, implementing fine-grained access control within opaque LLM inferences, and governing adaptive interactive agents. To address these challenges, we present UNSEEN, a coordinated cross-stack defense that combines an AR ACL (Access Control Layer) for identity-gated sensing, F-RMU-based LLM unlearning for sensitive profile suppression, and runtime agent guardrails for adaptive interaction control. We evaluate UNSEEN in an IRB-approved user study with 60 participants and a dataset of 360 annotated conversations across realistic social scenarios.

Distributed GNN training is dominated by remote feature fetching, which can be very costly. Multi-hop neighborhood sampling crosses partition boundaries and triggers fine-grained RPCs whose fixed initiation cost and GPU-stall latency waste energy. Prior systems try to reduce this overhead with presampling and static caching, but cache policies cannot react to runtime network variation. We show that under time-varying congestion, static caching can increase energy by up to 45% because a fixed rebuild schedule is insufficient. We present GreenDyGNN, which formulates cache window management as a sequential decision problem. GreenDyGNN performs intra-epoch cache rebuilds and uses a Double-DQN agent, trained in a calibrated simulator with domain-randomized congestion, to adapt rebuild window size and per-owner cache allocation at each boundary. An asynchronous double-buffered pipeline makes adaptation effectively free. Under congestion, GreenDyGNN cuts total energy by up to 43% over Default DGL and 4-24% over the best static policy, while closely matching the optimum under clean conditions.

Early detection and classifying brain tumors using Magnetic Resonance Imaging (MRI) images is highly important but difficult to extract in medical images. Convolutional Neural Networks (CNNs) are good at capturing both local texture and spatial information whereas Vision Transformers (ViTs) are good at capturing long-range global dependencies. We propose a new hybrid architecture that combines a SqueezeNet-style CNN branch with a MobileViT-style global transformer branch, through an Adaptive Attention Gate mechanism, in this paper. The gate learns dynamically per-sample, per-feature weights to weight the contribution of each branch, allowing context-sensitive merging of local and global representations. The proposed model has a test accuracy of 97.60, a precision of 97.30, a recall of 97.50, an F1-score of 97.40, and a macro-average area under the curve (AUC) of 0.9946 with a trained and evaluated on the Brain Tumor MRI Dataset (Kaggle). These scores are higher than single CNN and ViT baselines, and current competitive fusion methods, showing that dynamic feature weighting is an effective way to classify medical images.

Natural-language variation poses a key challenge in Lean autoformalization: semantically equivalent paraphrases of the same theorem statements can induce divergent formal outputs, yet it remains unclear whether this variation reflects semantic disagreements or shallower failures. We investigate this question in Lean 4 using 60 deterministic paraphrase rules applied to ProofNet\# and miniF2F. Across four GPT-family models and three open-weight 7B autoformalizers, we find that the observed paraphrase sensitivity reflects compilation-boundary failures rather than semantic divergence among successful formalizations. In particular, when both baseline and perturbed outputs compile, paired predictions are semantically equivalent under BEq+ and structurally near-identical under GTED. By contrast, paraphrasing substantially affects whether outputs compile, with failure modes varying across datasets and perturbation classes. Our results suggest that future training-time interventions should target the compile boundary rather than the semantic layer, and that benchmarks should separate compile-conditional equivalence from surface consistency.

Accurate molecular representations are critical for drug discovery, and a central challenge lies in capturing the chemical environment of molecular fragments, as key interactions, such as H-bond and π stacking, occur only under specific local conditions. Most existing approaches represent molecules as atom-level graphs; however, atom-level representations can hardly express higher-order chemical context (e.g., stereochemistry, lone pairs, conjugation). Fragment-based methods (e.g., principal subgraph, predefined functional groups) fail to preserve essential information such as chirality, aromaticity, and ionic states. This work addresses these limitations from two aspects. (i) OverlapBPE tokenization. We propose a novel data-driven molecule tokenization method. Unlike existing approaches, our method allows overlapping fragments, reflecting the inherently fuzzy boundaries of small-molecule substructures and, together with enriched chemical information at the token level, thereby preserving a more complete chemical context. (ii) h-MINT model. OverlapBPE induces many-to-many atom-fragment mappings, which necessitate a new hierarchical architecture. We therefore develop a hierarchical molecular interaction network capable of jointly modeling interactions at both atom and fragment levels. By supporting fragment overlaps, the model naturally accommodates the many-to-many atom-fragment mappings introduced by the OverlapBPE scheme. Extensive evaluation against state-of-the-art methods shows our method improves binding affinity prediction by 2-4% Pearson/Spearman correlation on PDBBind and LBA, enhances virtual screening by 1-3% in key metrics on DUD-E and LIT-PCBA, and achieves the best overall HTS performance on PubChem assays. Further analysis demonstrates that our method effectively captures interactive information while maintaining good generalization.

Large language models (LLMs) can still be jailbroken into producing harmful outputs despite safety alignment. Existing attacks show this vulnerability, but not the internal mechanisms that cause it. This study asks whether jailbreak success is driven by identifiable internal features rather than prompts alone. We propose a three-stage pipeline for Gemma-2-2B using the BeaverTails dataset. First, we extract concept-aligned tokens from adversarial responses via subspace similarity. Second, we apply three feature-grouping strategies (cluster, hierarchical-linkage, and single-token-driven) to identify SAE feature subgroups for the aligned tokens across all 26 model layers. Third, we steer the model by amplifying the top features from each identified subgroup and measure the change in harmfulness score using a standardized LLM-judge scoring protocol. In all three approaches, the features in the layers [16-25] were relatively more vulnerable to steering. All three methods confirmed that mid to later layer feature subgroups are more responsible for unsafe outputs. These results provide evidence that the jailbreak vulnerability in Gemma-2-2B is localized to feature subgroups of mid to later layers, suggesting that targeted feature-level interventions may offer a more principled path to adversarial robustness than current prompt-level defenses.

Knowledge workers face increasing challenges in synthesizing information from multiple documents into structured conceptual understanding. This process is inherently iterative: users explore content, identify relationships between concepts, and continuously reorganize their mental models. However, current approaches offer limited support. LLM-based systems let users query information but not shape how knowledge is organized; manual tools like mind maps support structure creation but lack intelligent assistance. This leaves an open opportunity: supporting collaborative construction where users and AI jointly develop an evolving knowledge representation. We present MindTrellis, an interactive visual system where users and AI collaboratively build a dynamic knowledge graph. Users can query the graph to retrieve document-grounded information, and contribute by introducing new concepts, modifying relationships, and reorganizing the hierarchy to reflect their developing understanding. In a user study where 12 participants created slide decks, MindTrellis outperformed retrieval-only baselines in knowledge organization and cognitive load, as measured by expert ratings of content coverage and structural quality.

Soil salinity is a major environmental challenge in coastal Bangladesh, threatening agricultural productivity and local livelihoods. This study develops a machine-learning-based framework to predict and map soil salinity in Satkhira district by integrating field observations with Landsat-derived spectral indices. A total of 205 soil samples collected during 2024-2025 were used to train an Extreme Gradient Boosting (XGBoost) model, and predictions were further improved using a Generalized Additive Model (GAM). Spatial cross-validation was applied to reduce autocorrelation bias, and bootstrap resampling was used to quantify prediction uncertainty. The results show strong spatial variability of soil salinity, with higher concentrations in the southern and central coastal regions and lower levels in the northern inland areas. Vegetation indices, particularly NDVI, along with salinity-related spectral indicators, were identified as key predictors. 10-year-window peak-exposure maps generated for 2014-2023 reveal recurrent high-salinity zones and a persistent, expanding footprint of moderate-to-high salinity exposure across the central parts of the district. Uncertainty analysis indicates higher variability in coastal zones and improved prediction stability when multi-year datasets are combined. The proposed framework provides a robust and scalable approach for long-term monitoring of soil salinity. It supports climate-resilient agriculture, land-use planning, and evidence-based decision-making in coastal Bangladesh.

Real-world data often exhibit long-tailed distributions with numerous noisy labels, substantially degrading the performance of deep models. While prior research has made progress in addressing this combined challenge, it overlooks the severe label-image mismatch inherent to high-noise settings, thereby limiting their effectiveness. Given that observed labels, though mismatched with images, still retain category information, we propose employing auxiliary text information from labels to address label-image inconsistencies in long-tailed noisy data. Specifically, we leverage the intrinsic cross-modal alignment in pre-trained visual-language models to correct the label-image inconsistencies. This supervisory signal, referred to as Weak Teacher Supervision (WTS), is unaffected by label noise and data distribution biases, albeit exhibits limited accuracy. Therefore, the activation of WTS is determined by evaluating the discrepancy between text-predicted labels and observed labels. Extensive experiments demonstrate the superior performance of WTS across synthetic and real-world datasets, particularly under high-noise conditions. The source code is available at https://anonymous.4open.science/r/WTS-0F3C.

As software systems grow in complexity, they must satisfy an increasing number of competing quality attributes, making it essential to balance them in a principled manner -- for example, a safety requirement for sensor-fusion verification may conflict with a tight planning-cycle budget. Multi-agent large language model frameworks support this balancing process by assigning specialized agents to different objectives. However, their conflict resolution is typically heuristic. Requirements are aggregated implicitly without explicit acceptance or rejection, limiting auditability in regulated domains. We present ArgRE, a multi-agent requirements negotiation system that embeds Dung-style abstract argumentation into the negotiation stage. Each proposal, critique, and refinement is modeled as an argument, conflicts are represented as directed attack relations, and the accepted set of arguments is computed under grounded and preferred semantics. The pipeline further integrates KAOS goal modeling, multi-layer verification, and standards-oriented artifact generation. Evaluation across five case studies spanning safety-critical, financial, and information-system domains shows that ArgRE provides argument-level traceability absent from existing frameworks. Independent evaluators rated its decision justifications significantly higher than those of heuristic synthesis (4.32 vs. 3.07, p < 0.001), indicating improved auditability, while semantic intent preservation remains comparable (94.9% BERTScore F1) and compliance coverage reaches 84.7% versus 47.6%--47.8% for baselines. Structural analysis further confirms that the default pairwise protocol yields acyclic graphs in which grounded and preferred semantics coincide, whereas cross-pair arbitration introduces controlled cyclicity, leading to predictable divergence between the two semantics.

Simulation is an indispensable tool for validating distributed IoT architectures before physical deployment, and iFogSim has emerged as one of the most widely adopted platform in the fog and edge computing research community. Yet the experience of using iFogSim for non-canonical, application-specific architectures remains incompletely documented, leaving practitioners without guidance on when the tool is appropriate, which scientific objectives it can address, and how to manage the modelling approximations it imposes. This article helps in providing that guidance through two complementary contributions. First, we present a structured state of the art covering iFogSim and iFogSim2, a taxonomy of ten scientific objectives that motivate IoT architecture simulation, and a comparative survey of eight simulation tools assessed against those objectives. Second, we report our experience of simulating a four-tier smart emergency response system for resource-constrained urban environments, covering a 25-node synthetic road topology, four experimental configurations, and quantitative results including end-to-end alert latency (near 205 ms), FPGA-accelerated Dijkstra path computation (x10 CPU speedup), concurrent incident conflict rates (75% under dual load), and path-cache acceleration (x197). The analysis is organised around five practitioner questions: whether iFogSim fits the target architecture, which objectives it covers natively versus partially, what modelling challenges arise and how their workarounds bias reported results, what changes to the iFogSim source code would close the identified gaps, and whether tool co-simulation can provide comprehensive coverage. Seven modelling challenges are documented with source-code-grounded root causes and explicit bias assessments; finally, seven developer recommendations are proposed as an actionable improvement roadmap for the iFogSim community.

Accurate prediction of drug-target binding affinity accelerates drug discovery by prioritizing compounds for experimental validation. Current methods face three limitations: sequence-based approaches discard spatial geometric constraints, structure-based methods fail to exploit hydrogen bond features, and conventional loss functions neglect prediction-target correlation, a key factor for identifying high-affinity compounds in virtual screening. We developed HBGSA (Hydrogen Bond Graph with Self-Attention), a 3.06M-parameter model that encodes hydrogen bond spatial features. HBGSA uses graph neural networks to model hydrogen bond spatial topology with self-attention enhancement and Pearson correlation loss. Experimental results on PDBbind Core Set and CSAR-HiQ dataset demonstrate that HBGSA outperforms baseline methods with strong generalization capability. Ablation studies confirm the effectiveness of hydrogen bond modeling and Pearson correlation loss.

In limited-data settings, a single endpoint mean of an evaluation metric such as the Continuous Ranked Probability Score (CRPS) is itself a random variable, yet it is routinely reported as if it were a stable property of the method. We study when this practice fails. Using 50 independent repetitions across six regression datasets, we show that CRPS variance trajectories differ substantially across methods and are not always well described by a smooth power-law decay. Methods with a learned heteroscedastic variance head, namely MAP and Deep Ensembles, can develop pronounced, reproducible variance peaks at intermediate training sizes on real datasets, whereas MC Dropout and Bayes by Backprop typically show smooth variance contraction. These peaks have direct practical consequences: at the variance peak on Seoul Bike, the relative RMSE of a single-seed MAP estimate reaches 93.6\%, and the probability of falling within \(\pm 10\%\) of the repeated-run mean drops to 5.9\%. We show that local CRPS variance provides a direct signal of single-seed estimation error, with Spearman correlations above 0.96 on every real dataset. Power-law fit quality and monotonicity together provide compact method-level summaries of trajectory regularity. Finally, replacing the standard heteroscedastic objective with \(β\)-NLL substantially reduces the irregular behavior, consistent with the view that the heteroscedastic training objective contributes to the instability. Practitioners should report trajectory summaries alongside endpoint means and concentrate repeated evaluation in high-variance regions.

Multimodal Federated Learning (MMFL) enables privacy-preserving collaborative training, but real-world clinical applications often suffer from within-modality missingness caused by sensor intermittency or irregular sampling. Existing methods implicitly represent unobserved data via architectural alignment or missing embeddings, often failing to recover the true distribution and yielding sub-optimal performance. We propose CondI, a federated framework explicitly addressing this missingness using conditional diffusion models. CondI employs a two-phase training pipeline: first, imputing unobserved temporal components using available multimodal context and conditional embeddings; second, optimizing modality-specific extractors and joint embedding spaces. During inference, imputed raw data pass through trained extractors to generate robust features, providing a holistic representation for downstream tasks. Explicit data imputation ensures models operate on complete semantic structures, significantly enhancing resilience against severe data incompleteness. Experiments on three clinical datasets (PTB-XL, SLEEP-EDF, MIMIC-IV) demonstrate CondI achieves comparable results to state-of-the-art baselines. Code: https://github.com/ZhengWugeng/CondI

Large Language Models (LLMs) based on Mixture-of-Experts (MoE) are pivotal in industrial applications for their ability to scale performance efficiently. However, standard MoEs enforce uniform expert sizes,creating a rigidity that fails to align computational costs with varying token-level complexity. While heterogeneous expert architectures attempt to address this by diversifying expert sizes, they often suffer from significant system-level challenges, specifically unbalanced GPU utilization and inefficient parameter utilization, which hinder practical deployment. To bridge the gap between theoretical heterogeneity and robust industrial application, we propose Mixture of Heterogeneous Grouped Experts (MoHGE) which introduces a two-level routing mechanism to enable flexible, resource-aware expert combinations. To optimize inference efficiency, we propose a Group-Wise Auxiliary Loss, which dynamically steers tokens to the most parameter-efficient expert groups based on task difficulty. To address the critical deployment challenge of GPU load balancing, we introduce an All-size Group-decoupling Allocation strategy coupled with an Intra-Group Experts Auxiliary Loss. These mechanisms collectively ensure uniform computation distribution across GPUs. Extensive evaluations demonstrate that MoHGE matches the performance of MoE architectures while reducing the total parameters by approximately 20% and maintaining balanced GPU utilization. Our work establishes a scalable paradigm for resource-efficient MoE design, offering a practical solution for optimizing inference costs in real-world scenarios.

Causal effect estimation from observational data requires careful adjustment for confounding. Classical estimators such as inverse probability weighting and augmented inverse probability weighting are effective under favorable model specification, but may become unstable when treatment assignment and outcome mechanisms are complex, non-linear, and high-dimensional. Machine learning and representation learning approaches improve flexibility, yet joint training can allow outcome-related information to influence treatment-side representations, which is undesirable from a causal perspective. We propose MOCA (Modular One-way Causal Attention), a transformer-based framework that separates treatment and outcome modeling through a modular design, and performs confounder adjustment using a one-way attention mechanism. A cutting-feedback strategy, implemented via gradient detachment, prevents the outcome loss from updating the treatment module. This design preserves directional information flow while retaining the representational power of transformer architectures for causal inference. Across multiple simulated scenarios, including linear, nonlinear, heavy-tailed, hidden confounding, and high-dimensional settings, MOCA shows competitive or improved performance relative to IPW, AIPW, X-learner, TARNet, and DragonNet. We further illustrate the method on the Infant Health and Development Program dataset and the Dehejia-Wahba dataset as real-world benchmarks. These results suggest that modular attention with one-way information flow provides a promising and interpretable direction for causal inference with modern deep learning models.

Existing multi-agent Large Language Model (LLM) frameworks for code generation typically use execution feedback and improve iteratively using Input/Output (I/O) test cases. However, this does not work for scientific workflows, where I/O test cases do not exist, and generating them requires solving the very problem at hand. To address this, we introduce MOSAIC, a training-free multi-agent framework for scientific code generation without I/O supervision. Instead of execution feedback, MOSAIC employs a student-teacher knowledge distillation framework that grounds generation through domain-specific examples and structured problem decomposition. To further mitigate hallucinations across chained subproblems, we introduce a Consolidated Context Window (CCW) for maintaining consistent reasoning across agents. Experiments on the SciCode benchmark show that MOSAIC improves accuracy, executability, and numerical precision over existing approaches while relying on lightweight models.

Standard evaluations of Bayesian deep learning methods assume that metric estimates are reliable, but we show this assumption fails under data scarcity. Method rankings are not only unreliable at small $n$, but also dataset-dependent in ways that point estimates cannot reveal: the same method comparison yields $P(\mathrm{MCD} \prec \mathrm{Ensemble}) = 1.000$ at $n = 50$ on one dataset and remains below $0.95$ even at $n = 500$ on another. Across the datasets we consider, no universal sample size threshold exists, which is precisely why dataset-specific posterior inference is necessary. To address this, we use a Bayesian hierarchical model with method-specific variances to treat evaluation metrics as random variables across data realizations, and we use a predictive Minimum Detectable Difference curve to assess whether an observed gap would be detectable at a given training size. Across six Bayesian deep learning methods and five regression datasets, our results show that uncertainty-aware evaluation is necessary in low-data settings, because current evidence for method superiority and predictive detectability at the same training size can diverge substantially. Our framework provides practitioners with principled tools to determine whether their evaluation data is sufficient before drawing conclusions about method superiority.

SystemVerilog Assertions (SVA) are essential for formal verification of digital hardware, yet their manual creation demands significant expertise in both the design under verification and temporal logic. Recent studies have explored using large language models (LLMs) to automate SVA generation, but existing approaches suffer from incorrect signal references, missing timing constraints, and lack of formal correctness guarantees. This paper presents ProofLoop, a tool-augmented ReAct agent that generates SVA from natural-language specifications using a solver-in-the-loop approach. The agent operates in two phases: Phase A autonomously gathers design context by invoking EDA and formal tools, including semantic search over an AST-indexed vector database and JasperGold structural queries, while Phase B generates SVA and iteratively refines it using JasperGold formal proof feedback over up to fixed (here 3) verification rounds. We evaluate ProofLoop on FVEval Design2SVA design benchmarks and demonstrate that this framework can achieve 93.7% syntax correctness and 82.0% functional correctness. An ablation study confirms that each component, i.e., retrieval-augmented generation (RAG), JasperGold tools, and the verification loop contributes significantly (and orthogonally).

Evaluating generative AI models is increasingly resource-intensive due to slow inference, expensive raters, and a rapidly growing landscape of models and benchmarks. We propose ProEval, a proactive evaluation framework that leverages transfer learning to efficiently estimate performance and identify failure cases. ProEval employs pre-trained Gaussian Processes (GPs) as surrogates for the performance score function, mapping model inputs to metrics such as the severity of errors or safety violations. By framing performance estimation as Bayesian quadrature (BQ) and failure discovery as superlevel set sampling, we develop uncertainty-aware decision strategies that actively select or synthesize highly informative inputs for testing. Theoretically, we prove that our pre-trained GP-based BQ estimator is unbiased and bounded. Empirically, extensive experiments on reasoning, safety alignment, and classification benchmarks demonstrate that ProEval is significantly more efficient than competitive baselines. It requires 8-65x fewer samples to achieve estimates within 1% of the ground truth, while simultaneously revealing more diverse failure cases under a stricter evaluation budget.

Indoor environments lack the spatial intelligence infrastructure that GPS provides outdoors; first responders arriving at unfamiliar buildings typically have no machine-readable map of safety equipment. Prior work on 3D semantic segmentation for public safety identified two barriers: scarcity of labeled indoor training data and poor recognition of small safety-critical features by native point-cloud methods. This paper presents INSIGHT, a zero-target-domain-annotation pipeline that projects 2D image understanding into 3D metric space via registered RGB-D data. Two interchangeable vision stacks share a common 3D back end: a SAM3 foundation-model stack for text-prompted segmentation, and a traditional CV stack (open-set detection, VQA, OCR) whose intermediate outputs are independently inspectable. Evaluated on all seven subareas of Stanford 2D-3D-S (70{,}496 images), the pipeline produces Pointcept-schema-compatible labeled point clouds and ISO~19164-compliant scene graphs with ${\sim}10^{4}{\times}$ compression; role-filtered payloads transmit in ${<}15$\,s at 1\,Mbps over FirstNet Band~14. We report per-point labeling accuracy on 7 shared classes, detection sensitivity for 15 safety-critical classes absent from public 3D benchmarks alongside code-capped deployable estimates, and inter-pipeline complementarity, demonstrating that 2D-to-3D semantic transfer addresses the labeled-data bottleneck while scene graphs provide building intelligence compact enough for field deployment.

The real-world adoption of portrait relighting is hindered by dataset domain gaps, camera sensitivity, and computational costs. We address these challenges with Hybrid Domain Knowledge Fusion, a paradigm that fuses the specialized strengths of synthetic, One-Light-at-A-Time (OLAT), and real-world datasets into a compact model. Our approach features specialized prior models hardened by domain-aware adaptation, followed by augmented knowledge distillation into a lightweight student model with multi-domain expertise. Our method demonstrates a 6x to 240x inference speedup while maintaining state-of-the-art (SOTA) visual quality in the experiments. Additionally, we construct a massive, high-fidelity synthetic dataset with diverse ground-truth intrinsics to support our training pipeline.

Scaling EEG foundation models requires pooling data across heterogeneous electrode montages, a prerequisite both for larger pretraining corpora and for downstream deployment. We present the first systematic comparison of four channel adaptation methods (Conv1d projection, spherical spline interpolation (SSI), source-space decomposition, and Riemannian re-centering) across five pretrained EEG foundation models (5M--157M parameters), five downstream tasks, and two training regimes with 10--15 random seeds each. We find that rigid-montage models (BENDR, Neuro-GPT) require external adaptation, while flexible models (EEGPT, CBraMod) match or exceed it natively when fine-tuned but benefit from external methods under frozen-encoder deployment. A probe-SFT asymmetry exists: external adaptation can cause severe negative transfer during fine-tuning of flexible models. The optimal method is architecture-dependent (Conv1d for BENDR, SSI/Riemannian for Neuro-GPT, source-space decomposition for depression detection), and 5M-parameter CBraMod outperforms models up to 31$\times$ larger on 4/5 datasets, consistent with independent findings that compact EEG-specific architectures can match larger models.

Automatically generating formal ontologies from unstructured natural language remains a central challenge in knowledge engineering. While large language models (LLMs) show promise, it remains unclear which architectural design choices drive generation quality and why current approaches fail. We present a controlled experimental study using domain-specific insurance contracts to investigate these questions. We first establish a single-agent LLM baseline, identifying key failure modes such as poor Ontology Design Pattern compliance, structural redundancy, and ineffective iterative repair. We then introduce a multi-agent architecture that decomposes ontology construction into four artifact-driven roles: Domain Expert, Manager, Coder, and Quality Assurer. We evaluate performance across architectural quality (via a panel of heterogeneous LLM judges) and functional usability (via competency question driven SPARQL evaluation with complementary retrieval augmented generation based assessment). Results show that the multi-agent approach significantly improves structural quality and modestly enhances queryability, with gains driven primarily by front-loaded planning. These findings highlight planning-first, artifact-driven generation as a promising and more auditable path toward scalable automated ontology engineering.

We present Code Broker, a multi agent system built with Google Agent Development Kit ADK that analyses Python code from files, local directories, or GitHub repositories and generates actionable quality assessment reports. The system employs a hierarchical five agents architecture in which a root orchestrator coordinates a sequential pipeline agent, which in turn dispatches three specialised agents in parallel a Correctness Assessor, a Style Assessor, and a Description Generator before synthesising findings through an Improvement Recommender. Reports score four dimensions correctness, security, style, and maintainability and are rendered in both Markdown and HTML. Code Broker combines LLM based reasoning with deterministic static-analysis signals from Pylint, uses asynchronous execution with retry logic to improve robustness, and explores lightweight session memory for retaining and querying prior assessment context. We position the paper as a technical report on system design and prompt or tool orchestration, and present a preliminary qualitative evaluation on representative Python codebases. The results suggest that parallel specialised agents produce readable, developer oriented feedback, while also highlighting current limitations in evaluation depth, security tooling, large repository handling, and the current use of only in memory persistence. All code and reproducibility materials are available at: https://github.com/Samir-atra/agents_intensive_dev.

Generative approaches to clustering provide information on geometric properties of clusters, whereas discriminative approaches provide boundaries between clusters. Ideas from both approaches are incorporated to present a fully unsupervised, probabilistic, and discriminative clustering method via a regularized mutual information objective function, wherein a mixture of mixtures of Gaussian and uniform distributions is used for formulation of the conditional model. Automatic selection of the number of components is established with the introduction of the regularizing term and a merge step, similar to those applied in reversible jump Markov chain Monte Carlo methods used in Bayesian clustering. Consequently, the turtle shell method -- a fully unsupervised clustering method capable of estimating non-linear boundary lines, automatically selecting the number of components, and capturing intuitive clusters in the presence of data abnormalities such as noise and/or irregular cluster shapes -- is introduced. We test this method on various simulated and real datasets commonly explored in clustering research, and extend the analysis to datasets arising from flow cytometry experiments.

Large-scale agentic systems run on distributed infrastructures where many software agents share physical hosts and are discovered via peer-to-peer mechanisms. Discovery must handle node-level churn from failures and host departures and agent-level churn from demand-driven activation, deactivation, and state changes. Their interaction reshapes classic trade-offs between structured and unstructured overlays. We study decentralized agent discovery under this two-level churn, assuming nodes host multiple agents, overlays are structured or gossip-based, and agents switch between warm and cold states. Using Kademlia as a structured and Cyclon+Vicinity as a gossip baseline, we compare stable, node-churn-only, agent-cooling-only, and combined regimes to see when routing efficiency, resilience, and service readiness align or favor different designs. Structured overlays are more robust and efficient in stable and node-churn regimes, while gossip-based overlays remain competitive and can be faster when readiness dominates.

The quality of diabetic retinopathy (DR) screening relies on the ability to correctly grade severity; however, many deep-learning (DL) classifiers cannot be easily interpreted in the clinical context. This study presents a methodology that combines strong discriminative models with multimodal explanations, converting retinal pixels into clinically interpretable outputs. Using the APTOS 2019 benchmark, we evaluated six representative CNN- and transformer-based backbones under a controlled protocol with stratified five-fold cross-validation. We then compared ensembling strategies (hard voting, weighted soft voting, stacking) and investigated a hybrid class-level fusion variant to exploit grade-specific advantages. For interpretability, we produced Grad-CAM++ visual attribution maps and short textual rationales using vision-language models (VLMs) conditioned on the fundus image and classifier outputs under conservative prompting constraints. Modern CNN backbones (ResNet-50 and ConvNeXt-Tiny) provided the strongest single-model baselines, with cross-validated QWK up to 0.919 and 0.914, respectively. Ensembling improved ordinal agreement, and weighted soft voting was the most consistent across folds (QWK 0.934 +/- 0.017). Hybrid class-level fusion was competitive but did not yield a statistically reliable improvement over standard fusion in paired fold comparisons (Holm-adjusted p >= 1.000). For explanation quality, Grad-CAM++ offered plausible but coarse localization, and VLM rationales were generally grade-consistent. Quantitatively, VLM variants showed a trade-off between clinical completeness and template-level semantic similarity (coverage 0.700 vs. BERTScore 0.072), while image-text alignment was comparable (CLIPScore approximately 0.34).

Vision-language-action (VLA) models can learn to perform diverse manipulation skills "out of the box," but achieving the precision and speed that real-world tasks demand requires further fine-tuning -- for example, via reinforcement learning (RL). We introduce a lightweight method that enables sample-efficient online RL fine-tuning of pretrained VLAs using just a few hours of real-world practice. We (1) adapt the VLA to expose an "RL token," a compact readout representation that preserves task-relevant pretrained knowledge while serving as an efficient interface for online RL, and (2) train a small actor-critic head on this RL token to refine the actions, while anchoring the learned policy to the VLA. Online RL with the RL token (RLT) makes it possible to fine-tune even large VLAs with RL quickly and efficiently. Across four real-robot tasks (screw installation, zip tie fastening, charger insertion, and Ethernet insertion), RLT improves the speed on the hardest part of the task by up to 3x and raises success rates significantly within minutes to a few hours of practice. It can even surpass the speed of human teleoperation on some of the tasks.

Large language model (LLM) agents are increasingly tasked with complex real-world analysis (e.g., in financial forecasting, scientific discovery), yet their reasoning suffers from stochastic instability and lacks a verifiable, compositional structure. To address this, we introduce Analytica, a novel agent architecture built on the principle of Soft Propositional Reasoning (SPR). SPR reframes complex analysis as a structured process of estimating the soft truth values of different outcome propositions, allowing us to formally model and minimize the estimation error in terms of its bias and variance. Analytica operationalizes this through a parallel, divide-and-conquer framework that systematically reduces both sources of error. To reduce bias, problems are first decomposed into a tree of subpropositions, and tool-equipped LLM grounder agents are employed, including a novel Jupyter Notebook agent for data-driven analysis, that help to validate and score facts. To reduce variance, Analytica recursively synthesizes these grounded leaves using robust linear models that average out stochastic noise with superior efficiency, scalability, and enable interactive "what-if" scenario analysis. Our theoretical and empirical results on economic, financial, and political forecasting tasks show that Analytica improves 15.84% accuracy on average over diverse base models, achieving 71.06% accuracy with the lowest variance of 6.02% when working with a Deep Research grounder. Our Jupyter Notebook grounder shows strong cost-effectiveness that achieves a close 70.11% accuracy with 90.35% less cost and 52.85% less time. Analytica also exhibits highly noise-resilient and stable performance growth as the analysis depth increases, with a near-linear time complexity, as well as good adaptivity to open-weight LLMs and scientific domains.

Large language model (LLM) agents often struggle in long-context interactions. As the agent accumulates more interaction history, context management approaches such as sliding window and prompt compression may omit earlier structured information that later steps rely on. Recent retrieval-based memory systems surface relevant content but still overlook the causal and logical structure needed for multi-step reasoning. We introduce ContextWeaver, a selective and dependency-structured memory framework that organizes an agent's interaction trace into a graph of reasoning steps and selects the relevant context for future actions. Unlike prior context management approaches, ContextWeaver supports: (1) dependency-based construction and traversal that link each step to the earlier steps it relies on; (2) compact dependency summarization that condenses root-to-step reasoning paths into reusable units; and (3) a lightweight validation layer that incorporates execution feedback. On the SWE-Bench Verified and Lite benchmarks, ContextWeaver improves performance over a sliding-window baseline in pass@1, while reducing reasoning steps and token usage. Our observations suggest that modeling logical dependencies provides a stable and scalable memory mechanism for LLM agents that use tools.

Automated methods for red teaming LLMs are an important tool to identify LLM vulnerabilities that may not be covered in static benchmarks, allowing for more thorough probing. They can also adapt to each specific LLM to discover weaknesses unique to it. Most current automated red teaming methods are intended for tackling safety and content moderation. Thus, they make use of content safety models as evaluators and optimize for circumventing them, and as such, have not been tested with other adversarial intents not typically captured by these. We propose a pipeline for training a red teaming model that can generalize to arbitrary adversarial goals, including objectives it has not been directly trained on, and that does not depend on the existence of a pre-existing evaluator available at training time. We demonstrate that finetuning small models, such as Qwen3-8B, using this pipeline results in a substantial improvement in their ability to generate attacks for both in and out of domain adversarial goals.

Frontier AI developers are increasingly deploying highly capable models internally to automate AI R&D, but these deployments currently face limited external oversight. It is essential, therefore, that developers provide evidence that internally deployed models are safe. While recent work has highlighted the risks of internal deployments and proposed broad approaches to transparency and governance, there remains little guidance on the specific information developers should disclose about them. We address this gap by identifying key information that companies should disclose about internally deployed models across four categories: capabilities, usage, safety mitigations, and governance. For each category, we analyse the key benefits and limitations of disclosure and consider how disclosure-related risks can be mitigated. Our framework could be used by developers to inform both public transparency documents, such as model system cards, and private periodic reports required under emerging frontier AI regulation.

Large language models (LLMs) show promise for molecular optimization, but aligning them with selective and competing drug-design constraints remains challenging. We propose C-Moral, a reinforcement learning post-training framework for controllable multi-objective molecular optimization. C-Moral combines group-based relative optimization, property score alignment for heterogeneous objectives, and continuous non-linear reward aggregation to improve stability across competing properties. Experiments on the C-MuMOInstruct benchmark show that C-Moral consistently outperforms state-of-the-art models across both in-domain and out-of-domain settings, achieving the best Success Optimized Rate (SOR) of 48.9% on IND tasks and 39.5% on OOD tasks, while largely preserving scaffold similarity. These results suggest that RL post-training is an effective way to align molecular language models with continuous molecular design objectives. Our code and models are publicly available at https://github.com/Rwigie/C-MORAL.

Clinical framing -- the linguistic manner in which clinical information is presented -- can influence patient understanding and decision-making, with important implications for healthcare outcomes. Obstetrics is a high-stakes domain in which physicians counsel patients on delivery mode choices such as vaginal birth after cesarean (VBAC) and repeat cesarean section (RCS), yet counseling language remains underexplored in large-scale clinical text analysis. In this work, we analyze physician counseling language in 2,024 obstetric history and physical narratives for a rigorously defined cohort of patients for whom both VBAC and RCS were clinically viable options. To control for confounding due to medical contraindications, we first construct a VBAC-eligible cohort using structured clinical data supplemented by a large language model (LLM)-based extraction pipeline constrained to grounded, verbatim evidence from free-text narratives. We then apply a zero-shot LLM framework to categorize counseling segments into predefined framing categories capturing how physicians linguistically present delivery options. Our analysis reveals a significant difference in counseling framing distributions between VBAC and RCS notes; risk-focused language accounts for a substantially larger share of counseling segments in RCS documentation than in VBAC, with category-level differences confirmed by statistical testing, highlighting the value of controlled LLM-based framing analysis in obstetric care.

Firms are deploying more capable AI systems, but organizational controls often have not kept pace. These systems can generate greater productivity gains, but high-value uses require broader authority exposure -- data access, workflow integration, and delegated authority -- when governance controls have not yet decoupled capability from authority exposure. We develop an analytical model in which a firm jointly chooses AI deployment and cybersecurity investment under this governance-capability gap. The central result shows a deployment paradox: in high-loss environments, better AI can lead a firm to deploy less when capability is deployed through broader authority exposure under weak governance. Optimal deployment also falls below the no-risk benchmark, and this shortfall widens with breach-loss magnitude and with the authority exposure attached to more capable systems. Governance investment that reduces breach-loss magnitude shrinks the paradox region itself, while breach externalities expand the range of environments in which deployment is socially constrained. Governance maturity is therefore not merely a constraint on AI adoption. It is a condition that shapes whether capability improvements translate into productive deployment.

We investigate whether explicit belief graphs improve LLM performance in cooperative multi-agent reasoning. Through 3,000+ controlled trials across four LLM families in the cooperative card game Hanabi, we establish four findings. First, integration architecture determines whether belief graphs provide value: as prompt context, graphs are decorative for strong models and beneficial only for weak models on 2nd-order Theory of Mind (80% vs 10%, p<0.0001, OR=36.0); when graphs gate action selection through ranked shortlists, they become structurally essential even for strong models (100% vs 20% on 2nd-order ToM, p<0.001). Second, we identify "Planner Defiance," a model-family-specific failure where LLMs override correct planner recommendations at partial competence (90% override, replicated N=20); Gemini models show near-zero defiance while Llama 70B shows 90%, and models distinguish factual context (deferred to) from advisory recommendations (overridden). Third, full-game evidence confirms inter-agent conventions (+128% over baseline, p=0.003) outperform all single-agent interventions, and individual belief-graph components must be combined to produce gains. Fourth, preliminary scaling analysis (N=10/cell, exploratory) suggests graph depth has diminishing returns: shallow graphs provide the best cost-benefit ratio, while deeper ToM graphs appear harmful at larger player counts (-1.5 pts at 5-player, p=0.029).

We propose a simple yet effective alternative to reward normalization in policy gradient reinforcement learning by integrating a 1D Kalman filter for online reward estimation. Instead of relying on fixed heuristics, our method recursively estimates the latent reward mean, smoothing high-variance returns and adapting to non-stationary environments. This approach incurs minimal overhead and requires no modification to existing policy architectures. Experiments on \textit{LunarLander} and \textit{CartPole} demonstrate that Kalman-filtered rewards significantly accelerate convergence and reduce training variance compared to standard normalization techniques. Code is available at https://github.com/Sumxiaa/Kalman_Normalization.

Predicting the outcomes of prospective clinical trials remains a major challenge for large language models. Prior work has shown that both traditional correlational predictors, such as random forests and logistic regression, and strong commercial LLMs achieve limited performance on this task. In this paper, we propose DeepImagine, a framework for teaching LLMs biomedical reasoning through successive counterfactual imagining. The central idea is to approximate hidden causal mechanisms of clinical trials by training models to infer how observed trial results would change under controlled perturbations of experimental conditions, such as dosage, outcome measures, study arms, geography, and other trial attributes. To support this objective, we construct both natural and approximate counterfactual pairs from real clinical trials with reported outcomes. For settings where strict counterfactual supervision is available, such as paired outcome measures or dose-ranging study arms within the same trial, we train models with supervised fine-tuning. For broader settings where only approximate counterfactual pairs can be retrieved, we optimize models with reinforcement learning using verifiable rewards based on downstream benchmark correctness. We further augment training with synthetic reasoning traces that provide causally plausible explanations for local counterfactual transitions. Using this pipeline, we train language models under 10B parameters, including Qwen3.5-9B, and evaluate them on clinical trial outcome prediction. We aim to show that DeepImagine consistently improves over untuned language models and traditional correlational baselines. Finally, we aim to show that the learned reasoning trajectories provide interpretable signals about how models represent trial-level mechanisms, suggesting a practical path toward more mechanistic and scientifically useful biomedical language models.

Mixed Binary Quadratic Programs (MBQPs) are an important and complex set of problems in combinatorial optimization. As solving large-scale combinatorial optimization problems is challenging, primal heuristics have been developed to quickly identify high-quality solutions within a short amount of time. Recently, a growing body of research has also used machine learning to accelerate solution methods for challenging combinatorial optimization problems. Despite the increasing popularity of these ML-guided methods, a large body of work has focused on Mixed-Integer Linear Programs (MILPs). MBQPs are challenging to solve due to the combinatorial complexity coupled with nonlinearities. This work proposes ML-guided primal heuristics for Mixed Binary Quadratic Programs (MBQPs) by adapting and extending existing work on ML-guided MILP solution prediction to MBQPs. We introduce a new neural network architecture for MBQP solution prediction and a new training data collection procedure. Moreover, we extend existing loss functions in solution prediction and propose to combine contrastive and weighted cross-entropy losses. We evaluate the methods on standard and real-world MBQP benchmarks and show that the developed ML-guided methods significantly outperform existing primal heuristics and state-of-the-art solvers. Furthermore, models trained with our proposed extension with combined losses outperform other ML-based methods adapted from MILPs and improve generalization in cross-regional inference on a real-world wind farm layout optimization problem.

Language models are increasingly deployed in interactive settings where users reason about facts over time rather than in isolation. In such scenarios, correct behavior requires models to maintain and update implicit temporal assumptions established earlier in a conversation. We study this challenge through the lens of temporal scope stability: the ability to preserve, override, or transfer time-scoped factual context across dialogue turns. We introduce ChronoScope, a large-scale diagnostic benchmark designed to isolate temporal scope behavior in controlled multi-turn interactions, comprising over one million deterministically generated question chains grounded in Wikidata. ChronoScope evaluates whether models can correctly retain inferred temporal scope when follow-up questions omit explicit time references, spanning implicit carryover, explicit scope switching, cross-entity transfer, and longer temporal trajectories. Through extensive evaluation of state-of-the-art language models, we find that temporal scope stability is frequently violated in controlled multi-turn settings, with models often drifting toward present-day assumptions despite correct underlying knowledge. These failures intensify with interaction length and persist even under oracle context conditions, revealing a gap between single-turn factual accuracy and coherent temporal reasoning under sequential interaction. We make our dataset and evaluation suite publicly available at https://github.com/yashkumaratri/ChronoScope

AI agents are increasingly deployed to execute tasks and make decisions within agentic workflows, introducing new requirements for safe and controlled autonomy. Prior work has established the importance of human oversight for ensuring transparency, accountability, and trustworthiness in such systems. However, existing implementations of Human-in-the-Loop (HITL) mechanisms are typically embedded within application logic, limiting reuse, consistency, and scalability across multi-agent environments. This paper presents a decoupled HITL system architecture that treats human oversight as an independent system component within the agent operating environment. The proposed design separates human interaction management from application workflows through explicit interfaces and a structured execution model. In addition, a design framework is introduced to formalize HITL integration along four dimensions: intervention conditions, role resolution, interaction semantics, and communication channel. This framework enables selective and context-aware human involvement while maintaining system-level consistency. The approach supports alignment with emerging agent communication protocols, allowing HITL to be implemented as a protocol-level concern. By externalizing HITL and structuring its integration, the system provides a foundation for scalable governance and progressive autonomy in agentic workflows.

Automated testing is crucial for maintaining open-source software quality. However, motivating contributors to include tests for code changes remains a challenge. While existing interventions, such as code coverage metrics and reviewer feedback, are often reactive and applied only after a pull request is opened, this study investigates whether documentation on testing can serve as a proactive measure to encourage testing behavior. In this work, we investigate the relationship between documentation on testing and contributor testing behavior. We introduce the Test Engagement Ratio (TER) to help understand testing frequency. Using data from 160 OSS repositories, we analyze the relationship between documentation comprehensiveness and TER. Our results show a weak but statistically significant positive correlation ($ρ=0.36$, $p<0.001$), which strengthens to a moderate relationship ($ρ=0.44$) in repositories with higher pull request activity. Documentation categories such as How to Run Tests and How to Write Tests show the strongest correlation with testing engagement. Furthermore, TER is found to be moderately correlated ($ρ=0.52$, $p<0.001$) with Test Code Ratio, providing preliminary evidence of its validity. Our findings suggest that documentation on testing may be associated with increased testing engagement. Future work will explore causality, documentation quality at a granular level, and cross-repository exposure effects.

We argue that current definitions of machine unlearning are underspecified for second-order optimizers. We compare first-order and second-order learners for their ability to handle the data deletion task with varying degrees of eigendecomposition to mimic the loss model memory. While both first and second-order methods realign with the ideal counterfactul in terms of performance and gradient, the second-order optimizer shows significant volatility in the optimizer state. This indicates residual information, supposedly deleted, that isn't detectable by first-order analysis. Various eigendecay treatments show that stability and information loss is regained only under controlled state pertubation where geometric information (or memory) is erased.

Systematic biases in Global Circulation Model (GCM) outputs limit their direct applicability in regional planning, necessitating bias correction. Correcting precipitation is particularly challenging due to its non-Gaussian distribution, intermittent nature, and non-linear extremes. However, traditional statistical methods cannot learn from big data and easily address systematic biases in the GCMs, and while machine learning does provide this flexibility, their black-box type functionality hinders us from understanding these biases completely which also further prevents generalization across different GCMs and locations, especially for precipitation. In this study, we propose a differentiable bias adjustment framework called δCLIMBA (or dCLIMBA), that learns a spatiotemporally adaptive parametric bias adjustment procedure between historical CMIP6 model outputs and reference reanalysis datasets (Livneh). Results demonstrate that the proposed method accurately corrects both the magnitude and distribution of extreme storm events, with particularly strong performance in capturing extremes. The quantile distribution of precipitation is well reproduced across diverse U.S. cities, and spatial patterns perform comparably to the widely used LOCA2 statistical downscaling technique. In addition, the framework showed future trend preservation unlike pure quantile based methods and LOCA2; and results from bias correction over unseen regions showed that the marginal biases were attenuated. This work presents a modular, computationally efficient and extensible bias correction approach that is physically informed, scalable, and compatible with both historical and future applications. Its flexibility makes it suitable for integration into Earth system post-processing pipelines and impact workflows.

Despite MoE models leading many benchmarks, supervised fine-tuning (SFT) for the MoE architectures remains difficult because its router layers are fragile. Methods such as DenseMixer and ESFT mitigate router collapse with dense mixing or auxiliary load-balancing losses, but these introduce noisy gradients that often degrade performance. In preliminary experiments, we systematically pruned experts and observed that while certain super experts are activated far more frequently, discarding less used experts still leads to notable performance degradation. This suggests that even rarely activated experts encode non-trivial knowledge useful for downstream tasks. Motivated by this, we propose an auxiliary-loss-free MoE SFT framework that combines bias-driven sparsification with always-active gated condenser experts. Rather than enforcing balanced activation across all experts, our method encourages task-relevant experts to remain active while pushing long-tailed experts toward inactivity. The condenser experts provide a persistent, learnable pathway that alleviates gradient starvation and facilitates consolidation of information that would otherwise remain fragmented across sparsely activated experts. Analysis further suggest that this design better preserves long-tailed expert information under sparse routing. Experiments on large-scale MoE models demonstrate that our approach outperforms state-of-the-art SFT baselines such as DenseMixer and ESFT, achieving average gain of 2.5%+ on both mathematical reasoning and commonsenseQA benchmarks.

Large language models (LLMs) have become central to modern AI workflows, powering applications from open-ended text generation to complex agent-based reasoning. However, debugging these models remains a persistent challenge due to their opaque and probabilistic nature and the difficulty of diagnosing errors across diverse tasks and settings. This paper introduces a systematic approach for LLM debugging that treats models as observable systems, providing structured, model-agnostic methods from issue detection to model refinement. By unifying evaluation, interpretability, and error-analysis practices, our approach enables practitioners to iteratively diagnose model weaknesses, refine prompts and model parameters, and adapt data for fine-tuning or assessment, while remaining effective in contexts where standardized benchmarks and evaluation criteria are lacking. We argue that such a structured methodology not only accelerates troubleshooting but also fosters reproducibility, transparency, and scalability in the deployment of LLM-based systems.

Android malware detectors built with machine learning often suffer from temporal bias: models are trained and evaluated without respecting apps' actual release times, inflating accuracy and weakening real-world robustness. We address this by constructing a time-stamped dataset of benign and malicious Android apps and introducing a timestamp-verification procedure to ensure temporal accuracy. We then propose a detection framework that uses Bootstrap Your Own Latent (BYOL) for self-supervised pre-training to learn obfuscation-resilient representations, followed by supervised classification. Under time-aware evaluation, the method attains 98% accuracy and 89% F1. We further characterize malware behavior by analyzing true positives and false negatives using VirusTotal and the MITRE ATT&CK framework. To support reproducibility and further innovation, we release our dataset and source code.

Two-stage recommender systems first choose a candidate generator and then rank items within the generated set. Because the generator decides which items are available to the ranker, changing the generator changes both the policy value and the data support used to estimate that value. This creates an offline selection problem that standard single-stage objectives do not capture: a policy may look good under a retrieval score or a raw off-policy value estimate, but still be unreliable if it depends on weakly supported generator-item pairs. We propose CASP (Coupled Action-Set Pessimism), a support-aware offline selector for finite libraries of two-stage recommender policies. CASP combines doubly robust value estimation with a support-burden penalty. We show that stagewise rules that ignore downstream continuation value can be arbitrarily suboptimal, and we derive population, finite-class, and reconstructed-propensity guarantees for conservative selection. In simulations and a reconstructed MovieLens 1M application, CASP selects lower-burden policies when estimated value and support credibility are in tension.

Accurate classification of tropical tree species from unoccupied aerial vehicle (UAV) imagery remains challenging due to high species diversity and strong visual similarity among species at typical image resolutions (centimeters per pixel). In contrast, models trained on close-up citizen science photographs captured with smartphones achieve strong plant species classification performance. Recent advances in UAV data acquisition now enable the collection of close-up images that are spatially registered with top-view aerial imagery and approach the level of visual detail found in smartphone photographs, with the trade-off that such high-resolution photos cannot be acquired for many trees. In this work, we evaluate the performance of existing methods using paired top-view and close-up UAV imagery collected in a species-rich tropical forest. Through fine-tuning experiments, we quantify the performance gap between vision foundation models and in-domain generalist plant recognition models across both image types (high-resolution close-up versus coarser-resolution top-view imagery). We show that classification performance is consistently higher on close-up images than on top-view aerial imagery, and that this performance gap widens for rare species. Finally, we propose that self-supervised representation alignment across these two spatial scales offers a promising approach for integrating fine-grained visual information into canopy-level species classification models based on top-view UAV imagery. Leveraging high-resolution close-up UAV imagery to enhance canopy-level species classification could substantially improve large-scale monitoring of tropical forest biodiversity.

Web-scale 3D asset collections are abundant, but rarely deployment-ready. Assets ship with arbitrary metric scale, incorrect pivots and forward axes, brittle geometry, and textures that do not support relighting, which limits their utility for embodied AI, robotics simulation, game development, and AR/VR. We present AmaraSpatial-10K, a dataset of over 10,000 synthetic 3D assets designed for downstream use rather than volume alone. Each asset is released as a metric-scaled, semantically anchored .glb with separated PBR material maps, a convex collision hull, a paired reference image, and rich multi-sentence text metadata. The dataset spans indoor objects, vehicles, architecture, creatures, and props under a unified spatial convention. Alongside the dataset, we introduce an evaluation suite for 3D asset banks. The suite comprises a continuous Scale Plausibility Score (SPS) with an LLM-as-Judge interval protocol, an LLM Concept Density score for metadata, an anchor-error metric, and a cross-modal CLIP coherence protocol, and we use it to audit AmaraSpatial-10K alongside matched subsets from Objaverse, HSSD, ABO, and GSO. Compared with Objaverse-sourced assets, we demonstrate that AmaraSpatial-10K substantially improves text-based retrieval precision (CLIP Recall@5 of 0.612 vs 0.181, a 3.4x improvement with median rank falling from 267 to 3), and we establish that it satisfies the spatial and semantic prerequisites for physics-aware scene composition and embodied-AI asset banks, leaving those downstream evaluations to future work. AmaraSpatial-10K is publicly available on Hugging Face.

Stochastic Gradient Descent (SGD) is a known stochastic iterative method popular for large-scale convex optimization problems due to its simple implementation and scalability. Some objectives, such as those found in complex-valued neural networks, benefit from updates like in SGD and Gradient Descent (GD) with a newly defined ``gradient'' that allows for complex parameters. This complex variant of the SGD/GD methods has already been proposed, but convergence guarantees without analyticity constraints have not yet been provided. We propose a variant of SGD (complex SGD) that allows for complex parameters, and we provide convergence guarantees under assumptions that parallel those from the real setting. Notably, these results extend to GD as well, and with the same set of assumptions, we confirm that some directional bias results extend from the real to the complex setting for kernel regression problems. We provide empirical results demonstrating the efficacy of the complex SGD in kernel regression problems utilizing complex reproducing kernel Hilbert spaces. In particular, we demonstrate we may recover superoscillation functions and Blaschke products from the Fock Space and Hardy Space, respectively, as the optimal functions for a particular choice of a loss function.

AI-powered generative models have significantly expanded the possibilities for editing, manipulating, and creating high-quality images. Particularly, images that falsely appear to originate from trusted sources pose a serious threat, undermining public trust in image authenticity. We propose DeepSignature, a novel approach that integrates the guarantees of digital signatures with the capabilities of deep neural networks. Neural networks are used both to generate content-encoding watermarks and to embed them imperceptibly into images while ensuring robust extraction. These watermarks are cryptographically verifiable, enabling source attribution and image integrity validation. DeepSignature is compatible with existing image formats and requires no special handling of signed images. It supports client-side verification, requiring only the signer's public key. Additionally, we introduce a novel latent-space verification approach to detect and localize tampering attempts. We evaluate DeepSignature in terms of imperceptibility, robustness to benign transformations, forgery detection, and its resilience against various attack scenarios. Our results highlight the inherent trade-offs between imperceptibility, robustness, and integrity verification. We demonstrate that DeepSignature reliably identifies significant forgery attempts -- achieving near 100\% in our experiments. Finally, we emphasize DeepSignature's modularity and tunable parameters, allowing adaptation to application-specific requirements. Code and model weights will be published.

This paper presents a complete, end-to-end on-device vision machine learning pipeline, comprising data acquisition, two-layer CNN training with Adam optimization, and real-time inference, executing entirely on a microcontroller-class device costing $15-40 USD. Unlike cloud-based workflows that require external infrastructure and conceal the computational pipeline from the practitioner, this system implements every step of the core ML lifecycle in approximately 1,750 lines of readable C++ that compiles in under one minute using the Arduino IDE, with no external ML dependencies. Running on the Seeed Studio ESP32-S3 XIAO ML Kit (8 MB PSRAM), the firmware achieves three-class 64x64 image classification in approximately 9 minutes per training run, with real-time inference at 6.3 FPS. Key contributions include: correct batch-level gradient accumulation; pre-computed resize lookup tables for inference; dual-format weight export for SD-free baked-in deployment; a three-tier weight priority system (SD binary > baked-in header > He-initialization) resolved automatically at boot; a single-constant network reconfiguration interface; and PSRAM-aware memory management suited to microcontroller constraints. All source code and reference datasets are released under the MIT License at https://github.com/webmcu-ai/on-device-vision-ai

Job Skill Named Entity Recognition (JobSkillNER) aims to automatically extract key skill information from large-scale job posting data, which is important for improving talent-market matching efficiency and supporting personalized employment services. To the best of our knowledge, this work presents the first Chinese JobSkillNER dataset for recruitment texts. We propose annotation guidelines tailored to Chinese job postings and an LLM-empowered Macro-Micro collaborative annotation pipeline. The pipeline leverages the contextual understanding ability of large language models (LLMs) for initial annotation and then refines the results through expert sentence-level adjudication. Using this pipeline, we annotate more than 20,000 instances collected from four major recruitment platforms over the period 2014-2025. Based on these efforts, we release Chinese-SkillSpan, the first Chinese JobSkillNER dataset aligned with the ESCO occupational skill standard across four dimensions: knowledge, skill, transversal competence, and language competence (LSKT). Experimental results show that the dataset supports effective model training and evaluation, indicating that Chinese-SkillSpan helps fill a major gap in Chinese JobSkillNER resources and provides a useful benchmark for intelligent recruitment research. Code and data are available at https://sites.google.com/view/cn-skillspan-resources .

In this paper, we propose a Physics-Informed Neural Network framework for time-dependent simulations of pollution propagation originating from moving emission sources. We formulate a robust variational framework for the time-dependent advection-diffusion problem and establish the boundedness and inf-sup stability of the corresponding discrete weak formulation. Based on this mathematical foundation, we construct a robust loss function that is directly related to the true approximation error, defined as the difference between the neural network approximation and the (unknown) exact solution. Additionally, a collocation-based strategy is introduced to speed up neural network training. As a case study, we investigate pollution propagation caused by snowmobile traffic in Longyearbyen, Spitsbergen, supported by detailed in-field measurements collected using dedicated sensors. The proposed framework is applied to analyze the effects of thermal inversion on pollutant accumulation. Our results demonstrate that thermal inversion traps dense and humid air masses near the ground, significantly enhancing particulate matter (PM) concentration and worsening local air quality.

Formalising informal mathematical reasoning into formally verifiable code is a significant challenge for large language models. In scientific fields such as physics, domain-specific machinery (\textit{e.g.} Dirac notation, vector calculus) imposes additional formalisation challenges that modern LLMs and agentic approaches have yet to tackle. To aid autoformalisation in scientific domains, we present FormalScience; a domain-agnostic human-in-the-loop agentic pipeline that enables a single domain expert (without deep formal language experience) to produce \textit{syntactically correct} and \textit{semantically aligned} formal proofs of informal reasoning for low economic cost. Applying FormalScience to physics, we construct FormalPhysics, a dataset of 200 university-level (LaTeX) physics problems and solutions (primarily quantum mechanics and electromagnetism), along with their Lean4 formal representations. Compared to existing formal math benchmarks, FormalPhysics achieves perfect formal validity and exhibits greater statement complexity. We evaluate open-source models and proprietary systems on a statement autoformalisation task on our dataset via zero-shot prompting, self-refinement with error feedback, and a novel multi-stage agentic approach, and explore autoformalisation limitations in modern LLM-based approaches. We provide the first systematic characterisation of semantic drift in physics autoformalisation in terms of concepts such as notational collapse and abstraction elevation which reveals what formal language verifies when full semantic preservation is unattainable. We release the codebase together with an interactive UI-based FormalScience system which facilitates autoformalisation and theorem proving in scientific domains beyond physics.https://github.com/jmeadows17/formal-science

Despite remarkable progress in Vision--Language--Action (VLA) models, a central bottleneck remains underexamined: the data infrastructure that underlies embodied learning. In this survey, we argue that future advances in VLA will depend less on model architecture and more on the co-design of high-fidelity data engines and structured evaluation protocols. To this end, we present a systematic, data-centric analysis of VLA research organized around three pillars: datasets, benchmarks, and data engines. For datasets, we categorize real-world and synthetic corpora along embodiment diversity, modality composition, and action space formulation, revealing a persistent fidelity-cost trade-off that fundamentally constrains large-scale collection. For benchmarks, we analyze task complexity and environment structure jointly, exposing structural gaps in compositional generalization and long-horizon reasoning evaluation that existing protocols fail to address. For data engines, we examine simulation-based, video-reconstruction, and automated task-generation paradigms, identifying their shared limitations in physical grounding and sim-to-real transfer. Synthesizing these analyses, we distill four open challenges: representation alignment, multimodal supervision, reasoning assessment, and scalable data generation. Addressing them, we argue, requires treating data infrastructure as a first-class research problem rather than a background concern.

Subtle visual anomalies such as hairline cracks, sub-millimeter voids, and low-contrast inclusions are structurally atypical yet visually ambiguous, making them both difficult to annotate and easy to overlook during active learning. Standard acquisition heuristics based on discriminative uncertainty or feature diversity often overselect dominant patterns while underexploring sparse yet important regions of the data space. This failure mode is especially severe in industrial defect inspection, where anomalies may be both low-prevalence and difficult to distinguish from surrounding structure. To resolve this, we propose GSAL, an active learning framework for object detection that combines a diffusion-based difficulty signal with a hierarchical semantic coverage prior. The diffusion component scores images and proposals using reconstruction discrepancy and denoising variability, prioritizing visually atypical or ambiguous examples. However, diffusion alone does not prevent acquisition from repeatedly favoring hard samples within dominant semantic modes. The semantic component therefore organizes candidate samples in a three-level concept graph and promotes coverage of underrepresented semantic regions while providing interpretable acquisition rationales. By balancing visual difficulty with semantic coverage, GSAL improves retrieval of subtle and rare targets that are often missed by uncertainty-only selection. Experiments on a proprietary thin-film defect, Pascal VOC and MS COCO dataset show consistent gains in label efficiency and rare-class retrieval over uncertainty-, diversity-, and hybrid-based baselines

Recent medical multimodal foundation models are built as multimodal LLMs (MLLMs) by connecting a CLIP-pretrained vision encoder to an LLM using LLaVA-style finetuning. This two-stage, decoupled approach introduces a projection layer that can distort visual features. This is especially concerning in medical imaging where subtle cues are essential for accurate diagnoses. In contrast, early-fusion generative approaches such as Chameleon eliminate the projection bottleneck by processing image and text tokens within a single unified sequence, enabling joint representation learning that leverages the inductive priors of language models. We present CheXmix, a unified early-fusion generative model trained on a large corpus of chest X-rays paired with radiology reports. We expand on Chameleon's autoregressive framework by introducing a two-stage multimodal generative pretraining strategy that combines the representational strengths of masked autoencoders with MLLMs. The resulting models are highly flexible, supporting both discriminative and generative tasks at both coarse and fine-grained scales. Our approach outperforms well-established generative models across all masking ratios by 6.0% and surpasses CheXagent by 8.6% on AUROC at high image masking ratios on the CheXpert classification task. We further inpaint images over 51.0% better than text-only generative models and outperform CheXagent by 45% on the GREEN metric for radiology report generation. These results demonstrate that CheXmix captures fine-grained information across a broad spectrum of chest X-ray tasks. Our code is at: https://github.com/StanfordMIMI/CheXmix.

Large Language Models (LLMs) are increasingly deployed to autonomously solve real-world tasks. A key ingredient for this is the LLM Function-Calling paradigm, a widely used approach for equipping LLMs with tool-use capabilities. However, an LLM calling functions incorrectly can have severe implications, especially when their effects are irreversible, e.g., transferring money or deleting data. Hence, it is of paramount importance to consider the LLM's confidence that a function call solves the task correctly prior to executing it. Uncertainty Quantification (UQ) methods can be used to quantify this confidence and prevent potentially incorrect function calls. In this work, we present what is, to our knowledge, the first evaluation of UQ methods for LLM Function-Calling (FC). While multi-sample UQ methods, such as Semantic Entropy, show strong performance for natural language Q&A tasks, we find that in the FC setting, it offers no clear advantage over simple single-sample UQ methods. Additionally, we find that the particularities of FC outputs can be leveraged to improve the performance of existing UQ methods in this setting. Specifically, multi-sample UQ methods benefit from clustering FC outputs based on their abstract syntax tree parsing, while single-sample UQ methods can be improved by selecting only semantically meaningful tokens when calculating logit-based uncertainty scores.

Reward models in RLHF are trained to score only the final token of a response - a choice that discards rich signal from every intermediate position and produces models whose token-level outputs are noise. We argue this is a missed opportunity: a well-trained reward model's output at any token should represent the conditional expectation of the final reward given the response so far. We introduce Temporally Coherent Reward Modeling (TCRM), which induces this property via two regularization terms on top of the standard Bradley-Terry loss, with minimizers provably equal to conditional expectations. The regularizers correspond to Monte Carlo and TD value-learning objectives, establishing a direct connection to RL value functions. TCRM requires zero changes to architecture, data, or inference, yet unlocks three capabilities from one principle: interpretable token-level reward trajectories (middle-token pairwise accuracy improved from 50% to 88.9%, final-token accuracy preserved); state-of-the-art PRM performance on ProcessBench (44.9% average F1) among models trained only on outcome data; and unified reward/value modeling in PPO, reducing peak GPU memory by 27% and step time by 19% with matching LLM quality.

We address Human Activity Recognition (HAR) utilizing Wi-Fi Channel State Information (CSI) under the joint requirements of causal interpretability, symbolic controllability, and direct operation on high-dimensional raw signals. Deep neural models achieve strong predictive performance on CSI-based HAR (CHAR), yet rely on continuous latent representations that are opaque and difficult to modify; purely symbolic approaches, in contrast, cannot process raw CSI streams. We propose a fully automatic and strictly decoupled pipeline in which CSI magnitude windows are compressed by a categorical variational autoencoder with Gumbel-Softmax latent variables under a capacity-controlled objective, yielding a compact discrete representation. The encoder is then frozen and used as a deterministic mapping to one-hot latent trajectories. Causal discovery is performed on these trajectories to estimate class-conditional temporal dependency graphs. Statistically supported lagged dependencies are translated into Linear Temporal Logic (LTL) rules, producing a fully symbolic and deterministic classifier based solely on rule evaluation and aggregation, without any learned discriminative head. Because rules are defined over discrete latent variables, antenna-specific rule sets can in principle be combined at the symbolic level, enabling structured multi-antenna fusion without retraining the encoder. Results from CHAR Latent Temporal Rule Extraction (CHARL-TRE) indicate competitive performance while preserving explicit temporal and causal structure, showing that deterministic symbolic classification grounded in unsupervised discrete latent representations constitutes a viable alternative to end-to-end black-box models for wireless HAR.

The TRUST democratic discourse analysis pipeline exposes its large language model (LLM) components to peer model identity through multiple structural channels -- a design feature whose bias implications have not previously been empirically tested. We provide the first systematic measurement of identity-dependent scoring bias across all active identity exposure channels in TRUST, crossing four model families with two anonymization scopes across 30 political statements. The central finding is that single-channel anonymization produces near-zero bias effects, because individual channels act in opposite directions and cancel each other out -- a result that would lead an evaluator to conclude that identity bias is absent when it is not. Only full-pipeline anonymization reveals the true pattern: homogeneous ensembles amplify identity-driven sycophancy when model identity is fully visible, while the heterogeneous production configuration shows the reverse. Model choice matters independently: one tested model exhibits baseline sycophancy two to three times higher than the others and near-zero deliberative conflict on ideological topics, making it structurally unsuitable for pipelines where genuine inter-role disagreement is the intended quality mechanism. Three practical conclusions follow. First, heterogeneous model ensembles are structurally more robust than homogeneous ones, achieving higher consensus rates and lower identity amplification. Second, full-pipeline anonymization is required for valid bias measurement -- partial anonymization is insufficient and actively misleading. Third, these findings have direct implications for the validation of multi-agent LLM systems in quality-critical applications: a system validated under partial anonymization or with a homogeneous ensemble may pass validation while retaining structural identity bias invisible to single-channel measurement.

Trust Region Bayesian Optimization (TuRBO) is an effective strategy for alleviating the curse of dimensionality in high-dimensional black-box optimization. However, inappropriate lengthscale design can cause the local Gaussian process (GP) model within the trust region to degenerate, leading to suboptimal performance in high dimensions. In this work, we show that TuRBO's local GP may remain either excessively complex or overly simple as the dimension $D$ and trust region side length $L$ vary. To address this issue, we propose a straightforward variant, AdaScale-TuRBO, which scales the GP lengthscale with both the problem dimension and trust region size, thereby preserving kernel geometry and maintaining consistent prior complexity. Empirically, we show that AdaScale-TuRBO can robustly outperform standard TuRBO and other popular high-dimensional BO methods on synthetic benchmarks and real-world trajectory planning tasks.

Each major technological revolution inverts a particular scarcity and rebuilds institutions around the shift. The near-consensus diagnosis of the AI revolution holds that AI collapses the cost of prediction while judgment remains scarce. This Opinion argues the inversion has now flipped: competent-looking judgment (selecting, ranking, attributing, certifying) is produced at scale and at marginal cost approaching zero, and four complements become scarce: verified signal, legitimacy, authentic provenance, and integration capacity (the community's tolerance for delegated cognition). Because judgment is the substance of institutions, the institutions built to manufacture legitimate judgment (courts, journals, licensing bodies, legislatures) now compete with the technology for the same functional role. The piece traces the pattern across scientific institutions, professional licensing, intellectual property, democratic legitimacy, and foundation-model concentration, and closes with a three-move agenda: reframe AI policy as institutional redesign, build provenance and verification as commons, and develop the formal apparatus for institutional composition under strategic agents.

Anemia affects over one billion people globally and remains severely under-diagnosed in low-resource regions where laboratory blood tests are inaccessible. This paper presents AnemiaVision, an end-to-end web-based system for non-invasive anemia screening from smartphone photographs of the palpebral conjunctiva and fingernail beds. The proposed pipeline fine-tunes a pre-trained EfficientNet-B3 backbone with a redesigned three-layer classifier head incorporating BatchNorm, GELU activations, and high-rate Dropout (0.45/0.35). Training employs four orthogonal accuracy-boosting techniques: TrivialAugmentWide for policy-free image augmentation, RandomErasing for spatial regularisation, Mixup (alpha=0.2) for inter-class smoothing, and cosine-annealing scheduling with linear warmup. Early stopping is governed by peak validation accuracy rather than validation loss to prevent premature termination on high-variance epochs. The deployed Flask application integrates persistent patient-history management backed by PostgreSQL on Render, with an automated database-migration entrypoint ensuring zero data loss across redeploys. Ablation experiments demonstrate that accuracy-first early stopping contributes +1.6% and Mixup contributes +2.8% to final validation accuracy. Overall, the proposed system achieves a validation accuracy of 96.2% and AUC-ROC of 0.98, compared with 44.9% validation accuracy and AUC-ROC of 0.58 from the three-epoch CPU-only baseline. Sensitivity for the anemic class reaches 0.96, making the system suitable as a first-line screening tool for community health workers in rural settings. The system is publicly accessible and source code is openly available.

Preference-based argumentation frameworks (PAFs) extend Dung's approach to abstract argumentation (AAFs) by encoding preferences over arguments. Such preferences control the transformation of attacks into defeats, and different approaches to doing so result in different reductions from a PAF to an AAF. In this paper we consider a PAF inverse problem which takes an argumentation graph, a labelling and a semantics as an input, and outputs a ``yes" or ``no" as to whether there is a preference relation between the arguments which can yield the desired labelling. This inverse problem has applications in areas including preference elicitation and explainability. We consider this problem in the context of the four most widely-used preference based reductions under the complete semantics. We show that in most cases, the problem can be answered in polynomial time.

Natural language data follows a power-law distribution, with most knowledge and skills appearing at very low frequency. While a common intuition suggests that reweighting or curating data towards a uniform distribution may help models better learn these long-tail skills, we find a counterintuitive result: across a wide range of compositional reasoning tasks, such as state tracking and multi-step arithmetic, training under power-law distributions consistently outperforms training under uniform distributions. To understand this advantage, we introduce a minimalist skill-composition task and show that learning under a power-law distribution provably requires significantly less training data. Our theoretical analysis reveals that power law sampling induces a beneficial asymmetry that improves the pathological loss landscape, which enables models to first acquire high-frequency skill compositions with low data complexity, which in turn serves as a stepping stone to efficiently learn rare long-tailed skills. Our results offer an alternative perspective on what constitutes an effective data distribution for training models.

History-dependent sampling can reduce long-run Monte Carlo variance by discouraging redundant revisits, but existing schemes typically encode history through empirical measure on finite state spaces, which is infeasible in high-dimensional discrete configuration spaces or ill-posed in continuous domains. We propose Score-Repellent Monte Carlo (SRMC) framework that summarizes trajectory history by a running average of score evaluations in $R^d$, where $d$ is the dimension of the score and state representation. This history is converted into a surrogate target through an exponential score tilt, indexed with $α$ that represents the strength of repellence in controlling the magnitude of the history-based repulsion. The surrogate family is normalization-free in the standard MCMC sense, yielding a generic wrapper: at each iteration, any base kernel targeting $π$ can instead be run on the current surrogate $π_{θ_n}$ while the history is updated online. We analyze the coupled evolution of the history recursion and Monte Carlo estimators using stochastic approximation with controlled Markovian noise, establishing almost sure convergence and a joint central limit theorem. We further identify regimes in which the asymptotic covariance decreases as $α$ increases, with scaling $O(1/α)$, extending the near-zero-variance effect of finite-state history-dependent samplers to general state spaces with constant memory. Experiments on continuous targets and discrete energy-based models demonstrate improved estimator variance and mode coverage, while retaining $O(d)$ memory usage and modest per-iteration overhead.

Multimodal environmental monitoring conventionally requires a suite of purpose-built transducers, each constrained to a predefined target. Here, we present MIND (Mycoponically Integrated Network Device), a platform that sustains living fungal mycelial networks on porous bioceramic substrates and reads their passive extracellular voltages. Without hardware modification, a single device produces distinguishable bioelectrical responses to 14 stimuli spanning chemical, optical, mechanical, thermal, and biological domains. We show that steady-state intensity responses follow Hill-type calibration functions conserved across five phylogenetically diverse fungal species, and that multichannel decoding recovers stimulus duration, spatial origin, and continuous position from the bioelectrical output. Strain selection tunes sensitivity without hardware redesign. The platform restores full electrophysiological function within 72 h of mechanical damage and has maintained calibration-quality readout for more than 11 months of continuous operation. These results position fungal electrophysiology as a measurement platform for sensing applications in which the full stimulus set, the electrode geometry, and the recovery requirements cannot be fully specified in advance.

Diffusion models produce high-quality synthetic data but suffer from slow inference. We propose 3D Variable-Step Denoising Diffusion Probabilistic Model (VS-DDPM) a framework engineered to maintain generative quality while accelerating inference by several factors. We tested our approach on four tasks (missing MRI, tumor removal, MRI-to-sCT, and CBCT-to-sCT) within the BraTS2025 and SynthRAD2025 challenges. Designed for high efficiency under hardware and time constrains imposed by both challenges. VS-DDPM achieved state-of-the-art (SOTA) performance in missing MRI synthesis, yielding Dice scores of 0.80, 0.83, and 0.88 for the enhancing tumor, tumor core, and whole tumor regions, respectively, alongside a structural similarity index (SSIM) of 0.95. For MRI tumor removal, the model attained a root mean squared error (RMSE) of 0.053, a peak signal-to-noise ratio (PSNR) of 26.77, and an SSIM of 0.918. While the framework demonstrated competitive performance in MRI-to-sCT and CBCT-to-sCT tasks, it did not reach SOTA benchmarks, potentially due to sensitivities in data pre and post-processing pipelines or specific loss function configurations. These results demonstrate that VS-DDPM provides a robust and tunable solution for high-fidelity 3D medical image synthesis. The code is available in https://github.com/andre-fs-ferreira/SynthRAD_by_Faking_it.

While the next-token prediction (NTP) paradigm enables large language models (LLMs) to express their intrinsic knowledge, its sequential nature constrains performance on specialized, non-generative tasks. We attribute this performance bottleneck to the LLMs' knowledge expression mechanism, rather than to deficiencies in knowledge acquisition. To address this, we propose Self-Knowledge Re-expression (SKR), a novel, task-agnostic adaptation method. SKR transforms the LLM's output from generic token generation to highly efficient, task-specific expression. SKR is a fully local method that uses only unannotated data, requiring neither human supervision nor model distillation. Experiments on a large financial document dataset demonstrate substantial improvements: over 40% in Recall@1 for information retrieval tasks, over 76% reduction in object detection latency, and over 33% increase in anomaly detection AUPRC. Our results on the MMDocRAG dataset surpass those of leading retrieval models by at least 12.6%.

The promise of data-driven materials discovery remains constrained by the scarcity of large, high-quality, and accessible experimental datasets. Here, we introduce a generalizable large language model (LLM)-powered pipeline for automated extraction and structuring of materials data from unstructured scientific literature, using concrete materials as a representative and particularly challenging example. The pipeline exhibits robust performance across a broad range of LLMs and achieves an $F_1$ score of up to 0.97 for diverse composition--process--property attributes. Within one hour, it extracts nearly 9,000 high-quality records with over 100 attributes screened from more than 27,000 publications, enabling the construction of the largest open laboratory database for blended cement concrete. Machine learning analyses underscore the importance of large, diverse, and information-rich datasets for enhancing both in-distribution accuracy and out-of-distribution generalization to unseen materials. The proposed pipeline is readily adaptable to other materials domains and accelerates the development of scalable data infrastructures for materials informatics.

Verification is becoming central to both reinforcement-learning-based training and inference-time control of large language models (LLMs). Yet current verifiers face a fundamental trade-off: LLM-based verifiers are expressive but hard to control and prone to error, while deterministic executable verifiers are reliable and interpretable but often limited in capability. We study the following question: given a development set of LLM outputs and labels for a target objective, such as correctness, can we automatically induce a minimal set of Python verifiers whose joint satisfaction closely matches that objective? We propose AutoPyVerifier, a framework that uses an LLM to synthesize candidate verifier functions and then refines them through search over a directed acyclic graph (DAG). By navigating the DAG, AutoPyVerifier systematically explores the space of deterministic executable verifiers and selects a compact verifier set whose joint satisfaction best approximates the target objective. Across mathematical reasoning, coding, function calling, and instruction-following benchmarks for several state-of-the-art LLMs, AutoPyVerifier improves target-objective prediction by up to 55.0 F1 points over the initial LLM-generated verifier sets. Additional analyses show that the most useful verification targets vary by benchmark and model, and that the DAG-based search shifts the learned verifier sets toward more structural and semantically grounded checks. We further show that exposing the discovered verifier set to an LLM as an external tool improves downstream accuracy by up to 17.0 points. We release our code

LLM-based agents for text-to-SQL often struggle with latency-performance trade-off, where performance improvements come at the cost of latency or vice versa. We reformulate text-to-SQL generation within the lens of software test coverage where the original query is prepared with a suite of test cases with simpler, atomic SQLs that are executed in parallel and together ensure semantic coverage of the original query. After iterating on test case coverage, the final SQL is generated only when enough information is gathered, leveraging the explored test case SQLs to ground the final generation. We validated our framework on a state-of-the-art benchmark for text-to-SQL, Spider 2.0, achieving a new state-of-the-art with 70.2% execution accuracy.

Understanding and representing complex climate variability is essential for both scientific analysis and predictive modeling. However, identifying meaningful climate regimes from raw variables is challenging, as they exhibit high noise and nonlinear dependencies. In this work, we explore the use of Masked Siamese Networks to discretize climate time series into semantically rich clusters. Focusing on daily minimum and maximum temperature, we show that the resulting representations: (i) yield clusters that reflect meaningful climate states under our modeling assumptions, offering a simplified representation for downstream use; (ii) enable sampling and analysis of specific climate scenarios; and (iii) exhibit statistical associations with El Niño events, underscoring their scientific relevance. Our findings highlight the potential of self-supervised discretization as a tool for climate data analysis and open avenues for incorporating richer climate indicators in future work.

Large language models (LLMs) have advanced rapidly, emerging as versatile tools across fields thanks to their exceptional language understanding, generation, and reasoning capabilities. However, performing LLM inference at the network edge remains challenging due to their large memory and compute demands. This survey outlines the challenges specific to LLM edge inference and provides a comprehensive overview of recent progress, covering system architectures, model optimization and deployment, and resource management and scheduling. By synthesizing state-of-the-art techniques and mapping future directions, this survey aims to unlock the potential of LLMs in resource-constrained edge environments.

Whole-body Positron Emission Tomography (PET) registration is essential for multi-parametric tumor characterization and assessment of metastatic disease progression. In deep learning-based deformable registration, the dense displacement field (DDF) regularizer is crucial for stabilizing optimization and preventing unrealistic deformations in large 3D volumes. A key challenge in whole-body deformable registration is anatomical heterogeneity, rigid structures (e.g., bones) should undergo stronger regularization, whereas soft tissues require more flexible deformation and weaker constraints. In this work, we propose a simple yet effective CT-guided spatially-varying regularization strategy for whole-body cross-tracer deformable PET registration. The key idea is to use the paired CT volume from the PET/CT acquisition to construct a voxel-wise regularization map for the DDF, replacing the conventional single global regularization weight. This yields anatomy-adaptive regularization strength across rigid and soft tissues. The proposed method is evaluated on a real clinical cross-tracer PET/CT dataset of 296 patients involving 18F-PSMA and 18F-FDG, showing that the proposed method achieves statistically significant improvements over weakly-supervised registration baseline in both whole-body registration performance and organ-wise alignment.

The integration of quantum machine learning with classical deep learning offers promising avenues for medical image analysis by mapping data into high-dimensional Hilbert spaces. However, effectively unifying these distinct paradigms remains challenging due to common optimization asymmetries. In this paper, a novel hybrid quantum-classical architecture for breast cancer diagnosis based on a dual-branch feature-extraction pipeline is proposed. Our framework extracts and unifies complementary representations from classical models and quantum circuits, exploring both trainable and deterministic (non-trainable) quantum paradigms. To integrate these embeddings, three progressive feature fusion strategies are introduced: Static Hybrid Fusion (SHF) for offline extraction, Dynamic Hybrid Fusion (DHF) for end-to-end co-adaptation, and a novel Temperature-Scaled Hybrid Fusion (TSHF). The TSHF strategy incorporates a learnable scalar, inspired by multimodal learning, that dynamically balances hybrid gradient dynamics and resolves optimization bottlenecks. Empirical validation on the BreastMNIST dataset confirms our hypothesis that unifying diverse feature representations creates a richer data context. The TSHF strategy, specifically when pairing a ResNet backbone with a trainable quantum circuit, achieved a peak accuracy of 87.82%, F1-score of 91.77%, and an AUC-ROC of 89.08%, outperforming purely classical baselines. These results demonstrate that the proposed hybrid framework improves classification accuracy and threshold reliability, providing a stable, high-performance architecture for the clinical deployment of quantum-enhanced diagnostic tools.

Diffusion models achieve remarkable success in time series generation. However, slow inference limits their practical deployment. We propose E$^2$-CRF (Error-Feedback Event-Driven Cumulative Residual Feature caching) to accelerate frequency domain diffusion models. Our method exploits two structural properties: (1) spectral localization, where signal energy concentrates in low frequencies, and (2) mirror symmetry, which halves the effective frequency dimension. E$^2$-CRF uses a closed-loop error-feedback system that adaptively caches transformer KV features across diffusion steps. We trigger recomputation using event-driven residual dynamics instead of fixed schedules. Our method selectively recomputes high-energy or rapidly-changing tokens while reusing cached features for stable high-frequency components. E$^2$-CRF achieves ~2.2 speedup while maintaining sample quality. We demonstrate effectiveness on 5 datasets. Our caching strategy naturally aligns with the diffusion process's structure-to-detail progression. We include sufficient-condition error and complexity bounds under standard regularity assumptions (Appendix), alongside empirical validation. Our code is available at https://github.com/NoakLiu/FastFourierDiffusion and is also integrated in https://github.com/NoakLiu/FastCache-xDiT.

Autonomous systems increasingly operate under partial observability where execution-relevant state is never fully accessible. Existing governance mechanisms -- trusted execution environments, oracle-signed state proofs, cryptographic attestation -- enforce the integrity of computation and state projections. We show this is structurally insufficient: an authenticated projection of state is necessary but never sufficient for execution validity. We introduce the Reconstructive Authority Model (RAM), which separates integrity from coverage. RAM defines a reconstruction gate that reasons over an explicit coverage envelope -- comprising proven state, declared assumptions, and an acknowledged unobservable residual -- and permits execution only when coverage is adequate for the action class. When coverage is insufficient, RAM narrows privileges dynamically or fails closed. Attestation proves trust in measurement; RAM proves adequacy of what is measured. We formalize RAM, prove necessity via two theorems (attestation insufficiency and RAM necessity) and three corollaries, and present a hybrid RAM+Attestation architecture with privilege-narrowing. Synthetic experiments (N=100,000, seed=42) show RAM achieves zero invalid execution rates at all coverage levels. Attestation-based systems exhibit IER=0.423 at low coverage and IER=0.233 even at full coverage, the latter arising from undefined-state handling failures undetectable by integrity checks alone. This reframes execution validity as a coverage reconstruction problem, distinct from and complementary to integrity guarantees provided by attestation.

Patent retrieval underpins critical decisions in innovation, examination, and IP strategy, yet progress has been hampered by the absence of benchmarks that reflect the diversity of real world search scenarios. We address this gap with two contributions. First, we introduce Sophiabench, a large-scale patent retrieval benchmark comprising 10,000 queries and 75,000 corpus documents stratified across ten years, eight IPC technology sections, and twelve filing jurisdictions. Unlike prior benchmarks, Sophia-bench tests retrieval using 12 different query types-from structured patent fields to AI-generated summaries-and evaluates results against citation-based ground truth enhanced with a novel domain-relevance metric (InScope). Together, these enable systematic measurement of how well models perform across query types, technology domains, and jurisdictions. Second, we introduce QaECTER, a 344M-parameter embedding model trained on patent citation graphs and multi-view self-alignment. Despite its compact size, QaECTER establishes a new state of the art for patent retrieval. It outperforms the \#1 model on the English retrieval text embedding benchmark (RTEB), a model 23x larger, as well as all existing patent specific models across every query type, IPC section, and jurisdiction on Sophia-bench, with gains of up to 7.2% average NDCG@10 over the next-best model. These results are confirmed on an independent external benchmark, where QaECTER surpasses all prior models without requiring task-specific instruction prompts. Both the benchmark and the model are designed for practical deployment in large-scale patent search systems.

Indoor positioning is an essential technology for a wide range of applications in GNSS-denied environments, including indoor navigation and IoT systems. Combining convolutional neural networks (CNNs) and magnetic field-based features offers a low-cost, infrastructure-free solution for precise positioning. While magnetic fingerprints are a promising approach for indoor positioning, models trained on raw 3D magnetometer data are highly sensitive to device orientation. We address this by using two rotation invariant features derived from the 3D magnetic field: the norm (Mn) and the projection onto the gravity axis (Mg). We train a lightweight 7-layer dilated CNN (MagNetS/XL) on magnetic sequences to directly regress (x, y) positions. Using the MagPie dataset (three buildings, handheld trajectories), we systematically evaluate fixed and random rotations of test and/or train data. Raw 3D inputs (Mx, My , Mz) exhibit isotropic error increases under fixed 90° rotations and further degrade with growing random rotations. In contrast, 2D (Mn, Mg) inputs maintain rotation invariant accuracy and surpass the 3D inputs once rotation exceeds building-specific thresholds for three reference buildings: 0° for Loomis (large), 5° for Talbot (medium), and 6° for CSL (small). MagNetXL achieves or exceeds state-of-the-art accuracy on the MagPie dataset, and MagNetS delivers similar performance with roughly one third of the parameters, favoring mobile deployment. These results show that the robustness gained from rotation invariant inputs outweighs the loss of input dimensionality in realistic usage, allowing mapping and localization without orientation alignment or added infrastructure.

Traditional data valuation methods based on ``row-count $\times$ quality coefficient'' paradigms fail to capture the nuanced, nonlinear contributions that data makes to Large Language Model (LLM) capabilities. This paper presents a dynamic data valuation framework that transitions from static accounting to utility-based pricing. Our approach operates on three layers: (1) token-level information density metrics using Shannon entropy and Data Quality Scores; (2) empirical training gain measurement through influence functions, proxy model strategies, and Data Shapley values; and (3) cryptographic verifiability through hash-based commitments, Merkle trees, and a tamper-evident training ledger. We provide comprehensive experimental validation on three real domains (instruction following, mathematical reasoning, and code summarization), demonstrating that proxy-based empirical gain achieves near-perfect ranking alignment with realized utility, substantially outperforming row-count and token-count baselines. This framework enables a fair Data-as-a-Service economy where high-reasoning data is priced according to its actual contribution to model intelligence, while providing the transparency and auditability necessary for trustworthy data markets.

Feature selection in high-dimensional genomic data ($d \gg n$) demands methods that are simultaneously accurate, sparse, and stable. Existing approaches either require manual threshold specification (mRMR, stability selection), produce unstable selections under data perturbation (Lasso, Boruta), or ignore biological structure entirely. We introduce StackFeat-RL, a meta-learning framework that optimises the hyperparameters of an iterative dual-criterion feature selection algorithm via REINFORCE policy gradients. The dual criterion, requiring both coefficient consistency and selection frequency, guards against two failure modes missed by single-criterion methods, while iterative accumulation provides convergence guarantees via the law of large numbers. On COVID-19 miRNA data (GSE240888, 332 features) and three Alzheimer's disease classification tasks (GSE84422, 13237 genes; Normal vs.\ Possible, Probable, and Definite AD), StackFeat-RL achieves the highest predictive accuracy among all evaluated methods, including ElasticNet, Boruta, mRMR, and stability selection, while requiring 3--4$\times$ fewer features. Keywords: feature selection, reinforcement learning, REINFORCE, elastic net, biomarker discovery, Alzheimer's disease, dual-criterion selection, protein interaction networks

LLM-as-a-Judge has become a dominant approach in automated evaluation systems, playing critical roles in model alignment, leaderboard construction, quality control, and so on. However, the scalability and trustworthiness of this approach can be substantially distorted by Self-Preference Bias (SPB), which is a directional evaluative deviation in which LLMs systematically favor or disfavor their own generated outputs during evaluation. Existing measurements rely on costly human annotations and conflate generative capability with evaluative stance, and thus are impractical for large-scale deployment in real-world systems. To address this issue, we introduce a fully automated framework to quantifying and mitigating SPB, which constructs equal-quality pairs of responses with negligible quality differences, enabling statistical disentanglement of discriminability from bias propensity without human gold standards. Empirical analysis across 20 mainstream LLMs reveals that advanced capabilities are often uncorrelated, or even negatively correlated, with low SPB. To mitigate this bias, we propose a structured multi-dimensional evaluation strategy grounded in cognitive load decomposition, which reduces SPB by 31.5\% on average.

Agent skills introduce a new and more severe form of indirect injection for LLM agents: unlike traditional indirect prompt injection, attackers can hide malicious instructions inside a dense, action-oriented skill that already functions as a legitimate instruction source. We study pre-execution skill-poison detection and show that successful skill poisoning induces a structured internal effect, attention hijacking, in which response-time attention shifts from trusted context to malicious skill spans and drives harmful behavior. Motivated by this mechanism, we propose RouteGuard, a frozen-backbone detector that combines response-conditioned attention and hidden-state alignment through reliability-gated late fusion. Across both real and synthetic open-source skill benchmarks, RouteGuard is consistently the strongest or most robust detector; on the critical Skill-Inject channel slice, it reaches 0.8834 F1 and recovers 90.51% of description attacks missed by lexical screening, showing that defending against skill poisoning requires internal-signal detection rather than text-only filtering

Federated cross-modal retrieval faces severe challenges from heterogeneous client data, particularly non-IID semantic distributions and missing modalities. Under such heterogeneity, a single global model is often insufficient to capture both shared cross-modal knowledge and client-specific characteristics. We propose RCSR, a personalization-friendly federated framework that integrates prototype anchoring, retrieval-centric semantic routing, and optional client-specific adapters. Built on a frozen CLIP backbone, RCSR leverages lightweight shared adapters for global knowledge transfer while supporting efficient local personalization. Prototype anchoring helps unimodal clients align with global cross-modal semantics, and a server-side semantic router adaptively assigns aggregation weights based on retrieval consistency to mitigate alignment drift during heterogeneous updates. Extensive experiments on MS-COCO, Flickr30K, and other benchmarks show that RCSR consistently improves global retrieval accuracy and training stability, while further enhancing client-level retrieval performance, especially for clients with incomplete modalities. Code is available at https://github.com/RezinChow/RCSR-Retrieval-Centric-Semantic-Routing.

Multimodal Large Language Models (MLLMs) have shown promising potential in diverse understanding tasks, e.g., image and video analysis, math and physics olympiads. However, they remain blank and unexplored for Small Object Understanding (SOU) tasks. To fill this gap, we introduce SOUBench, the first and comprehensive benchmark for exploring the small objects understanding capability of existing MLLMs. Specifically, we first design an effective and automatic visual question-answer generation strategy, constructing a new SOU-VQA evaluation dataset, with 18,204 VQA pairs, six relevant sub-tasks, and three dominant scenarios (i.e., Driving, Aerial, and Underwater). Then, we conduct a comprehensive evaluation on 15 state-of-the-art MLLMs and reveal their weak capabilities in small object understanding. Furthermore, we develop SOU-Train, a multimodal training dataset with 11,226 VQA pairs, to improve the SOU capabilities of MLLMs. Through supervising fine-tuning of the latest MLLM, we demonstrate that SOU-Train can effectively enhance the latest MLLM's ability to understand small objects. Comprehensive experimental results demonstrate that, the proposed SOUBench, along with the SOU-VQA and SOU-Train datasets, provides a crucial empirical foundation to the community for further developing models with enhanced small object understanding capabilities. Datasets and Code: https://github.com/Hanfj-X/SOU.

Alzheimer's disease (AD) is a progressive neurodegenerative disorder and a major cause of dementia. Structural MRI is widely used to analyze AD-related brain atrophy; however, most deep learning methods rely on computationally expensive 3D convolutional neural networks (CNNs), limiting deployment in resource-constrained settings. This work introduces two main contributions. First, we propose a pipeline that converts T1-weighted MRI into anatomically informed 2D point clouds using Anatomical Priority Sampling (APS), producing ADNI-2DPC, the first neuroanatomically labeled MRI-derived point cloud dataset. Second, we present NeuroAPS-Net, a lightweight geometric deep learning model that incorporates anatomical priors via region-aware feature encoding and ROI token aggregation. Experiments on ADNI-2DPC demonstrate that NeuroAPS-Net achieves competitive classification accuracy while significantly reducing inference latency and GPU memory compared to state-of-the-art point cloud methods. These results highlight the potential of anatomically guided point cloud learning as an efficient and interpretable alternative to voxel-based CNNs for AD classification.

Modern container ships face higher wind loads due to increased windage areas, making accurate predictions of wind loads essential for mooring design. Existing empirical models, largely developed for container ships with smaller windage areas and simpler geometrical configurations than those of modern large-scale vessels, often lack accuracy and do not account for the influence of nearby structures. This study proposes a multi-fidelity surrogate modelling framework for the prediction of wind-load coefficients, combining empirical correlations with simplified and detailed CFD models for ships in open-sea and harbor environments. The approach relies on recursive co-kriging to consistently fuse information across fidelity levels, enabling accurate predictions at a reduced computational cost. A sensitivity analysis is used to identify the most influential geometric parameters, and the resulting reduced parameter space is explored through sequential sampling to efficiently construct the training database. The surrogate models are validated over a wide range of loading configurations and for two distinct harbor environments. The results demonstrate that the multi-fidelity approach significantly improves prediction accuracy compared to single-fidelity models, while substantially reducing the reliance on high-fidelity simulations. In particular, the proposed framework captures the dependence of wind loads on key geometric parameters and consistently outperforms traditional empirical correlations, providing a robust and efficient tool for engineering applications.

Generative recommendation (GR) offers superior modeling capabilities but suffers from prohibitive inference costs due to the repeated encoding of long user histories. While cross-request Key-Value (KV) cache reuse presents a significant optimization opportunity, the massive scale of individual user states creates a storage explosion that far exceeds physical GPU limits. We propose MTServe, a hierarchical cache management system that virtualizes GPU memory by leveraging host RAM as a scalable backup store. To bridge the I/O gap between tiers, MTServe introduces a suite of system-level optimizations, including a hybrid storage layout, an asynchronous data transfer pipeline, and a locality-driven replacement policy. On both public and production datasets, MTServe delivers up to 3.1* speedup while maintaining near-perfect hit ratios (>98.5%).

Existing document OCR largely targets plain text or Markdown, discarding the structural and executable properties that make LaTeX essential for scientific publishing. We study page-level reconstruction of scientific PDFs into compilable LaTeX and introduce TexOCR-Bench, a benchmark, and TexOCR-Train, a large-scale training corpus, for this task. TexOCR-Bench features a multi-dimensional evaluation suite that jointly assesses transcription fidelity, structural faithfulness, and end-to-end compilability. Leveraging TexOCR-Train, we train a 2B-parameter model, TexOCR, using supervised fine-tuning (SFT) and reinforcement learning (RL) with verifiable rewards derived from LaTeX unit tests that directly enforce compilability and referential integrity. Experiments across 21 frontier models on TexOCR-Bench show that existing systems frequently violate key document invariants, including consistent section structure, correct float placement, and valid label-reference links, which undermines compilation reliability and downstream usability. Our analysis further reveals that RL with verifiable rewards yields consistent improvements over SFT alone, particularly on structural and compilation metrics.

We identify and formalize a novel security risk: Context-Fragmented Violations (CFVs) - a class of policy breaches where individual agent actions appear locally safe and reasonable, yet collectively violate organizational policies because critical policy facts are siloed in different departments private contexts. Existing prompt-based alignment mechanisms and monolithic interceptors are poorly matched to violations that span contextual islands. We propose Distributed Sentinel, a distributed zero-trust enforcement architecture that introduces the Semantic Taint Token (STT) Protocol. Through lightweight sidecar proxies, our system propagates security state across organizational boundaries without exposing raw cross-domain data, enabling Counterfactual Graph Simulation for cross-domain policy verification. We construct PhantomEcosystem, a comprehensive benchmark comprising 9 categories of realistic cross-agent violation scenarios with adversarially balanced safe controls. On this benchmark, Distributed Sentinel achieves F1 = 0.95 with 106ms end-to-end latency (16ms verification + 90ms entity extraction on A100), compared to 0.85 F1 for prompt-based filtering and 0.65 for rule-based DLP. To empirically validate the need for external enforcement, we evaluate eight frontier LLMs in execution-oriented multi-agent workflows with per-agent domain world models. All models exhibit substantial violation rates (14-98%), with cross-domain data flows showing systematically higher violation rates than same-domain flows. These results indicate that self-avoidance is unreliable and that multi-agent security benefits from a centralized enforcement layer operating above individual agents.

GBM is a highly aggressive primary malignancy in adults, necessitating personalized therapeutic strategies due to its inherent molecular heterogeneity. MGMT promoter methylation is a pivotal prognostic biomarker for anticipating response to temozolomide-based chemotherapy. Although various AI frameworks have been developed for non-invasive MGMT prediction, spatial heterogeneity of methylation status and the high-dimensional and correlated nature of MRI data frequently constrain discriminative feature learning and generalizability of classical models. To circumvent these limitations, a specialized IA-QCNN architecture is proposed, based on the principles of quantum mechanics, including superposition and entanglement, and enabling more efficient representation learning in high-dimensional Hilbert space. The framework establishes a methodological bridge between GBM radiogenomics and quantum deep learning by integrating energy-based slice selection, importance-aware weighting, ring-topology quantum convolution, and folding-based pooling layers. When the model predicts MGMT promoter methylation status using both mpMRI and T1Gd images, experimental results demonstrate that the IA-QCNN achieves high accuracy despite its low number of trainable parameters while effectively minimizing the overfitting problem observed in classical models. Quantitative analyses reveal that the T1Gd modality possesses higher discriminative power than mpMRI, establishing a clinically significant sequence preference. Furthermore, the model exhibits exceptional robustness in hybrid noise environments, effectively utilizing noise as a regularization mechanism to enhance predictive performance. Consequently, the specialized IA-QCNN architecture provides a robust and computationally efficient alternative to classical approaches in the analysis of heterogeneous radiogenomic data.

When answering questions about images, humans naturally point, label, and draw to explain their reasoning. In contrast, modern vision-language models (VLMs) such as Gemini-3-Pro and GPT-5 only respond with text, which can be difficult for users to verify. We present SketchVLM, a training-free, model-agnostic framework that enables VLMs to produce non-destructive, editable SVG overlays on the input image to visually explain their answers. Across seven benchmarks spanning visual reasoning (maze navigation, ball-drop trajectory prediction, and object counting) and drawing (part labeling, connecting-the-dots, and drawing shapes around objects), SketchVLM improves visual reasoning task accuracy by up to +28.5 percentage points and annotation quality by up to 1.48x relative to image-editing and fine-tuned sketching baselines, while also producing annotations that are more faithful to the model's stated answer. We find that single-turn generation already achieves strong accuracy and annotation quality, and multi-turn generation opens up further opportunities for human-AI collaboration. An interactive demo and code are at https://sketchvlm.github.io/.

Offline reinforcement learning (RL) can learn effective policies from fixed datasets, but deployment objectives may change after training, and in many applications the trained actor cannot be retrained because of data, cost, or governance constraints. We study deployment-time adaptation for frozen offline actors using Product-of-Experts (PoE) composition with a goal-conditioned prior. Our main practical finding is graceful degradation rather than universal performance gain: under degraded or random priors, precision-weighted composition remains anchored to the frozen actor, while additive and prior-only adaptation collapse, and a KL-budget selector often recovers a near-oracle operating point. We also make explicit a closed-form identity in the frozen-actor setting: for diagonal-Gaussian actors and priors, PoE with coefficient alpha yields the same deterministic policy as KL-regularized adaptation with beta = alpha / (1 - alpha), with posterior covariances differing only by a global scalar factor. Empirically, across four D4RL environments (3,900 MuJoCo episodes), we observe a 4/5/3 HELP/FROZEN/HURT split. Extending the analysis to six harder cells and two AntMaze diagnostics reveals an actor-competence ceiling: medium-expert remains HURT in all 9 cells at every tested alpha, while AntMaze with a behavior-cloned frozen actor yields zero success for all composition rules. Overall, PoE and KL-regularized adaptation are best viewed as a single actor-anchored safety mechanism for deployment-time steering.

Automated red-teaming methods for large language models typically optimize attack prompts within a fixed, human-designed strategy, leaving the attack strategy itself unchanged. We instead optimize the strategy. We propose AutoRISE, a method that searches over executable attack programs rather than individual prompts. At each iteration, a coding agent edits a strategy and a fixed evaluation harness scores the resulting attacks, returning both a scalar objective and per-example diagnostics that guide subsequent edits. This allows structural changes, including new attack components and altered control flow, that prompt-level methods do not directly express. We also release two benchmark suites developed on disjoint target sets and evaluate on 11 models from five families against seven established jailbreak datasets. Across held-out models, AutoRISE improves average attack success rate by 17.0 points over the strongest baseline, and improves attack success by up to 16 points on frontier targets with low baseline success rates. Ablations against parametric and strategy-library baselines suggest that these gains arise from unrestricted program search, particularly compositional techniques and control-flow edits. AutoRISE operates in a black-box, inference-only setting, requiring no fine-tuning, human annotation, or GPU compute.

We study the expressive power of message-passing aggregate-combine-readout graph neural networks (ACR-GNNs). Particularly, we focus on the first-order (FO) properties expressible by this formalism. While a tight logical characterisation remains a difficult open question, we make two contributions towards answering it. First, we show that sum aggregation and readout suffice for GNNs to capture FO properties that cannot be expressed in the logic C2 on both directed and undirected graphs. This strengthens known results by Hauke and Wał{\k e}ga (2026) where aggregation and readout functions are specially crafted for the task. Second, we identify two natural ways of restoring characterisability (with regard to C2) for ACR-GNNs. One option is to limit local aggregation (without imposing restrictions on global readout), whilst the second is to run ACR-GNNs over graphs of bounded degree (but unbounded size). In both cases, the FO properties captured by GNNs are exactly those definable by a formula in graded modal logic with global counting modalities. Our results thus establish an innate lower- and upper-bound in terms of how far (fragments of) C2 can be taken to characterise GNNs, and imply that is indeed the unbounded interaction of aggregation and readout that pushes the logical expressive power of GNNs above C2.

In many cyber-physical systems, especially in critical applications such as aeroplanes, data to train anomaly detection and diagnosis algorithms is lacking due to data protection issues and partial observability. To combat this inherent lack of data, we introduce a high-fidelity, physics-informed co-simulation of a common aircraft main-fuel-pump system modelled in \textsc{MATLAB/Simulink Simscape Fluids}. We also describe its generated time-series data with health and fault mode annotations. To show feasibility of our benchmark, we apply an unsupervised Recurrent Variational Autoencoder (RNN-VAE) for anomaly detection and a SOM-VAE for operating mode discretization, trained to separate healthy and faulty conditions.

Visual planning represents a crucial facet of human intelligence, especially in tasks that require complex spatial reasoning and navigation. Yet, in machine learning, this inherently visual problem is often tackled through a verbal-centric lens. While recent research demonstrates the promise of fully visual approaches, they suffer from significant computational inefficiency due to the step-by-step planning-by-generation paradigm. In this work, we present EAR, an editing-as-reasoning paradigm that reformulates visual planning as a single-step image transformation. To isolate intrinsic reasoning from visual recognition, we employ abstract puzzles as probing tasks and introduce AMAZE, a procedurally generated dataset that features the classical Maze and Queen problems, covering distinct, complementary forms of visual planning. The abstract nature of AMAZE also facilitates automatic evaluation of autoregressive and diffusion-based models in terms of both pixel-wise fidelity and logical validity. We assess leading proprietary and open-source editing models. The results show that they all struggle in the zero-shot setting, finetuning on basic scales enables remarkable generalization to larger in-domain scales and out-of-domain scales and geometries. However, our best model that runs on high-end hardware fails to match the zero-shot efficiency of human solvers, highlighting a persistent gap in neural visual reasoning.

This article applies postphenomenological theory to the field of cybersecurity risk management, arguing that formal risk models function as mediating artifacts that shape how security practitioners or analysts perceive, interpret, and act on threats. Based on Don Ihde's taxonomy on human-technology relationships and Peter-Paul Verbeek's extended mediational framework, the Contextual and Multimodal Hazard Impact Index (CIIM), an original dynamic risk model presented as an empirical case study, is analyzed. CIIM is formally defined as CIIM(t+1) = [A T(t) V(t) E(t)] / R(t) + {alpha} P(t), where the condition R(t) 0 is not treated as a computational artifact to be smoothed out, but as a genuine systemic collapse that signals singularity. This design choice constitutes a deliberate phenomenological move, allowing organizational fragility to be made visible in a way that previous CVSS-based and probabilistic models conceal. In addition, we examine how CIIM's time projection (t+1) and its hybrid machine learning architecture, combining LSTM/GRU, XGBoost, and Reinforcement Learning, produce a new form of technological intentionality that structures practitioner or analyst attention and ethical deliberation. The article concludes by establishing implications for the ethical design of cybersecurity instrumentation and for the post-phenomenological methodology itself, proposing the concept of 'phenomenology of collapse' as a contribution to the empirical philosophy of technology.

Reasoning LLMs often spend substantial tokens on long intermediate reasoning traces (e.g., chain-of-thought) when solving new problems. We propose to summarize and store reusable reasoning skills distilled from extensive deliberation and trial-and-error exploration, and to retrieve these skills at inference time to guide future reasoning. Unlike the prevailing \emph{reasoning from scratch} paradigm, our approach first recalls relevant skills for each query, helping the model avoid redundant detours and focus on effective solution paths. We evaluate our method on coding and mathematical reasoning tasks, and find that it significantly reduces reasoning tokens while improving overall performance. The resulting lower per-request cost indicates strong practical and economic potential for real-world deployment.

Existing benchmarks for systematic reviewing remain limited either in scale or in disciplinary coverage, with some collections comprising only a modest number of topics and others focusing primarily on biomedical research. We present Webis-SR4ALL-26, a large-scale, cross-disciplinary corpus of 301,871 systematic reviews spanning all scientific fields as covered by OpenAlex. Using a multi-stage pre-processing pipeline, we link reviews to resolved OpenAlex metadata and reference lists and extract, when explicitly reported, structured method artifacts relevant to retrieval and screening. These artifacts include reported search strategies (Boolean queries or keyword lists) that we normalize into executable approximations, as well as reported inclusion and exclusion criteria. Together, these layers support cross-domain benchmarking of retrieval and screening components against review reference lists, training and evaluation of extraction methods for review artifacts, and comparative meta-science analyses of systematic review practices across disciplines and time. To demonstrate one concrete use case, we report large-scale baseline retrieval signals by executing normalized search strategies in OpenAlex and comparing retrieved sets to resolved reference lists. We release the corpus and the pre-processing pipeline, along with code used for extraction validation and the retrieval demonstration.

We introduce MMTR-Bench, a benchmark designed to evaluate the intrinsic ability of Multimodal Large Language Models (MLLMs) to reconstruct masked text directly from visual context. Unlike conventional question-answering tasks, MMTR-Bench eliminates explicit prompts, requiring models to recover masked text from single- or multi-page inputs across real-world domains such as documents and webpages. This design isolates the reconstruction task from instruction-following abilities, enabling a direct assessment of a model's layout understanding, visual grounding, and knowledge integration. MMTR-Bench comprises 2,771 test samples spanning multiple languages and varying target lengths. To account for this diversity, we propose a level-aware evaluation protocol. Experiments on representative MLLMs show that the benchmark poses a significant challenge, especially for sentence- and paragraph-level reconstruction. The homepage is available at https://mmtr-bench-dataset.github.io/MMTR-Bench/.

LLM architecture research generally aims to maximize model quality subject to fixed compute/latency budgets. However, many applications of interest such as edge and on-device deployment are further constrained by the model's memory footprint, thus motivating parameter-efficient architectures for language modeling. This paper describes a simple architecture that improves the parameter-efficiency of LLMs. Our architecture makes use of looped Transformers as a core primitive, which reuse Transformer layers across depth and are thus more parameter-efficient than ordinary (depth-matched) Transformers. We organize the looped Transformer into three blocks--begin, middle, and end blocks--where each block itself consists of multiple Transformer layers, and only the middle block is applied recurrently across depth. We augment the looped middle block with hyper-connections (Xie et al., 2026), which expand the residual stream into matrix-valued residual streams. Hyper-connections are applied only after each loop, and therefore add minimal new parameters and compute cost. Across various model scales, we find that our Hyper-Connected Looped Transformer (Hyperloop Transformer) is able to outperform depth-matched Transformer and mHC Transformer baselines despite using approximately 50% fewer parameters. The outperformance persists through post-training weight quantization, thus positioning Hyperloop Transformers as an attractive architecture for memory-efficient language modeling.

Hyperdimensional computing (HDC), also referred to as vector symbolic architectures (VSA), represents information with high-dimensional vectors and a compact algebra of primitives. This paper establishes an explicitly unitary embedding from discrete bipolar HDC/VSA vectors to coherent broadband waveforms and develops a common wave-domain realization of the core HDC/VSA primitives within that embedding. Under the resulting RFC/UWE stack, bundling becomes linear superposition, permutation becomes coherent phase evolution, binding is reproduced by nonlinear spectral mixing together with an engineered aliasing step that restores circular-convolution structure, and similarity is recovered as a calibrated differential-power readout. Full-wave FDTD studies validate the physically nontrivial parts of this program, including array-level readout in a mutually coupled setting and the binding pipeline under realistic propagation. In a documented $N=1000$ mutually coupled-array calibration, the predicted interaction effect appears with the expected sign pattern and order of magnitude, yielding a coupled Correlation Contrast Ratio of approximately $8.7 \times 10^{-5}$. The result is a wave-geometric duality for HDC/VSA: existing symbolic operations admit a physically grounded waveform realization, while coherence, isolation, and readout sensitivity remain the central engineering constraints for future hardware.

Scientific research relies on accurate information retrieval from literature to support analytical decisions. In this work, we introduce a new task, INformation reTRieval through literAture reVIEW (IntraView), which aims to automate fine-grained information retrieval faithfully grounded in the provided content in response to research-driven queries, and propose IntrAgent, an LLM-based agent that addresses this challenging task. In particular, IntrAgent is designed to mimic human behaviors when reading literature for information retrieval -- identifying relevant sections and then iteratively extracting key details to refine the retrieved information. It follows a two-stage pipeline: a Section Ranking stage that prioritizes relevant literature sections through structural-knowledge-enabled reasoning, and an Iterative Reading stage that continuously extracts details and synthesizes them into concise, contextually grounded answers. To support rigorous evaluation, we introduce IntraBench, a new benchmark consisting of 315 test instances built from expert-authored questions paired with literature spanning five STEM domains. Across seven backbone LLMs, IntrAgent achieves on average 13.2% higher cross-domain accuracy than state-of-the-art RAG and research-agent baselines.

We develop a theoretical framework for designing quantum couplers based on Dirac materials that can modulate the polarization of transmitted quasiparticles without significantly perturbing their propagation. We analyze in detail the conditions required for perfect transmission (Klein tunneling) together with controlled polarization transformation of the incoming states. We then discuss an explicit model of a quantum coupler composed of AA-stacked bilayer graphene nanoribbons with armchair edges and a localized interlayer interaction. Perfect transmission through the desired polarization channels is examined for both narrow and wide couplers. We show that the transmission of polarized states can be finely tuned by external fields.

The quiet-Sun coronal electron-temperature ratio $R \equiv T_\mathrm{EUV}/T_B \approx 2.4$, stable across an eight-year solar cycle, is read here as a measurement of relative entropy between two diagnostic projections of the coronal electron distribution onto the one-parameter Maxwellian family. The EUV ionization temperature is a moment-matching projection against a Bethe-type ionization kernel; the radio brightness temperature is the Rayleigh-Jeans source function of thermal bremsstrahlung. For a kappa distribution in the mean-energy convention, Fleishman & Kuznetsov (2014) give the radio-side projection in closed form as $T_B = T_\mathrm{core}$; the EUV side returns $T_\mathrm{eff}$ up to a shape-dependent correction within the Dudík et al. (2014) intensity-ratio envelope. At $κ= 2.5$ the Kullback-Leibler divergences between the true distribution and its two Maxwellian projections evaluate to $0.32$ and $1.20$ nats, and their difference satisfies $ΔD_\mathrm{KL} = (3/2)[R_0 - \ln R_0 - 1] = (3/2) d_\mathrm{IS}(T_\mathrm{eff}, T_\mathrm{core})$, where $R_0 \equiv κ/(κ- 3/2)$ is the ideal closed-form ratio and $d_\mathrm{IS}$ is the Itakura-Saito distance. The identity is offered as an analytical reference for observational systems in which two diagnostics project different moments of a common non-equilibrium distribution; the eight-year stability of $R$ expresses a stability of that projection structure.

The erratic non-Hermitian skin effect (ENHSE), emerging from the interplay between disorders and locally nonreciprocal yet globally reciprocal couplings, has reshaped the conventional bulk-boundary correspondence through its disorder-dependent localization properties. Here, we experimentally observe the dynamical phenomena of ENHSE in phononic crystals with disordered imaginary gauge fields. The erratic localization occurs in the bulk independent of the excitation position, with the main and satellite peaks precisely located at the local maxima of the cumulative gauge field in accordance with random-walk extreme-value statistics. Remarkably, the selective manipulation of satellite peaks can be realized by tuning the staggered disorder strengths in a dimerized chain. These findings can deepen the understanding of non-Hermitian physics and establish a new route for disorder-engineered non-Hermitian wave control.

Non-Hermitian random matrices provide a useful framework for understanding universal characteristics of dissipative quantum chaotic systems with loss or gain. We consider a model of two such system represented by two independent $N\times N$ complex Ginibre matrices interacting via a deterministic matrix $c{\bf 1}_N$, where $c$ is the complex coupling parameter whose magnitude $|c|$ controls the interaction strength. We characterize quantitatively how the eigenvectors of the whole system, initially localized in one of the individual subsystems for $|c|=0$, eventually spread over the full system with growing interaction strength. The resulting asymptotic formula describing such spread in the limit $N\to \infty$ is very explicit and provides a full picture of the gradual ergodization of eigenvectors as a function of the coupling parameter $|c|$ in the whole transition regime. As a by-product of our method we also compute the mean eigenvalue density for our model at the origin of the spectral bulk $z=0$ in the fully ergodic regime, when the coupling is scaled with the matrix size as $c=\sqrt{N}\tilde{c}$. We find that as $N\to \infty$ the limiting density at the origin vanishes beyond the critical value $|\tilde{c}|=1$, signalling of a split of the density support in the complex plane into two disjoint domains.

We investigate the momentum-space entanglement between two Dirac quasiparticles in a double-layer honeycomb lattice coupled via a planar electromagnetic cavity. We model the low-energy excitations as massive Dirac fermions in $(1+2)$ dimensions and derive the Bethe-Salpeter equation using the ladder approximation. We use a Born-level approximation around a free two-body quasiparticle state, where the interaction is mediated by the cavity photon propagator. From the reduced sublattice density matrix, we compute a momentum-resolved von Neumann entropy. Within the perturbatively controlled regime, the entropy remains small, while phenomenological self-energy dressing drives a crossover to strong enhancement of the entanglement entropy. Stationary entanglement is obtained only when the quasiparticle coherence time exceeds the photon propagation time between the layers. The maximum-entropy regime appears to be a viable method for achieving Bell-like states. These results demonstrate how self-energy renormalization, virtual particle exchange, and spinor geometry combine to reshape the entanglement landscape of Dirac materials.

The theoretical treatment of depletion gels with central interactions often involves expanding the free energy around a stress-free reference state to derive a constitutive relation between global stress and strain. The premise upon which the previous continuum theories are based, i.e., the stress-free reference state and the affine deformation, both of which do not hold in the context of amorphous gel materials. Gels never reach a true global minimum in the potential energy landscape and contain local regions of significant compressive and tensile stress, interspersed with zero-stress regions. Hence, expansion of free energy around a stressed reference state will produce scalar terms in harmonic expansion, the effects of which are qualitatively different from the terms appearing in the expansion around an unstressed reference state. In this study, we demonstrate the limitations of traditional continuum theories and propose simple constitutive relations that better capture the mechanical response of gel materials. The robustness of the proposed relations is established through large-scale numerical simulations of depletion and frictional gels across a vast parameter space.

Van Hove singularities (vHSs) strongly amplify electron interactions and can stabilize correlated phases in topological bands. Here we report signatures of topological magnetism in large-angle twisted bilayer MoTe2 driven by the interplay of vHSs, strong correlations, and valley topology. In a 4.8 degree device, electrostatic tuning to a vHS produces a spontaneous anomalous Hall hot spot near nu = -1. Combined transport and reflective magnetic circular dichroism measurements indicate that this regime is not governed by magnetization alone, but instead emerges from a correlated intervalley-coherent antiferromagnetic state that evolves with doping into a canted phase. With increasing magnetic field, the Hall response develops an additional finite-field component consistent with a topological Hall effect from a noncoplanar spin texture, before transitioning into a C = -1 Chern insulator. Our results establish tunable vHSs in moire topological bands as a route to chiral magnetism and engineering topological phase transitions.

This paper presents a symmetric unified transport (UT) compact model for MOSFETs that bridges drift-diffusion (DD) and ballistic transport (BT) regimes. The proposed model self consistently accounts for both current and charge across the DD-BT transition. Quantum capacitance and carrier transport are incorporated into the charge density formulation. Drain side velocity saturation and the source side thermal velocity limit are unified within a single framework using a physically motivated high field scattering length, enabling accurate modeling from DD square law behavior to the ballistic limit. In addition, a physical channel charge and capacitance model is developed to capture capacitance reduction in the quasi ballistic regime, which is not considered in standard compact models. The model is verified using theoretical analysis and experimental data from MOSFETs with multiple channel lengths, achieving accurate fitting using only physical transport parameters. The formulation is continuous and symmetric, and it passes both DC and AC symmetry tests.

We present the first analytic theory with elegant and closed-form analytical solutions to explore the cosolvency effect in polymer brushes, where polymer chains that are poorly soluble in two pure solvents become fully soluble in certain mixtures thereof. This effect is key to designing stimulus-responsive smart materials but has not previously been addressed by analytic theory for polymer brushes. Our theoretical framework reveals that preferential adsorption of cosolvent induces an effective repulsion between monomers solvated by cosolvent and those solvated by solvent. The equilibrium solvation of polymer chains by cosolvent gives rise to a concentration-dependent $χ$-function, which captures the effective interactions within the brush and reproduces the reentrant behavior characteristic of the cosolvency effect. The model predicts a discontinuous soluble transition followed by a re-collapse transition at higher cosolvent concentrations. Analytical treatment within a minimal free-energy model for the case of two symmetric poor solvents shows that the swelling and re-collapse transitions share the same thermodynamic origin. For low-density brushes, we derive an analytical approximation and delineate the phase diagram of parameter space in which discontinuous transitions occur. For cosolvency to take place, the theory specifies a minimum strength for preferential solvation and the associated repulsive coupling. Furthermore, it demonstrates that, contrary to previous models, repulsive interactions between cosolvent and solvent in the bulk are not required. This work lays the groundwork for the rational design of smart stimulus-responsive materials based on the cosolvency effect in polymer brushes, a capability which was not previously established.

Large nonlinear recurrent neural networks with random couplings generate high-dimensional, potentially chaotic activity whose structure is of interest in neuroscience, machine learning, ecology, and other fields. A fundamental object encoding the collective structure of this activity is the $N \times N$ covariance matrix. Prior analytical work on the covariance matrix has been limited to low-dimensional summary statistics, not the full high-dimensional object for a specific realization of the couplings. Recent work proposed an ansatz in which, at large $N$, the covariance matrix for a typical quenched realization takes the same form as that of a linear network with the same couplings, driven by independent noise, with mean-field order parameters setting the effective transfer function and the noise spectrum. Here, we derive this ansatz using the two-site cavity method, providing two distinct derivations that offer complementary perspectives. The first decomposes each unit's activity into a linear component and a nonlinear residual, and shows that cross-covariances between residuals at distinct sites are strongly suppressed, so that residuals act as independent noise within a linear network. The second writes a self-consistent matrix equation for the covariance matrix. A naive Gaussian closure for the joint statistics of activity at distinct sites gives the wrong equation; the cavity method separates Gaussian and non-Gaussian contributions, which enter at the same order, and produces the correct one. We verify the predictions numerically across a range of network sizes. These results extend linear equivalence from feedforward high-dimensional nonlinear systems, where the weights being analyzed are independent of their inputs, to recurrent networks, where the activities depend on the same couplings that generate them.

Unlike for tunneling Josephson junctions, for which the current-phase relation is given by the sine function, with the critical current ($I_c$) and normal-state resistance ($R_N$) following the relation $I_cR_N=(π/2)\,Δ_0/e$ (where $Δ_0$ is the superconducting gap and electron charge is $-e$), mesoscopic Josephson junctions show more complex current-phase relations, with the skewness $S>0$, what is related to the presence -- in case the leads are in the normal state -- of transmission probabilities taking the values comparable to $1$. Here, we show that these features also appear for a superconductor-graphene-superconductor (S-g-S) junction in the disk-shaped (Corbino) geometry, when the magnetic field is adjusted such that $I_c\rightarrow{}0$ and $R_N\rightarrow{}\infty$. In such a case, the product $I_cR_N\approx{}1.85\,Δ_0/e$, and the skewness $S\approx{}0.14$. The results obtained from quantum-mechanical mode-matching analysis for the Dirac-Bogoliubov-De-Gennes equation are compared with simpler model assuming incoherent scattering between two circular interfaces separating the sample and the leads.

Reconstructing magnetization in nanoscale magnetic thin films is essential for developing next-generation memory, sensors, and various spintronic technologies. However, this remains challenging due to the ill-posed nature of the stray field inverse problem, i.e., there are infinitely many magnetization solutions to a given stray field distribution. Here, we demonstrate that a physics-informed deep image prior (DIP) framework, using a simple convolutional autoencoder conditionally achieves a reasonable qualitative and quantitative reconstruction of complex in-plane magnetization patterns from scanning NV magnetometry. We find that the orientation of user-defined masks implemented to restrict the reconstruction solution space dramatically affects convergence. The optimal alignment of the mask improves the reconstruction signal-to-noise ratio by up to $\SI{3}{\decibel}$, thereby also serving as a diagnostic tool. The DIP approach requires no pre-trained datasets and is considered computationally less intensive as compared to supervised learning approaches. We analyze both Landau and dipole domain structures in lithographically patterned Permalloy nanostructures by incorporating experimentally-guided spatial constraints. Complementary magnetic force microscopy measurements were carried out to support the Scanning NV measurements.

Semiconductor quantum dots (QDs) offer outstanding quantum-optical properties, making them highly attractive for quantum information technologies. However, combining wide-range electrical tunability, efficient photon extraction, elevated-temperature operation, monolithic silicon integration, and telecom-wavelength compatibility remains a major challenge. Here, we demonstrate electrically contacted circular Bragg grating (eCBG) resonators incorporating InGaAs QDs directly grown on silicon, enabling bright single-photon emission in the telecom O-band. Deterministic electron-beam lithography and a ridge-based vertical p--i--n diode architecture enable precise device integration and electrical control of individual emitters. The QD--eCBGs exhibit a quantum-confined Stark shift of approximately 16 nm (11 meV) at 4 K, representing a record for QDs embedded in nanophotonic structures at telecom wavelengths. This is achieved alongside a photon extraction efficiency of $(21.7 \pm 3.0)\%$ into the first lens, while maintaining excellent radiative properties and high single-photon purity, with $g^{(2)}(0)=0.0078 \pm 0.0012$ below saturation and $g^{(2)}(0)=0.0183 \pm 0.0021$ at saturation under pulsed excitation. Robust antibunching persists up to 77 K, with $g^{(2)}(0)=0.0663 \pm 0.0056$, enabling operation with liquid-nitrogen or compact Stirling cryocoolers. Furthermore, spatially separated QD--eCBGs can be electrically tuned into spectral resonance without degrading photon statistics. These results establish a silicon-compatible, electrically addressable telecom O-band quantum light platform combining wide spectral tunability, high single-photon purity, and elevated-temperature operation, providing a scalable route toward practical photonic quantum networks.

We identify a boundary-robust topological signature of open Floquet lattices: although nonadiabatic boundaries strongly reshape the transmission lineshape, the integrated left--right transmission asymmetry saturates to a plateau set by the bulk Floquet winding number. Its origin is a deep-bulk branch-population principle: in the long-sample limit, each propagating Floquet--Bloch branch is generically populated with unit weight, since true Floquet bound states are nongeneric. The robust observable is therefore the cumulative transmission imbalance rather than the boundary-sensitive transmission profile. We propose direct detection by cold-atom transmission spectroscopy. For electronic transport, the same asymmetry admits contact-model-dependent electrical readouts: a coherent Floquet--Landauer--Büttiker interpretation predicts a near-\(2ef\) response in weak SAW devices, whereas a blocking-factor post-processing yields a qualitatively different signal.

A system of interacting atoms is represented as an union of two subsystems, one of which is the system of atoms, and the other is an auxiliary scalar covariant field, which is equivalent to a given static interatomic potential of general form only in the non-relativistic approximation. It is shown that the auxiliary field is a superposition of Klein-Gordon fields, the parameters of which are related to singular points of the Fourier transform of the corresponding interatomic potential. The general form of the relativistic Hamiltonian system of interacting atoms is established. It is shown that the exact calculation of the relativistic partition function of a system of interacting atoms, taking into account the field degrees of freedom, reduces to renormalizing the parameters of the auxiliary field. It is established that the field degrees of freedom lead to a divergence in the total energy of a classical relativistic system - an analogue of the ultraviolet catastrophe. Quantization of the auxiliary field eliminates this divergence. The existence of a phase transition within the framework of relativistic quantum statistical thermodynamics has been proven.

Transition metal dichalcogenides (TMDs) attract significant attention as potential building blocks in next-generation electronic devices. On the other hand, a comprehensive understanding of how various defects affect local electronic properties under realistic operational conditions is yet to be formed. Here, we present results of electronic spectroscopy experiments performed on individual defects in the prototypical TMD molybdenum disulfide (MoS2) under ambient conditions, by way of conductive atomic force microscopy (C-AFM). Data acquired in the form of consecutive, high-resolution current maps at various bias voltages allow the assessment of local conductivity and differential conductance as a function of bias voltage for individual defects, the effects of which range from single atomic sites to several nanometers in lateral size. Characteristic behavior in spectroscopy data allows the categorization of observed defects into distinct groups, and their chemical identification as n-type and p-type transition metal substitutions of molybdenum atoms, as well as oxygen substitutions of sulfur atoms.

A process that images or measures bond energies in the critical Ising model can be in distinct measurement ``phases'', depending on the precision of measurement. We study the transition into the strong-measurement phase using replica field theory (an epsilon expansion around six dimensions) and numerical simulations in two dimensions. The results reveal multiscaling of correlation functions at the critical point, and a striking enlarged symmetry of the replica description. This is an analog of the Nishimori phenomenon in the Ising spin glass, in a distinct replica limit. The enlarged symmetry is present microscopically for certain measurement protocols, but more generally can emerge in the infrared, and it fixes the exact value of the exponent for the Edwards-Anderson correlator both in 2D and near the upper critical dimension. We also examine the epsilon expansion for models with power-law interactions and/or long-range measurement.

The organization of higher-order interactions plays a central role in shaping collective dynamics, yet a general structural principle governing contagion on hypergraphs remains lacking. Here we introduce a nesting coefficient that quantifies how lower-order interactions are embedded within higher-order ones, defining a continuum between simplicial complexes and random hypergraphs. Using a higher-order susceptible-infected-susceptible model, we show that increasing nesting lowers the activation threshold and suppresses discontinuous transitions, while weak embedding favors explosive behavior. We further demonstrate that correlations between nesting and interaction order modulate the onset of activity while only weakly affecting transition discontinuity. Analysis of synthetic and empirical networks reveals that nesting strongly predicts hysteresis, establishing it as a key structural determinant of phase transitions in higher-order systems.

One of the central goals in the emergent field of altermagnetism is the unambiguous experimental identification and characterization of altermagnetic order across a variety of compounds. This motivates exploring tools that can clearly distinguish altermagnets from antiferromagnets, based on symmetry signatures, and offer access to the dominant orbital character (e.g., $d$-wave vs. $g$-wave) of the magnetic order parameter. In this work, we theoretically explore the potential of noise magnetometry for this task, studying contributions from both magnons and itinerant electrons in different regimes and scenarios. While altermagnetism and antiferromagnetism also lead to different noise spectra for magnons, we find the most striking and symmetry-sensitive signatures in the charge fluctuations of itinerant altermagnets. Both for the homogeneous bulk case and in the presence of strain and/or around domain walls, we identify noise contributions that are only permitted by symmetry in the altermagnet and, thus, provide a unique signature of altermagnetism. Furthermore, the angular dependence of noise around domain walls also offers access to the orbital character of the altermagnet. On a more technical note, we discuss the role and relevance of lattice effects related to the dipole tensor. We hope that our work will help pave the way towards the clear experimental identification of altermagnetism across a wide range of candidate materials.

We study quantum tunneling through a potential barrier whose height fluctuates in time and is modeled by Gaussian white noise. We map the stochastic dynamics onto an equivalent time-independent Lindblad equation for the density matrix, allowing fully analytical solutions. For Schrödinger particles, noise introduces dissipation that suppresses Fabry-Pérot oscillations and yields an exponentially decaying transmission. Applying the same formalism to graphene, we demonstrate that noise induces a complex longitudinal wavevector within the barrier, leading to a strong suppression of transmission and Klein tunneling, even at normal incidence. Our approach promises improved control over Klein tunneling. These results demonstrate that noisy barriers can act as tunable dissipative elements, offering a pathway to enhanced control of electron transport in graphene-based devices. We also briefly discuss how our results could guide the design of graphene quantum dots for potential use in spin qubit devices.

We investigate the transient electromagnetic field radiated by a pulsed vertical electric dipole above a lossy half-space and identify its time-domain signatures associated with the Zenneck wave. Starting from the classical Sommerfeld representation, we derive a causal time-domain formulation based on the double-deformation technique, with successive contour deformations in the transverse-wavenumber and frequency planes. This yields an explicit decomposition of the field into source-pole, loss-pole, modal-pole, and residual steepest-descent contributions. The resulting expressions exactly satisfy causality and are validated against a reference solution obtained through a standard double inverse transform. The analysis shows that one modal contribution, generated by the frequency-plane deformation and related to the frequency-domain Zenneck pole, exhibits reduced-time invariance and a spatial attenuation consistent with a surface-wave component. Under suitable source and observation conditions, this term can dominate the field over a broad and physically relevant finite late-time interval. At the same time, for the considered damped-sinusoidal excitation, the strict asymptotic tail at fixed distance remains algebraic of order $t^{-5/2}$, with contributions from both the residual continuous spectrum and the modal-pole family. These results provide a rigorous and physically interpretable time-domain manifestation of the frequency-domain Zenneck wave in the pulsed half-space problem.

We present the comparative experimental analysis of frequency dependencies of linear (Lamé) and nonlinear (Murnaghan) elastic moduli of polystyrene, PMMA and polycarbonate. The measurement methodology, based on the acousto-elastic effect, provided data on variations of these moduli in block samples of the polymers in the frequency range of 0.45-3 MHz. In all the three polymers the linear Lamé moduli demonstrated moderate rise with frequency, most pronounced rise was observed in modulus $λ$ of PMMA in about 35%. The frequency dependencies of Murnaghan moduli were considerably nonlinear. At higher frequencies above ~1 MHz no significant variations of the Murnaghan moduli occurred, while at lower frequencies the absolute values of the moduli $l$ and $m$ demonstrated rapid rise, more pronounced for the modulus $l$. At the same time the absolute values of the modulus $n$ decreased and demonstrated a tendency to become positive at lower frequencies. Both linear and nonlinear moduli of PMMA had higher values than those of PC and PS, with the latter two demonstrating close values of both types of moduli. The potential origins of the differences in nonlinear elastic properties of the three polymers are discussed.

Profound finite size effects are observed in both the moment and ordering temperature in thin Co$_{70}$Zr$_{30}$ layers. The results are consistent with the presence of interface regions with reduced magnetic interactions and moment. The extension of this region is determined to be around 1 nm thick at each interface. Above and near the apparent critical temperature, the magnetic properties can be understood in terms of Griffith phases.

We numerically study topological effects of electromagnetic (EM) waves in a two-dimensional (2D) non-Hermitian photonic crystal (PhC) composed of lossy magneto-optical materials. In this system, not only the EM wavefunctions but also the complex eigenfrequencies exhibit nontrivial topological properties. We demonstrate that the non-Hermitian skin effect, protected by point gaps in the complex eigenfrequency spectrum, emerges at both the edges and corners of truncated structures. This phenomenon has no counterpart in Hermitian systems. In addition, we identify non-Hermitian topological edge states originating from the nontrivial topology of the bulk bands. While most previous studies of non-Hermitian topology have focused on tight-binding models, our work addresses a continuous photonic system, providing a more realistic platform and offering a concrete route toward experimental realization of non-Hermitian effects.

We study how conservation laws shape the spreading of quantum coherence in many-body dynamics. Focusing on $U(1)$-symmetric random circuits, charge-and-dipole conserving circuits, as well as ergodic Hamiltonian dynamics, we probe coherences both globally, via the participation entropy, and locally, via the relative entropy of coherence. Combining exact vector evolution, matrix product state simulations, and replica tensor networks methods, we find that conservation laws replace the logarithmic saturation of unconstrained circuits with slow hydrodynamic relaxation of the global coherence measures. Locally, symmetry-constrained circuits show a clean rise-peak-fall structure whose peak time grows algebraically with subsystem size. In contrast, ergodic Hamiltonians broaden the peak into an extended plateau at larger subsystems, highlighting a qualitatively distinct mechanism. Coherence thus emerges as a sensitive probe of symmetry-constrained thermalization, linking quantum resource dynamics to many-body transport.

Recently, it was predicted that nonreciprocal magneto-electric effect in antiferromagnetic multilayered metamaterials occurs in two distinct versions. One is the conventional axion response, while another one is dual axion response captured by electrodynamics with magnetic charge. Here we investigate a model condensed matter system of spins coupled through antiferromagnetic exchange interaction and derive its effective electromagnetic properties. We predict that this system gives rise to the emergent dual axion field, support our conclusion by numerical simulations and put forward candidate materials.

We study the statistics of rewired random regular graphs within the canonical ensemble, in which the average number of triangles is controlled by the fugacity $λ$. Using mean-field arguments, we estimate the location of the transition point $λ^{-}$. We focus primarily on the triangle-enriched phase above $λ^{-}$, where the graph becomes a loosely connected web of clusters. This inter-cluster network is scale-free and exhibits a power-law vertex degree distribution $P(d) \sim d^{-γ}$, with $γ\approx 2$ independent of degree and size of the graph. We attribute this behavior to an "emergent preferential attachment" for triangle motifs, derive the exponent $γ$ with a mean-field approach, and speculate on a possible connection between the typical topology of inter-cluster triangles and Efimov states in a conformally invariant potential.

Phonons can carry angular momentum and exhibit chirality through the circular polarization of atomic motion. This enables a phonon-mediated spin Seebeck effect (SSE) via the conversion of phonon angular momentum into electron spin angular momentum. In this Letter, we develop a theoretical framework for calculating the spin current in a normal-metal (NM)-chiral-insulator (CI) heterostructure within the nonequilibrium Green's function formalism. We discuss the influence of (i) the thermal bias across the NM-CI interface, (ii) the chemical potential of the NM, and (iii) the insertion of an additional interfacial layer, on the spin transport properties. We identify two remarkable nonlinear spin transport phenomena: negative differential SSE and spin-current rectification. The negative differential SSE arises from the competition between the thermal bias and the thermally excited electron density. The spin-current rectification suggests the possibility of realizing a thermally controlled spin diode. We also find that the spin transport behavior is closely associated with an effective interfacial spectral density. This work provides a novel route toward thermally controlled spintronic devices using chiral phonons.

Optical cavities play a central role in photonic and quantum technologies by enhancing light-matter interactions. In semiconductor microcavities, achieving high quality (Q) factors typically relies on sophisticated epitaxial growth techniques, such as molecular beam epitaxy, which offer atomic-scale precision but are costly and limited in material compatibility. For dielectric microcavities, high Q factors can be achieved using dielectric Bragg mirrors. However, conventional deposition techniques for the top mirrors, such as plasma-enhanced chemical vapor deposition or sputtering, can damage embedded emitters. This limitation is particularly severe for van der Waals materials, especially atomically thin semiconductors. Moreover, the conventional top-mirror deposition can cover or degrade predefined metal contacts. Recovering electrical access typically requires additional lithography and etching steps. Here, a deterministic dry-transfer approach is developed to fabricate complete dielectric microcavities using both top and bottom SiO_2/TiO_2 Bragg mirrors without post-growth lift-off processes, reaching a Q factor ~ 4x10^3. Using a WS_2 monolayer as the active medium, clear signatures of strong exciton-photon coupling are observed at both room temperature and cryogenic temperatures. These results demonstrate an efficient cavity fabrication approach that preserves the integrity of the emitter of layered materials, enabling next generation integrated photonic devices.

We develop an angular momentum representation and implementation of the Clifford+T set of unitaries for qutrits. We show that local gates from this set can be realized by the sole use of suitable rotations and one-axis-twisting operations, which are at most quadratic in the angular momentum operators and thus can be experimentally realized in many quantum systems. Controlled rotations are shown to only require linear angular momentum couplings and, as a consequence, the full qutrit Clifford+T set is shown to be expressed solely in terms of two-body angular momentum couplings, rotations and one-axis-twisting operations. By employing the Jordan-Schwinger map, we show an analogous implementation in terms of bosonic modes, improving on the number of modes with regard to a previous scheme. Moreover, we employ the cross-Kerr interaction in order to obtain any qutrit Clifford+T gate for bosonic modes. We illustrate our findings with simple schemes for preparing entangled states of interest.

Environmental noise can enhance transport, an effect known as environmental noise-assisted quantum transport. Most theoretical studies focus on optimizing system parameters under spatially uniform system-environment coupling. Here, instead, we optimize the environmental noise itself by allowing for site-dependent dephasing. We investigate steady-state transport in one-dimensional lattices with either ramped or disordered energy landscapes, considering both short- and long-range coherent tunneling. In the absence of environmental effects, in the thermodynamic limit these systems can exhibit localization, and thus suppressed transport, arising from destructive interference. Using a Lindblad master equation framework, we implement local dephasing optimized to maximize steady-state population flux. We find that for ramp potentials, short-range tunneling favors selective dephasing on alternating sites, whereas long-range tunneling benefits from a dephasing profile that increases with distance from the injection site. In energetically disordered systems, strongly detuned sites require enhanced local dephasing under short-range tunneling to facilitate transport. In all cases, we find that site-optimized dephasing allows higher transport efficiency than uniform dephasing, and it is accompanied by increased spatial delocalization of the steady state. Our results provide microscopic insight into the interplay between coherent dynamics and environmental noise. Dephasing broadens energy levels locally, helping to overcome detuning and destructive interference. More generally, we establish spatially-structured environmental noise as a strategy for controlling both quantum transport and state coherence in open systems.

We derive equations of motion for capital deepening in a competitive economy directly from accounting identities, without assuming a production function. A profit imperative $η^* \equiv (w/κ+ 1/τ)/(1-f_p)$ sets the minimum viable capital productivity, where $η= Y/K$ [yr$^{-1}$] is capital productivity, $κ= K/L$ is capital per worker, $w$ is the wage rate, $τ$ is the capital lifetime, and $f_p$ is the production tax share. Four coupled relaxation equations govern $κ$, $η$, the frontier productivity $η_{\rm new}$ of new investment, and the labor share $q \equiv w/y$, with the sandwich constraint $η^* \leq η_{\rm new} \leq η$ maintained as an exact invariant. The frontier equation separates two physically distinct channels: a structural cheapening channel ($μ$, always active, drives $η_{\rm new}$ downward) and a productivity channel ($φ$, historically zero). Calibration against BEA 2-digit NAICS sector data (1998--2023) confirms $φ= 0$ for all identifiable sectors over 25 years; the 75-year postwar record extends this finding across four capital lifetimes. A step $φ= 0.01$\,yr$^{-1}$ -- a 1\%/yr improvement in new-capital productivity, modest but historically unprecedented -- nearly doubles the aggregate growth rate within one capital lifetime, a falsifiable prediction with a precise observable signature: upward-curving $η(t)$ in BEA sector data. Firms near the zero-profit threshold have a cash martingale, predicting establishment exit rate $\sim t^{-1/2}$; convolved with the Zipf firm-size distribution~\cite{WP}, this yields firm exit rate $\sim t^{-1/2}\!\log t$ with apparent exponent $b = 0.295 \pm 0.03$, confirmed against BDS data with no free parameters.

Crystalline noble metal flakes are emerging as versatile platforms in nanophotonics, enabling a broad range of optical phenomena and applications. Their atomically flat surfaces, high crystallinity, and superior optical quality open new avenues in advanced plasmonics, quantum light generation, and hybrid photonic systems. In contrast to conventional polycrystalline metal films, which typically suffer from higher optical losses due to grain boundaries, surface roughness, and structural disorder, these monocrystalline flakes provide minimal scattering and enhanced performance. They serve as templates for precise nanostructuring through techniques like focused-ion beam (FIB) milling and are crucial for advanced applications in sensing and optoelectronics. Additionally, they facilitate frontier research in quantum plasmonics, enabling fundamental studies of nonlocal optical effects and the generation of nonclassical light. Furthermore, the well-defined $\{111\}$ facets of these flakes host Tamm--Shockley surface states that support 2D plasmons coexisting with bulk modes. At near-infrared wavelengths and beyond, crystalline flakes act as nearly ideal metallic mirrors, featuring surface roughness limited only to atomic terrace steps, making them highly suitable for integration with 2D materials in hybrid photonic architectures. This review surveys the key roles these flakes play, highlighting recent developments and discussing future prospects while emphasizing their unique benefits in addressing fundamental and applied challenges in modern nanophotonics.

The distribution of income and wealth in developed economies exhibits a robust two-class structure: an exponential (Boltzmann--Gibbs) bulk covering $\sim\!97\%$ of the population, and a power-law (Pareto) tail in the upper $\sim\!3\%$. We derive this structure from first principles via an explicit mechanistic chain: Gibrat's law for firm growth implies a Zipf firm-size distribution; maximum entropy applied to within-firm wages, combined with mixture aggregation across firms, yields a Boltzmann--Gibbs income distribution with temperature $T_y$ for employees; additive-noise wealth dynamics with a reflecting wall at zero produce a Boltzmann--Gibbs employee wealth distribution with temperature $T_w$. For firm owners, multiplicative capital returns produce a Pareto wealth tail with exponent $α_w = 1/θ$, where $θ$ encodes how total returns scale with firm size. The empirical value $α_w \approx 1.30$ \cite{Yakovenko2009} is reproduced with no tuned parameters from the observed firm value scaling $V = V_0(s/s_0)^{0.77}$~\cite{Axtell2001}, and simultaneously yields the first quantitative estimate of the returns-per-employee size exponent: $ζ= θ- 1 \approx -0.23$. For empirical values $ν\approx 0.3$, $c \approx 0.81$, $k \approx 0.15$ (BEA long-run savings rate $\approx 5\%$), the model gives $T_w/T_y \approx 1.7\,\text{yr}$, i.e.\ lower-class households hold roughly 1--2 years of income as wealth, with the precise ratio depending on savings and tax rates and testable cross-country. As a parameter-free empirical test, firms near zero profit have a cash martingale whose first-passage time gives establishment exit rate $\sim t^{-1/2}$; convolving with the Zipf firm-size distribution yields firm-level exit rate $\sim t^{-1/2}\!\log t$, with apparent exponent $b = 0.295 \pm 0.03$, confirmed against BDS firm-age data with no free parameters.

This study investigates the nanoscale conductivity of encapsulated monolayer graphene at temperatures down to 5 K and magnetic fields of up to 1 T. We use the scattering-type scanning near-field optical microscopy (s-SNOM) technique to probe magnetic-field-dependent responses from graphene close to charge neutrality in the terahertz spectral region. We observe the near-perfect high-$q$ reflector behavior of graphene but with subtle changes by the presence of magnetic fields. Measurements align with calculations of the magneto-optical conductivity and the near-field spectroscopic contrast that describes the field-tunable cyclotron resonance of Dirac fermions. Our result provides an initial step toward understanding temperature and magnetic-field effects on nanoscale terahertz transport in two-dimensional quantum materials.

Common belief, confirmed by existing experiments, is that arbitrarily weak disorder should lead to spatial localization of eigenmodes of scalar wave equations when wave propagation is two-dimensional (2D). We predict that contrary to this belief, a localization-delocalization transition can take place for light scattered by two-level atoms placed at random positions in the middle plane of a parallel-plate 2D waveguide fed by its fundamental transverse-magnetic (TM) mode (electric field polarized perpendicular to the waveguide and atomic planes). This transition, driven by near-field dipole-dipole interactions between atoms, occurs upon increasing the areal number density of atoms beyond some critical value. A finite-size scaling analysis of the transition yields an estimate of its critical exponent $ν$ = 1.4 $\pm$ 0.2.

This work presents a unified viscoelastic-viscoplastic continuum framework for modeling rate-dependent granular flows across regimes. The formulation incorporates two distinct rate-dependent mechanisms, namely micro-inertia and viscoelastic dissipation, within a single continuum description. A central contribution is an explicit link between the coefficient of restitution and a continuum viscosity, derived from an analysis of wave attenuation in granular assemblies, thereby establishing a direct connection between particle-scale collision physics and macroscopic damping. This relation is introduced while retaining inertia-dependent plastic flow governed by the classical $μ(I)$ rheology. The constitutive model is constructed by meticulously partitioning elastic and viscous responses within the model and corresponding stress-update routine, such that viscous dissipation governs wave propagation and collisional processes without altering the plastic flow rule. The framework is implemented within the material point method to simulate transient processes involving large deformations, material separation, and subsequent reconsolidation. A range of numerical examples, including steady, transient, vibrational, and impact-driven flows, demonstrates that the model captures wave propagation, diffusion, and rate-dependent granular behavior within a unified continuum setting.

Block-sequential continual learning demands that a single model both protect prior solutions from catastrophic forgetting and efficiently infer at inference time which prior solution matches the current input without task labels. We present Functional Task Networks (FTN), a parameter-isolation method inspired by structural and dynamical motifs found in the mammalian neocortex. Similar to mixture-of-experts, this method uses a high dimensional, self-organizing binary mask over a large population of small but deep networks, inspired by dendritic models of pyramidal neurons. The mask is produced by a three-stage procedure: (1) gradient descent on a continuous mask identifies task-relevant neurons, (2) a smoothing kernel biases the result toward spatial contiguity, (3) and k-winner-take-all binarizes the resulting group at a fixed capacity budget. Like mixture-of-experts, each neuron is an independent deep network, so disjoint masks give exactly disjoint gradient updates, providing structural guarantees against catastrophic forgetting. This three-stage procedure recovers the sub-network of a previously-trained task in a single gradient step, providing unsupervised task segmentation at inference time. We test it on three continual-learning benchmarks: (1) a synthetic multi-task classification/regression generator, (2) MNIST with shuffled class labels (pure concept shift), and (3) Permuted MNIST (domain shift). On all three, FTN with fine grained smoothing (FTN-Slow) results in nearly zero forgetting. FTN with a large kernel and only 2 iterations of smoothing (FTN-Fast) trades off some retention for increased speed. We show that the spatial organization mechanism reduces the effective mask search from the combinatorial top-k subset problem in O(C(H,K)) to the complexity of a near-linear scan in O(H) over compact cortical neighborhoods, which is parallelized by the gradient-based update.

Functional connectivity varies across individuals due to genetic and environmental factors, yet classical twin models typically confound non-shared environment with measurement error and are largely limited to resting-state analyses. We hypothesized that: i) explicitly modeling measurement error from repeated fMRI sessions enables more accurate application of classical twin models (ACE/ADE) to functional connectivity; ii) model applicability depends on scan-length and parcellation granularity; iii) genetic and environmental effects on functional connectomes show differentiated functional modules across conditions. We extended ACE/ADE models to include a repeated-scan derived error term by analyzing monozygotic and dizygotic twins from the Young-Adult Human Connectome Project dataset. Genetic and environment variance components were estimated for all functional couplings across resting-state and task conditions, integrated across conditions using a minimum-error criterion, and analyzed using multilayer community detection across resolution scales. Functional couplings segregated into distinct categories characterized by shared environmental, additive, dominant, or epistatic influences, with a substantial fraction not meeting twin-model assumptions. Integrating across conditions revealed hierarchical community structure in genetic and environmental components observed across community resolution scales. Incorporating measurement error into twin models improves interpretability and applicability at the functional connectome level, revealing that genetic and environmental influences are structured into coherent, multiscale brain networks.

Information theory is a powerful framework to capture aspects of dynamical systems with multiple degrees of freedom. Mathematically, the dynamics can be represented as a continuous curve $\mathcal{C}$ on a suitable hyperplane in flat space and the Fisher information provides the norm of an infinitesimal displacement along this curve. In many applications, however, we do not have direct access to $\mathcal{C}$. Instead, we have to reconstruct the latter from a time-series of measurements (obtained as samples of size $n$), which are represented by an ordered set of points $\widehat{\mathcal{C}}$ on the same hyperplane. In this work, we calculate the bias of the Fisher information for large $n$, which provides a quantitative estimation for how accurately the dynamics of a system can be reconstructed from a given set of sampled data. Based on this result, we show that a clustering of the degrees of freedom reduces the bias and thus improves the accuracy with which the new system can be described with the same data. Inspired by a recent proposal for such a clustering, we provide a quantitive assessment of the loss of information, which allows to estimate how much information about the dynamics of a system can reliably be extracted based on a given set of data. We illustrate our findings in the case of a simple compartmental model. Although the latter is inspired by epidemiology, the results of this work are applicable to very general dynamical models with multiple degrees of freedom.

Stride-to-stride fluctuations in human walking carry a fractal correlation structure that reverses sign under external cueing: self-paced gait is persistent, whereas metronomic or visually cued gait is anti-persistent. Three decades of detrended fluctuation analysis (DFA) have established this reversal as a scaling-exponent shift, but DFA cannot distinguish genuine long-memory dynamics from short-memory autoregressive moving-average (ARMA) processes that produce the same apparent exponent. We fit the full eight-model ARFIMA(1,d,1) family to stride interval and stride speed series from three independent datasets (N = 70 subjects) spanning overground walking, fixed-speed treadmill walking, metronomic and visual cueing, and graded positional constraint. Model evidence is aggregated through BIC-based Schwarz weights, and the fractional differencing parameter d together with the autoregressive and moving-average coefficients phi and theta are estimated by Bayesian model averaging. Three findings emerge. Long-memory specifications decisively outweigh ARMA alternatives under both persistent and anti-persistent conditions, establishing cued gait anti-persistence as a genuine fractional phenomenon. DFA alpha overestimates d + 0.5 by 0.25 to 0.34 alpha units owing to short-memory components that DFA conflates with long-memory persistence, establishing ARFIMA-based decomposition as the more informative estimator. The estimated (d, phi, theta) parameters are consistent with a corrective sensorimotor model in which a fractal intrinsic generator, a reactive feedback correction, and a motor-delay component together shape stride-interval fluctuations, with the strength of the correction varying according to the type and tightness of external constraint. A unified mechanistic account of these parameter ranges across rhythmic, spatial, and unconstrained conditions remains an open question.

Citizen science is transforming how cognitive scientists study the human mind, and video games are at the heart of this shift. By embedding experimental tasks into engaging, game-like experiences, researchers can reach large, diverse populations while collecting rich behavioral data outside the lab. In this review, we explore how citizen science video games bridge the gap between players and participants, turning entertainment into large-scale cognitive research. Drawing on recent projects such as Sea Hero Quest and The Music Lab, we outline the key benefits of this approach: scalability, ecological validity, and public engagement. We also examine the challenges of designing games that are scientifically rigorous, ethically sound, and meaningful for both researchers and players. Through professional game developer insights, we highlight what it takes to develop a successful citizen science video game for cognitive science, and why this approach is still rare in the literature.

Motivation: Multi-omics integration can improve cancer subtyping, but modality informativeness and noise vary across cancer types and patients. Existing graph-based methods optimize modality weights jointly with the classification objective and therefore lack independent reliability estimates, so low-quality omics distort patient similarity graphs and amplify noise through message passing. Results: We propose CMGL, a two-stage framework that estimates per-sample modality reliability through evidential deep learning and uses the frozen confidence scores to guide cross-omics fusion and graph construction. On four MLOmics cancer-subtype tasks and the 32-class pan-cancer task, CMGL consistently improves over the strongest baseline, surpassing it by 4.03% in average accuracy on the four single-cancer tasks. Its representations recover the PAM50 intrinsic subtypes of breast invasive carcinoma (BRCA), and the BRCA-trained model transfers without fine-tuning to kidney renal clear cell carcinoma (KIRC), stratifying patients into prognostically distinct groups.

We calculate the moments and response functions of a nonlinear random recurrent neural network in the large $N$ limit. Our approach does not require averaging over synaptic weights and gives the first nontrivial term in a $1/\sqrt{N}$ expansion of general intensive-order correlation functions, proving a recent conjecture by Shen and Hu as a special case. Our results provide an analytical link between synaptic connectivity, correlations in spontaneous activity, and the response of a network to small perturbations.

Comparative analyses of phylogenetic trees typically require identical taxon sets, however, in practice, trees often include distinct but overlapping taxa. Pruning non-shared leaves discards phylogenetic signal, whereas tree completion can preserve both taxa and branch-length information. This work introduces a polynomial-time algorithm for set-wide completion of phylogenetic trees with partial taxon overlap. The proposed method identifies and extracts maximal completion subtrees that frequently appear across the source trees and constructs a weighted majority-rule consensus. Branch lengths are scaled using rates derived from common leaves. Each consensus subtree is inserted at the position that minimizes the quadratic distance error measured against information from the source trees, with candidate positions restricted to the original branches of the target tree. We demonstrate that the algorithm runs in polynomial time and preserves distances among the original taxa, yielding a unique completion that is order-independent with respect to the processing order of target trees. An experimental evaluation on amphibians, mammals, sharks, and squamates shows that the proposed method consistently achieves the lowest distance to the subset reference trees across subsets among all methods, in both topology and branch lengths. An open-source Python implementation of the proposed algorithm and the biological datasets utilized in this study are publicly available at: https://github.com/tahiri-lab/overlap-treeset-completion/.

Developing robust and clinically reliable EEG biomarkers requires evaluation frameworks that explicitly address cross population generalization in multi site settings such as Parkinsons disease (PD) detection. Models trained under i.i.d. assumptions often capture population specific artifacts rather than disease relevant neural structure, leading to poor generalization across clinical cohorts. EEG further amplifies this challenge due to low signal to noise ratio and heterogeneous acquisition conditions. We propose a population aware evaluation framework to assess the robustness and clinical reliability of EEG biomarkers under distribution shift. Using an n gram expansion strategy, we enumerate all cross population train test configurations across five independent cohorts, resulting in 75 directional evaluations. A nested cross validation design with integrated channel selection ensures prospective biomarker identification without population leakage. Results show that cross population transfer is asymmetric and that both accuracy and biomarker stability improve with increasing training population diversity, achieving up to 94.1% accuracy on held out cohorts. A theoretical analysis based on mixture risk optimization and hypothesis space contraction explains these trends, showing that multi population training promotes population robust representations. This work establishes a principled framework for learning robust, generalizable, and clinically reliable EEG biomarkers for multi site biomedical applications.

Large-scale neuroscience is generating rich datasets across animals, brain areas and behavioral contexts, yet our modeling efforts remains fragmented across isolated experiments. We argue that understanding behavior requires integrative neurocybernetic models: understandable dynamical models that capture the closed-loop coupling of brain, body and environment, treat the brain as a controller pursuing latent objectives, represent structured variation across scales, and scale to heterogeneous datasets. Such models shift the goal from predicting neural recordings in isolation to inferring the organizing principles that govern neural and behavioral dynamics. We outline a practical route toward this goal by combining nonlinear state-space models and meta-dynamical extensions with scalable inference, knowledge distillation, mixed open- and closed-loop training, and connectomics-informed architectures. By pooling complementary constraints from recordings, behavior, perturbations and anatomy, integrative neurocybernetic models can provide statistical amplification, few-shot generalization, and mechanistic insight into shared dynamical structure, individual variation, and the control objectives that govern behavior. This agenda offers a model-centric path from fragmented data to a mechanistic science of how brains produce behavior.

The foundation of cognitive flexibility and higher-order intelligence lies in the functional structure and activity of brain networks, which can be dynamically configured by both external environments and internal states. However, decoding these dynamics from high-dimensional neural data remains a challenge. In this study, we propose a computational framework using Recurrent Neural Networks (RNNs) with neural dynamic constraints to model source-localized resting-state EEG data from $114$ participants. We aim to clarify the "triple brain network configurations" driven by exogenous and endogenous factors, including external stimuli, information processing tasks, and spontaneous activities. Our model identifies the parietal network as a critical hub supporting these multiple configuration patterns. Furthermore, we reveal that the anterior and posterior parietal regions exhibit distinct functional specializations under different stimulus modalities. By formalizing a triple configuration framework, this work separates latent factors of brain dynamics and underscores the computational significance of parietal regions in orchestrating higher-order intelligence.

Collective decision-making arises from individual agents integrating their own personal observations with information obtained from social partners. In many biological systems that exhibit collective decision-making, the process by which social information is produced, transmitted, and used is subject to two key constraints. First, individuals often do not observe the internal states or personal observations of their neighbors; instead, they observe neighbors' discrete actions. Second, agents often have limited attention, such that, at any given moment, only a subset of social partners influences decisions. Using methods from nonlinear dynamics, we show that either of these constraints can cause collective accuracy to become extremely sensitive to the weight individuals place on the information they receive from others. This sensitivity arises from the spontaneous formation of echo chamber-like states in which individuals receive and transmit homogeneous social messages. Under such conditions, collectives become locked in self-reinforcing states that prevent them from tracking changes in the environment. We reveal the mathematical basis of this phenomenon, and show that it emerges not only in generic models of collective decision-making but also in models developed to describe specific biological systems, including neural circuits, eusocial insect colonies, and mobile animal groups. Finally, we identify biologically plausible mechanisms through which individuals may reduce the risk of echo chamber formation and achieve robust yet sensitive collective decisions without requiring fine-tuning parameters. Our results reveal how fundamental constraints on communication shape the dynamics and reliability of collective decisions across diverse biological systems.

Compartmental epidemic models, grounded in mass-action kinetics, often assume homogeneous mixing. Although this neglects network structure, recent results show that for Poisson random graphs, the classical SIR model, especially the susceptible decay curve, matches the susceptible decay dynamics of its network counterpart. Motivated by this, we investigate whether the extended SIRI model with relapse from the recovered class admits a similar correspondence. SIRI dynamics arise in sevaral scenarios like spread of diseases with reactivation and behavioral contagion with relapse. We derive parameter relationships under which the pairwise SIRI model on a Poisson network closely follows the mass-action ODE trajectories. When transmission per contact is small relative to recovery, the susceptible and infectious trajectories of both systems align. This establishes conditions under which nonlinear SIRI dynamics on networks can be effectively approximated by tractable mean-field equations.

Progress in vision research has been slower downstream than upstream of primary visual cortex (V1). Traditional frameworks have largely overlooked a central constraint: only a tiny fraction of retinal input is recognized. Thus, to a first approximation, vision is better formulated as looking and seeing through a bottleneck. Looking, mainly by the peripheral visual field, selects visual information to enter this bottleneck, largely via gaze shifts that center selected contents at fovea. Seeing, mainly by the central visual field, recognizes this content. Converging evidence suggests that V1 initiates the bottleneck and contributes to looking by generating a bottom-up saliency map that guides saccades exogenously, and that top-down feedback along the visual pathway, targeting mainly the representation of the central visual field, refines seeing. Progress will accelerate through falsifiable theories that explicitly link behavior with neural substrates, and by experimental designs that avoid forced fixation and precisely track gaze.

Homologous proteins evolve from a common ancestral sequence, constrained by intricate patterns of co-evolving residues. Accurate reconstruction of evolutionary histories remains a challenge, primarily due to the inability of the existing approaches to capture long-range coevolutionary ties and lack of a precise metric to represent the evolutionary distance between sequences. Standard approaches are based on p-distance or substitution-corrected measures such as Jukes-Cantor. These methods saturate in cases of deep evolutionary divergence, losing all evolutionary signal after enough time. We present HyperEvoGen, a Poincaré variational autoencoder with adversarial training, hyperbolic latent geometry, and a compound loss function that learns evolutionarily meaningful representations from single-family alignments. The arrangement of protein sequences in HyperEvoGen's hyperbolic embedding aims to preserve phylogenetic structure and produce latent distances which scale with true evolutionary divergence. HyperEvoGen enables fast, scalable modeling of protein evolution while preserving hierarchical relatedness in a geometry-aware representation. On Potts-coupled simulation benchmarks, it produces more accurate ancestral reconstructions than conventional baselines, and offers higher-quality sequence generation with less training time than Potts models. This combination of accuracy and throughput supports large-family evolutionary studies and accelerates design-oriented applications.

Different strains competing for a common pool of susceptible individuals is a key problem in mathematical epidemiology. To address this problem, we investigate a two-strain model within a Susceptible-Infected-Recovered (SIR) framework. While classical deterministic theory predicts that the basic reproduction number fully determines selection, we show that stochastic effects play a key role in the dynamics. We discover that stochastic fluctuations can reverse the deterministic advantage even far from the quasi-neutral regime. Further, we find that stochasticity drastically reduces fixation times from years, in the deterministic case, to days. The fixation time is non-linearly proportional to the noise intensity and the distance from the quasi-neutral regime, following a universal rule obtained from a scaling law. The nature of the problem and the equations allow us to interpret the competition as a dynamical evolution around an effective potential, with the potential barrier corresponding to the unstable manifold associated with the coexistence. Even in a stable situation of dominance of one strain, the noise can induce crossings through the potential. We find that the reversal can occur even far from the quasi-neutral regime with significant probability.

The escalating data rate demands of future wireless communications necessitate the deployment of extremely large aperture arrays (ELAA) at base stations. In such configurations, wireless channel characteristics are profoundly influenced by the near-field wavefront curvature of electromagnetic waves. While numerous channel estimation algorithms have been proposed for standard spherical wavefronts, practical electromagnetic waves reflected by curved surfaces exhibit anisotropic wavefront curvature. This anisotropy causes a severe deviation from the ideal spherical model, leading to substantial accuracy degradation in existing channel estimation techniques. To tackle this challenge, this paper proposes a novel, low-complexity channel estimation algorithm tailored for the anisotropic wavefront channel (AWC), termed MACAW. We first mathematically formulate the AWC model. Subsequently, rather than relying on conventional matching-based framework, we develop an matching-free algorithm based on harmonic analysis. This approach successfully circumvents the prohibitive computational overhead associated with massive dictionary construction and matching pursuits. Simulation results demonstrate that the proposed algorithm significantly outperforms existing approaches in AWC scenarios, while maintaining effectiveness in spherical wave channels. Furthermore, achieving an identical estimation accuracy, our method drastically reduces pilot overhead and time complexity, and exhibits enhanced robustness in low signal-to-noise ratio (SNR) regimes.

This paper investigates a joint hybrid digital-analog beamforming and trajectory design for a cell-free multi-static integrated sensing and communication (ISAC) system supported by multiple unmanned aerial vehicles (UAVs).Specifically, these UAVs cooperatively serve ground users and perform multi-static sensing to detect the target.We formulate a weighted sum-rate (WSR) maximization problem by jointly optimizing the hybrid beamformers and the UAV trajectories.This joint design explicitly accounts for practical constraints, including transmit power budgets, sensing signal-to-noise ratio (SNR) requirements, UAV kinematic constraints, and both continuous and discrete phase shifters.In particular, we reformulate the original complex problem into a solvable form that can be addressed using the penalty dual decomposition (PDD) method.Simulation results demonstrate that the proposed design achieves performance close to that of the fully digital (FD) scheme and significantly outperforms other schemes.Furthermore, leveraging UAV mobility and multi-static cooperation provides crucial spatial degrees of freedom, effectively avoiding WSR degradation under limited transmit power or strict sensing requirements.

In large antenna arrays, hardware power consumption becomes a dominant design constraint, making energy efficiency (EE) a first-class objective alongside spectral efficiency (SE). Microwave linear analog computer (MiLAC)-aided beamforming, whose front end is a passive reciprocal stream-to-antenna network, addresses this tension by reducing the active radio-frequency chain count to the stream number, at a moderate SE cost. Despite this promise, no EE optimization framework has been established for MiLAC-aided beamforming that accounts for digital-to-analog converter quantization noise and post-quantized transmit power. We fill this gap for downlink multiuser multiple-input single-output (MU-MISO) systems by formulating quantization-aware EE maximization over the MiLAC-feasible beamformer and characterizing the resulting SE-EE tradeoff. Three contributions follow. First, we prove a row-space optimality property of the effective MiLAC-aided beamformer, yielding an equivalent reduced-dimension reformulation whose complexity scales with the stream number rather than the antenna number. Second, we develop a low-complexity Dinkelbach-weighted minimum mean-square error algorithm aided by projected gradient descent that is guaranteed to converge to a stationary point. Third, we cast the SE-EE tradeoff as a multi-objective problem and trace its Pareto boundary via a weighted-sum method that combines an alternative reduced-dimension coordinate with auxiliary-variable successive convex approximation, yielding convex per-iteration subproblems with guaranteed convergence. Numerical results on a DeepMIMO v4 deployment show MiLAC-aided beamforming substantially improves EE over digital and hybrid benchmarks at a moderate SE cost and significantly expands the achievable SE-EE operating region.

Direct Device-to-Satellite (D2S) communications, which enable direct satellite connectivity with unmodified user equipment (UE), not only expand global coverage but also reshape the evolution of future access networks. However, D2S links face fundamental challenges due to inherently low signal-to-noise ratios (SNRs) and limited spatial multiplexing gains arising from near line-of-sight propagation, both of which severely constrain achievable spectral efficiency. Despite the lack of spatial multiplexing, this work shows that aggressive non-orthogonal transmissions, where multiple users (e.g., four) transmit concurrently over the same frequency resources, even to a single receive antenna, can unlock substantial capacity gains that remain entirely unexploited by existing systems. Realizing these gains in practice, however, requires receiver architectures that, to the best of our knowledge, have not yet been developed. To this end, we introduce NL-COMM-Sat, an efficient and flexible framework that overcomes this limitation by enabling aggressive non-orthogonal signal transmissions. In contrast to conventional non-orthogonal multiple access (NOMA) schemes, NL-COMM-Sat supports more than two UEs per receive antenna on the same frequency resource. The framework revisits optimal receiver design principles and proposes computationally efficient processing schemes that translate previously unexplored theoretical gains into tangible throughput improvements, even under realistic channel estimation errors and high-mobility Doppler conditions. Our evaluation shows that NL-COMM-Sat achieves up to a 2x increase in spectral efficiency compared to orthogonal multiple access and NOMA baselines across all considered SNR and Doppler regimes, even with a single-antenna receiver and user speeds of up to 500 km/h.

Electroencephalography (EEG) is highly susceptible to artifact contamination, such as electrooculographic (EOG) and electromyographic (EMG) interference, which severely degrades signal quality and hinders reliable interpretation in applications including neurological diagnosis, brain-computer interfaces (BCIs), etc. Effective EEG denoising remains challenging because different artifact sources exhibit diverse and temporally varying distributions, together with distinct spectral characteristics across frequency bands. To address these issues, we propose BandRouteNet, an adaptive frequency-aware neural network for EEG denoising that jointly exploits band-specific processing and full-band contextual modeling. The proposed model performs band-wise denoising to explicitly capture frequency-dependent artifact patterns. Within this framework, we introduce a routing mechanism that adaptively determines where and to what extent denoising should be applied across temporal locations within each frequency band. In parallel, a full-band conditioner directly processes the original noisy EEG to extract global temporal context, producing both conditional parameters for modulating the band-wise pathway and a coarse-grained signal-level refinement to supplement the final reconstruction. Extensive experiments on the EEGDenoiseNet benchmark dataset demonstrate that BandRouteNet outperforms other methods under EOG, EMG, and mixed-artifact conditions in terms of Relative Root Mean Square Error (RRMSE) and Signal-to-Noise Ratio Improvement (SNR$_{\text{imp}}$) under unified experimental settings, while remaining highly parameter-efficient with only 0.2M trainable parameters. These results highlight its strong potential for high-performance EEG artifact removal in resource-constrained applications.

Quantitative analysis of the kinematic chain in sports motion is essential for performance evaluation and injury prevention. Conventional methods such as the kinematic-sequence (KS) and continuous relative phase (CRP) are confined to adjacent joint pairs and lack a unified framework for whole-body coordination, while segmental power-flow analysis requires force plates and inertial parameters that restrict it to laboratory environments. We apply Complex Hilbert Principal Component Analysis (CHPCA) separately to each motion phase (backswing and downswing) on markerless 3D pose estimation data, extracting the dominant whole-body phase pattern as a single complex eigenvector. The pipeline further includes a fully automatic signal-based phase segmentation (no priors on strike count or rest location) and an extension to 1,079 body-surface mesh vertices, so that the kinematic chain is represented as a continuous phase field across the body. On 14 hammer-striking trials of a single subject, the framework reveals (i) a trunk-anchored global phase architecture, (ii) a functional asymmetry between preparation and execution phases quantified by Mode-1 contribution (45.5% vs. 70.5%) and inter-trial Spearman consistency (0.38 vs. 0.58), and (iii) a consistent reorganisation across both skeletal joints and mesh vertices ($p < 10^{-10}$ on 1,079 vertices). As a methodological consistency check, pairwise phase differences from the Mode-1 eigenvector are compared against CRP on all 190 joint pairs by a permutation test ($ρ= 0.473$, $p = 0.0005$). A correspondence analysis between Mode-1 amplitude and kinetic-energy mobilisation variance further shows a strong positive correlation in the downswing ($ρ\approx 0.71$ on both skeleton and mesh) and no correlation in the backswing, indicating that the proposed framework bridges kinematic and kinetic descriptions of coordination through phase structure.

Public warning systems (PWS) in cellular networks enable authorities to broadcast emergency alerts to all mobile phones in a geographic region in the event of threats such as earthquakes or severe weather. If an attacker can imitate these alerts and transmit a forged warning containing fake news or phishing links, the impact could range from public panic to user compromise. In this work, we present the first open-source 5G emergency alert spoofing attack, implemented by modifying the openairinterface (OAI) radio access network (RAN) code and executed using a software-defined radio, complemented by a custom network management system to automate network and warning configuration. We conduct a detailed analysis of how different smartphones behave under various conditions. Our findings show that while devices readily display spoofed alerts, the alerting mechanism enables multiple practical attack scenarios beyond simple warning display. Finally, to address this threat, we propose and implement a lightweight cross-cell verification mechanism in OAI, in which the device compares the received warning with neighboring cell broadcasts to flag single-source alerts as suspicious.

Resource allocation in integrated sensing and communication (ISAC) systems needs to be optimized to balance the requirements of the communication and sensing modules considering complicated cross-layer data traffic and queue status in dynamic multi-user environments. This paper studies the beam allocation for cross-layer ISAC that achieves low-latency communication and minimizes sensing parameters estimation error. To handle the complex coupling between practical data buffer dynamics and varying wireless channels, we propose a deep reinforcement learning (DRL)-assisted approach. Rather than relying on explicit channel state information, the DRL-assisted beam allocation reduces feedback overhead by leveraging sensing observations. Simulation results verify that the DRL framework effectively takes buffer status into account and adapts to the wireless environment while allocating resources. The proposed multi-beam scheme improves overall throughput with only modest delay increases. Finally, the DRL-assisted beam management achieves both communication and sensing performance close to that of the genie-aided benchmark with perfect angle-of-departure (AoD) knowledge. These contributions advance the state-of-the-art intelligent resource management for ISAC systems.

Microwave linear analog computer (MiLAC)-aided transmit beamforming, which processes transmitted symbols entirely in the analog domain, has recently emerged as a promising alternative to fully digital or hybrid beamforming architectures for single-user multi-antenna systems. However, recent studies have shown that deploying a single lossless and reciprocal MiLAC at the transmitter cannot achieve the same capacity as fully digital beamforming in multi-user scenarios. To address this limitation, we propose a novel two-layer MiLAC-aided beamforming architecture at the transmitter for a downlink multi-user multiple-input single-output (MISO) network. Leveraging microwave network theory, we first prove that lossless and reciprocal two-layer MiLAC-aided beamforming can achieve the same performance as digital beamforming, and we derive a closed-form mapping from digital beamforming to two-layer MiLAC analog beamforming. Furthermore, we formulate a sum-rate maximization problem and develop an efficient optimization framework to jointly optimize the power allocation and the scattering matrices for the proposed two-layer MiLAC architecture. Numerical results validate our theoretical findings and demonstrate that two-layer MiLAC achieves the same sum-rate performance as fully digital beamforming.

The escalating climate crisis and ecosystem degradation demand intelligent, low-cost sensors capable of robust, long-term monitoring in real-world environments. Absolute dissolved oxygen (DO) concentration is a key parameter for predicting climate tipping points. Inexpensive optoelectronic sensors based on microstructured polymer films doped with phosphorescent dyes could be readily deployable; however, signal drift and marine biofouling remain major challenges. Here, we introduce a sensing paradigm that combines camera-based DO sensors with a visual transformer (ViT)-based physics-informed neural network (PINN) for high-fidelity sensing under biofouling conditions. Training and testing data were obtained from an algae-laden water tank over 14 days to capture accelerated biofouling. The ViT-PINN, which embeds the Stern-Volmer (SV) equation into the loss function, reduces mean average error (MAE) by 92% and 89% compared to classical statistical and ML approaches, achieving ~2 umol/L absolute error. A deep ensemble further quantifies predictive uncertainty, enabling self-diagnostic sensing.

We propose Speech Enhancement based on Drifting Models (DriftSE), a novel generative framework that formulates denoising as an equilibrium problem. Rather than relying on iterative sampling, DriftSE natively achieves one-step inference by evolving the pushforward distribution of a mapping function to directly match the clean speech distribution. This evolution is driven by a Drifting Field, a learned correction vector that guides samples toward the high-density regions of the clean distribution, which naturally facilitates training on unpaired data by matching distributions rather than paired samples. We investigate the framework under two formulations: a direct mapping from the noisy observation, and a stochastic conditional generative model from a Gaussian prior. Experiments on the VoiceBank-DEMAND benchmark demonstrate that DriftSE achieves high-fidelity enhancement in a single step, outperforming multi-step diffusion baselines and establishing a new paradigm for speech enhancement.

Separating multiple graph signals from a single observed mixture is an inherently ill-posed problem that traditionally relies on restrictive and handcrafted priors. This letter addresses this challenge by proposing an unsupervised learnable spectral filtering framework. Our approach reconstructs latent components by passing a fixed random input through learnable spectral filters, operating within the low-frequency eigenspace of each source-specific graph Laplacian. The architecture implicitly biases the recovered signals toward smooth patterns by confining reconstruction to these low-frequency subspaces. This acts as a structural prior, establishing a principled bridge between classical graph spectral analysis and modern neural decomposition. Numerical experiments confirm that this framework successfully isolates individual sources using solely the observed mixture and the underlying graph topology.

Radio Frequency Fingerprinting (RFF) is a key technology for identity authentication in wireless networks. However, due to the rapid dynamics of Autonomous Aerial Vehicles (AAVs) in low-altitude wireless networks, RFF models require parameter updates to maintain authentication performance, posing a major challenge to existing schemes. Conventional retraining approaches for handling departed or compromised AAVs are computationally prohibitive and risk retaining polluted features, which compromises both authentication security and user privacy. To address these limitations, we propose an Input-Perturbation-based RFF Unlearning (IPRU) scheme. By optimizing a universal Fingerprint Forget Vector (FFV) as a lightweight input perturbation, IPRU successfully erases the fingerprints of target AAVs without modifying the RFF model parameters, achieving an effective balance between efficient unlearning and preserved authentication performance. A combinatorial optimization strategy further enables multi-AAV forgetting on demand. The simulation results demonstrate that IPRU achieves 1.41% unlearning accuracy, 99.41% remaining accuracy, and 100% resistance to membership inference attack, while running 5.79X faster than retraining and 2.1X faster than the baseline scheme.

This paper considers multiple extended object tracking based on Poisson multi-Bernoulli mixture (PMBM) filtering, which gives the closed-form Bayesian solution for standard multiple extended object models with Poisson birth. To efficiently address the challenging extended object data association problem in PMBM filtering, we develop implementations of the extended object PMBM filter using blocked Gibbs sampling. By formulating the PMBM density on an augmented state space with auxiliary variables and leveraging the Poisson object measurement model, we first derive a joint posterior over potential objects, previous global hypotheses, and current measurement association variables, together with its corresponding factorization. This factorized representation leads to blocked Gibbs samplers that efficiently generate high-weight global hypotheses and thereby provide an efficient implementation of the PMBM update step. We further introduce a collapsed Gibbs sampling variant, in which the Bernoulli object existence variables are marginalized out, yielding higher sampling efficiency, especially for the initiation of newly detected objects. The proposed methods, implemented under the gamma Gaussian inverse-Wishart model, are compared with an extended object Poisson multi-Bernoulli filter based on particle belief propagation. Simulation results demonstrate that the proposed approaches achieve comparable tracking performance while requiring substantially less runtime.

Line spectral estimation (LSE) involves estimating both spectral frequencies and their associated complex amplitudes. Existing Fisher-information-based benchmarks are local and therefore do not capture either the threshold behavior of frequency estimation or the propagation of frequency errors to subsequent amplitude reconstruction. This paper develops explicit performance benchmarks for LSE from two complementary perspectives: ordered frequency estimation and plug-in amplitude reconstruction. On the frequency side, we develop a computable Ziv-Zakai bound (ZZB)-type benchmark under an ordered prior by combining a generalized-likelihood-ratio-test (GLRT)-based surrogate for the unavailable pairwise kernel with an ordered-prior correction. The resulting benchmark recovers the ordered a priori bound at low signal-to-noise ratio (SNR) and the marginalized frequency-side Cramer-Rao bound (CRB) at high SNR. On the amplitude side, we derive a local transfer characterization for the sequential plug-in estimator and obtain a computable benchmark for the induced amplitude error. The resulting framework explicitly characterizes threshold behavior on the frequency side and error propagation on the amplitude side. Numerical results support the proposed benchmarks across different SNR regimes, snapshot numbers, and model orders.

The shift to the radiative near field region due to large antenna arrays necessitates beamforming that accounts for both angle and range, evolving mobility management into a joint angular range tracking challenge. Conventional schemes rely on rigid pilot payload structures with dedicated training slots, which interrupt data transmission and degrade spectral efficiency. To address this, we propose a pilot-free beam tracking framework leveraging Thompson sampling(TS). Within each sliding window, the user trajectory is modeled by local low-order polynomials in angle and range, and the motion parameters are estimated by maximum likelihood with uncertainty quantified via the Fisher information matrix. TS adaptively probes uncertain trajectory regions using beams that simultaneously serve as payload beams. Simulations demonstrate that the proposed framework maintains reliable connectivity while eliminating the overhead of dedicated pilot-based beam sweeping.

Electrooculogram (EOG) is a non-invasive bio-signal generated by the potential difference between the retina and cornea during eye movement, and is widely utilized in Human-Computer Interaction (HCI) systems. Expanding the range of detectable eye movements enhances system capability. However, increasing the number of classes typically degrades classification performance. While AI-based approaches can mitigate this limitation, their complexity increases significantly when operating on single-cycle EOG signals. Although single-cycle signals offer advantages such as low latency, reduced power consumption, and improved responsiveness, they are inherently limited by reduced informational content and higher susceptibility to noise. Ensuring low latency remains critical for real-time HCI applications, where system response must remain below human reaction thresholds. In this experimental study, using explainable AI, we address these challenges by developing 1-dimensional (1D) and cascaded ANN and CNN architectures capable of highly accurate classification across ten EOG classes (Stare, Blink, Up, Down, Right, Left, Up-left, Up-right, Down-left, and Down-right) using single-cycle signals, while simultaneously achieving latency substantially lower than human reaction time. The study achieved an accuracy of around 99% for all the models with a latency of 38.6 ms for the 1D ANN, and 82.85 ms for the cascaded CNN. These findings confirm that cascaded neural network architectures, can effectively balance high classification accuracy and low latency for single-cycle, multi-class EOG-based HCI systems under limited data availability.

This paper investigates the cross-frequency structure of background clutter induced by random dispersive media in single-snapshot FDA-MIMO-GPR. Representative media are modeled by the Cole--Cole formulation to relate dispersive constitutive behavior to the reference propagation environment and observation-domain statistics. A normalized incremental contrast function is introduced under a reference-medium framework, and a single-snapshot background-response expression with first-order propagation-kernel feedback is derived. Based on this expression, a cross-frequency coupling strength of the leading-order background covariance is defined. Numerical results show that, in weakly dispersive scenes, the proposed analysis remains consistent across constitutive mapping, the zeroth-order propagation skeleton, first-order distorted-Born truncation, propagation-kernel feedback, and single-channel response closure. The proposed metric distinguishes uncoupled and explicitly coupled constructions, remains stable under pure energy scaling, responds clearly to correlation length and relaxation-location parameters, and corresponds directly to the error of the frequency block-diagonal approximation. Additional experiments show that the resulting cross-frequency structure affects whitening and principal-subspace extraction. In scenes with pronounced relaxation, abrupt breakdown under strong perturbations and high-error plateaus indicate that the present theory is mainly applicable within the validity range of first-order feedback.

Satellites provide seamless coverage and are critical for emergency communications during natural disasters. However, their performance is constrained by limited spectrum and high deployment cost. To address these issues, we propose a fluid antenna system (FAS)-based solution that enables dynamic signal adaptation. Building on this concept, a compact ultra-massive antenna array (CUMA) is introduced, where multiple ports are simultaneously activated to coherently combine signal components. This design mitigates interference while reducing cost, as each fluid antenna requires only a single RF chain yet achieves significant improvement in the received signal-to-interference-plus-noise ratio (SINR). We consider a satellite CUMA network where all ground users share the same satellite for uplink transmission, and CUMA is employed to suppress inter-user interference. Closed-form expressions for the received signal power, interference power, and their distributions are derived. Based on these results, the outage probability is obtained in a unified form along with an accurate approximation, and the ergodic rate is characterized. Our analysis identifies the conditions under which CUMA outperforms maximum ratio combining in satellite systems. Notably, with sufficiently compact fluid antenna configurations, the received signal becomes deterministic, indicating that system performance is dominated by interference statistics. Moreover, increasing the number of ports yields a linear beamforming gain. Numerical results further compare orthogonal and non-orthogonal multiple access CUMA, showing that the latter achieves superior performance under wideband conditions.

This paper presents a deep unfolding-supported coordinated multipoint beam pattern synthesis (DUCoMP-BPS) scheme to overcome the high complexity, poor adaptability, and limited scalability of traditional cell-free anti-jamming beamforming. In the proposed design, access points (APs) independently determine analog beamforming using local angle information, while the central processing unit (CPU) performs cooperative digital beamforming with only a single AP-CPU interaction, significantly reducing fronthaul overhead. To further improve efficiency, a deep unfolding strategy transforms the costly step size search in analog beamforming into a trainable parameter, where an offline-trained complex-valued neural network enables fast and adaptive online inference. Simulation results show that the complexity of DUCoMP-BPS scales linearly with the number of APs, reduces single-AP analog beamforming runtime by about 67% compared to conventional optimization, and achieves superior nulling performance over purely data-driven approaches. Hardware feasibility is validated on an Advanced RISC Machine-Field Programmable Gate Array (ARM-FPGA) heterogeneous platform, where algorithm-hardware co-verification and hardware-software decoupling enable efficient parallelism and low-latency execution. Finally, anechoic chamber measurements under practical hardware imperfections confirm robust beamforming performance, demonstrating the strong potential of DUCoMP-BPS for real-world deployment.

Smart glasses are emerging as a promising interface between humans and artificial intelligence (AI) agents, enabling first-person perception, contextual awareness, and real-time assistance. However, continuous offloading of visual data from wearable devices to cloud-based vision-language models (VLMs) is fundamentally constrained by limited wireless bandwidth and energy resources. This paper proposes an intention-aware semantic agent communication framework for AI glasses, where data transmission is guided by user intention rather than raw pixel fidelity. In the proposed architecture, AI glasses act as an edge semantic agent while a server-side VLM executes high-level cognition and reasoning. The user intention can be inferred by the server-side VLM through the current transmitted content and the historical prompts. Driven by specific user intentions, the glasses adaptively preserve textual content, document layout, or object semantics before transmission. We evaluate three representative scenarios with different lightweight preprocessing tools on the AI glasses. Simulation results demonstrate that intention-aware preprocessing significantly achieves more than 50% bandwidth reduction depending on the current task while maintaining task performance. Moreover, semantic transmission exhibits graceful degradation under low SNRs. The findings demonstrate that aligning communication resources with user intention is essential for robust and efficient wearable AI agent systems.

Consistency models (CMs) learn a consistent mapping from multiple noise levels to the data endpoint and can therefore perform generative inference in one or a few steps. This property makes them attractive as learned priors for low-latency inverse problems. Multiple-input multiple-output (MIMO) channel estimation under limited pilot overhead can be formulated as a high-dimensional linear inverse problem with an explicit measurement matrix, where data consistency alone is often insufficient for stable angular-domain channel recovery. This paper applies the plug-and-play consistency model (PnP-CM) framework to pilot-aided MIMO channel estimation. The PnP-CM inference procedure enforces the pilot observation model in the data-consistency update and invokes a pretrained CM denoiser in the prior update, thereby recovering the angular-domain channel vector within a small number of outer iterations. Preliminary experiments validate the feasibility of using CMs as low-latency channel-estimation priors and show that adaptive parameter scheduling and cross-scenario robustness remain important directions for further improvement.

The growing number of Internet-of-Things (IoT) based artificial intelligence (AI) applications deployed at resource-constrained network edge call for ultra-reliable and low-latency data processing pipelines from distributed front-end sensors to remote inference units. Meanwhile, brain-inspired neuromorphic computing featuring spiking neural networks (SNNs) have arisen as a new paradigm for energy-efficient AI inference. However, significant energy and time expenses incurred in high-complexity transceivers that combat fading and multi-user interference hinder implementations of multi-user neuromorphic inference for edge intelligence. To address this challenge, we consider in this paper a broadband multi-user remote inference system that integrates event-based sensing and time-hopping (TH) on-off keying (OOK) based ultra-wideband (UWB) communications for reliable neuromorphic inference. Specifically, we propose a novel two-timescale repetition coding that leverages intra-frame pulse sparsity for low-latency repetition. We also develop two neuromorphic inference schemes based on: (i) digital spike encoding that recovers each pixel of the event-frame by threshold-adaptive detection via an SNN based sparsity estimator; and (ii) analog spike encoding that converts noisy correlator outputs at the receiver into analog-valued inputs for end-to-end (E2E) classification. Finally, numerical results validate the effectiveness of the proposed coding schemes, and reveal a signal-to-noise ratio (SNR)-dependent performance crossover between the two inference schemes, indicating that analog spike encoding based schemes are preferable with mild or high SNR while digital spike encoding based schemes remain robust in low SNR regime.

Unified 2D and 3D radio map construction supports network planning, wireless digital twins, and unmanned aerial vehicle (UAV) applications. In urban environments, blockage, reflection, and diffraction make accurate construction expensive for physics-based solvers. Autoregressive next-token prediction offers a single sequential formulation that can cover both 2D and 3D generation, but standard raster ordering ignores the spatial structure of radio propagation. When generation follows propagation, each token is predicted from propagation-relevant history rather than spatially arbitrary context, which provides more causally informative conditioning and lowers conditional uncertainty. We propose PILOT, a pretrained autoregressive framework that replaces raster scan with a wavefront sequence expanding outward from the transmitter. Each prediction step is guided by an environment-aware instruction that spatially aligns environment features with the queried radio map region. The same framework extends to 3D radio maps through height-slice stacking while a gradient loss enforces vertical continuity. On standard 2D benchmarks, PILOT achieves the lowest NMSE among all baselines. For volumetric generation, it reduces NMSE by 78% relative to the diffusion baseline at roughly $2500\times$ faster inference. It also outperforms methods that rely on 10% sparse measurements and achieves the best zero-shot results in the cross-domain evaluation.

To enable larger apertures in multipleinput multipleoutput MIMO systems the trihybrid MIMO architecture offers a promising lowcost and lowpower solution by introducing reconfigurable antennas as a third layer of precoding on top of conventional digital and analog processing In this paper we develop a unified signal processing framework for trihybrid MIMO that explicitly captures the electromagnetic EM characteristics of diverse reconfigurable antenna technologies We first propose a generic inputoutput model that incorporates the reconfigurable antenna layer into an effective channel representation revealing a fundamental coupling between the channel precoder and radiated power Building on this model we formulate a general optimization problem that jointly accounts for digital analog and antennadomain precoding under hardware and power constraints We then instantiate this framework across seven representative reconfigurable antenna architectures including parasitic arrays dynamic metasurface antennas fluidpixel antennas polarizationreconfigurable antennas stacked intelligent metasurfaces pinching antenna systems and nonradiating wires To systematically compare these heterogeneous architectures we introduce a new metric the reconfigurability efficiency factor REF which quantifies the performance gains achievable through antenna reconfiguration under realistic constraints Numerical results demonstrate the tradeoffs among aperture size power consumption hardware complexity and spectral efficiency Our results establish that EMlevel reconfiguration reshapes the signal processing design space highlighting the need for new architectures and algorithms that jointly optimize across digital analog and electromagnetic domains This work reveals that electromagnetic reconfiguration couples the channel and precoder

Two-fold redundant sparse arrays possess inbuilt redundancy to tackle single-element failures. This property enables them to perform accurate direction of arrival (DOA) estimation even during single sensor faults. However, recent literature suggests that some TFRSAs suffer from hidden dependencies whereby a single sensor fault at peculiar positions within the array cause discontinuities (holes) in the difference coarray (DCA). This violates the very idea of providing two-fold redundancy. Such hidden dependencies could prove catastrophic in many critical applications such as defense, autonomous driving, and biomedical imaging. Despite this issue, no formal tools or techniques exist to ascertain whether a given array configuration is truly twofold redundant or not. To address this gap, we provide a comprehensive framework and a first-ever graphical user interface (GUI). The GUI has been built using the features in MATLAB app designer and tested with known examples available in sparse array literature. Several numerical examples have been discussed to check the tool's response in each scenario. We conclude that the GUI is functionally accurate and can be an indispensable tool for sparse array designers in making informed choices about array configurations prior to real deployment.

As a practical physical implementation of pinching-antenna systems, leaky coaxial cable (LCX) enables distributed radiation in more general wireless environments, particularly for lower-frequency applications. In this paper, a leaky-coaxial pinching-antenna system, referred to as the LCX pinching-antenna system, is investigated, and adjustable slot apertures are introduced, such that the slot size can be continuously adjusted rather than being restricted to binary activation. Specifically, the aperture adjustment is modeled as amplitude scaling of the channels induced by the corresponding slots, or equivalently, as power coefficients associated with different slots. Accordingly, analytical results are derived to quantify the performance gain of continuous aperture adjustment over binary slot activation and to reveal the impact of channel coherence on the achievable data rate improvement. Furthermore, static and dynamic time-division multiple access (TDMA) schemes are considered, and the corresponding sum rate maximization problems are formulated and efficiently solved by quadratic transform based optimization, combined with successive convex approximation and alternating updates. Simulation results demonstrate that the proposed design can significantly outperform conventional fixed-antenna systems, traditional LCX schemes, and binary slot activation in terms of both achievable sum rate and outage probability.

Directional Selective Fixed-Filter Active Noise Control (D-SFANC) can effectively attenuate noise from different directions by selecting the suitable pre-trained control filter based on the Direction-of-Arrival (DoA) of the current noise. However, this method is weak at tracking the direction variations of non-stationary noise, such as that from a moving source. Therefore, this work proposes a Predictive Directional SFANC (PD-SFANC) method that uses a Convolutional Recurrent Neural Network (CRNN) to capture the hidden temporal dynamics of the moving noise and predict the control filter to cancel future noise. Accordingly, the proposed method can significantly improve its noise-tracking ability and dynamic noise-reduction performance. Furthermore, numerical simulations confirm the superiority of the proposed method for handling moving sources across various movement scenarios, compared to several representative ANC baselines.

We investigate the problem of jointly testing a pair of composite hypotheses and, depending on the test result, estimating a random parameter under distributional uncertainties. Specifically, it is assumed that the distribution of the data given the parameter of interest, is subject to uncertainty. Both, a Bayesian formulation and a Neyman-Pearson-like formulation, are considered. It is shown that the optimal policy induces an $f$-similarity that must be maximized to identify the least favorable distributions. Besides the general results, the implementation is investigated using a band-type uncertainty model. For designing the minimax procedures, existing algorithms are modified to increase convergence speed while maintaining numerical stability. The proposed theory is supplemented by numerical results for both formulations.

The Terahertz (THz) band (0.1-10 THz) has emerged as a critical frontier for future communication systems, offering ultra-wide bandwidths that enable Terabits-per-second (Tbps) wireless links and high-precision sensing and imaging. However, practical deployment of THz systems is hindered by unique challenges, including intricate channel characteristics, high-dimensional and large-scale optimization problems, and highly dynamic network environments. Artificial Intelligence (AI) serves as a transformative enabler to address these challenges, providing robust capabilities for precise modeling, advanced signal processing, complex optimization, real-time decision-making, and prediction, among others. Reciprocally, the unprecedented bandwidth and high-resolution sensing capabilities of THz networks provide a promising physical infrastructure for AI, facilitating training, inference, and data collection. This survey presents a systematic and comprehensive overview of AI-driven solutions across the entire THz communication network and the symbiosis of AI and THz networks. To begin with, a foundational overview of AI technologies tailored for wireless communications is presented. Subsequently, AI-based innovations are investigated, spanning from hardware design, channel modeling, physical layer optimization, up to higher-layer network protocols and advanced THz services, including mobile edge computing and sensing-empowered applications. In parallel, the capacity of THz networks to serve AI is examined, underscoring a profound paradigm shift towards a mutual symbiosis where AI and THz co-evolve and empower each other. Finally, by synthesizing these state-of-the-art advancements and identifying open research directions, this survey highlights the potential of AI in copilot with development of THz communication systems.

Optical wireless communication (OWC) is a promising technology for supporting data intensive services in indoor environments due to its large unregulated spectrum, high spatial reuse, and potential for multigigabit data rates. In particular, vertical cavity surface emitting laser (VCSEL) based systems enable highly directional transmission, allowing efficient spatial separation of users and improved link performance. However, the use of narrow optical beams also makes system performance highly sensitive to user mobility and device orientation, as movement directly affects beam alignment and optical channel gain. Consequently, power allocation strategies that ignore mobility dynamics often provision excess optical power to maintain reliable connectivity, resulting in inefficient energy use. In this work, a power control framework for dynamic indoor OWC networks that explicitly accounts for mobility driven channel variation is developed. It uses a hybrid Gauss Markov and learning based approach that captures both user movement continuity and behaviour driven orientation changes. The mobility states are then used to guide power allocation decisions. Simulation results show that incorporating mobility aware channel prediction enables more accurate power allocation, and improves energy efficiency compared with conventional power control schemes in dynamic indoor environments.

This paper discusses the left and right ranks of quaternion matrices with Hankel structure. While they are in general different for arbitrary quaternion matrices, we show that the left and right ranks of quaternion Hankel matrices are equal. Moreover, we establish the relation between Hankel matrices and the existence of linear recurrence relations with quaternion coefficients.

We present a novel framework for line-of-sight (LoS) delay-Doppler (DD) estimation in dense scattering propagation environments. We present two time-frequency (TF) domain pilot sequences inspired by the Zadoff-Chu sequence that exhibit desirable autocorrelation properties. Further, we present a twisted convolution-based approach for LoS DD estimation directly from the TF-domain received signal, avoiding an additional TF to DD transformation, which is commonly found in literature. Numerical results from simulations demonstrate that the proposed framework significantly outperforms traditional single-carrier Zadoff-Chu sequences in both delay and Doppler estimation over a wide range of Rician fading factor and SNR values.

This paper proposes and analyzes Riemannian optimization algorithms on the manifold of unitary and symmetric matrices, denoted ${\cal {U}}_s$, which naturally models the scattering matrices of passive and reciprocal devices such as beyond-diagonal reconfigurable intelligent surfaces (BD-RISs). Despite its relevance, the geometry of ${\cal {U}}_s$ has remained largely unexplored, and existing BD-RIS optimization methods either ignore the symmetry constraint or rely on costly Takagi-based parameterizations. We first provide a rigorous geometric characterization of ${\cal {U}}_s$, deriving its tangent space, a simple retraction, and closed-form expressions for geodesics. Building on these results, we develop two Riemannian manifold optimization (MO) algorithms tailored to ${\cal {U}}_s$: a line-search (LS) based scheme and a phase-optimization (PO) update along geodesics. We then apply the proposed framework to BD-RIS-assisted multiple-input multiple-output (MIMO) links, addressing sum-gain maximization, rate maximization, and minimum mean-square error problems, where they outperform existing approaches. Furthermore, we show that when the number of BD-RIS elements exceeds the total number of antennas, the optimal scattering matrix is low-rank, which motivates and enables efficient low-rank variants of the proposed algorithms.

This paper studies an integrated sensing and communication (ISAC) system where a multi-antenna base station (BS) communicates with multiple single-antenna users in the downlink and senses the unknown and random angle information of a target based on its prior distribution information and the received echo signals. We focus on a challenging scenario with heterogeneous unknown parameters where the target's reflection coefficient is also unknown with no prior information. We consider a general transmit beamforming structure with both communication beams and dedicated sensing beams, where the communication users can cancel the interference caused by the pre-determined sensing signals. By adopting the periodic posterior Cramer-Rao bound (PCRB) to quantify a lower bound of the mean-cyclic error (MCE) for sensing the periodic angle parameter, we optimize the transmit beamforming to minimize the periodic PCRB, subject to individual communication user rate constraints, which is a non-convex problem. By leveraging the semi-definite relaxation (SDR) technique and Lagrange duality theory, we derive the optimal solution and prove that at most one dedicated sensing beam is needed. Numerical results validate our analysis and effectiveness of the proposed beamforming design.

Wave-domain processing is an emerging paradigm where signal processing operations are partially shifted from the digital to the electromagnetic (EM) domain. Leveraging reconfigurable EM devices, this approach aims to reduce complexity, energy consumption, and latency in next-generation wireless systems employing holographic MIMO. This paper establishes fundamental theorems on the controllability of generic reconfigurable EM devices, where wave processing is achieved through the dynamic configuration of passive scatterers. Specifically, we derive necessary and sufficient conditions for controllability as a function of geometry and mutual coupling between elements. Finally, we provide a detailed discussion and numerical results characterizing the interplay between the number of elements, physical size, degrees of freedom, and directivity.

Reconfigurable antenna systems (RASs), such as fluid antennas and movable antennas, are poised to play a pivotal role in sixth-generation (6G) systems by dynamically adapting the antenna elements for system performance enhancement. However, unlocking their full potential requires channel models that accurately capture the influence of antenna configurations on the radiation, propagation, and reception of signals. Existing channel models suffer from several limitations, such as neglecting polarization effects, being restricted to specific antenna types, or relying on oversimplified assumptions. In this paper, we propose a general electromagnetic (EM)-based channel model grounded in spherical vector wave expansion (SVWE). The proposed EM-based channel model captures the impact of antenna position and orientation on the channel gain, thereby making it particularly well-suited for RASs. The effectiveness and accuracy are validated through comparisons with commercial simulation software, demonstrating excellent agreement in predicted channel gains. Moreover, it is shown that antenna orientation is a critical factor governing communication performance, and that dynamically adjusting the antenna orientation yields up to 70% improvement in achievable communication rate compared to a fixed-antenna configuration.

Affine frequency division multiplexing (AFDM) has emerged as a promising integrated sensing and communication (ISAC) waveform due to its intrinsic chirp signalling nature. Nevertheless, practical AFDM-based ISAC still faces two key obstacles, namely, high ambiguity function (AF) sidelobes and high peak-to-average power ratio (PAPR). By leveraging the reserved chirp-subcarrier (RCS) symbols, we develop a unified AFDM waveform design framework for AF shaping and/or PAPR control. The proposed framework supports three modes: AF shaping via weighted integrated sidelobe level (ISL) minimization, PAPR minimization, and joint AF shaping and PAPR control under a prescribed PAPR constraint. To solve the formulated nonconvex problem and to accommodate the discrete-phase constraints on the optionally optimized pre-chirp parameters, a joint ISL-PAPR-discrete-phase majorization-minimization (JIPD-MM) algorithm is developed. Simulation results verify the effectiveness of the proposed framework under all three design modes. The joint mode further demonstrates that the prescribed PAPR constraint can be effectively satisfied while still achieving meaningful ISL reduction. These gains are also reflected in improved weak-target detectability under multitarget scenarios and lower bit error rate (BER) under power-amplifier (PA) nonlinearity.

Independent component analysis (ICA) estimates a demixing matrix that can recover statistically independent sources from linear mixtures. FastICA is a popular ICA algorithm due to its efficiency, but its performance strongly depends on a user-chosen nonlinear function matched to the source distribution. When the source distribution is unknown, this function must be guessed at, and incorrect guesses can lead to significant drops in performance. We remove the need to guess by estimating a suitable function directly from the observed data. Our experiments show that the separation error stays close to the best fixed choice across synthetic mixtures comprising heavy-tailed or discrete sources while retaining a FastICA-like runtime.

Integrated sensing and communication (ISAC) is poised to be a defining feature of 6G networks, promising to transform cellular base stations (BSs) into ubiquitous radar sensors. However, a significant gap exists between the theoretical promise of ISAC and the commercial reality of legacy cellular communication infrastructure. Existing communication networks are constrained by fragmented spectrum, blockage-prone environments, and cost-prohibitive high-rate analog-to-digital converters (ADCs). These limitations stifle the high-resolution sensing required for emerging applications. This article advocates a shift from dependence on physical resources to computational synthesis and introduces a unified full stack virtualization framework that upgrades legacy networks with minimal hardware changes, spanning signal generation, propagation, and acquisition. Specifically, we virtualize signal generation via space-time -frequency synthesis across distributed BSs to synthesize a larger effective aperture and a wider effective bandwidth. We then virtualize signal propagation by leveraging environmental multipath and digital maps to reinterpret reflections as massive virtual arrays. Finally, we virtualize signal acquisition using sub Nyquist strategies to bypass sampling bottlenecks. We demonstrate that by trading computation for hardware, commercial networks can achieve fine-grained sensing without expensive retrofitting.

PRIMEX (prime-based graph encoding and extraction) is a recently proposed framework for scalable distributed fusion. In PRIMEX, the information pedigree of state estimates or probability density functions is encoded using the information codes, enabling lightweight arithmetic for redundancy removal and data integration. Building on PRIMEX and its memoryless fusion strategy based on a least-squares approximation, in this paper we present two efficient distributed tracking algorithms: a consensus-based PRIMEX method that fuses information from all neighbors, and a greedy gossip-based PRIMEX method that fuses with the most informative neighbor. To further increase communication efficiency, we incorporate an event-triggered mechanism, in which transmission decisions are driven by information novelty measured using differences between the information codes. The proposed methods are evaluated and compared with covariance intersection and centralized fusion in a distributed single target tracking scenario. Simulation results show that PRIMEX-based methods remain competitive in tracking accuracy while improving communication efficiency.

In this paper, we investigate the thermodynamic and radiative properties of a regular black hole sourced by a dark matter halo described by the Einasto density profile. The closed-form expressions for the Hawking temperature, specific heat capacity, sparsity parameter of Hawking flux, and the spectral energy emission rate were obtained. All these are examined as a function of the characteristic scale parameter $α$ of the dark matter distribution and compared with the standard Schwarzschild results. We show that the presence of a dark matter halo suppresses both the Hawking temperature and energy emission rate relative to the standard black hole result. Crucially, the specific heat capacity can be positive for a finite range of horizon radii, signaling a thermodynamically stable phase; the boundary of this stable region defines a Davies-type phase transition. The sparsity parameter is higher than in the standard black hole, indicating that the dark matter environment makes the Hawking flux even more intermittent.

Numerical analysis to determine the form of the scalar power spectrum in Warm Inflationary paradigm is inevitable. One further needs numerical techniques to analyse any Warm Inflation model with the current observational data through the MCMC codes that are available publicly, like COSMOMC or Cobaya. We present \texttt{SWIM} (Stochastic Warm Inflation Module) written in C++ and Python, that not only helps generate the Warm Inflationary scalar power spectrum, either semi-analytically or fully numerically, but also is integrated with Cobaya enabling the user to constrain the model parameters with current CMB data and thus to put any Warm Inflation model to test. \texttt{SWIM} numerically solves the standard stochastic perturbation equations of Warm Inflation without any approximations, uses machine learning techniques to speed up the MCMC analysis while analysing the fully numerical power spectrum that significantly reduces the computational cost, and is able to accommodate any Warm Inflation model with any form of inflationary potential and dissipative coefficient for numerical analysis. We show that \texttt{SWIM}, in most of the cases, outperforms other numerical codes on Warm Inflation that are designed to yield only the semi-analytical power spectrum as far as the runtimes are concerned. We further point out that there can be situations where the semi-analytical way of determining the scalar power spectrum in Warm Inflation can fall short, and one needs the full numerical power spectrum for parameter estimation given the observational data. In such cases, \texttt{SWIM} is the only code available so far that is designed to perform the task. Hence, \texttt{SWIM} offers a complete numerical platform for thorough analysis of Warm Inflation models against the current cosmological data. \texttt{SWIM} has been made publicly available at https://github.com/umg-kmr/SWIM.

Quantum computers could solve problems beyond the reach of classical devices, but this potential depends on quantum error correction (QEC) to protect fragile quantum states from noise. A central challenge in QEC is decoding: inferring likely physical errors from syndrome patterns generated by repeated stabilizer measurements. Existing decoders, including graph-based and neural approaches, typically return a single correction hypothesis and therefore discard the richer posterior structure of the error distribution conditioned on the observed syndrome. Here we recast QEC decoding as posterior inference using discrete denoising diffusion, exploiting the analogy between stochastic error accumulation and the forward diffusion process. We introduce DiffQEC, a generative decoder that combines a syndrome processor for multi-round spatial-temporal syndrome histories with syndrome feature modulation to condition denoising on the observed syndrome throughout inference. On experimental data from Google's superconducting quantum processor, DiffQEC reduces logical error rates by up to 10.2% relative to minimum-weight perfect matching and by about 5% relative to tensor-network decoding. These improvements persist for larger code distances up to 17 under depolarizing noise and for logical circuits of increasing depth. Beyond accuracy, the learned posterior provides confidence estimates for post-selection and reveals physically meaningful error structure, establishing posterior generative decoding as a practical framework for QEC.

It was pointed out in [JSW+25] that widely-studied optimization problems such as D-regular max-k-XORSAT can be reduced to decoding of LDPC codes, using quantum algorithms related to Regev's reduction. LDPC codes have very good decoders, such as Belief Propagation (BP), and this therefore makes D-regular max-k-XORSAT an enticing target for this class of quantum algorithms. However, BP was found insufficient to achieve quantum advantage. Here, we develop an intrinsically quantum decoding technique, which decodes classical LDPC codes subject to coherent superpositions of bit flip errors. For average-case instances of D-regular max-k-XORSAT drawn from Gallager's ensemble, this quantum decoder strongly outperforms classical belief propagation at many values of k and D. For some (k,D) the approximate optima achievable using this decoder surpass both Prange's algorithm and simulated annealing. However, we stop short of achieving quantum advantage because we identify an enhancement to Prange's algorithm that recovers a precise tie, much as a precise tie was observed between the standard version of Prange's algorithm and a more limited version of locally-quantum decoding in [CT24].

In this work, we construct a three-dimensional non-relativistic Chern--Simons supergravity theory with both curvature and torsion within the Mielke--Baekler framework. We show that a consistent non-relativistic supergravity formulation requires starting from a $\mathcal{N}=2$ supersymmetric extension of the Mielke--Baekler algebra and implementing a non-relativistic expansion via the semigroup expansion method, rather than a naive contraction. This procedure allows one to overcome the usual difficulties of non-relativistic supergravity constructions, ensuring closure of the superalgebra and the existence of a non-degenerate invariant bilinear form. The resulting model is characterized by two parameters $(p,q)$, which interpolate between different non-relativistic supergravity theories, including the extended Bargmann, Newton--Hooke, and torsional models. Our results provide a unified framework for non-relativistic supergravity with torsion and open new avenues for exploring supersymmetric extensions of Galilean and Carrollian gravity theories.

We investigate the spin dynamics of a dipole-coupled system by comparing a direct solution of the Schrodinger equation for quantum spins with simulations of classical spins. Although classical spins have long been used in microscopic spin dynamics simulations, we demonstrate that their results differ significantly from those of quantum spins. Using Free Induction Decay as a benchmark, we find that while the overall patterns are qualitatively similar, significant discrepancies emerge at both short and long timescales. We trace these differences to fundamental distinctions in the two descriptions.

We provide evidence of quantum kernel advantage under noiseless simulation in binary insurance classification on MIMIC-CXR chest radiographs using quantum support vector machines (QSVM) with frozen embeddings from three medical foundation models (MedSigLIP-448, RAD-DINO, ViT-patch32). We propose a two-tier fair comparison framework in which both classifiers receive identical PCA-q features. At Tier 1 (untuned QSVM vs. untuned linear SVM, C = 1 both sides), QSVM wins minority-class F1 in all 18 tested configurations (17 at p < 0.001, 1 at p < 0.01). The classical linear kernel collapses to majority-class prediction on 90-100% of seeds at every qubit count, while QSVM maintains non-trivial recall. At q = 11 (MedSigLIP-448 plateau center), QSVM achieves mean F1 = 0.343 vs. classical F1 = 0.050 (F1 gain = +0.293, p < 0.001) without hyperparameter tuning. Under Tier 2 (untuned QSVM vs. C-tuned RBF SVM), QSVM wins all seven tested configurations (mean gain +0.068, max +0.112). Eigenspectrum analysis reveals quantum kernel effective rank reaches 69.80 at q = 11, far exceeding linear kernel rank, while classical collapse remains C-invariant. A full qubit sweep reveals architecture-dependent concentration onset across models. Code: https://github.com/sebasmos/qml-medimage

Quasinormal modes (QNMs) underpin black-hole ringdown modeling and spectroscopy, where higher overtones can contribute at early times and neighboring modes can become locally organized in the complex-frequency plane. Motivated by this, we continuously track Kerr overtones along a spin scan and study a raw neighboring-overtone quantity: the complex-frequency gap between adjacent overtones. We find robust interior minima whose spin locations drift from pair to pair, even within the same \((s,\ell,m)\) sector. We explain this by reformulating minimum setting as a local zero-setting problem for the complex separation between the two modes. Differentiating the squared gap yields a denominator-free real-projection diagnostic (the real part of the product between the complex separation, with conjugation, and its spin derivative). The sampled minimum is governed by an approximate local zero of this diagnostic, so minimum drift becomes the drift of its dominant zero crossing. This also yields a geometric picture: the minimum is a local radial turning event of the separation vector, while angular motion in the complex plane may persist. Finally, an expanded but deliberately restricted representative set (including same-family continuations, external positive transfers, and smooth no-trigger controls) supports this local picture for the triggered cases examined here. At the same time, the detailed local crossing environment remains intrinsically pair dependent.

Rare earth ion (REI)-doped oxide thin films are attractive for the application of quantum interconnects due to their stable optical levels and scalability$^{1-3}$. Among them, Er$^{3+}$ doped CaWO$_{4}$ is promising because it possesses narrow optical linewidth transitions and a long spin coherence time$^{4-6}$. The electron spin coherence is limited at high temperatures by paramagnetic impurities and by the presence of the 14.3% $^{183}$W nuclear spin. To further increase the spin coherence time at millikelvin temperatures, where the paramagnetic impurities are frozen out, our approach is to synthesize chemically and isotopically purified thin films as a host material. We first grow non-isotopically enriched Er$^{3+}$ doped CaWO$_{4}$ thin films, which exhibit a 214(13) MHz photoluminescence (PL) inhomogeneous linewidth, indicating the thin film has high crystalline quality. We then grow isotopically enriched CaWO$_{4}$ thin films using an isotopically purified $^{186}$WO$_{3}$ source. Time of flight secondary ion mass spectrometry (ToF-SIMS) was used to measure the relative concentration of W isotopes. $^{183}$W, the only W isotope that has a net nuclear spin and is the major cause of spin decoherence, was at a relative abundance of 1.2%, a factor of 10 lower than natural abundance. We also observed PL emission from single ions after integrating nano-photonic devices with the thin film. These results establish isotopically engineered CaWO$_{4}$ thin films as a promising platform for future studies of nuclear-spin-limited coherence and for scalable rare-earth-ion-based quantum nanophotonic devices.

A challenge with the Quantum Approximate Optimisation Algorithm (QAOA), and variational algorithms in general, is finding good variational parameters, a task which in itself can be NP-hard. Recent work has sought to de-variationalise QAOA by picking well-informed guesses for the variational parameters. The Linear Ramp QAOA (LR-QAOA) achieves this by using parameter schedules inspired by the quantum adiabatic algorithm. We go a step further and use spectral gap information from an adiabatic Hamiltonian, with the QAOA mixer Hamiltonian as our initial Hamiltonian, to make smooth ramps which we call Spectral Gap Informed Ramps (SGIR-QAOA). SGIR-QAOA schedules perform slow evolution where the spectral gap of the adiabatic Hamiltonian is small. We show that SGIR-QAOA has performance improvements over LR-QAOA on Grover's problem at constant depth and that SGIR-QAOA requires shorter depths to achieve the same optimal solution probability. We then show that these performance benefits extend to a problem with potential practical applications -- the Maximum Independent Set (MIS) problem. Finally, we demonstrate the scalability of the SGIR-QAOA method using extrapolated spectral gap information for scales that the spectral gap cannot be exactly evaluated, and show that the advantage appears to persist under mild depolarising noise.

The Hamiltonian for a system of itinerant particles on a two-dimensional lattice in a uniform magnetic field reduces the translational symmetry to a magnetic translation group, because of the need to choose a particular gauge for the vector potential. Nonetheless, in many situations all physical observables of the ground state remain entirely translation invariant. In this work, we introduce a projected entangled-pair state (PEPS) wavefunction with a pattern of virtual flux tensors, for which all physical expectation values are translation invariant by construction, possibly within an enlarged unit cell reflecting any symmetry breaking in the target state. Moreover, we show that the usual contraction and optimization methods for translation-invariant PEPS can be used, with the magnetic flux per plaquette only entering as a continuous parameter in the tensor network contractions. Therefore, our approach provides a method for simulating an interacting many-body system in a uniform magnetic field independently of the gauge choice for the vector potential and bypassing the need to consider extended magnetic unit cells.

We investigate the excitation of a three-level ladder-type atom by a unidirectional field with a pair of indistinguishable photons. Starting from an analytical expression for the two-photon absorption probability, we determine the two-photon state that maximizes the population of the upper atomic state at a chosen time and show that, in the limit of an infinitely long pulse, perfect excitation is possible. The optimal state is identified as the time-reversed counterpart of the two-photon state emitted in spontaneous cascade decay. We then compare this ideal excitation strategy with experimentally accessible families of states, including symmetrized Gaussian product states, temporally correlated Gaussian states, and coherent pulses. We analyze how the optimal excitation conditions depend on the ratio of atomic decay rates and on the separation of the atomic transition frequencies. For indistinguishable photons described by Gaussian pulses, quantum interference may shift the maxima of the marginal spectral distribution away from the atomic resonances and qualitatively modify the optimal excitation strategy. Our results clarify the role of indistinguishability and correlations in two-photon absorption and provide guidance for designing realistic excitation schemes in quantum-optical light-matter interfaces .

We investigate the nonequilibrium dynamics of a two-dimensional rotating Bose gas confined in a symmetric anharmonic trap, employing the multiconfigurational time-dependent Hartree method for bosons (MCTDHB). We study states ranging from vortex-free configurations to multicharged (giant) vortices, prepared by tuning the rotation frequency, and analyze their response to sudden interaction and trap quenches. In vortex-free states, interaction quenches induce regular breathing--like dynamics, whereas in the presence of giant vortices they lead to symmetry-breaking surface excitations. In contrast, trap deformations that excite quadrupole-like modes produce stable oscillations in vortex-free condensates but trigger rapid, irregular, and effectively chaotic splitting dynamics in multicharged vortices. To characterize these processes beyond conventional density and phase observables, we employ information-theoretic measures, including marginal and joint entropies, mutual information, and Kullback-Leibler (KL) divergence, supplemented by an angular-resolved KL measure that captures symmetry breaking and azimuthal localization. We find that chaotic splitting is accompanied by a pronounced growth of information-theoretic indicators, signaling the buildup of many-body correlations and increasing complexity in the system dynamics. Our results demonstrate the extreme sensitivity of giant vortices to excitation protocols and establish information-theoretic measures as a powerful framework to quantify correlations and complexity in rotating quantum gases.

At the core of the quantum Internet lie quantum repeaters that enable remote end-to-end entanglement generation. Fundamentally, the entanglement generation rate and fidelity of quantum repeaters constitute the bottleneck for end-to-end performance. To achieve high rates, quantum repeaters employ quantum memory multiplexing. In a high-rate standard repeater, each memory sequentially generates an entanglement with its neighboring nodes and then applies entanglement swapping. This, however, results in low fidelity due to decoherence of the first-formed entanglement in the sequential generation process. By allocating different numbers of memories to simultaneously form entanglements with the left and right adjacent nodes, quantum repeaters reduce high waiting times and achieve high fidelity. In such a repeater, a mismatch problem arises due to the difference between the probabilistic number of generated entanglements on both sides. Consequently, some entanglements remain stored until opposite entanglements are available. The mismatch problem reduces the repeater rate and particularly the entanglement fidelity. In this paper, we consider the mismatch problem in an asymmetric repeater with different distances to its adjacent nodes. To mitigate the mismatch problem, we derive a dynamic optimal memory allocation. Under the optimal allocation, we derive statistical lower bounds on the achievable rate and fidelity. We demonstrate that the optimal allocation significantly improves the fidelity while maintaining a comparable rate to the standard repeater. In contrast, our results show that fixed memory allocation may be detrimental to the fidelity.

Quantum error mitigation (QEM) is essential for extracting reliable results from near-term quantum devices, yet practical deployments must balance mitigation strength against runtime overhead under time-varying noise. We introduce \emph{GSC-QEMit}, a telemetry-driven, \textbf{context--forecast--bandit} framework for \emph{adaptive} mitigation that switches between lightweight suppression and heavier intervention as drift evolves. GSC-QEMit composes three coupled modules: (G) a Growing Hierarchical Self-Organizing Map (GHSOM) that clusters streaming telemetry into operating contexts; (S) an uncertainty-aware subsampled Gaussian-process forecaster that predicts short-horizon fidelity degradation; and (C) a cost-aware contextual multi-armed bandit (CMAB) that selects mitigation actions via Thompson sampling with explicit intervention cost. We evaluate GSC-QEMit on benchmark circuit families (GHZ, Quantum Fourier Transform, and Grover search) under nonstationary noise regimes simulated in Qiskit Aer, using an instrumented testbed where action labels correspond to graded mitigation intensity. Across Clifford, non-Clifford, and structured workloads, GSC-QEMit improves average logical fidelity by \textbf{+9.0\%} relative to unmitigated execution while reducing unnecessary heavy interventions by reserving them for inferred noise spikes. The resulting policies exhibit a favorable fidelity--cost trade-off and transfer across the evaluated workloads without circuit-specific tuning.

Gravitational wave (GW) observations provide a unique opportunity to test Einstein's General Relativity (GR) in the strong-field regime. While GR predicts only two tensor polarization modes, generic metric theories allow up to six independent modes. We perform a parameterized test of GR using the parameterized post-Einsteinian (PPE) framework applied to GW170817, incorporating for the first time the polarization angle constraints from the gamma-ray burst afterglow alongside other electromagnetic (EM) counterpart information. We extend the GR waveform by adding a scalar breathing mode and modifications to the tensor modes, introducing three non-GR parameters. We perform Bayesian inference for both quadrupole $\ell = |m|= 2$ and dipole $\ell = |m|= 1$ angular harmonics, with two frequency evolution models. For $\ell = |m|= 2$ , we find mild preference for a scalar mode (scalar amplitude deviates from zero at $\sim 2 σ$), while for the $\ell = |m|= 1$, we find no preference for a scalar mode. The EM constraint on the polarization angle places very tight bounds on non-GR parameters; for instance, in the case $\ell = |m| = 2$, the bound on the scalar (tensor) amplitude modification parameter improves by roughly $60\%$ $(30\%)$, highlighting the impact that long-term follow up of GW events can have on tests of gravity.

The paper presents a scheme to detect entanglement arising from the quantum nature of gravity between a spin qubit and photons, using Stokes parameters. One of the crucial tests of the general theory of relativity is the bending of light due to the curvature. Recently, a quantum counterpart of this experiment to test the quantum nature of the gravitational interaction has been proposed, in which the spin-2, massless graviton yields entanglement between matter and a photon sector. Hence, it provides one of the most crucial experimental signatures for testing the quantum nature of gravity in a lab, since only spin-2-induced entanglement can yield the correct deflection of light due to matter. Here, we propose a positive partial-transpose (PPT) witness criterion for witnessing such an entanglement. We scan the entangled states in this context by studying the overlap of the final state, which is proportional to the entanglement phase. We exploit the Stokes observables to measure the photon state and the spins in the matter sector, thereby constructing a witness for the quantum nature of gravity in this setup. To quantify this entanglement, we will couple the photon to a local oscillator, whose phase need to be controlled to probe the orthogonal components of the macroscopic interference in the laser beam. We have shown that for a non-maximally entangled state mediated by the quantum nature of gravity, the witness attains a maximal negativity of $-0.052$. Our findings indicate that this witness effectively detects entanglement within the range $0.71 \leq |γ| < 1$, where $γ$ is the overlap between the two coherent states of the photon, providing a clear signature of quantum correlations.

Zero-noise extrapolation (ZNE) mitigates errors in near-term quantum devices by extrapolating measurements obtained at amplified noise levels to estimate noise-free expectation values. In practice, commonly used extrapolation models are fitted without enforcing physical constraints, which can yield predictions outside the valid range of quantum observables. In this work, we introduce physically bounded variants of polynomial, exponential, and polynomial--exponential extrapolation models by explicitly parameterizing the zero-noise estimate and constraining it during optimization. We evaluate the approach using a large synthetic benchmark comprising 180,000 circuits and approximately 3.6 million ZNE experiments generated under realistic device noise models derived from IBM quantum backends. We also perform preliminary validation on real quantum hardware using GHZ and W-state circuits. Across the synthetic benchmark, bounded extrapolation substantially reduces unphysical predictions and improves the stability of exponential- and polynomial--exponential-family models, whereas polynomial models show little difference between bounded and unbounded variants. Hardware experiments show similar qualitative behaviour: bounded models generally avoid pathological extrapolations and often provide a more reliable balance between accuracy and usable coverage. At the same time, the results highlight practical limitations of current devices, including stronger-than-expected noise effects and variability not fully captured by simulation models. These results suggest that enforcing physical constraints during extrapolation improves the reliability of ZNE and that this approach can be incorporated into existing workflows with minimal modification.

Quantum optimal control (QOC) provides a systematic framework for achieving high-fidelity operations in quantum systems and plays a central role in tasks such as gate synthesis, state transfer, and pulse design. Existing QOC methods broadly fall into two categories: gradient-based and gradient-free algorithms. The associated optimization landscape is often high-dimensional, non-convex, and populated by numerous local minima, making efficient gradient-free search strategies essential. To address this, we introduce a gradient-free matrix product state/tensor train (MPS/TT) sampling heuristic for discrete quantum optimal control. In our approach, the MPS defines a score function over the space of discrete control parameters, which in turn induces a sampling distribution over candidate control sequences. This distribution is iteratively refined through selection of better performing sequences and local tensor updates to bias the search toward high-performing sequences. We evaluate the method on a range of benchmark problems, including single-qubit state transfer, Bell-pair preparation, qutrit gate implementation, and open-system population transfer. Across these tasks, the method exhibits stable convergence behavior and competitive empirical performance relative to established gradient-free baselines. These results suggest that tensor network sampling offers a viable heuristic framework for discrete quantum control.

Entanglement distillation is the process of converting noisy entangled states into maximally entangled pure states via local operations and classical communication. A long-standing, unresolved question is which entangled states are amenable to distillation, known as the distillability problem. We solve this for Bell-diagonal qutrits with Weyl structure, and present a noise-robust scheme for entanglement distillation. In particular, we find that violating the positive partial transposition (PPT) criterion is necessary and sufficient for the 1-distillability of these states. For this, we construct a Schmidt rank 2 eigenvector of the partially transposed density matrix associated with its unique, three-fold degenerate negative eigenvalue. This feature makes the derived entanglement distillation protocol resilient to white-noise effects on the quantum states. Our results thus make noisy entangled qutrit pairs more accessible for future quantum technologies.

This work introduces a geometrical method for analyzing transient gravitational waves recorded at interferometric observatories. This approach is intended to aid in assessing the performance and sensitivity of next-generation detector configurations, such as Cosmic Explorer, Einstein Telescope, and the South American Gravitational-wave Observatory.

We analyze device-dependent correlation sets generated by fixed local dichotomic measurements for two-qubit systems in the $(2,m,2)$ Bell scenario. We consider three fundamental state spaces for the composite system: the separable state space, the standard quantum state space, and the maximal tensor-product state space, which contains beyond-quantum states compatible with local quantum measurements. We formulate the corresponding correlation sets for general fixed dichotomic measurements and, in the traceless case, derive particularly simple explicit formulae for their support and gauge functions. These functions furnish dual operational characterizations of the three correlation sets: the support functions give optimal witnesses for entanglement and beyond-quantum states, whereas the gauge functions quantify the robustness of these detections against depolarizing noise. We further derive convex-hull representations that elucidate the extremal structures of the correlation sets and the physical states realizing them, showing in particular that extremal quantum correlations are realized by maximally entangled states. The fundamental limits of these dual operational tasks are governed solely by the smaller of the numbers of linearly independent measurement directions available to Alice and Bob. When both parties have three linearly independent measurement directions, our entanglement criterion detects Werner states up to the optimal PPT threshold $p_{\mathrm{crit}}=2/3$. For beyond-quantum-state detection, a nontrivial separation from the quantum set occurs only under the same measurement condition; in that case, the same optimal noise threshold is attained for an extremal state in the maximal tensor-product state space.

Noise in contemporary quantum hardware is highly non-uniform across qubits and couplers, giving rise to localized low-noise "islands" within otherwise noisy device topologies. As quantum workloads scale, executions are increasingly forced to traverse high-noise regions, degrading algorithmic fidelity. Circuit cutting provides a route to circumvent such regions by decomposing large circuits into smaller subcircuits, but its practicality is limited by exponential sampling overhead and the lack of systematic guidance on how cut strategies should align with heterogeneous hardware noise. In this work, we present a hardware-noise-aware circuit cutting framework that explicitly exploits the spatial non-uniformity of noise in quantum devices. Rather than proposing a new cut-finding algorithm, we formalize the problem of device-constraint selection under realistic hardware noise and show that this choice critically determines both execution overhead and effective noise. Using a unified gate- and wire-cutting formulation, we demonstrate that small, hardware-informed relaxations in the device constraint yield exponential reductions in execution overhead while preserving alignment with low-noise hardware regions. Across representative workloads, our method achieves an average reduction in the number of circuit executions ranging from 5-54x for 20-qubit circuits, and enables tractable circuit cutting for 50-qubit circuits and application-level benchmarks where conventional strategies incur prohibitive overhead. These results establish noise-aware device-constraint selection as a necessary ingredient for making circuit cutting resource-efficient and practically deployable on contemporary quantum hardware.

Extreme Mass Ratio Inspirals (EMRIs) are among the key targe sources for the space-based gravitational wave (GW) detectors. The waveforms of the EMRIs are highly sensitive to the types of the central supermassive black hole (SBH) and can serve as a novel sensitive tool to probe the background spacetime. In this work, we compute GWs radiated from EMRIs in the backgrounds of Kerr black hole and rotating black hole with Dehnen-type dark matter halo (DMBH). Following the Teukolsky prescription, we obtain the perturbed equations for curvature tensor from the Newman-Penrose (NP) equations, and for the DMBH we obtain the radial and angular equations through separation of variables. To solve the equations with numerical method we apply the Sasaki-Nakamura (SN) transformation to convert the Teykolsky-type equation into the SN equation. We study the radiation reaction of GWs by computing the energy flux and angular momentum flux at infinity and at the horizon. The orbital evolution is then derived from the total fluxes. We extract the two polarizations of GWs by solving the equation numerically. By comparing the waveforms of Kerr and DMBH, it is found that the DM halo induces noticeable changes in both the amplitude and phase of GWs. We compute the strain of GW detector with the response function and evaluate the mismatch between the waveforms of Kerr and DMBH. The results show that the mismatch increases with the mass parameter of DM halo and the spin of the SBH.

Quantum batteries (QBs) have emerged as a promising route for fast energy storage and on-chip power supply in quantum devices. Given the limited analytical understanding of open Floquet QBs, we employ the kicked-Ising model as a tractable platform to systematically study its performance under realistic conditions, including finite temperature effects and environmental dissipation. Starting from Gibbs states of the transverse-field Ising model, we incorporate thermal and decoherence effects along the evolution, using both analytical and numerical approaches. Taking ergotropy as a central figure of merit, we characterize the injected and extractable energy, and identify regimes where charging remains robust despite environmental effects. Our results provide a systematic framework for assessing QB performance under thermal and dissipative effects.

In the noisy intermediate-scale quantum (NISQ) regime, quantum devices contain hardware-specific noise sources which restrict device-invariant error mitigation strategies. We explore transfer learning approaches to apply noise models learned on one quantum device to a different device with the help of a small amount of data. We create a real-hardware dataset from two IBM quantum devices, ibm_fez (source) and ibm_marrakesh (target), comprising 170 noisy and ideal circuit output distributions, with device calibration features added. We train a residual neural network on the source device to map noisy to ideal outcomes. The zero-shot transfer test shows a KL divergence of 1.6706 (up from 0.3014), establishing device specificity. With K = 20 fine-tuning samples, KL drops to 1.1924 (28.6% improvement over zero-shot), recovering 34.9% of the gap between zero-shot and in-domain KL. Ablation studies reveal that the major cause of mismatches across devices is CX gate error, followed by readout error. The results show quantum noise can be learned and fine-tuned with minimal samples, and provide a plausible approach to cross-device quantum error mitigation.

Phase retrievability of a quantum channel asks whether pure states can be reconstructed from suitable measurements. In this paper, we study this problem from three complementary viewpoints: quantum information theory, operator-valued frames, and the physical realization through quantum interferometry. We first show that a quantum channel is phase retrievable if and only if its complementary channel is pure-state informationally complete. This structural characterization leads to several consequences for phase retrievability, including criteria involving the dimension of the complementary operator system, Choi-rank type bounds, and specific results for entanglement breaking channels and twirling channels. We then introduce an interferometric coupling in which two arm channels are coherently recombined through port operators \(M_i(θ)=A_i+e^{iθ}B_i\). Unlike classical mixing, this construction produces interference cross terms that can enlarge the complementary operator system and thereby enhance phase retrievability. From the frame theory viewpoint, the interferometer realizes a coherent coupling of operator-valued frames. To quantify this effect, we introduce injectivity indices for completely positive maps. The examples in Section~5 show that coherent interference can significantly improve phase retrieval behavior even when the arm channels are individually not phase retrievable.

Quantum interior point methods (QIPMs) promise polynomial speed-ups over classical solvers for linear programming by outsourcing the solution of Newton linear systems to quantum linear solvers (QLSAs). However, asymptotic speed-ups do not necessarily translate to practical advantages on realistic problem instances. In this work, I evaluate whether practical advantage of a standard hybrid QIPM pipeline can already be excluded relative to the classical open-source solver HiGHS on a broad and diverse collection of LP instances spanning eight problem families, including public benchmark libraries, such as MIPlib, and relaxations of combinatorial optimisation problems. Following the hybrid benchmarking paradigm initiated by Cade et al., I derive rigorous lower bounds on the quantum runtime under a series of highly benevolent assumptions and compare them against classical runtimes. I equip the QIPMs with the best-performing functional QLSA, the Chebyshev-based method, as identified by Lefterovici et al., and evaluate two Newton system formulations proposed by Mohammadisiahroudi et al.: the modified normal equation system and the orthogonal subspace system. The exclusion analysis yields a consistent negative picture: across all instances and for any realistic quantum cycle duration, the quantum runtime lower bounds already exceed the classical runtimes, establishing that these hybrid QIPMs will offer no practical advantage over good classical solvers for realistic linear programming instances.

We study massless scalar, electromagnetic, and Dirac perturbations of the exact asymptotically flat regular black hole supported by a phantom Dirac--Born--Infeld scalar. Using the Padé-improved WKB method, with a time-domain Prony check for the scalar fundamental mode, we compute representative quasinormal frequencies and find that larger regularity shifts the spectrum toward smaller oscillation frequencies and damping rates, whereas the quality factor changes only weakly. The spectral shifts remain well above the estimated numerical uncertainty, demonstrating that the DBI regularity scale leaves a robust spin-dependent imprint on ringdown.

In this work, we extend the class of previously introduced non-Euclidean neural quantum states (NQS) which consists only of Poincaré hyperbolic GRU, to new variants including Poincaré RNN as well as Lorentz RNN and Lorentz GRU. In addition to constructing and introducing the new non-Euclidean hyperbolic NQS ansatzes, we generalized the results of our earlier work regarding the definitive outperformances delivered by hyperbolic Poincaré GRU NQS ansatzes when benchmarked against their Euclidean counterparts in the Variational Monte Carlo (VMC) experiments involving the quantum many-body settings of the Heisenberg $J_1J_2$ and $J_1J_2J_3$ models, which exhibit hierarchical structures in the forms of the different degrees of nearest-neighbor interactions. Here, in particular, using larger systems consisting of 100 spins, we found that all four hyperbolic RNN/GRU NQS variants always outperformed their respective Euclidean counterparts. Specifically, for all $J_2$ and $(J_2,J_3)$ couplings considered, including $J_2=0.0$, Lorentz RNN NQS and Poincaré RNN NQS always outperformd Euclidean RNN NQS, while Lorentz/Poincaré GRU NQS always outperformed Euclidean GRU NQS, with a single exception when $J_2=0.0$ for Poincaré GRU NQS. Furthermore, among the four hyperbolic NQS ansatzes, depending on the specific $J_2$ or $(J_2, J_3)$ couplings, on four out of eight experiment settings, Lorentz GRU and Poincaré GRU took turns to be the top performing variant among all Euclidean and hyperbolic NQS ansatzes considered, while Lorentz RNN, with up to three times fewer parameters, was capable of not only surpassing the Euclidean GRU eight out of eight times but also outperforming both Lorentz GRU and Poincaré GRU four out of eight times, to emerge as the best overall hyperbolic NQS ansatz.

The space-borne gravitational-wave (GW) detectors will open a new mass and redshift regime, allowing us to observe massive black hole binaries (MBHBs) throughout the Universe. A subset of these systems is expected to produce electromagnetic (EM) counterparts, offering a unique opportunity to follow the continuous evolution of massive black holes through joint GW and EM observations. Realizing this potential, however, requires low-latency, high-throughput data-analysis pipelines that can extract reliable source parameters and sky localizations from space-borne data streams fast enough to trigger EM follow-up. In this work we develop a fast, normalising flow-based inference pipeline designed for early-warning analysis of MBHB signals in a TianQin-like configuration. Our method combines a learned embedding of the detector time series with a neural spline flow (NSF) to perform amortized Bayesian inference, producing posterior samples for the main source parameters in roughly one minute per event. For a representative MBHB whose merger occurs $\sim 15$ minutes after the end of the analyzed GW observation, the pipeline achieves pre-merger sky localizations of order $\sim 20~\mathrm{deg}^2$, recovers the same number of sky modes as a reference parallel-tempered Markov chain Monte Carlo (PTMCMC) analysis, and yields parameter uncertainties of comparable scale, while still operating within a practically useful pre-merger warning window. These results demonstrate that NSF-based inference can deliver accurate, near-real-time parameter estimation for space-borne MBHB GW signals, and that the resulting early-warning localizations are sufficiently precise to make rapid EM follow-up.

This paper introduces the concept of exhaustively parametrised, feasibility-respecting quantum circuits for constrained combinatorial optimisation problems. Such circuits can reach, given the right parameter values, every feasible solution with certainty -- including the optimum -- with a fixed number of parameters, while avoiding infeasible solutions altogether. This is in sharp contrast to conventional quantum alternating operator ansatz schemes, which are merely guaranteed to reach the optimum asymptotically. We introduce an abstract pipeline for constructing exhaustively parametrised, feasibility-respecting circuits from a transitive group action on a problem's feasible set. Our constructions rely on the simple combination of the group action with group representation and the novel notion of generating sequences: group elements in fixed order, possibly with repetitions, that generate the entire group. That is, we trace expressivity of parametrised quantum circuits back to the most fundamental concepts of group theory. We apply this pipeline to two concrete examples for the travelling salesperson problem, thus showing that exhaustively parametrised, feasibility-respecting circuits are not an empty definition. Furthermore, we provide numerical proof-of-principles on instances with up to nine cities, comparing the suitability of our constructions for parameter optimisation purposes against established mixers.

We introduce a new regularization scheme for divergent integrals in quantum field theory. The framework is based on the structural decomposition of the integrand asymptotic expansion, which distinguishes between contributions that drive UV singularities and those that remain finite. This asymptotic regularization method isolates the genuinely singular sector and enables a consistent subtraction of divergences while maintaining covariance and gauge symmetry. In single-scale theories, we show that the renormalized quantities exhibit a non-local logarithmic dependence uniquely determined by the UV asymptotics, offering a derivation of logarithmic terms that is independent of standard renormalization-group flows. Because it relies only on asymptotic structure rather than on standard relativistic power counting, the method is naturally applicable to theories with modified dispersion relations and non-standard UV scaling. Although formulated here for ultraviolet divergences, the underlying strategy extends straightforwardly to infrared singularities.

In realistic quantum information processing tasks, quantum states are inevitably affected by environmental noise, leading to decoherence and degradation of useful quantum resources. The coherence fraction, which serves as an important figure of merit for several quantum protocols, may decrease significantly after the action of a noisy channel. Such degradation can result in unsatisfactory performance in real-world applications. In this work, we investigate whether catalysis can be used to pre-process the input state to enhance the coherence fraction of an output state from a quantum channel. Specifically, we study whether using a processed state $ρ_s'$ as the input to a quantum channel $Λ$, instead of the original state $ρ_s$, can yield an output state $Λ(ρ_s')$ whose coherence fraction exceeds that of $Λ(ρ_s)$. We analyze the conditions under which such an improvement is possible. We also provide a practical application of our setup for the phase discrimination task. Furthermore, we establish a necessary and sufficient condition for an incoherent state preserving CPTP(Completely Positive Trace Preserving) map $\mathcal{E}$ to be a particular type of Strictly Incoherent Operation (SIO). This characterization provides a new structural understanding of SIO and clarifies its role in coherence manipulation. Our results offer practical insights into coherence preservation and enhancement in noisy quantum processes and may be useful for optimizing quantum information protocols under realistic conditions. We also provide numerical examples to support our claims.

We study numerical integration on $[0,1]$ by quantum amplitude estimation (QAE), focusing on the cost of constructing the amplitude oracle. Although QAE improves the statistical component of the integration error, this advantage is relevant only when the integrand has low encoding complexity. We introduce a hierarchy of grid function classes $\mathcal{G}_n^{(d)}$, defined by requiring the angle map $Θ_g:\{0,1\}^n\to[0,π]$ to be multilinear of degree at most $d$. Membership is classically checkable in $O(n2^n)$ time by the Walsh--Hadamard transform. For $g\in\mathcal{G}_n^{(d)}$, the encoding operator factorises into $\sum_{k=0}^d\binom{n}{k}$ multi-controlled $R_Y$ gates, interpolating between an affine $O(n)$ regime and the generic exponential regime. Combining this structure with classical discretisation estimates for $g\in C^α[0,1]$, we obtain a depth-versus-accuracy trade-off: gate count $O((\log(1/\varepsilon))^d\varepsilon^{-1})$ suffices to achieve $\varepsilon$-accuracy with constant probability. For $d=1$ this becomes $O(\varepsilon^{-1}\log(1/\varepsilon))$, improving over classical Monte Carlo for every $α\ge1$. We also prove an unconditional separation: $\mathcal{G}_n^{(1)}$ contains functions of Sobolev regularity $s<1/2$ for which the quantum oracle cost is $O(1/\varepsilon)$, whereas classical deterministic or randomised quadrature requires $Ω(\varepsilon^{-1/s})$ evaluations. These results identify explicit integrand classes for which the full cost of QAE-based integration, including state preparation, is asymptotically better than classical methods. Experiments on SpinQ Triangulum and IBM Kingston illustrate the hierarchy at $n=2$: circuits inside $\mathcal{G}_n^{(d)}$ run successfully, while those exceeding the Triangulum coherence budget fail as predicted.

Configuring variational quantum algorithms for combinatorial optimization remains a difficult, expert-driven process requiring coordinated choices over solver family, ansatz, objective, and optimizer. We present AutoQResearch, an LLM-guided closed-loop experimentation framework that casts this task as sequential policy search over a curated design space. Instead of a single static configuration, the framework searches for adaptive solver-control policies that condition future decisions on diagnostics such as feasibility, optimality gap, and convergence stagnation. The system operates through a structured workflow: an LLM agent edits a small policy surface under a fixed evaluation harness, candidate policies are screened using cheap scout evaluations, and only the strongest candidates are promoted to full confirmation. This enables controlled autonomous exploration while guarding against proxy overfitting and unstable selection. We evaluate the framework on Maximum Independent Set (MIS) and the Capacitated Vehicle Routing Problem (CVRP). On MIS instances (16--64 vertices), discovered policies substantially outperform static baselines and reveal scale-dependent behavior: CVaR objectives are effective at small scale, while QRAO-based qubit compression provides the most effective explored scaling path. On CVRP curricula (8--12 customers) and a held-out E-n13-k4 benchmark, the framework discovers adaptations involving sampling budget, penalty design, and hybrid repair protocols, yielding high-quality solutions. Methodologically, we find that staged confirmation is essential: cheap proxy evaluations can materially misestimate policy quality and even invert candidate rankings. Overall, the paper positions AutoQResearch as a benchmarked quantum--GenAI co-design workflow for autonomous solver discovery in variational quantum optimization.

We establish a framework for realizing back-action-evading (BAE) measurements and quantum non-demolition (QND) variables in linear quantum systems. The key condition, a purely imaginary Hamiltonian with a real or imaginary coupling operator, enables BAE measurements of conjugate observables. Symmetric coupling further yields QND variables. For non-compliant systems, coherent feedback is designed to engineer BAE measurements. Crucially, the QND interaction condition simultaneously ensures BAE measurements and promotes the coupling operator to a QND observable.

Two-mode squeezed states as paradigmatic entangled resources have broad applications in quantum information processing. Here, we study the generation of stable optical-microwave squeezing in structured environments within a hybrid electro-optomechanical system, where a mechanical oscillator is simultaneously coupled to an optical cavity mode and a microwave mode of an LC resonator. Specifically, an effective Hamiltonian that captures the optical-microwave squeezing interaction is constructed by combining strongly modulated driving fields applied to both photonic modes with a mechanical parametric amplifier. Based on this effective model, the dynamical evolution of two-mode squeezing in structured environments is analyzed. It is remarkably shown that the non-Markovian noise can substantially enhance the squeezing level in comparison to the Markovian case, and that two-mode squeezing can persist even in the absence of external driving fields under non-Markovian conditions, thereby mitigating the detrimental effects of anti-squeezing. Furthermore, the persistence of the two-mode squeezed state is enhanced when the environmental spectral densities of the microwave and optical modes are identical. Our work provides a theoretical framework for generating and persisting two-mode squeezing in structured environments.

In a previous paper by some of the authors (Gen. Rel. Grav. 56, 116, 2024), we introduced a novel paradigm with which to understand gravitational phenomena. We called it the Harmonic Background Paradigm (HBP). In this paradigm, gravity amounts to an effective causality deformation with respect to a more fundamental causality, which always encompasses the former through a causal cone hierarchy. In that paper, the fundamental idea was described in detail but fully elaborated only when restricted to the linear gravitational approximation. In this work, we discuss and conjecture how this idea could be extended to the full non-linear regime. We identify a connection between the cone hierarchy and a property of gravity that can be described as a screening mechanism of negative-energy gravitational clouds surrounding (but never overcoming) positive-energy seeds. We illustrate our ideas by applying them to spherically symmetric matter distributions. The paper concludes with a discussion of some key implications and directions for future research, including some remarks beyond General Relativity.

We propose a gate-based quantum algorithm for the prediction step of Bayesian state estimation based on the Fokker-Planck equation on a discretized position-velocity state space. The probability density is encoded in the amplitudes of a quantum state, enabling a compact representation of high-dimensional distributions. Exploiting the circulant structure of finite-difference operators, the evolution is realized in the spectral domain using quantum Fourier transforms and phase rotations. A key result is that the drift component can be implemented exactly in amplitude space, leading to an accurate reproduction of the classical transport dynamics. In contrast, the diffusion term does not admit a linear representation in amplitude space due to the nonlinear relation between probability density and wave function. To enable a quantum implementation, we introduce a unitary surrogate based on a Wick rotation, transforming diffusion into a dispersive phase evolution. This yields a fully unitary propagation that can be implemented efficiently on a gate-based quantum computer. The proposed method is evaluated numerically for different scenarios and shows strong agreement with the exact solution of the Fokker-Planck equation. The approach demonstrates the potential of quantum computing for Bayesian state estimation, as the representable state space grows exponentially with the number of qubits. This allows the efficient representation and propagation of probability densities that would otherwise require complex tensor decompositions on classical hardware, making the method a promising candidate for high-dimensional filtering problems.

Currently, quantum computing and artificial intelligence are driving revolutionary advancements in computational science. This study pioneers the integration of quantum kernel networks on smoothed particle hydrodynamics (SPH). SPH has matured into a highly versatile meshfree/particle method, exceptionally suited for tracking spatiotemporal trajectories and dynamic modeling phenomena. We developed a hierarchy of Lagrangian quantum network models built upon an improved quantum multilayer perceptron (QMLP). Specifically, a sequential hybrid quantum-classical framework is constructed, utilizing Pauli-Z expectation values over traditional probability outputs to ensure robust gradient-based optimization and mitigate barren plateaus. It combines smoothing kernels with quantum learning, establishing a novel quantum intelligent SPH paradigm. The framework is validated through some continuous benchmarks on eurypalynous quantum neural networks, static multi-level nebula vortex interference reconstructions and transient scalar field advectional tests. Numerical results demonstrate that while pure elementary quantum circuits struggle with parameter-specific generalization in unstructured domains, the proposed hybrid crossed-QMLP seamlessly matches the fitting accuracy of classical SPH in quantum optimized space. Although this approach currently faces limitations in computational efficiency and hardware implementation, it nonetheless paves the way for a novel investigation into quantum SPH, by mapping unstructured Lagrangian particle topologies into integrated quantum circuits.

Asymptotic symmetries are expected to leave subtle but physically meaningful imprints on quantum probes of gravity, yet their manifestation in near-horizon dynamics remains incompletely understood. We examine this question for a closed bosonic string propagating in the near-horizon geometry of a five-dimensional Schwarzschild black hole subjected to a generalized Bondi-van der Burg-Metzner-Sachs (BMS) supertranslation. The extended nature of the string makes it especially sensitive to the resulting anisotropic geometric distortions, and this sensitivity appears most clearly in the angular sector of the worldsheet dynamics. Under the gauge and falloff conditions adopted here, the temporal and radial sectors remain unaffected by the supertranslation, while the angular deformation breaks the original SO(4) symmetry of the background. The radial equation is governed by modified Bessel modes, with a nonvanishing radial conserved current, indicating transport-like propagation. Radial squeezing driven by gravity and anisotropic angular spreading induced by supertranslations provide a dynamical realization of string spreading near the horizon. Thus this analysis demonstrates that probe string dynamics encodes nontrivial signatures of BMS-induced deformations, providing a dynamical probe of symmetry structures in higher-dimensional black hole spacetimes.

This work studies quantum algorithms to solve high-dimensional stochastic differential equations (SDEs) $\mathrm{d} \mathbf{X}_t = A(t) \mathbf{X}_t \mathrm{d} t + B(t) \mathrm{d} \mathbf{W}_t$. Aiming for a speed-up in the dimension $N$ of $\mathbf{X}_t$, we generate quantum states that encode $\mathbf{X}_t$ in the amplitudes, while most of the existing quantum methods for SDEs employ binary encoding. A key challenge is the amplitude encoding of the noise term, and we address this by utilizing the quantum circuit implementation of a pseudorandom number generator (PRNG). We propose two methods: the Dyson series-based method and the Euler-Maruyama (EM)-based method. In the former, we express the noise term via the Dyson series approximation of the time evolution operator, while in the latter, it is approximated using the EM time discretization. Both methods use the quantum linear systems solver to generate the amplitude-encoding state of $\mathbf{X}_t$, making only ${\rm polylog}(N)$ queries to the PRNG circuit and the block-encodings of $A$ and $B$. Additionally, going beyond state preparation, we present methods to estimate expectations of functions of $\mathbf{X}_t$ using the state.

Nonreciprocal relaxation matrices can have skin-localized right eigenmodes, but their imprint on a mixed steady state is not fixed by the density profile alone. We develop an exact steady-state theory for number-conserving Gaussian fermion chains and show that the dominant natural orbital of the correlation matrix provides a mode-resolved diagnostic of hidden skin order. The steady-state correlator admits a biorthogonal decomposition in terms of the left and right eigenmodes of the relaxation matrix $X$ and the source matrix $Y$. This formula separates three ingredients: slow rapidity denominators, source loading by left eigenmodes, and real-space geometry from right eigenmodes. For a local pump, the pump position is read by the left modes, whereas the selected profile is drawn by the right modes. In a single-slow-mode regime, the dominant natural orbital locks to the Euclidean-normalized slow right mode. The density can follow the same boundary trend, but it is a less selective incoherent sum over occupied natural orbitals. We verify this selection law in a nonreciprocal Hatano--Nelson chain and show that, in a nonreciprocal SSH chain, the selected natural orbital crosses over from a topological edge candidate to a slow bulk-skin candidate. These results identify natural-orbital locking as a steady-state diagnostic of nonreciprocal localization in Gaussian open fermion chains.

Effective field theory (EFT) provides a systematic framework to describe possible deviations from general relativity through higher-curvature corrections to the gravitational action, capturing low-energy effects of an underlying fundamental theory. In this work, we investigate quasinormal modes (QNMs) and both weak and strong gravitational lensing in static, spherically symmetric spacetimes, focusing on the behavior of null geodesics near the photon sphere. Adopting the strong deflection limit formalism developed by Bozza, we derive the logarithmic divergence structure of the deflection angle and explicitly separate the divergent and regular contributions. Within a simplified setup with $2M=1$, we analyze how deviations from general relativity, parametrized in an EFT framework, modify key observables such as the photon sphere radius, the critical impact parameter, and the coefficients governing the strong deflection expansion. We show that these quantities encode direct information about higher-curvature corrections to the gravitational action. Our results demonstrate that strong-field observables provide a sensitive probe of EFT corrections, and that precision measurements of gravitational lensing and QNM spectra could place constraints on EFT couplings beyond general relativity, offering a novel observational window into quantum gravity-inspired effects.

We present a unified framework for the discussion of black hole thermodynamics of $d$-dimensional static black holes with spherical, toroidal or compact hyperbolic horizon topology satisfying $g_{tt}g_{rr}=-1$ in Schwarzschild gauge. To that end, we consider any such black hole as a solution to an integrable $2$-dimensional effective dilaton theory and thereby as a vacuum solution to an extended notion of $d$-dimensional quasi-topological gravity. We show that the generating function determining $f(r) = -g_{tt} $ in the integrated equation of motion provides the thermodynamic mass in a generalised first law with entropy computed as the Wald entropy. The framework presented here can be applied to singular and regular black holes with flat or anti-de Sitter asymptotics.

We study single-copy stabilizer learning, the problem of identifying a stabilizer group of dimension $n-t$ from an $n$-qubit quantum state $ρ$. We obtain two complementary results. First, in the average case, logarithmic-depth local Clifford circuits suffice to efficiently learn almost all stabilizer groups with $t=O(\log n)$, instead of the linear-depth measurements required in previous approaches. We support this result with numerical simulations for systems of up to 100 qubits. Second, we show that, in the worst case, any adaptive single-copy measurement scheme requires a number of samples that scales exponentially in $t$. Together with existing results on two-copy learning, our findings suggest that, for large $t$, identifying Pauli symmetries of a quantum system exhibits a quantum advantage in the learning setting.

The quantization of superconducting transmission-line resonators is usually introduced phenomenologically by modeling the resonator as an effective LC circuit and imposing canonical commutation relations on macroscopic variables such as charge and flux. Although this approach is highly successful, it leaves open why these macroscopic variables should obey quantum commutation relations and how this behavior emerges from the superconducting state. In this work, starting from the microscopic pairing Hamiltonian underlying BCS superconductivity, we derive the low-energy effective Hamiltonian of a circuit-QED architecture containing a superconducting transmission line with distributed capacitive and inductive elements. We establish quantitative relations between macroscopic observables, including current and voltage, and the spatially local superconducting phase, as well as the microscopic parameters of the electron-phonon system. We then extend the third quantization of the superconducting order parameter, introduced in Paper (I) for the global phase, to the spatially local case. This gives a macroscopic field quantization of the superconducting phase. We show that, after restriction to the low-energy excitation subspace, the local superconducting phase becomes a genuine quantum dynamical variable. Thus, the quantum behavior of transmission-line resonators need not be postulated at the macroscopic level, but follows from the third quantization of the superconducting order parameter. These results suggest that capacitive and inductive superconducting circuit elements share the same microscopic origin, providing a unified framework for superconducting circuit quantization.

Magnetic clock transitions (CTs), defined by vanishing first-order sensitivity of the transition frequency to magnetic field fluctuations, provide a powerful route to suppress decoherence in donor spin systems. Here, we present the observation of magnetic field CTs from an ensemble of near-surface $^{75}$As ($I = 3/2$) spins in silicon using low-field ($< 10$~mT) continuous-wave electrically detected magnetic resonance (EDMR). As the CT condition is approached, pronounced linewidth broadening is observed, consistent with a donor Hamiltonian informed linewidth model. These results establish low-field EDMR as a sensitive probe of CTs in near-surface donor systems relevant to silicon-based quantum devices.

Scalable quantum computing is fundamentally bottlenecked not by qubit count or fabrication yield, but by a rigid temporal mismatch: macroscopic classical coordination latency ($τ_c$) inevitably grows with system diameter, while microscopic quantum coherence ($τ_q$) remains strictly bounded. Beyond a critical scale, this mismatch breaches the classical control light cone, triggering a superlinear geometric penalty ($ε> 0$) that renders monolithic synchronization physically impossible. We formalize the resulting structural phase transition through a governing scaling law, $1+ε> γ$, which mandates modular decomposition and a shift from global unitaries to Local Operations and Classical Communication (LOCC). To manage the resulting resource contention under strict coherence budgets, we introduce a layered semantic architecture and a time-aware Reserve--Commit protocol. By embedding predictive temporal pre-validation, the protocol acts as an architectural semantic classifier: it preemptively aborts transactions that exceed the causal horizon and explicitly converts scheduling-induced failures into location-known erasure metadata, directly relaxing hardware fidelity thresholds for downstream QEC decoders. Under near-term transduction targets ($η_{\mathrm{trans}} \sim 0.1$), we project a crossover scale at $N_c \sim 10^5$--$10^6$ physical qubits. This threshold marks a profound architectural convergence: the footprint required for modularity aligns precisely with early fault-tolerant utility, establishing time-aware distributed orchestration, rather than monolithic expansion or centralized classical control, as the physical imperative for utility-scale quantum computing.

The Kerr-nonlinear parametric oscillator (KPO) provides a foundational semiclassical model for cat-state quantum hardware. Standard analyses of the KPO typically rely on autonomous, frozen-time approximations to describe the stabilization of macroscopic coherent states. However, state preparation and gate manipulation are driven by explicitly time-dependent microwave pulses, so the operational dynamics are inherently nonautonomous. In this paper, we show that static algebraic equilibrium pictures are incomplete for describing both state formation and gate-induced transport in the Kerr-cat qubit. For nonautonomous state preparation, we analyze the ramped resonant model by combining a linear nonautonomous stability analysis with a local invariant-graph reduction near the vacuum trajectory. This yields a quintic reduced normal form in the critical direction and identifies two symmetric post-threshold moving branches that organize the local state-formation dynamics. The associated diagnostics separate the reduced branch dynamics from the full two-dimensional phase-twist relaxation observed in the hardware coordinates. For gate execution, we model a fast pulse as a weak aperiodic perturbation of the conservative resonant figure-eight separatrix and apply Melnikov's method to derive a leading-order transport criterion. In this framework, transient lobe dynamics emerge as a semiclassical mechanism for non-adiabatic leakage, and the resulting amplitude-width threshold curve provides a leading-order geometric indicator for the onset of gate-pulse-induced transport.

We investigate inflation in modified teleparallel gravity within a scalar-tensor framework. We focus on two viable extensions of the Teleparallel Equivalent of General Relativity: a power-law model and an exponential model, which introduce controlled deviations from standard teleparallel gravity through a correction parameter $α$. Inflation is driven by a string-inspired fiber inflation potential that naturally realizes a transient ultra slow-roll (USR) phase. We solve the background equations numerically and compute the evolution of cosmological perturbations within the modified teleparallel framework. We show that both models generate an amplification of the primordial curvature power spectrum on small scales due to the USR phase, while remaining compatible with cosmic microwave background constraints at large scales. The modified gravity sector introduces corrections to the slow-roll parameters, tensor spectral index, and tensor-to-scalar ratio through derivatives of the torsion function, leading to potentially observable signatures distinct from canonical inflation. We further analyze the implications of enhanced scalar perturbations for primordial black hole (PBH) formation and demonstrate that modified teleparallel gravity provides a theoretically consistent and phenomenologically rich framework for producing PBHs during inflation.

We present a generalized phenomenological parameterization of the deceleration parameter $q(z)$ that incorporates an effective radiative component (ERC) in addition to a localized late-time contribution. The proposed framework extends previous two-parameter $q(z)$ reconstructions by explicitly regulating the high-redshift behavior while preserving the late-time transition dynamics. We constrain the free parameters $(h, q_0, z_c, z_e)$ using late-time observational data from cosmic chronometers (CC), Pantheon+ Type Ia supernovae (SNIa), H\,\textsc{ii} galaxies (HIIG), and intermediate-luminosity quasars (QSO). For the full data combination (CC+SNIa+HIIG+QSO), we obtain $q_0 = -0.25^{+0.04}_{-0.04}$ and a transition redshift $z_T \simeq 0.80$, indicating a currently accelerating Universe with a transition occurring earlier than in the $Λ$CDM model. Within the redshift range probed by the data, the reconstructed $q(z)$ deviates from the $Λ$CDM trend, suggesting a possible reduction of the late-time acceleration. Furthermore, the reconstruction favors a relatively high value of the Hubble parameter, $h = 0.729 \pm 0.006$. The ERC remains weakly constrained by late-time data but ensures a smooth and monotonic evolution of $q(z)$, $j(z)$, and $w_{\rm eff}(z)$ across a wide redshift range. Within the observed interval, the model effectively reproduces the late-time behavior of the previous parametrization, while providing a controlled extension toward early epochs. Our results show that current low- and intermediate-redshift data are compatible with a reduced late-time acceleration.

We study when block-coupled regular graphs can realize prescribed complex quantum-like bit states as exact synchronized eigenstates. Two regular subgraphs $G_A$ and $G_B$ supply normalized all-ones eigenvectors $V_A$ and $V_B$, and algebraically regular bipartite couplings reduce the full graph-supported operator exactly to a $2\times 2$ effective block on $\mathcal S=\operatorname{span} \{ \lvert 0\rangle, \lvert 1\rangle \}$. Within this reduction we prove that two natural symmetric complexifications are not universal under a real-spectrum requirement: complex symmetric coupling with real diagonal regularities forces the target computational basis amplitude ratio $r=ω_2/ω_1$, for $\lvert ψ\rangle = ω_1\lvert 0\rangle + ω_2\lvert 1\rangle$, to satisfy $r^2\in\mathbb{R}$, while real symmetric coupling with complex diagonal regularities forces $r+1/r\in\mathbb{R}$. Replacing complex symmetry by Hermitian coupling removes this phase obstruction. For any nonbasis target state, any prescribed real eigenvalue, and any prescribed nonzero signed spectral gap, a Hermitian weighted coupling realizes the target exactly. Additionally, an independently tuned directed-coupling model gives a second universality mechanism. We then pass from continuous effective parameters to finite weighted graphs with entries in $\{0, \pm1, \pm i\}$ (the fourth roots of unity and zero), characterize the balanced discrete coupling lattice by perfect matchings, and show that exact discrete Hermitian realizations are dense in the synchronized pure-state space. These results give a universality taxonomy for complex QL-bits and identify Hermitian conjugate pairing as the robust structural mechanism that supports arbitrary complex amplitudes with real two-level spectra.

Variational quantum circuits (VQCs) are a leading approach to quantum machine learning on near-term devices, yet it remains unclear which circuit architecture yields the best accuracy-parameter trade-off on classical tabular data. We present a systematic empirical comparison of four VQC families -- multi-layer fully-connected (FC-VQC), residual (ResNet-VQC), hybrid quantum-classical transformer (QT), and fully quantum transformer (FQT) -- across five regression and classification benchmarks. Our key findings are: \textbf{(i)}~FC-VQCs achieve 90-96\% of the $R^2$ of attention-based VQCs while using 40-50\% fewer parameters, and consistently outperform equal-capacity MLPs (mean $R^2{=}0.829$ vs.\ MLP$_{720}$'s $0.753$ on Boston Housing, 3-seed average); \textbf{(ii)}~FC-VQC's Type~4 inter-block connectivity provides partial cross-token mixing that approximates the role of attention -- explicit quantum self-attention yields only marginal gains on most datasets while significantly increasing parameter count; \textbf{(iii)}~expressibility saturates at circuit depth~${\approx}\,3$, explaining why shallow VQCs already cover the Hilbert space effectively; \textbf{(iv)}~LayerNorm on the fully quantum transformer improves classification accuracy, suggesting normalization is important when all operations are quantum; \textbf{(v)}~in our noise study on Boston Housing, FQT degrades gracefully under depolarizing noise while QT collapses. All results are validated across three random seeds. These findings provide practical architectural guidance for deploying VQCs on near-term quantum hardware.

Electronic final states generated by sudden changes of the Hamiltonian are studied here, with emphasis on nuclear charge variation in $β$ decay. A $λ$-parametrized family $\hat H(λ)$ that continuously connects the initial and final Hamiltonians, so that the electronic response can be represented as a continuous deformation in Hilbert space, is introduced. Within the sudden approximation, transition amplitudes are written as overlaps between eigenstates of distinct Hamiltonians. To relate non-orthogonal one-electron basis sets in a stable way, the paper uses a practical transport scheme based on overlap metrics and truncated singular value decomposition (SVD). This mapping is interpreted as a discrete counterpart of continuous transport along the $λ$ path. The formalism is first developed for the one-electron case, where analytic structure and selection rules are made explicit, and then generalized to many-electron systems via nonorthogonal determinant overlap expressions. The resulting formulation gives transition probabilities in bound and continuum channels in a way that is both numerically stable and easy to interpret.

We place several known indicators of Kochen--Specker contextuality -- the KCBS correlator $χ$, the contextual fraction $\CF$, the Shannon-entropic $n$-cycle inequality of Chaves and Fritz, and the operational commutator witness $D$ of Paper~I -- into a single projector-geometric framework organized around the overlap matrix $\Tcal_{ij} = d^{-1}\tr[(\hat P_i \hat Q_j)^2]$, where $\hat P_i$ and $\hat Q_j$ are the joint-eigenspace projectors of the two compatible observable pairs within a measurement context. The state-independent scalar content of $\Tcal$ is carried by two independent contractions: the mutual information energy $E = \sum_{ij}\Tcal_{ij}$ of Paper~I (equivalently, its logarithmic form $S_2 = -\log_2 E$), and the Maassen--Uffink extremal overlap $c_\MU = \max_{i,j}|\langle a_i,b_i | c_j,b_j\rangle|$. We prove that $S_2$ is non-increasing under coarse-graining, that $S_2(\Gcal) = 0$ is a necessary configuration-level condition for observable contextuality, and that the additive composition $S_2(\Gcal) = \sum_αS_2(\Gcal_α)$ is exact for the KCBS pentagon. We further show that in the spin-$1$ realization of the KCBS pentagon, a shared $m_s=0$ eigenstate in each context forces $c_\MU = 1$, rendering every Maassen--Uffink-type bound trivial -- a structural mechanism that makes explicit why outcome-entropic uncertainty relations based on $c_\MU$ are silent on KCBS contextuality, while $S_2 \approx 2.7266$~bits throughout. Applied to KCBS and CHSH, the framework identifies regimes in which every state-dependent witness considered here is silent yet $S_2(\Gcal) > 0$ by an amount set by the projector geometry alone.

Representability determines when a two-particle reduced density matrix (2-RDM) corresponds to a physical quantum state, enabling many-particle quantum calculations with 2-RDMs rather than the wave function. In this Letter, we present a solution of the representability problem for quantum systems without particle-number conservation. The physically allowed set of 2-RDMs can be characterized from a geometrically `orthogonal' set, the polar cone. We derive explicit linear equations for the two-body operators in the polar cone -- the intersection of the $p$-positive cone with the two-body operator space -- to obtain a systematic hierarchy of representability conditions that do not depend on higher RDMs or the wave function. Moreover, by augmenting these conditions with the particle-number variance, we obtain a unified framework for treating both particle-number-conserving and nonconserving systems. We illustrate with a spin system and molecular H$_4$.

A derivation of the relative velocity used in the definition of the relativistic cross-section is given in terms of manifestly Lorentz invariant quantities. Along the way we find that there is a certain arbitrariness in the usual definition of cross-section.

To date, frequency-dependent squeezed light has been used to reduce quantum noise in interferometric gravitational wave detectors by 6.1 dB (a factor of two). Future upgrades and detectors aim to both reduce quantum noise by 10 dB (a factor of three) and to increase the circulating power in the interferometer arm cavities. Achieving these goals will be extremely challenging due, in part, to the degradations to the squeezed state caused by mode mismatch between the internal interferometer optical cavities and between the auxiliary external cavities. It is therefore imperative to gain a detailed understanding of all sources of mismatch and to obtain experience in mitigating their effects in the current detectors in order to improve astrophysical sensitivity now and in the future. Two types of internal mismatch are identified which are due to the thermal aberrations generated when the test mass optics absorb a small fraction of the circulating arm power. It is found that the dynamics responsible for the degradations caused by the mismatch between the quadratic part of the wavefront of two modes has a characteristic low-pass frequency dependence while the dynamics of the mismatch due to all higher order thermal aberrations has a high-pass behavior. As a consequence, the two types of mismatch are predominantly responsible for different squeezing degradations -- some of which are significant for the current detectors and some of which will only be important for future detectors with longer arms. The behavior of these two types of internal mismatch are described and the implications for detector design, operation, and characterization are discussed.

The challenge of programming classical computers to play traditional, competitive games against human players has helped to advance classical hardware and software. Quantum computers have the potential to play games in a unique way: programmed only with the rules of a game, they should be able to implicitly represent all future paths of a game leading to wins, losses, or draws, and sample from this path set to identify moves that maximize the likelihood of a win. This permits skilled play without hard-coded or machine-learned strategy. As a proof of principle, we present early results obtained after programming the D-Wave quantum annealer with the rules of tic-tac-toe, enabling it to play against a human opponent. We anticipate that, as it has for classical computers, game-playing will serve as an important real-world benchmark for quantum computers.

Young interference is usually read as the gradual statistical accumulation of random detection events. Here we show that a time-reversed Young (TRY) geometry has a different statistical character: the fringe is not a marginal distribution of detector positions, but a conditional response indexed by a programmed source coordinate. With a fixed detector and a scanned source basis, the observable is an operational hybrid correlator between detector signal and source label. The resulting interference is deterministic at the response-function level, while noise enters only through estimation precision. We formulate this distinction using Fisher information, estimator variance, and noise scaling, clarifying why TRY naturally supports calibration, lock-in readout, null-fringe sensing, and source-plane superresolution.

We perform a comprehensive investigation of the early-to-late time cosmic evolution within the framework of $f(Q,L_m)$ gravity, characterized by a non-minimal coupling between non-metricity and matter. The model is further tested against a combined set of observational data, including DESI DR2 BAO, previous BAO measurements, cosmic chronometers (CC), and gravitational-wave (GW) standard sirens, using a Markov Chain Monte Carlo (MCMC) approach. Further by incorporating the Big Bang Nucleosynthesis (BBN) freeze-out constraint, we place stringent limits on the model parameters, ensuring consistency with early-Universe physics. The resulting constraints exhibit strong agreement with observations, with the model successfully describing the transition from decelerated to accelerated expansion. The evolution of the effective equation-of-state parameter, together with statefinder diagnostics and energy conditions, reveals a quintessence-like nature and confirms the physical viability of the model. Overall, the $f(Q,L_m)$ framework emerges as a viable alternative to $Λ$CDM, closely reproducing its predictions while allowing controlled deviations in the expansion history.

We investigate the nonequilibrium dynamics of core-hole excitations in a one-dimensional fermionic few-body system consisting of a spin-polarized Fermi bath coupled to a single heavy mobile impurity. The bath is initially prepared in a particle-hole configuration by emptying a selected bath single-particle orbital, while the impurity is displaced with respect to the center of the bath confinement potential. The quench dynamics are initialized by suddenly switching on the impurity-bath interaction. To resolve the resulting dynamics, we combine two complementary \textit{ab initio} approaches, namely the Multi-Layer Multi-Configuration Time-Dependent Hartree method for mixtures and a multi-channel Born-Oppenheimer framework. We show that the postquench response is governed by the interaction strength, impurity confinement, mass imbalance, and the location of the initially prepared hole within the Fermi sea. The density evolution and impurity center-of-mass motion reveal a competition between mixing and demixing of impurity and bath, while the von Neumann entropy demonstrates the buildup of pronounced many-body correlations. Most importantly, the occupation dynamics of the initially emptied orbital identifies deep core holes as substantially more robust against refilling than bulk or edge vacancies. Our results establish core-hole excitations as robust dynamical many-body features in trapped ultracold fermions and provide a controlled route towards probing orthogonality response, correlation buildup, and hole refilling in real time.

We present a novel architecture-aware transpilation method for exact general unitary gate synthesis on superconducting quantum hardware. Our approach is tightly integrated with the optimized block-ZXZ decomposition, exploiting its recursive structure to make hardware-aware decisions at each level of the recursion rather than treating transpilation as an independent post-processing step. The method introduces three key techniques: a greedy qubit mapping strategy that minimizes pairwise distances between physical qubits, an adaptive Gray code selection combined with qubit swapping that optimizes the construction of uniformly controlled Rz gates for the target topology, and a heuristic for reducing CNOT gates by exploiting the structure of long-range CNOT ladders. We benchmark our method against TKet, Qiskit, and Pennylane on the 20-qubit IQM Garnet (square lattice) and the 156-qubit IBM Marrakesh (heavy-hex) architectures with qubit counts ranging from 3 to 11. Our method achieves CNOT count reductions of up to 36 percent on the IQM Garnet and up to 34 percent on the IBM Marrakesh compared to the best competing transpiler, while simultaneously achieving transpilation speedups of up to 553x. Furthermore, our method is the only one capable of transpiling circuits beyond 10 qubits within a 30-minute time limit across both architectures.

The Barnett effect is usually understood through an effective magnetic field generated by mechanical rotation, while its reciprocal Einstein--de Haas effect describes the transfer of spin angular momentum to mechanical motion. We show that this effective-field picture changes qualitatively once the mechanical degree of freedom itself is quantized. To demonstrate this, we introduce an exactly solvable minimal spin-rotor model in which a spin-$1/2$ is coupled to a quantum rotor. In a fixed angular-momentum sector, the model reproduces the conventional Barnett splitting and remains formally equivalent to a Zeeman problem. For a superposition of rotor sectors, however, the Barnett field becomes operator-valued and the resulting dynamics generates coherent spin-rotor entanglement. This is directly visible in the reduced spin purity, rotor coherence, and entanglement entropy. Our results identify a minimal quantum setting in which the Barnett effective-field picture departs from its classical form and acquires a reciprocal manifestation through spin-dependent rotor coherence.

We consider static, spherically symmetric configurations of nonlinear electromagnetic fields with Lagrangians $L(f)$, where $f = F_{μν} F^{μν}$, in general relativity (GR) and other metric theories of gravity. The corresponding exact solutions are well known in the framework of GR in cases where only an electric charge ($q_e$) or a magnetic charge ($q_m$) are present, but only a few solutions in particular examples of $L(f)$ are known for dyonic systems with both nonzero $q_e$ and $q_m$. We study the properties of such systems in the special case of equal electric and magnetic charges and, assuming a correct Maxwell limit of $L(f)$, show that there always exists such a configuration of the electromagnetic field that the invariant $f$ is zero in the whole space. It leads to the existence of the corresponding families of solutions both in GR in the presence of other sources of gravity (like fluids or scalar fields) and in a wide range of extended theories of gravity (e.g., scalar-tensor and $F(R)$ gravity), in which the nonlinear electromagnetic field behaves in an especially simple manner.

In this paper, we investigate the photon regions, shadow, and observational constraints of a Kerr-Newman-like black hole in Bumblebee gravity within a plasma medium. By employing a specific non-homogeneous power-law plasma model to ensure the separability of the Hamilton-Jacobi equation, we derive the null geodesic equations, analyze the photon regions, and construct the black hole shadow. Furthermore, we introduce two sets of shadow observables to systematically analyze the distinct effects of each physical parameter (spin $a$, charge $Q_0$, Lorentz-violating parameter $\ell$, and plasma parameter $k$) on the shadow geometry. Specifically, we find that $a$ and $\ell$ mainly enhance the distortion of the shadow, whereas $Q_0$ and $k$ primarily lead to its radial shrinkage. Additionally, a brief evaluation of the energy emission rate shows that an increase in these parameters generally suppresses the emission peak. Finally, by modeling M87* as a charged rotating black hole in Bumblebee gravity surrounded by plasma, we can constrain the physical parameters using observations from the Event Horizon Telescope (EHT). While the angular diameter $θ_d = 42 \pm 3 \, μ\text{as}$ narrows the viable parameter space, the circularity deviation $ΔC \lesssim 0.1$ and axis ratio $1 < D_x \lesssim 4/3$ obey the EHT limits. This suggests that the charged rotating black hole in Bumblebee gravity surrounded by plasma might be a candidate for real astrophysical black holes.

The current noisy intermediate-scale quantum (NISQ) era is characterized by substantial errors and noise, which limit the practical feasibility of deep, many-qubit circuits. To address these constraints, quantum circuit cutting has emerged as a promising tool. Recently, there has been significant research on methods for performing such cutting effectively. In this work, the duality between quantum circuits and classical graphs - specifically, directed acyclic graphs (dags) - is leveraged to analyze the complexity of finding an optimal circuit-cutting configuration that minimizes the number of cuts. After developing a rigorous graph-theoretic framework, the complexity of identifying cut locations that partition a given quantum circuit into smaller fragments is characterized. The corresponding graph-combinatorial task is then defined, and the resulting partition problem is shown to be NP-complete. Furthermore, even a simplified version of the problem, restricted to circuits composed only of one- and two-qubit gates, is shown to be NP-complete. Finally, based on these constraints, an algorithm grounded in satisfiability modulo theories (SMT) is proposed to find optimal cuts when the number of qubits per partition is bounded. This work therefore provides a complexity-theoretic characterization of cut placement and a practical solver for bounded-size decompositions.

We study the small-$c$ expansion of general relativity in ADM variables up to next-to-next-to-leading order (NNLO). We show that, in the stationary sector, this formulation renders the field equations more tractable for explicit solution building. The stationary sector exhibits both strong- and weak-gravity branches, whose structure becomes richer at NNLO. In the strong-gravity branch, we first obtain exact vacuum solutions of NLO Carroll gravity, including the Lense--Thirring and rotating C-metric backgrounds. At NNLO, we then construct the corresponding Lense--Thirring-type and C-metric-type exact vacuum geometries. These solutions also arise from the small-$c$ expansion of the Kerr and rotating C-metric geometries around the strong-gravity background, up to $\mathcal{O}(J)$ at NLO and up to $\mathcal{O}(J^3)$ at NNLO. In the weak-gravity branch, we find exact Hartle--Thorne-type solutions with an independent quadrupole moment, together with exact spin-squared corrections and a mixed quadrupolar-rotating solution. We further extend the $\ell=0,2$ sector by including higher multipoles up to $\ell=4$, where $\ell$ denotes the multipole index. These results show that the full NLO/NNLO theory admits a richer stationary vacuum sector than the magnetic Carroll truncation. More broadly, the ADM formulation provides a practical framework for constructing and analyzing stationary backgrounds in the small-$c$ expansion of general relativity, and may also offer a useful framework for organizing rotational and higher-multipole deformations relevant to compact astrophysical objects such as slowly rotating black holes and neutron stars.

We study a one-dimensional non-Hermitian Stark chain in which nonreciprocal hopping, a linear potential, and linearly graded hopping act simultaneously. The central question is how boundary pumping and field-induced confinement are reorganized when the hopping amplitude itself grows with position. We show that the graded term separates the two localization channels at the level of the large-position asymptotics. An exact diagonal similarity transformation removes the bond asymmetry and converts the usual exponential skin factor into an algebraic boundary accumulation with exponent $η=γ/F_2$. The transformed symmetric chain then reduces asymptotically to a constant-coefficient recurrence, giving the Stark threshold $|F_1|=2|F_2|$. The original right eigenstates acquire the unified envelope $ψ_j^R\sim j^ηφ_j$, with oscillatory, double-root, and exponentially localized branches across the threshold. This form also yields two finite-size scales, one measuring the logarithmic screening of nonreciprocity and the other balancing the algebraic skin factor against the exponential Stark tail. A joint localization map in the $(γ,F_1/F_2)$ plane verifies this structure. The edge polarization bends near the Stark threshold and weakens on the localized side, while the inverse participation ratio of the most localized eigenstates rises rapidly for $F_1/F_2>2$. Using a normalized Gaussian projector appropriate for non-unitary evolution, we further show that the same threshold enhances half-chain entanglement growth after a charge-density-wave quench. These results identify graded hopping as a controlled mechanism for screening nonreciprocity, resetting Stark asymptotics, and organizing the finite-size crossover between algebraic skin accumulation and Stark localization.

Recent developments have seen the application of finite Gaussian basis sets to the $α(Zα)^{n\geq3}$ vacuum polarization. The energy shift for $s$ and $p$ electron states have been tabulated and their convergence investigated. In this work, we extend this problem to the multi-electron case. Hartee-Fock potentials obtained self-consistently are used to treat the vacuum polarization for lithium-like systems and are found to be in good agreement with comparable results in the literature. The results presented in this work demonstrate the use of Gaussian basis sets for atomic potentials whose Green's functions expressions cannot be simply obtained via analytic or numerical methods.

Ergodicity breaking in isolated systems has emerged as an important frontier in the study of quantum many-body physics. While generic Hamiltonians are expected to obey the eigenstate thermalization hypothesis (ETH), recent studies on Hilbert space fragmentation (HSF) have revealed possible robust nonthermal behavior even in disorder-free systems. Although numerous models exhibiting strong HSF are already known, existing analyses are typically model dependent, and a general organizing principle remains elusive. In this work, we introduce a simple mechanism for achieving strong HSF in one-dimensional integer spin chains, which we term "peak-valley (PV) fragmentation". The key idea is to devise a simple local rule which ensures the spin states in the computational basis can be labeled by a set of emergent good quantum numbers corresponding to the heights and depths of alternating peaks and valleys in a geometrical representation. We demonstrate that some known examples of strong HSF models, as well as their variants which break the HSF property, can be understood within the framework of PV fragmentation. Our approach also enables systematic construction of new fragmented models in higher-spin systems, and allows us to identify higher-order HSF models.

Spectral problems governed by differential operators underpin a wide range of physical systems, yet remain computationally challenging because their spectra depend sensitively on continuous parameters and often demand repeated evaluations across parameter space. Here we present $\texttt{DeepOPiraKAN}$, an open source physics informed neural network architecture for spectral analysis. By combining operator learning with enhanced optimization stability, it captures the underlying parameter-to-spectrum mapping in a single model, avoiding repeated spectral solutions at isolated points in parameter space. As a representative and stringent benchmark, we apply this framework to the computation of quasinormal modes of Kerr black holes. A single trained network accurately resolves modes with $(\ell,m)\in \{(2,0),(2,1)\}$ and overtones up to $n=7$ across the full spin range, achieving relative errors of $\mathcal{O}(10^{-6})$ for the fundamental mode and gradually increasing to $\mathcal{O}(10^{-4})$ for higher overtones, benchmarked against the Leaver's method. This level of accuracy is already significant for black hole spectroscopy and practical ringdown modelling for current and future observatories. More broadly, these results highlight the potential of $\texttt{DeepOPiraKAN}$ as a general and scalable framework for parameter dependent spectral problems across complex physical systems.

This chapter provides an overview of chaotic billiard lasers as a prominent branch of quantum chaos. These lasers offer an ideal experimental platform for demonstrating the principles of quantum chaos within a physical system. We begin by introducing the fundamental principles of chaotic ray dynamics in optical microcavities, where the transition from regular to fully chaotic dynamics fundamentally alters the underlying wavefunctions and lasing properties. A central focus is placed on "chaos-assisted light emission," which serves as a practical manifestation of "chaos-assisted tunneling" -- a hallmark phenomenon in the study of quantum chaos. We discuss both theoretical frameworks and experimental validations, demonstrating how chaotic orbits facilitate the coupling between evanescently localized modes and far-field emission. Furthermore, exploring how the presence of a gain medium influences established results from quantum chaos research remains a fundamental and intriguing problem in physics. To address this, we establish a rigorous and comprehensive derivation of the Maxwell-Bloch equations for two-dimensional microcavity lasers, specifically examining their application to fully chaotic, stadium-shaped billiard lasers. By bridging the gap between nonlinear lasing processes and chaotic wavefunctions, this chapter highlights the unique potential of chaotic billiards for controlling light-matter interactions and shaping the next generation of unconventional coherent light sources.

We study spherically symmetric gravitational collapse of an inhomogeneous fluid with anisotropic energy momentum tensor (EMT) in Rastall gravity. Considering a linear equation of state (EoS) for the fluid profiles, i.e., $p_r=w_rρ$ and $p_θ=w_θρ$, we try to build and investigate non-singular collapse scenarios for which, the spacetime singularity that appears in the homogeneous case~\cite{ahz2019}, is absent. We therefore set the Rastall parameter in such a way that the effective radial pressure vanishes. This helps us to obtain a class of exact nonsingular solutions in which the matter shells undergo a {collapse} process in a contracting regime, reach a bounce point, and then enter an expanding phase. We further investigate formation of trapped surfaces during the dynamical evolution of the collapsing body. {It is found} that for the obtained solutions, {the} trapped surface formation can be avoided and consequently, the bounce event is not covered by the apparent horizon. Validity of weak energy condition (WEC) is also examined for the obtained solutions.

Topological quasiparticles of light, such as classical and quantum optical skyrmions, have so far relied on fully coherent or pure quantum states whose topology is encoded in the entanglement between polarization and two-dimensional spatial modes. Here we show that skyrmionic topology can emerge directly within the density matrix of a mixed quantum state. We introduce a framework in which a coherence-Stokes vector defines a topological texture over the density matrix, enabling the realization of quantum skyrmions using only a pseudospin and a real or synthetic one-dimensional space of modes. For a single photon, the density matrix is analogous to the coherence matrix of classical light, and can be encoded using partially coherent electromagnetic fields. We further analyze the topological texture encoded in a bipartite entangled photon pair, showing how skyrmions arise not only in the full bi-photon system, but also in its reduced subspaces with any pseudospin-mode combination simultaneously. We then explore the robustness of such skyrmions to environmental noise, and discover their persistence in mixed-quantum states of multiple photons. Finally, we propose a feasible experimental route to generate and measure such skyrmions using integrated photonic networks, and suggest avenues for similar implementations in other quantum systems. Our work paves the way for robustly encoding classical information on partially coherent light or on mixed quantum states and for encoding topological charges on many-body quantum systems.

Controlling and generating ultrafast free-electron wavepackets via laser is pivotal for photon-induced near-field electron microscopes (PINEM) and also for developing compact, coherent free-electron radiation sources. Here, we present a quantum echo-enabled high-harmonic generation (QEEHG) scheme that manipulates the quantum phase of electron wavepackets to produce tunable, coherent high-harmonic radiation at ultrashort wavelengths. This framework, inspired by the EEHG concept for free-electron lasers by Stupikov et al. (2009), leverages multiphoton PINEM scattering followed by dispersive chirp sections to induce quantum interference among photon sidebands. Such interference selectively enhances a targeted harmonic order - for instance, the 60th harmonic at 13.3nm from an 800nm seeding - while suppressing unwanted radiations. The optimization of harmonic orders and its non-classical spectral characteristics are analyzed. This quantum echo technique establishes a promising paradigm for compact coherent sources and provides new perspectives for quantum wavefunction shaping in ultrafast electron microscopy and diffraction.

We study the entanglement dynamics of $n=2,3,4$-qubit Bell- and GHZ-type states under an amplitude-damping channel (ADC). We quantify multipartite entanglement using the genuine multipartite concurrence (GMC) and evaluate its utility through the optimal teleportation fidelity. For $2$-qubit states, we analyze the standard (Bennett) teleportation protocol. For $3$- and $4$-qubit states, we study controlled quantum teleportation (CQT) with one and two \emph{controllers}, respectively. Entanglement sudden death (ESD) denotes the abrupt, finite-time disappearance of entanglement caused by decoherence in contrast to asymptotic decay. To counteract ESD, we apply local NOT ($\hatσ_x$) operations on $m$ of the $n$ qubits ($m \leq n$) and derive analytic formulae, revealing that a single-NOT operation often suffices to alter ESD into asymptotic decay when handling GMC. In contrast, teleportation fidelity can decay more rapidly for single-NOT flipped states, whereas flipping all qubits is more useful for preserving teleportation fidelity in certain regimes, highlighting that the amount of entanglement alone does not guarantee teleportation utility. Remarkably, in the case of GHZ-type states, ADC-evolved mixed biseparable states can be exploited successfully in the CQT protocol. Further, using the GHZ-symmetric parametrization, we map the 2- and 3-qubit ADC-evolved mixed states onto a $(x,y)$ plane, revealing their SLOCC (Stochastic Local Operations and Classical Communication) entanglement classes. We also explicitly check the Bell-CHSH nonlocality hierarchy in the 2-qubit teleportation alongside localizable-entanglement diagnostics for 3-qubit CQT. Our results clarify the distinct roles of global versus localizable bipartite correlations and suggest simple, experimentally accessible unitary controls for preserving useful quantum resources in noisy channels.

Calibrating the role of entanglement in quantum algorithms is a crucial task in the development of quantum computing. Most existing studies have primarily focused on how the static properties of entanglement-such as its magnitude and phase-affect key performance metrics. In this work, we instead explore the relationship between the dynamical behaviors of entanglement and the execution of variational quantum algorithms from a geometric perspective. We find that, in contrast to conventional Hamiltonian dynamics where the evolution process is dominated by the dynamical phase, quantum state evolution in quantum algorithms is primarily governed by the geometric phase with the trajectory determined by the parameter-dependent Hilbert space geometry. In the problem-agnostic Hardware-Efficient Ansatz (HEA), entanglement dynamics and state evolution are decoupled. Conversely, in the problem-inspired Hamiltonian Variational Ansatz (HVA), the dynamical phase contribution is enhanced, allowing entanglement to function as a dynamical resource: more entanglement consumption correlates directly with faster quantum state evolution.

Orbital angular momentum (OAM)-entangled states produced by spontaneous parametric down-conversion (SPDC) are considered ideal for realizing high-dimensional entangled states, which have several advantages for quantum technologies. However, the limited sensitivity of current two-photon OAM detectors is a major roadblock not only for realizing such technologies but also for resolving foundational questions, such as OAM conservation in SPDC. The current theoretical understanding is that OAM is not conserved in Type-II SPDC but is conserved in Type-I. Experimentally, although non-conservation in TypeII has not been demonstrated, conservation in Type-I has been reported frequently and has become an underlying assumption for techniques generating high-dimensional OAM entangled states. In this work, we experimentally demonstrate a high-sensitivity two-photon OAM detector, using which, contrary to the current understanding, we report non-conservation of OAM in Type-I SPDC. We attribute this to a spatial walk-off effect and prove it using a framework free of standard phase-matching approximations.

We reconsider the problem of regularizing the divergent series $\sum_{n=1}^{\infty}n^α$ for $\operatorname{Re}α>-1$, and offer a regularization prescription that yields the Riemann zeta regularization as a special case. The development of the regularization is framed as a two-step problem. The first step is prescribing a regularization of the divergent sum $\sum_{n=1}^{\infty}n^m$ for every non-negative integer $m$; and the second step is the extension of the sum for non-integer $α$. The extension is obtained under the consistency condition that the regularized sum for integer $m$ emerges continuously from the sum for non-integer $α$. The scheme is specified by a differential generator $L=L(\mathrm{d}/\mathrm{d}t)$ through which a generalized spectral function (GSF), $K_L(t)$, is constructed. Under the condition that the GSF has a holomorphic complex extension $K_L(z)$ with $z=0$ as a pole, the case for integer $m$ takes the regularized value $\sum_{n=1}^{\infty} n^m = (2πi)^{-1}\oint_C L^m K_L(z) z^{-1}\mathrm{d}z$, where $C$ is a closed contour enclosing only the pole of $K_L(z)$ at the origin. On the other hand, under the consistency condition, the case for non-integer $α$ takes the value $\sum_{n=1}^{\infty}n^α=(2πi)^{-1}\int_{\tilde{C}} L^α K_L(z) z^{-1}\mathrm{d}z$, where $L^α$ is the fractional extension of $L^m$ and $\tilde{C}$ is an appropriate deformation of the contour $C$. Here, we obtain the regularization corresponding to the generator $L=h(t) \mathrm{d}/\mathrm{d}t$, where $h(t)$ has the analytic extension $h(z)$ such that $1/h(z)$ is an entire function. We find that the regularized sum is equal to the Riemann zeta regularized value plus terms determined by the generator $L$.

This work introduces a novel quantum algorithm for gradient-based edge detection that operates entirely within the quantum circuit model. Grayscale images are encoded using the Novel Enhanced Quantum Representation (NEQR), allowing exact arithmetic on pixel intensities. Directional gradients are computed by generating superpositions of neighboring pixels via cyclic shift operations and performing subtraction with an exact quantum arithmetic circuit. To refine accuracy, we introduce a direction-aware shifting mechanism that aligns edges with the darker side of intensity transitions. Our novel Quantum Partitioning Algorithm enables efficient in-place thresholding of edge candidates. This work exhibits polynomial-time improvements and optimizes the ancilla count compared to previous NEQR-based quantum edge detection algorithms. These results demonstrate a resource-efficient and fully quantum approach to edge detection, highlighting a practical quantum advantage in image processing.

Taiji is a gravitational wave detection mission in space initiated by the Chinese Academy of Sciences, which will open the millihertz window through a heliocentric triangular constellation of three drag-free spacecraft. Its ultimate sensitivity is determined partly by the residual acceleration noise of the gravitational reference sensors (GRS), within which the coupling between the test-mass and the fluctuating environmental magnetic field constitutes one of the key stray-force contributions. Following the path established by the LISA and TianQin teams, high-precision ground characterization of remanent magnetic moment $\vec{m}_r$ and volume susceptibility $χ$ of the test masses is a central step in the Taiji pre-launch test program. A persistent challenge for this characterization is the non-stationary, colored background noise inherent to torsion-pendulum facilities, which systematically biases classical Ordinary Least Squares (OLS) and Kalman filter (KF) estimators. We propose an AI-enhanced Differentiable Weighted Least Squares (AI-WLS) framework that fuses a dilated one-dimensional residual network, acting as a dynamic noise evaluator, with a fully differentiable analytical physical solver. This architecture preserves the exact linear mapping from the magnetic parameters to the torque response while autonomously identifying and suppressing contaminated data segments. Validated on real measured noise from the Changchun Institute of Optics, Fine Mechanics and Physics torsion-pendulum facility developed for Taiji, which achieves a torque sensitivity of order $10^{-13}\,\mathrm{N\cdot m\,Hz^{-1/2}}$, the AI-WLS framework bounds the maximum absolute estimation errors at $4.46\times 10^{-10}\,\mathrm{A\cdot m^2}$ for $\vec{m}_r$ and $7.8\times 10^{-8}$ for $χ$, satisfying Taiji's ground-test requirements on all these parameters simultaneously.

We consider a generalization of the two-body contact interaction for nonrelativistic particles confined to a one-dimensional box, in which the interaction is decentered, i.e., the particles interact only when they are separated by a distance c. In contrast to the harmonically trapped system, this model is nonintegrable. Despite this, we demonstrate that the system exhibits partial solvability due to the presence of dark states, i.e., bosonic or fermionic states unaffected by the interaction. These states form exactly solvable subspaces embedded within an interacting spectrum. We characterize the stationary properties of the system, identify the conditions for the appearance of dark states, and show how they structure the spectrum and delineate interacting and noninteracting sectors.

We investigate quantum average correlations and complementarity relations based on metric-adjusted skew information. Several natural averaging procedures are considered, including complete families of mutually unbiased bases, all orthonormal bases, operator orthonormal bases, and twirling channels induced by the unitary group. All these approaches lead to the same closed expression, which identifies the resulting average correlation as an intrinsic quantity independent of the averaging scheme. By defining measures of wave and particle features via metric-adjusted skew information, we establish complementarity relations among wave and particle features, quantum entropy, and average correlation. These results provide a unified framework for investigating quantum average correlations and complementarity relations in terms of metric-adjusted skew information.

This paper studies the quantification and structural properties of quantum average correlation based on average coherence. Motivated by two mathematically equivalent approaches to define average coherence: one by averaging over complete sets of mutually unbiased bases, and the other by integrating over all orthogonal bases under the Haar measure, we define an average correlation for bipartite systems as the difference between global and local skew information. This correlation measure is shown to satisfy essential properties including non negativity, contractivity under local quantum channels, and local unitary invariance. We further prove the equivalence between the average correlation defined via mutually unbiased bases and that defined via unitary groups. Finally, we derive a complementarity relation that connects wave-particle duality with the average correlation between a system and its environment.

Flexible manipulation of quantum correlation resources enables the implementation of diverse quantum tasks based on hybrid quantum networks, where atom-magnon and optomagnonic entanglements and steerings play important roles. In this work, we propose an effective scheme to generate and manipulate quantum entanglements and steerings based on a hybrid optomagnomechanical system, which is composed of a polarizer, an optical cavity with YIG bridge as one end, and an atomic ensemble in it. According to the results of the parameter dependence of various quantum correlations, we can selectively generate bipartite and genuine tripartite entanglements and deterministically manipulate the concrete situation of bipartite, multipartite steerings, and collective pentapartite steering, by adjusting the polarization direction of the driving laser and the Tavis-Cummings coupling strength. Our all-optical controlled scheme is flexible, convenient, compact, and experimentally feasible, because multiple coupling channels can be tuned simultaneously. This work provides a new perspective for implementing specialized quantum tasks, such as hierarchical ultra-secure multi-user quantum communications.

Quantum fully homomorphic encryption (QFHE) promises secure delegated quantum computation but has been impeded by the prohibitive quantum resource demands of existing constructions. This paper introduces a unified framework that achieves an \textbf{exponential improvement} in efficiency by synergistically integrating three theoretical tools: \textbf{modular arithmetic programs (MAP)}, the \textbf{garden-hose model}, and \textbf{measurement-based quantum computation (MBQC)}. Our central innovation is a novel MAP tailored to the algebraic structure of Learning-with-Errors (LWE) decryption. Unlike generic approaches that incur exponential overhead, our MAP computes the inner product $\langle \boldsymbol{sk}, \boldsymbol{c} \rangle \bmod q$ by tracking a partial sum modulo $q$, requiring only $O(\log q)$ bits of state width. This yields branching programs of width $O(\log λ)$ and length $O(λ\log λ)$, thereby reducing the size of the essential quantum gadget from $O(λ^{2.58})$ to $O(λ\log^2 λ)$ EPR pairs -- a concrete improvement factor of $2^{15}$ to $2^{18}$ for standard security parameters. Critically, we demonstrate that LWE decryption is not a \textbf{symmetric function}, necessitating our specialized MAP design beyond prior symmetric-function optimizations. The framework provides a direct mapping from the MAP to an efficient gadget via the garden-hose model, with MBQC furnishing the deterministic control flow for homomorphic evaluation. The resulting QFHE scheme supports \textbf{fully classical clients}, relies solely on the \textbf{classical LWE assumption} (avoiding circular security or quantum hardness assumptions), and maintains compactness. This work dramatically lowers the quantum resource barrier for practical QFHE, paving the way for realistic privacy-preserving quantum cloud computing.

High-precision quantum parameter estimation is fundamental to the advancement of quantum metrology. Although reservoir engineering provides a powerful approach to improve estimation by tailoring system-environment interactions, the role of the squeezing phase and correlations arising from the sequential utilization of the same squeezed reservoir remains inadequately explored. In this work, we employ a correlated squeezed-thermal reservoir to enhance the precision of estimating the phase parameter $φ$ and the correlation factor $μ$, both individually and simultaneously. We show that the squeezing phase $Φ$ is crucial for achieving quantum-enhanced precision, with optimal phase-matching conditions that depend strongly on $μ$. Specifically, we derive the near-optimal phase-matching relations aimed at maximizing the quantum Fisher information (QFI) for both $φ$ and $μ$, as well as minimizing the total variance $Δ_{\rm sim}$ in joint estimation. Furthermore, we show that the joint estimation variance is dominated by $F_φ$, which motivates our search for the phase-matching conditions that minimize $Δ_{\text{sim}}$. Through the ratio $R$ of variances, we demonstrate that joint estimation conserves quantum resources and maintains high precision when the squeezing phase is optimized for $F_φ$, despite the inherent incompatibility of the parameters. These findings provide practical insights into reservoir engineering strategies for high-precision quantum sensing and information processing.

Causal discovery from observational data underpins applications in finance, climate modeling, and machine learning. Constraint-based causal discovery reduces structure learning to a sequence of conditional independence (CI) tests, where each test decides independence by estimating conditional mutual information $I(X;Y \mid Z)$ to additive precision $τ$ and thresholding against it. Classically this requires $Θ(1/τ^{2})$ samples per test, a cost that dominates in the high-precision regime typical of weak dependencies. We present QKLA (Quantum Kullback--Leibler Amplitude estimation), a quantum algorithm that encodes a clipped log-density ratio as a bounded amplitude and applies amplitude estimation to recover the KL divergence. Given coherent oracle access to the joint distribution, QKLA achieves a quadratic precision improvement, needing only $\mathcal{O}((L/τ)\log(1/δ))$ queries, where $L$ is the log-ratio clip bound. Embedded in the PC algorithm, this compounds to an $\widetildeΩ(1/(Lτ))$ reduction in total queries for the full causal discovery procedure. We validate the theory in three experiments. A gate-level state-vector simulation of the full QKLA circuit confirms the predicted $\mathcal{O}(1/M)$ error decay. Across $K=20$ random binary distributions, classical and quantum error scalings match theory to slope accuracy $\pm 0.005$. On two benchmark networks (\textsc{Asia}, 8 nodes; \textsc{Synthetic-12}, 12 nodes), quantum PC matches classical skeleton-recovery F1 while using $2.5$--$3.0\times$ fewer oracle queries at $τ= 5\cdot 10^{-3}$ bits and up to $12\times$ fewer at $τ= 10^{-3}$ bits.

Signal photons emitted by quantum nodes typically fall outside the low-loss telecom window of optical fibers, leading to severe transmission losses. Quantum frequency conversion (QFC) offers an effective optical interface that bridges quantum nodes with telecom-band channels, enabling long-distance quantum communication. In this work, we demonstrate a compact, fiber-integrated QFC system with low noise and a high signal-to-noise ratio (SNR). Using a periodically poled lithium niobate (PPLN) waveguide, input photons at 637.2 nm are down-converted to telecom photons at 1588.3 nm. Our system achieves a total conversion efficiency of approximately 9%, with pump-induced noise suppressed to 154 Hz. For input photon rates of 32.7, 118.0, and 327.7 kHz, the corresponding SNRs are 12.3, 43.9, and 117.8, respectively. We further develop a theoretical model to simulate the entanglement fidelity between nitrogen-vacancy (NV) center spins and the frequency-converted telecom photons. At the emission rate of an NV center, our QFC system maintains an expected fidelity exceeding 52% over a transmission distance of 100 km. These findings highlight the potential of our QFC system for scalable, long-distance quantum networking.

We investigate whether Scalar-Tensor-Vector Gravity in its weak-field modified gravity form can account for the cluster-scale inverse-square force law inferred from recent kinematic Sunyaev-Zeldovich measurements of cluster pairwise motions. The starting point is the X-COP cluster fit of STVG-MOG, for which a representative baryonic cluster mass $M\sim 10^{15}M_\odot$ together with parameters $α\sim 9.11$ and $μ\sim 0.196~{\rm Mpc}^{-1}$ provides a successful description of cluster dynamics without particle dark matter. We extrapolate this fit to the separation range $30$ to $230~{\rm Mpc}$, relevant for the pairwise kSZ analysis. Since the Yukawa transition length $μ^{-1}\simeq 5.1~{\rm Mpc}$ is much smaller than these separations, the STVG-MOG acceleration law reduces to an effective inverse-square form. This explains why the theory can satisfy the observed Newtonian behavior while remaining distinct from MOND-like long-distance modifications. We derive the corresponding pairwise velocity curve and show that, after fitting a single overall kSZ amplitude, the extrapolated STVG-MOG prediction reproduces the measured trend of the pairwise kSZ data. The analysis shows that the X-COP cluster fit and the cosmological-scale kSZ force-law result are mutually consistent within STVG-MOG.

Slepian developed the Prolate Spheroidal Wavefunction (PSW) spatial-mode basis, which forms the normal modes of the Fresnel-propagation kernel of a free-space optical communications channel bookended by hard-circular apertures. The zero-th order PSW mode has the highest power-transfer eigenvalue, exciting which on the transmitter side therefore maximizes the transmissivity for single-spatial-mode communications. We show that the transmissivity performance of this fundamental PSW mode can be obtained by an aperture-truncated Gaussian beam of an optimized beam waist, despite the two mode shapes deviating from one another in the near-field regime.

We present an invariant characterization of the Kerr spacetime, and utilize the invariant structure of the spacetime to derive a function whose zeros identify a special family of null geodesics. Each member of this family is tangent to every photon surface in the Kerr photon region, offering a method of invariantly characterizing photon surfaces in axially symmetric spacetimes and thereby a providing a computational tool for efficiently computing the geodesic equations for any part of the photon region. The invariant that identifies all of the spherical photon orbits is parameterized by a Lorentz parameter, where the parameter is effectively an inclination angle of the spherical photon orbits through the equatorial plane. We also show how the invariant determines the constants of motion for all spherical orbits in the photon region. Finally, we briefly derive invariants which identify the other geometrically important surfaces such as the ergosurfaces and local horizons.

We develop a full-covariance formalism for joint pulsar timing array (PTA) and astrometric analyses of a nanohertz stochastic gravitational-wave background and apply it to forecasts of graviton-mass constraints. Analytic covariance expressions are derived for auto- and cross-channel overlap reduction function estimators, including signal-signal, noise-noise, and signal-noise contributions, and are validated against numerical simulations. In a current-like forecast with sensitivities comparable to NANOGrav 15 yr and Gaia, the constraint remains PTA-dominated, while a future-like forecast with sensitivities comparable to the SKA and Theia/Gaia-NIR improves the joint graviton-mass bound by about one order of magnitude. These results demonstrate that PTA-astrometry multichannel inference provides a viable avenue for placing stronger constraints on the graviton mass.

We have developed a non-unitary extension of Grover's search algorithm by changing the hidden geometry of Hilbert space carried by diffusion operator. Our algorithm finds the solution for search problem by performing a unique bigger rotation rather than small rotations in order polynomial times in the size $N$ of search space. We analyze the complexity of implementing the non-unitary operation and we observed that the price paid by performing this rotation is due the normalization. In Kraus operator approach we need $O(N)$ repetition of the algorithm to have a chance of measuring a solution in a post-selection, this is no better than the classical solution. However, the quantum singular value transform in addition with block encoding and Chebyshev polynomial approximation, we got complexity $O(\sqrt{N})$ and reach the Grover's bound with an extra resource of one single qubit, compared with the standard Grover's algorithm.

The quasinormal mode spectrum is a unique fingerprint linking gravitational-wave observations to extra-dimensional geometry. In this Letter, we show that thick braneworlds exhibit a spectral butterfly effect: infinitesimal deformations of the effective potential trigger dramatic migrations of quasinormal modes, challenging the presumed stability of this fingerprint. Frequency-domain instabilities depend sensitively on the perturbation's location and strength. In the time domain, near-brane perturbations primarily modify the early ringdown, while far-brane perturbations generate clean late-time echoes. Crucially, the graviton zero mode remains localized, preserving four-dimensional gravity. Despite this pronounced spectral fragility, the observable early-stage signal under current detector sensitivities is still dominated by the original fundamental mode. Hence, thick braneworlds display a nontrivial coexistence of a fragile spectrum and a resilient ringdown, supporting the continued use of the standard fingerprint in present-day gravitational-wave astronomy while revealing its hidden sensitivity.

We explore the viability of scalar field dark matter halo models through the rotation curve analysis of the Andromeda galaxy (M31), taking into account a realistic description of its baryonic structure. The mass model includes a stellar disk described by the Freeman profile and two alternative bulge configurations: a classical single de Vaucouleurs bulge and a two-component structure consisting of inner and main bulges modeled by exponential sphere profiles. The dark matter halo is modeled using three scalar field motivated models: fuzzy dark matter (FDM), Bose-Einstein condensate and multistate scalar-field dark matter. The model parameters are determined through the Levenberg-Marquardt nonlinear least-squares fitting, and the relative performance of the models is evaluated using the Bayesian Information Criterion which allows a direct comparison with previous phenomenological halo studies performed for the same galaxy. We find that the two-bulge baryonic configuration ensures a better statistical description of the M31 rotation curve, independently of the adopted halo model. The results also suggest that, within scalar field dark matter scenarios, smooth cored halos, such as FDM, provide the most consistent description of the M31 kinematics.

Quantum optimization algorithms promise advantages for difficult problems but are costly to simulate and analyze on classical machines. Recently, constrained quantum optimization has been investigated through the lens of Quantum Zeno dynamics, an approach which constrains the search to a subspace by means of quantum measurements. Exploiting that quantum measurements are projections, we propose a model reduction approach and show that simulations can be conducted in a lower-dimensional space. As possible applications, we demonstrate exponential state-space reduction of constrained quantum optimization in case of random 3-SAT and agent coordination problems over graphs.

It is well known that bosons and fermions exhibit opposite behaviors when experiencing interference, in the sense that bosons have a tendency to bunch whereas fermions have a tendency to antibunch. Recently, this complementarity was mathematically characterized in [arXiv:2312.17709] by means of an identity relating the transition probabilities of both types of particles in a linear interferometer. Here, we show that such a complementarity still holds even when particles become partially distinguishable, for example, when they have slightly different polarizations or time delays. Namely, we establish a relation that combines bosonic and fermionic multiparticle interferences in an arbitrary linear interferometer, in the presence of partial distinguishability. Incidentally, this also provides a new mathematical identity relating the permanent and determinant of tensors of order 3. Importantly, this complementarity has direct operational consequences in quantum metrology. Indeed, we show that the correlation matrices for bosonic and fermionic particle number distributions at the output of the interferometer obey a simple sum rule: their sum equals twice the correlation matrix for classical particles. This, in turn, constraints the achievable quantum Fisher information in phase-estimation protocols, highlighting a trade-off whereby greater indistinguishability enhances bosonic sensitivity whereas reduced indistinguishability can benefit fermionic schemes.

The irreversible emergence of classical behavior from a reduced quantum description via a canonical intrinsic decomposition of the density operator is analyzed. In the intrinsic reference basis (IRB), defined for a fixed physical conjugation K (determined by measurement convention, system symmetry, or secular approximation) by diagonalizing the real symmetric part of the state, the density operator separates into a diagonal population sector and a real antisymmetric coherence sector. For the class of Markovian open-system dynamics whose Lindblad operators are diagonal in the IRB, we prove that the quadratic coherence functional is a Lyapunov functional under pure-dephasing or interaction-picture evolution, with each intrinsic coherence component decaying exponentially at a computable rate. This yields a canonical state-dependent operational classicality criterion via the normalized cohesion index, an explicit logarithmic classicalization time controlled by the slowest dephasing rate, and a demonstration that the IRB projectors emerge as dynamically stable pointer sectors under IRB-selective evolution. Suppression of intrinsic coherences is exactly equivalent to suppression of fringe visibility in the corresponding interferometric sector; for a balanced two-path setup the cohesion index coincides with the fringe visibility, making the classicality criterion directly testable with standard interferometric equipment. The approach complements environment-induced einselection: it is applicable whenever a coarse-grained reduced description is available, independently of whether the microscopic system-environment coupling is known.

We develop a theory of quantum sufficiency on real *-subalgebras and real Jordan algebras. In contrast to the conventional formulation, which is based on families of states, complex completely positive coarse-grainings, and Radon-Nikodym cocycles associated with faithful reference states, our framework allows models consisting of general self-adjoint operators, including derivatives of states. Within this framework, square-root likelihood ratios and symmetric logarithmic derivatives arise naturally as fundamental self-adjoint likelihood-type objects. This makes it possible to treat ordinary quantum statistical models and local quantum statistical structures within a unified setting. We introduce sufficient real positive maps and show that sufficient complex *-subalgebras, sufficient real *-subalgebras, and sufficient real Jordan algebras correspond respectively to complex completely positive maps, real completely positive maps, and real positive maps. We characterize minimal sufficient real *-subalgebras by the likelihood-ratio set together with rho-modular invariance, and show that the real Jordan algebra generated by the likelihood-ratio set and the projected reference state is the minimal sufficient real Jordan algebra. We also obtain Koashi-Imoto type decompositions for sufficient real *-subalgebras and sufficient real Jordan algebras. Our formulation admits degenerate reference states and separates the likelihood-ratio aspect of sufficiency from its genuinely quantum modular aspect. These results suggest that real Jordan structure provides a natural framework for the statistical aspect of quantum theory beyond the conventional complex *-algebraic setting.

In this paper, numerical simulations of four-mode continuous-variable cluster states with different topologies in the framework of measurement-based quantum computation are presented. By utilizing the symplectic representation and covariance matrix, the process of generating cluster states with linear, square, and T-shaped topologies has been systematically modeled. The simulation results show that the cluster graph structure is directly reflected in the pattern of quadrature correlations; in other words, the theoretical nullifier relations of the cluster states are reproduced in the final covariance matrices. Increasing the squeezing parameter leads to the strengthening of the target correlations and the suppression of unwanted components arising from anti-squeezing; such that the off-diagonal elements of the covariance matrix in the linear and square topologies increase to significant values, and in the T-shaped topology a stronger central correlation (similar to GHZ-like behavior in the continuous-variable domain) is observed. In order to quantitatively analyze these structural differences, a metric titled CCR (Correlation Concentration Ratio) is introduced that quantifies the concentration of effective correlations on the graph edges relative to the total correlations of the system. This index enables direct comparison of different topologies from the perspective of structural entanglement distribution and provides a framework for evaluating the efficiency of cluster graphs in MBQC architectures. The results show that CCR can be used as a practical tool for designing and selecting optimal topologies in larger clusters and more complex structures.

Motivated by alleged inconsistencies in the scientific and educational literature, Asenbaum, Overstreet and Kasevic \href{https://doi.org/10.1088/1402-4896/ad340c}{(2024)} aim to clarify some fundamental concepts in the physics of gravitation. To this end they reexamine the first experimental measurement of the gravitational redshift by Pound and Rebka in 1960, claiming that it did not in fact measure the gravitational redshift predicted by Einstein almost half a century earlier, but rather a Doppler shift originating from non-gravitational reaction forces. We show that their conclusion arises from a misunderstanding of the reference systems involved, along with an unphysical interpretation of non gravitational forces. Thus, our work restores the Pound and Rebka experiment to its rightful place in Physics.

This paper introduces a conceptual framework of technology-dependent ternary quantum gates that could be implemented and fabricated into various superconducting and photonic quantum systems for operating 3-valued quantum bits (qutrits). The "technology-dependent" term means that such ternary quantum gates are on-purpose designed analogy to the contemporary binary quantum gates. Conceptually, the final built technology-dependent one-qutrit and two-qutrit ternary quantum gates are Chrestenson, Z3, 01, 02, 12, +1, +2, and their corresponding controlled gates, respectively.

We investigate the performance of Grover's algorithm on three different generations of IBM Heron QPUs. On Heron family of IBM QPUs the success probabilities for three, four and five qubits without dynamical decoupling is better than results reported for previous generations of QPUs. The success probability as function of number of iterations of Grover operator is considered. A study of the improvement of results of Grover's algorithm for five qubit case with the help of topological dynamical decoupling is considered. For a six qubit case on Heron r3 QPU a clear result for finding the sought-after bitstring is reported for theoretically suboptimal number of iterations of Grover operator with the help of dynamical decoupling.

Preheating after inflation proceeds through parametric resonance, leading to efficient particle production in scalar field models. In this work, we investigate the structure of parametric resonance in the $φ^4$ chaotic inflationary model during the preheating phase by performing a fully numerical analysis of the coupled dynamical equations governing the inflaton field and the mode function of the produced particles, thereby avoiding the approximations commonly employed in earlier studies. Our results reveal resonance patterns that differ significantly from those obtained with approximate analytical treatments. In the weak coupling regime, short-wavelength modes rapidly settle into oscillations with nearly constant amplitude, while the corresponding occupation numbers approach saturation. However, the long-wavelength modes exhibit gradual amplitude growth, with occupation numbers transitioning into a non-linear oscillatory regime. As the coupling strength increases, the dynamics becomes increasingly nonlinear, leading to the emergence of stochastic behavior. In the strong coupling regime, short-wavelength modes display a step-like (staircase) evolution in the occupation number, indicative of intermittent bursts of particle production. However, the long-wavelength modes exhibit a more gradual, monotonic growth with small superimposed fluctuations. These findings highlight the rich, coupling-dependent, structure of parametric resonance in the quartic inflationary model and underscore the importance of exact numerical treatment in accurately capturing preheating dynamics.

Magneto-optical traps are central to atomic and molecular quantum technologies and precision tests of fundamental physics, where both sensitivity and bandwidth scale strongly with atom number and loading rate. We demonstrate that employing multiple, closely spaced optical frequency components in the cooling light of a $^{87}$Rb magneto-optical trap -- without utilizing any additional slowing techniques -- can double the steady state atom number and increase the loading rate by up to a factor of 4, compared to a conventional single-frequency implementation. Subsequently, we capture up to $1.0(1)\times10^{10}$ atoms with a loading rate of up to $1.3(2)\times 10^{11}\,\mathrm{atoms\,s^{-1}}$ from a thermal background. Numerical simulations reproduce the observed trends and predict substantially larger gains for increased trap sizes beyond our experimental bounds. By re-examining earlier studies of multi-frequency atom capture in the context of modern experimental hardware and emerging applications, we show that previously identified limitations can be avoided and establish multi-frequency cooling as a practical and scalable route to high-flux cold-atom sources. These results have immediate applications in portable atom-based quantum sensing, where higher bandwidth and precision can be achieved without forgoing compactness, and in atom-interferometric tests of fundamental physics, which benefit from access to larger-mass quantum systems.

We investigate formulations of quantum field theories whose kinetic terms involve fractional or continuous powers of the d'Alembert operator. The primary requirements are perturbative unitarity and a well-defined classical limit with a finite number of initial conditions. A direct approach consists of continuing the correlation functions from Euclidean space to Minkowski spacetime using the fakeon prescription for the fractional part of the power. Alternative formulations arise through decomposition, in which the fractional part is represented as a continuum of ordinary fakeons. These options are infinite in number and yield inequivalent Minkowskian theories with the same Euclidean counterpart. We demonstrate these features at tree level and for bubble diagrams. We also point out potential pitfalls in the calculations. Finally, we show how to treat continuous powers of covariant d'Alembertians in fractional gauge and gravity theories. The Ward and Cutkosky identities hold in all formulations.

Spectral collapse in the two-photon quantum Rabi model (tpQRM) has long been regarded as incompatible with quantum criticality due to the absence of a vanishing excitation gap. We show that, in the anisotropic tpQRM, spectral collapse constitutes a genuine continuous quantum phase transition governed by a single soft mode. The excitation gap within the same parity closes as $ε_{sp} \sim |g - g_c|^{zν}$ with $zν= 1/2$, placing the system in the same universality class as the standard QRM, while the gap between different parities reflects symmetry-induced level splitting rather than a critical excitation. This soft mode defines a unique energy scale that controls both equilibrium and nonequilibrium properties, including macroscopic observables, quantum Fisher information, and Kibble-Zurek dynamics. These results establish spectral collapse as an experimentally accessible realization of quantum criticality in a few-body system and demonstrate that universality is fully determined by the soft-mode structure rather than by microscopic details.

This letter presents an assessment of quantum gravity effects on extreme-mass-ratio inspirals (EMRIs) for the rotating quantum Oppenheimer-Snyder (qOS) black hole. Employing the adiabatic evolution, we compute the gravitational wave (GW) dephasing, which quantifies the cumulative phase shift induced by the quantum correction α . We further generate the augmented analytic kludge (AAK) waveform and investigate the faithfulness between the waveforms with and without the quantum parameter α for different values of a. Our results reveal that the quantum gravity effect induces detectable imprints in LISA, while the presence of rotation suppresses these signatures. This suggests that rotational degrees of freedom must be carefully accounted for when probing quantum gravity with EMRI observations.

Any protocol to process quantum information has to conclude with a measurement, aimed at producing a specific set of probabilities of measurement outcomes. In this work, we investigate the time, energy and importantly the genuine quantum resources necessary for transforming a set of measurement probabilities generated by a positive-operator-valued measure (POVM), to a target set of measurement probabilities. To this end, we first show that the speed of measurement probabilities, defined as the average rate of the surprisal of measurement outcomes, is constrained by the genuine quantum fluctuations contained in the measurement probabilities. Interestingly, this quantum speed limit can act as a witness for bipartite quantum correlations by selecting an optimal local projective measurement. Furthermore, we obtain a minimum time to transform an initial measurement probabilities to a target measurement probabilities, and apply this result to analyzing the cost of generating a local athermality in terms of genuine quantum uncertainty.

Trotterization is a technique that allows one to approximate a time evolution of a Hamiltonian by repeatedly evolving the individual terms of the Hamiltonian one-at-a-time for small time durations. Bounds on the error of this approximation exist; however, they are typically loose and moreover, it is known that the true error can be greatly influenced by the order in which the terms of the Hamiltonian are evolved. In this work, we consider various ordering strategies that exploit the commutation structure of the Hamiltonian, in addition to a few other baseline ordering strategies. These commutation-based strategies involve dividing the terms of the Hamiltonian into groups where all the terms within each group commute with one another. These groupings can be obtained by using graph coloring techniques on what we call the "commutation graph" of the Hamiltonian. We prove various results regarding the structure and properties of such commutation graphs for certain classes of Hamiltonians. We also empirically calculate the (true) Trotter error using these ordering strategies on various 1D and 2D Heisenberg-style systems.

In this paper, we study the non-Gaussianity of the eigenstates of the Pegg-Barnett phase observable. By computing the Wigner functions of the eigenstates, we confirm that they take negative values in specific regions of the phase space. The Pegg-Barnett phase-operator eigenstates lie on a finite-dimensional Hilbert space. Thus, we examine how their negativity volumes depend on the dimension of the Hilbert space. Moreover, we present a quantum-optical circuit that generates these eigenstates and identify single-photon detection as the origin of their non-Gaussianity. To investigate a more realistic experimental implementation, we introduce imperfect single-photon detectors with non-unit efficiency into the circuit and evaluate the dependence of the detection probability, the output-ideal fidelity, and the negativity volume of the approximate eigenstate output from the circuit on the detector efficiency. Finally, as a practical application, we consider a phase-estimation experiment of an arbitrary unknown state by injecting both the unknown state and a known Pegg-Barnett eigenstate into a 50-50 beam splitter and individually counting the numbers of photons emitted from its two output ports.

The speed of quantum evolution is limited under finite energy resources. While most quantum speed limits (QSLs) are formulated in terms of quantum states, they can be extended to the evolution operator itself, and thus impose fundamental limits on how quickly logical gate operations can be implemented on a quantum computer. Here, we derive a general, tight QSL that holds for any unitary evolution under the constraint that the spectral width of the Hamiltonian is bounded. We apply this result to obtain QSLs for several standard quantum gates, including Hadamard, CNOT, and Toffoli gates, finding that the QSL can vary significantly across different gates, including ones with the same entangling power. These findings can be understood geometrically using the Space Curve Quantum Control formalism, which maps unitary evolution to space curves in Euclidean space. In this formalism, the problem of finding QSLs is recast as the problem of finding minimal-length curves obeying a curvature bound. We find that time-optimal gates map to helices of varying dimensions, and that QSLs can be understood from the perspective of a bottleneck principle in which the operator that evolves the slowest governs the minimal gate time.

Szegedy's quantum walk gives a generic quadratic speedup for reversible classical Markov chains, but extending this mechanism to quantum Gibbs sampling has remained challenging beyond special cases. We present a walk-free quantum algorithm for preparing purified Gibbs states with a quadratic improvement in spectral-gap dependence for a broad class of quantum Gibbs samplers that satisfy exact Kubo-Martin-Schwinger detailed balance. Our main structural result is an explicit factorization of the corresponding parent Hamiltonian into noncommutative first-order operators. This turns purified Gibbs-state preparation into a singular-value filtering problem and enables a quantum singular value transformation algorithm with quadratically improved gap dependence under standard coherent-access assumptions. The framework applies to several efficiently implementable Gibbs samplers beyond the Davies setting. We also introduce an auxiliary dissipative dynamics based on the same factorization, which can be used to generate warm starts in the doubled Hilbert space in metastable regimes.

We propose a protocol that transfers entanglement from an entangled atomic two-level-system (TLS) resource to a pair of free electrons in an energy-sideband ladder via local electron-TLS interactions. In a controlled rotating-wave regime, closed-form reduced states are derived. TLS heralding then prepares a maximally entangled electron state in a two-dimensional single-excitation manifold, with a simple dependence on the initial TLS resource entanglement. Numerical integration of the full bilinear Hamiltonian quantifies the impacts of detuning and pulse shaping and identifies the leading beyond-rotating-wave corrections. The results establish a heralded route to entangled free electrons and will facilitate further advances in quantum electron optics.

This paper examines interacting dark energy models within the Chevallier-Polarski-Linder (CPL) parametrization, emphasizing both theoretical structure and observational viability. Two commonly adopted interaction terms are considered: $Q = βH ρ_{de}$ and $Q = βH ρ_c$. We derive exact analytic solutions that describe how the dark sector evolves. These solutions involve incomplete gamma functions and reveal a non-trivial mathematical structure that is often missed in numerical analyses. We perform a Bayesian analysis using current cosmological observations, including the Hubble parameter (OHD), Type Ia supernovae (SNIa), baryon acoustic oscillations (BAO), and cosmic microwave background (CMB) data. Relative to the non-interacting CPL scenario, the interacting model with $Q = βH ρ_{de}$ yields a modestly improved fit, as indicated by the Akaike Information Criterion (AIC). However, the Bayesian Information Criterion (BIC) penalizes increased model complexity, leading to a continued preference for $Λ$CDM. In contrast, the interaction model that depends on dark matter density does not provide observational support. The preferred interacting scenario indicates that the dark energy equation of state evolves dynamically, transitioning from an effective phantom regime at high redshift to quintessence-like behavior at late times. Further analysis indicates the potential for a transient phase of cosmic acceleration in the future. These findings suggest that interacting dark energy models within the CPL framework enrich the standard cosmological model by introducing more diverse phenomenology while maintaining consistency with current observations.

It is known that the classical information like strings of bits can be copied. In 1982, Wootters and Zurek proposed the quantum no-cloning principle [1]. No-cloning principle says that it is impossible to make an identical copy of an arbitrary unknown quantum pure state by using unitary evolution. In this paper, we propose a general no-cloning principle. We call U|psi>, where U is any unitary operator, a U-copy of the state |psi>. The general no-cloning principle states that it is impossible to make a U-copy of an arbitrary unknown quantum pure state by using unitary evolution.

Strongly-correlated electrons in transition-metal oxides give rise to intriguing emergent phenomena, including high-temperature superconductivity in cuprates. While simplified one-band Hubbard models capture some aspects, explicitly describing the interplay of copper and oxygen orbitals -- as in the three-band Emery model -- is essential to capture the full phenomenology of cuprates. Quantum simulators based on ultracold atoms offer a promising route to study such systems in a controlled setting, but realizing realistic multi-orbital Hubbard models remains challenging. Here we propose an optical superlattice architecture that implements the three-band Emery model with ultracold fermions. By combining lattice beams with controllable interference, we engineer orbital degrees of freedom that reproduce key features of the cuprate band structure, while enabling independent control of orbital-dependent interactions and charge-transfer energy. We show that single-particle quantum walks can benchmark the resulting tight-binding model. Using determinant quantum Monte Carlo, we further investigate thermodynamic properties in the undoped regime and find a finite-temperature metal-insulator crossover accompanied by the onset of antiferromagnetic correlations accessible in current experiments. Finally, we apply a Hamiltonian learning protocol enabling to infer effective single-band Hubbard models from experimental realizations of Emery models. Our results provide a practical pathway to simulate multi-orbital Hubbard physics with quantum gas microscopes.

This paper presents solutions to Einstein's equation -- and the numerical methods used to construct them -- that describe simple cosmological models on manifolds with compact non-orientable spatial slices. These solutions have been constructed on a selection of manifolds having positive, negative, and vanishing spatial scalar curvatures. One example is shown to be indistinguishable locally from a homogeneous Friedman cosmological model, others are constructed with significant inhomogeneities. Together these examples are used to explore the strengths and the limitations of the numerical methods used in this study, and to test the code used to implement them.

We study the Dunkl anharmonic oscillator (Kerr medium) Hamiltonian from an algebraic approach of the $SU(1,1)$ group. In order to obtain the exact energy spectrum of this problem, we write its Hamiltonian in terms of the Dunkl creation and annihilation operators, which close the $su(1,1)$ Lie algebra. This allows us to exactly solve this Hamiltonian and obtain its parity-dependent energy spectrum. Then, we investigate the quantum dynamics of the system, particularly the collapse and revival phenomena, by using an initial state given by a superposition of even and odd Dunkl coherent states. We compute the field quadrature and the survival probability, showing that the Dunkl parameter $μ$ modulates the fractional revivals and produces perfect state reconstructions at half-periods for specific deformation values. We analyze the quadrature variance to show that the Dunkl deformation generates interference-induced squeezed states around $t \approx π$. The standard Kerr medium dynamics are exactly recovered in the limit $μ\rightarrow 0$.

Driven atomic Josephson junctions allow one to monitor phase-coherent dynamics with unprecedented control and flexibility of the system's physical conditions. While cold-atom manifestations of the Josephson effect have been extensively studied in a wide variety of settings, atomic Josephson junctions in synthetic electromagnetic fields remain largely unexplored. Here, we show that synthetic magnetic fields can induce Fraunhofer-like modulations of the critical current in atomic Josephson junctions. Although this effect presents analogies to the Fraunhofer patterns found in superconducting devices, distinctive features emerge due to the neutral nature of the superfluid. We investigate the underlying spatial interference mechanisms and elucidate the role of Josephson vortices in the formation of spatially modulated current distributions based on numerical simulations. Our results open up new avenues for matter-wave circuits to deepen our understanding of spatial coherence in Josephson junctions, which are fundamental to the development of novel quantum technologies.

We extend kinematic flow to momentum-integrated loop-level cosmological correlators, focusing on banana loops of conformally coupled scalars in power-law cosmologies and, in de Sitter, on arbitrary mixtures of massless and conformally coupled scalars. Exploiting their dual description as tree-level exchanges of unparticles, we show that the associated correlators are described by a finite set of master integrals obeying a first-order system of differential equations. The corresponding basis is constructed from tubings of marked graphs and is distinguished by the appearance of nested tubes and an arborescence ordering of the vertices. We derive the connection matrices from four combinatorial rules -- activation, merger, swap, and copy. The last two are unique to unparticle exchanges: they induce richer mixing among basis functions and introduce new kinematic letters. Our framework extends systematically to arbitrarily complicated configurations, including necklace diagrams, and establishes unparticle exchange as a distinct class of kinematic flow.

Symmetry-resolved spectroscopies, such as angle-resolved photoemission spectroscopy and polarization-resolved Raman, are central for quantum material characterization, yet remain challenging at nanoscale dimensions and low frequencies. Here, we propose correlated quantum dephasometry, which enables symmetry resolved quantum noise spectroscopy of materials at nanoscale and low ($\sim$MHz) frequencies via correlated dephasing of two spin qubits near materials. Our approach leverages the finite-range spatial structures of nonlocal near-field noise correlations to isolate rotational symmetry of the material response in momentum space beyond single qubit capabilities. We apply our approach to two-dimensional (2D) superconductors, and predict clear fingerprints that discriminate s-, d-, and g-wave symmetry of the superconducting gap. To highlight the generality, we further show that the same framework resolves 2D s-wave antiferromagnets and d-wave altermagnets. Our results establish correlated quantum dephasometry as a nanoscale, low-frequency complement for symmetry-resolved spectroscopy applicable to a broad class of quantum materials.

We study free Carrollian quantum field theories from an algebraic perspective and explore their implications for flat space holography. As explicit examples, we construct the electric and magnetic Carrollian Weyl algebras obtained from Carroll limits of the relativistic scalar field and analyze their states, including vacuum and thermal configurations. For the massive electric theory, we find a regular Carroll-invariant vacuum state and a regular KMS state, yielding a consistent Carrollian thermodynamic system. By contrast, the massless electric and magnetic theories are more subtle: depending on the quantization, they admit either no regular distinguished vacuum or only nonregular Carroll-invariant ground states, while still supporting nonregular thermal states. We further analyze alternative classes of states in the massless electric theory, including spatially homogeneous quasifree pure states and Sorkin--Johnston states.Motivated by these results, we discuss consequences for flat space holography. We construct a well-defined quasifree state relevant for Carrollian holography whose Hilbert-space representation factorizes into a standard Fock sector and a nonseparable zero-mode sector, thereby highlighting the role of infrared degrees of freedom in the boundary theory.

Several theoretical and astrophysical problems - including gravitational-wave modeling for extreme mass-ratio inspirals - require accurate time-domain solutions of the spin-weight $s=-2$ Teukolsky equation in Boyer-Lindquist coordinates. Because such simulations are performed on finite computational domains, they typically introduce an artificial outer boundary where nontrivial boundary conditions must be imposed. If these conditions are inaccurate, then spurious reflections and slowly-growing unphysical modes may corrupt long-time evolutions. We develop and implement exact radiation outer boundary conditions for the Bardeen-Press equation (a harmonic moment of the $a=0$ Teukolsky equation), making the artificial boundary transparent at any finite radius. We also construct near-to-far field teleportation kernels that map field data recorded at finite radius $r_1$ to the data reaching $r_2 > r_1$. The possible choice $r_2 = \infty$ corresponds to asymptotic waveform evaluation, that is propagation of the data to future null infinity. We show that both boundary and teleportation kernels are well approximated by exponential sums, with associated error bounds. Implemented in a time-domain solver, our kernel-based boundary conditions eliminate unphysical late-time growth and give the correct late-time decay rates, affording efficient long-duration simulations for waveform modeling and related blackhole perturbation calculations.

We investigate quasinormal modes, shadow observables, weak gravitational lensing, and neutrino--antineutrino annihilation for a static, spherically symmetric black hole that carries a nonlinear-electrodynamics magnetic charge and is embedded in a Hernquist dark-matter halo. The geometry is controlled by the black-hole mass $M$, magnetic charge $g$, and halo parameters $(α,β)$, and provides a simple analytic setting in which compact-object and environmental deformations can be studied simultaneously. We derive the scalar, electromagnetic, and axial gravitational master equations and compute the corresponding quasinormal spectra using a high-order WKB expansion supplemented by Pade resummation. The magnetic charge raises the real oscillation frequency and slightly increases the damping rate, whereas the Hernquist halo shifts the spectrum in the opposite direction; for suitable parameters the two effects partially cancel at the level of individual modes. We then connect the eikonal spectrum with the photon sphere and shadow radius, emphasizing the distinction between comparisons performed at fixed bare mass and at fixed asymptotic mass $\mathcal{M}=M+α$. At fixed asymptotic mass, the residual NED and halo-concentration terms reduce the shadow and the weak-deflection angle relative to Schwarzschild at the first nontrivial order. Finally, we formulate neutrino-pair annihilation in the same background, including the angular factor, Tolman-redshifted $T^9$ kernel, integrated deposition rate, and reduced shell profile. The magnetic sector suppresses the annihilation efficiency, while the halo sector enhances it through its lowering of the lapse. These results show that ringdown, imaging, lensing, and high-energy deposition probe the same underlying competition between near-horizon magnetic structure and extended dark-matter environment, but with different parameter weights.

Near-term hardware is constrained by high error rates, small qubit counts, and relatively low output fidelity, making the execution of large, high performance quantum circuits difficult. Circuit partitioning (or circuit cutting) has emerged as a promising approach to circumvent these limitations by decomposing circuits into smaller subcircuits at two-qubit interaction points. However, it remains unclear which classes of circuits benefit the most from partitioning and under what hardware conditions it is most effective. In this work, we evaluate the suitability of quantum circuits for partitioning from three perspectives: improving fidelity, enabling distributed execution, and scaling to larger circuit sizes. Specifically, we compare uncut circuit execution against two circuit partitioning approaches: Qiskit's automatic cut finding technique and a custom performance optimized circuit cutting method. We also measure these across GHZ, QFT, brickwork, and random quantum circuits ranging from 4 to 14 qubits, using mean absolute error of expectation values and overall output fidelity. Our results show that partitioning benefits larger, highly interconnected circuits, with our custom method reducing error by up to 55\% and improve fidelity for GHZ circuits, but degrading performance for brickwork circuits at larger scales.

In certain modified theories of gravity, non-minimal couplings between matter and geometry lead to the nonconservation of the energy-momentum tensor. By interpreting this as an effective dissipative process, we formulate a general class of f(R, Matter) theories with the Herglotz variational principle, a variational approach designed for dissipative systems. We demonstrate that, for an appropriate choice of the Herglotz contribution, the resulting Herglotz extension of f(R, Matter) gravity restores the covariant conservation of the energy-momentum tensor.

We develop analytical and algorithmic techniques that enable efficient simulation of a broad class of noisy stabilizer circuits. We derive closed-form expressions of expectation values for tensor product of Paulis in circuits with non-deterministic Pauli measurements, yielding an efficient strong simulation method that avoids explicit density matrix construction and enables direct noise parameter sweeps. We introduce a circuit compression framework that reduces the per-sample cost of weak simulation in general noisy stabilizer circuits, including deterministic measurements, by separating parameter-independent preprocessing from sampling. Finally, we extend the analytical framework beyond its standard domain to include a small number of deterministic measurements, general rotations, and non-diagonal noise channels. Our results provide a unified framework for both strong and weak simulation of noisy stabilizer circuits and corresponds to an extension of the noisy stabilizer formalism introduced in \cite{PhysRevA.107.032424}. They offer applications ranging from calculation of the expectation values of entanglement witnesses, determination of reduced states, to energy evaluation.

We study truncations of hierarchical equations of motion (HEOM) for finite-dimensional open quantum systems. We prove that for finite-dimensional approximations constructed with a Schur-complement type of terminator, the spectrum converges to that of the full HEOM as the truncation depth increases. We also prove that this approximation is free of spectral pollution: sufficiently deep truncations do not produce spurious unstable modes, provided the exact HEOM is stable. We illustrate the results for the spin-boson model.

We extend the CDPR lattice reduction algorithm from ideal to module lattices, leveraging the trace orthogonality of the power basis to decompose the module into rank-1 submodules and applying CDPR independently to each. This base module reduction achieves a Hermite factor $\exp(\tilde{O}(\sqrt{n}))$ matching the ideal case, with a module reduction factor $O(1)$ independent of the rank, under a balance hypothesis automatically satisfied for MLWE-distributed bases. To control precision, we introduce CRT-scaled rounding at totally split primes, reducing the Gram-Schmidt rounding error and yielding a bounded-precision implementation. We further reformulate the CDPR sign-selection subproblem as a mixed-integer linear program, determining the optimal balanced discrepancy to be a universal constant $δ^*\approx 0.4407$. All results build on the class number one condition $h_k^+=1$ established in Part I of this series.

We study the optimal transport problem on globally hyperbolic spacetimes associated with Orlicz-type Lorentzian cost functions of the form $u \circ \ell$, where $u$ is a suitable monotonically increasing and concave function, and $\ell$ is the time separation. Our work encompasses and generalises the case $u(x) = u_p(x) = p^{-1}x^p$ for $p \in (0,1)$, as well as the more recent $p < 0$, which have been the only examples considered so far in the literature. A fundamental notion for our purposes is the property of $u$-separation for a pair of measures, which generalises McCann's $p$-separation and for which we are able to obtain strong duality to the full Orlicz-type optimization problem. In our main results, we characterise timelike Ricci curvature lower bounds via the convexity of the relative entropy along geodesics arising from the Orlicz-type optimal transport with cost $u \circ \ell$, which is a far-reaching generalization of McCann's seminal work in the case $u = u_p$, $p \in (0,1)$.

This work describes a few-body dynamics method based on the Faddeev integral equations in momentum space for determining the total cross sections of fusion and breakup reactions with two- and three-body final channels in the continuum, employing a cluster representation of the colliding nuclei. Total cross sections were obtained for the reactions $^3\text{He}(T,D)^4\text{He}$, $^3\text{He}(T,np)^4\text{He}$, $^3\text{He}(T,nD)^3\text{He}$, $^3\text{He}(^3\text{He},2p)^4\text{He}$, $^3\text{He}(^3\text{He},pD)^3\text{He}$, $^7\text{Li}(^3\text{He},\phantom{0}^4\text{He})^6\text{Li}$, $^7\text{Li}(^3\text{He},D^4\text{He})^4\text{He}$, and $^7\text{Li}(^3\text{He},T^3\text{He})^4\text{He}$, in which both the projectile and the target nucleus were treated in a cluster representation. The work also implements a two-potential method to determine the Coulomb $t-$matrix and to account for Coulomb effects in short-range dynamics in momentum space. Calculations of the initial-state Coulomb interaction were performed; furthermore, an estimate was obtained for the magnitude of the off-shell effect of the Coulomb $t-$matrix, as well as the magnitude of the atomic electron anti-screening effect on the Coulomb interaction of the colliding nuclei. The calculated contribution of the cluster mechanism to the total cross section of the considered fusion reactions in the kinetic energy range $T\in[1~\text{keV},20~\text{MeV}]$ is in good agreement with known experimental data.

Long-range interactions are the source of many equilibrium and out-of-equilibrium quantum many-body phenomena. Analog simulators based on ionic, atomic, superconducting, and molecular systems provide a natural platform to obtain these interactions using vibration- and photon-mediated processes. Recent experimental advances, such as their integration in multi-mode cavities and waveguides, or the use of Raman-assisted transitions, enable dynamical control over both the strength and the spatial range of these interactions, thereby rendering them programmable. Here, we develop a hybrid classical-quantum toolbox that exploits this tunability to enhance many-body state preparation in analog simulators beyond fixed-connectivity architectures. Our approach is based on classical pre-compilation in homogeneous small systems, whose optimized parameters are extrapolated iteratively to larger system sizes, and then refined on the quantum hardware using noise-aware hybrid re-optimization and error-mitigation techniques. We benchmark this strategy across several fermionic, spin-1/2, and spin-1 models, demonstrating orders-of-magnitude improvements in fidelity and energy estimation for system sizes ranging from 100 to 1000 particles. Finally, we show that the combination of such high-fidelity programmable state preparation techniques with tunable-range out-of-equilibrium dynamics enables controlled studies of many-body thermalization in regimes accessible to current experimental platforms. Our results establish programmable long-range interactions as a powerful resource for next-generation analog quantum simulators.

We present the standard double slit experiment with non-relativistic particles in the Heisenberg Picture of quantum mechanics. Our motivation is threefold. First and foremost, and contrary to some claims in the literature, we show that there is no need to talk about non-locality when explaining the interference fringes. Secondly, we emphasise the fact that even in the non-relativistic regime, and in order to preserve locality, we should define the position and momentum observables of a particle as functions of both space and time (and not just time). Thirdly, our presentation compares the projective measurements in the Heisenberg picture with the "Church of the Larger Hilbert Space", the latter of which is seldom discussed in the Heisenberg picture of quantum mechanics.

Future quantum architectures are expected to be modular, with quantum processors connecting multiple quantum processing units (QPUs) via photonic interconnects. In topological quantum error correction, such as color codes, this creates seam boundaries where parity checks require remote CNOT operations using heralded Bell pairs. These non-local checks are slower and noisier than bulk local checks because entanglement generation is probabilistic, causing data qubits to accumulate idle noise while waiting for remote operations. A natural way to reduce this overhead is to skip some seam measurements; however, doing so makes seam syndrome information stale and can degrade decoding. The central scheduling problem is therefore to determine how frequently seam checks should be measured so as to balance remote-operation and waiting noise against syndrome staleness. To address this trade-off, we develop a scheduling module that integrates directly into standard syndrome-extraction circuits. We consider two policies: Skip-Seam-$τ$ (SS-$τ$), which measures all bulk checks every round while measuring seam checks once every $τ$ rounds and copying the most recent syndrome in skipped rounds, and Adaptive Skip-$τ$ (AST), which selects $τ$ as a function of code distance and entanglement generation rate (EGR). We evaluate these policies on triangular color codes under circuit-level noise in Stim, including idling errors induced by Bell-pair generation delays. Our simulations show that SS-tau and AST reduce remote-operation overhead and can lower the logical error rate (LER) relative to the Measure-All (MA) baseline. For physical error rate $p = 10^{-3}$, we identify an EGR regime in which both SS-$τ$ and AST exhibit behavior consistent with fault-tolerant scaling, with LER decreasing as code distance increases. Across these regimes, SS-$τ$ and AST outperform MA.

Quantum computing may speed up numerical problems involving large matrices that are demanding for classical computers, and active research on this possibility is ongoing. In this work, we propose quantum algorithms for the exact simulation of a normalised correlated Gaussian random vector $|x\rangle=\vec{x}/\lVert\vec{x}\rVert$, $\vec{x}\sim\mathcal{N}(0,Σ)$, and its exponentiation $|e^{\vec{x}} \rangle= e^{\vec{x}}/\lVert e^{\vec{x}}\rVert$. When an $O(\mathrm{polylog} N)$-gate-depth quantum data loader for the covariance matrix $Σ\in\mathbb{R}^{N\times N}$ is available, preparing $|x\rangle$ and $|e^{\vec{x}}\rangle$ require $\widetilde{O}\left(\frac{\lVertΣ\rVert_F}{λ_{\max}}κ^{1.5}\right)$ and $\widetilde{O}\left(\lVert\vec{x}\rVert\frac{\lVertΣ\rVert_F}{λ_{\max}}κ^{1.5}\right)$ elementary gate depth respectively, where $\lVertΣ\rVert_F$, $λ_{\max}$, $κ$ denote the Frobenius norm, maximal eigenvalue, and condition number of $Σ$. Motivated by financial applications, we provide an end-to-end resource analysis when $\vec{x}$ represents a sample path of a Riemann-Liouville or standard fractional Brownian motion, or of a stationary fractional Ornstein-Uhlenbeck process. As a concrete example, we construct the quantum state encoding the rough Bergomi variance process and analyse the extraction of the integrated variance via quantum amplitude estimation. Under specific conditions, the dependence of $\lVertΣ\rVert_F/λ_{\max}$ and $κ$ on $N$ is small, and subcubic complexity in $N$ is achieved, indicating a quantum advantage over classical Cholesky-based sampling methods. To our knowledge, this constitutes the first quantum algorithmic framework for the amplitude encoding of exponentiated Gaussian processes, providing foundational primitives for quantum-enhanced financial modelling.

We explore a cosmological model based on Rényi entropic corrections to the Friedmann equations and constrain it using DESI, P-BAO, CC, and gravitational-wave observations. Unlike earlier works, we obtain a direct and stringent constraint on the Rényi parameter $λ$ from late-time cosmic acceleration, with the resulting value even satisfying recent Big Bang Nucleosynthesis and baryogenesis bounds. The model predicts values of $H_0$ and $Ω_{m0}$ that are consistent with current observational data. It provides a successful description of late-time acceleration with a quintessence-like behavior, smoothly approaching the $Λ$CDM limit without crossing the phantom divide. Furthermore, the statical comparisons along with the evaluation of energy conditions and stability analyses reinforce its viability as a robust alternative to the standard cosmological model.

Black holes with resonant hair are static, spherical, electrically charged solutions of the Einstein-Maxwell-(gauged-)scalar system. Scalar self-interactions are mandatory for their existence. Initial dynamical studies restricted to spherical symmetry suggested stability; more recently, fully non-spherical dynamical studies revealed instabilities, at least for a particular class of self-interactions. Here, we provide a more detailed study of this instability together with a different decay channel, depending on the chosen solutions. Moreover, considering a second model, we provide evidence that the instabilities may be generic for different classes of self-interactions. We conclude these solutions are dynamically unstable and split into a bosonic lump and a bald black hole (via fission) or implode to the latter (via absorption). In both cases, the non-spherical dynamics seems to be key.

The Euclidean Kerr metric is conformal, in two distinct ways, to a Kahler metric, with conformal factors determined by the repeated eigenvalue of the two chiral halves of the Weyl curvature. A Lorentzian analogue holds, where the conformally related metric is complex but retains key features of Kahler geometry. We show that this hidden Kahler structure provides a geometric explanation for the existence of decoupled equations for curvature scalars, such as the Teukolsky equations. The essential mechanism is that, on a Kahler background, self-dual 2-forms are parallel with respect to a natural covariant derivative, so differential operators acting on them preserve their decomposition and do not mix components. In this way, decoupling is seen to be a direct consequence of Kahler geometry. We make this mechanism explicit in two ways. First, we show that the spin-k Teukolsky operator can be obtained from a Laplace-type operator associated with the Kahler metric by a similarity transformation. Second, for electromagnetic perturbations, we use the conformal invariance of Maxwell's equations delta F = 0 to show that they imply d delta F = 0, where delta is the co-differential of the Kahler metric. This operator automatically decouples, and the resulting equations for the extremal components coincide with the spin-one Teukolsky equations.

We develop a consistent mathematical framework for studying two flavor neutrino oscillations in presence of non-Hermitian dynamics. We consider two approaches : (a) bi-orthonormal inner product defined by a positive-definite metric operator $\mathcal{G}$ and (b) the density matrix prescription by Brody and Graefe [Phys. Rev. Lett. 109, 230405 (2012)]. For the $\mathcal{PT}$-symmetric case, we show that the $\mathcal{G}$ metric approach does not work well (probabilities are not conserved) both in $\mathcal{PT}$-unbroken as well as $\mathcal{PT}$-broken regime. Hence, we adopt the density matrix prescription by Brody and Graefe which is a positive semi-definite map. In the density matrix prescription, we note that probability in the steady state limit is not necessarily $1/2$ thereby indicating non-Markovian behavior.

We consider the ghost-free dRGT massive gravity with two of its three possible mass terms. This theory has five gravitational degrees of freedom. On Minkowski spacetime these modes have helicity-2, -1 and -0 and propagate on the Minkowski lightcone in the high-frequency limit. However for a general background the degrees of freedom corresponding to the helicity-1 and -0 modes have characteristics different to that of the metric lightcone. Here we prove in all generality, that the two degrees of freedom corresponding to the helicity-2 mode always propagate on the metric lightcone for any background in the high-frequency limit, which has significant relevance for current and future observational tests of the theory.

Entanglement has been proposed as a means to improve the sensitivity of sensing weak signals. While the degree of this quantum advantage is well understood in noiseless settings, the situation is more complex under realistic conditions, where the system is subject to decoherence. In this case, the enhancement depends on the specific noise characteristics. Previous treatments of colored noise typically assume that the noise is uncorrelated between successive experiments. Here, we consider the scenario in which the noise exhibits correlations across multiple shots. We derive a simple fundamental limit to the sensitivity based on the fact that the sensitivity cannot be better than the signal-to-noise ratio seen by the probe. Focusing on Ramsey spectroscopy with probes affected by pure classical dephasing, we show that, for suitable spatial and temporal noise correlations, entangled probes achieve a better scaling of the sensitivity with the number of probes than separable states. This demonstrates that entanglement can provide a substantial improvement for Ramsey spectroscopy subject to correlated noise.

We propose a quantum-enhanced sensing scheme for the detection of wave-like dark matter and high-frequency gravitational waves using two-dimensional ion crystals in a Penning trap. The protocol employs spin-motion squeezed states to improve the signal-to-noise ratio and enable a super-Heisenberg scaling with respect to the number of ions over a broad parameter range. We analyze the sensitivity of the protocol to representative wave-like dark matter candidates, including the axion-like particle and the dark photon, as well as to high-frequency gravitational waves, taking into account the decoherence and dephasing of the ion spins. Our results indicate that two-dimensional ion crystals and this new protocol provide a promising platform for probing previously unexplored parameter space in searches for light dark matter and high-frequency gravitational waves.

We investigate the spacetime geometry of a black hole solution coupled to a nonlinear Born-Infeld-type electrodynamics of the Kruglov form, taking into account the effective geometry governing photon propagation. Our analysis focuses on the role of parameter $q$, which controls deviations from Maxwell electrodynamics. We study observational signatures including light deflection, the black holes shadow, and images of a thin accretion disk. We find that the effective photon geometry induces significant modifications to null trajectories, leading to systematic $q$-dependent shifts in the deflection angle, shadow radius, and lensing observables. In particular, smaller positive values of $q$ enhance light bending, while negative values produce the opposite trend. Moreover, the efective geometry admits stable photon orbits outside the event horizon in certain regions of parameter space, together with nonstandard (reversal direction of) angular behavior of photon trajectories. These features leave distinct imprints on accretion disk images, especially in the structure and visibility of photon rings.

Entanglement is a key resource for many quantum applications. Understanding fundamental properties of entangled states is an important step towards their practical exploitation. We characterize entanglement in the context of Gaussian and non-Gaussian processes and identify entangled states that cannot be produced by any Gaussian evolution acting on separable states. This distinction exposes intrinsic limitations of Gaussian operations and highlights the role of non-Gaussian operations as an important resource. We apply this framework to develop a certification method that connects entanglement theory with quantum non-Gaussianity - an advanced quantum feature that is essential for several quantum applications. Our approach tailors the certification to experimentally relevant states that can be produced in current quantum optics experiments. We demonstrate this certification for recognizing the genuine non-Gaussian entanglement of various entangled Fock states and hybrid entangled states.

Broad, skewed iron K$α$ emission lines in the X-ray spectra of accreting black holes encode key information about the spacetime geometry of the innermost disk. While the Kerr metric is standard for spin measurements, horizonless alternatives like traversable "Kerr-like" wormholes can mimic many black hole signatures, challenging current data interpretations. We develop a relativistic reflection framework incorporating Kerr-like wormhole geometries to predict iron line distortions and assess the feasibility of distinguishing event horizons from wormhole throats.Using a custom ray-tracing subroutine, we implement two \textsc{XSPEC} modules: \texttt{kwline} for $δ$-function profiles and \texttt{kwconv} for full reflection spectra, parameterized by spin, throat radius, and shape-function coefficients. We compute a dense grid of line profiles and generate synthetic \textit{NuSTAR} spectra with realistic response matrices. By fitting these simulations with canonical Kerr models, we quantify deviations attributable to wormhole geometries.We find that Kerr-like wormholes produce narrower Fe K$α$ lines with suppressed red wings as the throat parameter $λ$ increases. In 50 ks \textit{NuSTAR} simulations ($λ=0.9, a_*=0.998$), simple convolutional models (\texttt{kerrconv}) can mimic the wormhole spectrum. However, self-consistent models like \texttt{relxillCp} result in statistical failure, yielding structured residuals and unphysical parameter pegging (e.g., emissivity $q_{\rm in} \to 10$). We conclude that large-throat wormholes are detectable in high-quality X-ray spectra if analyzed with fully consistent reflection models rather than post-processing approximations.

We study the integrability conditions of the conformal Killing equations for the Eisenhart lift of a scalar field in a flat Friedmann-Lema\^ıtre-Robertson-Walker universe. We show that the potential found in our earlier work is already the most general local potential that admits a non-trivial conformal Killing vector in the sector independent of the cyclic Eisenhart coordinate. The determinant condition of the prolonged conformal Killing equations reduces to a nonlinear second-order differential equation for $h=V'/V$. We solve this equation locally and find two branches. The regular branch reproduces exactly the family of potentials obtained before, while the singular branch lies on the locus where the determinant equation cannot be written locally in normal form with respect to $h''$ and is incompatible with the full conformal Killing equations. Hence the ansatz used in our earlier work is exhaustive.

Device-Independent Quantum Key Distribution (DIQKD) provides information-theoretic security by relying solely on the violation of Bell inequalities, eliminating the need to trust the quantum devices. However, practical implementations of DIQKD are highly sensitive to noise. Efficient error correction during the classical post-processing stage is important for improving the fidelity. This work investigates the impact of noise on the Clauser-Horne-Shimony-Holt (CHSH) value and evaluates the effectiveness of Cascade error correction. The protocol is applied iteratively to correct errors via parity checking and binary search procedures. Simulation results show that noise significantly degrades the CHSH value, reducing the strength of nonlocal correlations required for secure DIQKD. Nevertheless, Cascade reduces the error ratio, and most corrections occur within the first several rounds. These findings highlight the importance of classical error correction in improving DIQKD systems.

Hybrid quantum optimization for vehicle routing faces a practical bottleneck: direct QUBO encodings of CVRP quickly exceed near-term qubit and gate budgets, while quantum evaluations are expensive, noise-limited, and sensitive to backend and circuit configuration. We address this gap with an end-to-end decomposition pipeline that converts CVRP into bounded-width quantum subproblems and treats quantum execution as a decision problem within the optimization loop. Starting from a Fisher--Jaikumar assignment linearization, we apply Lagrangian relaxation to dualize customer-assignment couplers, yielding independent per-vehicle knapsack subproblems that admit QUBO/Ising evaluation. To replace brittle subgradient tuning, we learn a multiplier-update controller using expert-guided pretraining followed by reinforcement-learning fine-tuning, with rewards based on execution-realized progress and route reconstruction. We also introduce a constrained contextual bandit as a hardware-aware execution layer that selects backend and circuit configuration with feasibility screening, enabling adaptation across heterogeneous noisy resources and parallel multi-QPU scheduling. Computational results on multiple CVRPLIB families show that the decomposition yields stable bounded-width subproblems across instance sizes, learned multiplier updates improve end-to-end routing quality relative to classical subgradient control under matched budgets, and hardware-mode configuration reduces median optimality gaps relative to static execution choices in our test set. We do not claim quantum advantage. Instead, the contribution is a practical end-to-end framework for scaling hybrid quantum CVRP optimization through OR decomposition, learning-augmented dual control, and adaptive hardware-aware execution.

Optimal quantum linear equation solvers provide complexity $O(κ\log(1/ε))$, where $κ$ is the condition number and $ε$ is the allowable error. The optimal solver using a discrete adiabatic approach [PRX Quantum 3, 040303 (2022)] has large analytically proven constant factors for the upper bound on the complexity. The constant factors were later found to be about 1,200 times smaller in numerical testing [Quantum 9, 1887 (2025)]. This meant it is about an order of magnitude more efficient than using a randomised approach from [PRX Quantum 6, 040373 (2025)], which has far smaller analytically proven constant factors. Recently, a ``Shortcut'' method has been found to provide an optimal solver which also has small proven constant factors. In the present work, we conduct a comprehensive numerical analysis comparing this method with the adiabatic solver for two families of random linear systems. We find that, in the case where the solution norm is unknown, the adiabatic solver provides slightly better performance. If the solution norm is known, then the shortcut method provides significantly better performance for non-Hermitian matrices.

We establish a thermodynamic-optical duality that directly bridges the semiclassical quantum evaporation of black holes with their classical macroscopic geometry. By employing a diffeomorphic inversion, we re-parameterize the intrinsic black hole mass entirely in terms of the observable shadow radius $R_{sh}$. This mapping allows the formulation of the classical weak deflection angle, Hawking temperature, and integrated semiclassical luminosity, bypassing the unobservable bare mass. Applying this methodology to the standard Kerr, Kerr-MOG, and rotating Horndeski spacetimes, we reveal distinct, model-specific phenomenological signatures. For a statistically fixed shadow radius constrained by Event Horizon Telescope (EHT) observations of M87*, the standard Kerr geometry yields a baseline luminosity scaling of $L \propto R_{sh}^{-2}$. In modified gravity regimes, the duality breaks the degeneracy between bare mass and modified field strengths: the MOG repulsive vector field enhances classical deflection while strictly suppressing quantum luminosity, whereas Horndeski scalar hair introduces a unique logarithmic augmentation to astrometric lensing and drives up to a $\sim 52\%$ deviation in Hawking emission under current EHT limits. By strictly anchoring theoretical observables to empirical interferometric boundaries, this framework provides a computationally efficient avenue for testing the Kerr hypothesis and probing fundamental fields in strong-field gravity.

While the spatial arrangement of individual units is essential for the physical implementation of quantum batteries, geometry-dependent interactions are rarely explicitly incorporated into existing theoretical models. To address this, we propose a planar many-body quantum-battery architecture consisting of coupled resonators. By introducing a distance-dependent function to modulate both the inter-battery coupling and tunneling, we investigate the open-system charging dynamics in the strong-coupling regime using a Redfield master-equation approach. Using ergotropy as the primary figure of merit, we demonstrate that the charging performance is highly sensitive to the inter-battery distance, nearest-neighbor coupling strength, and environmental conditions. Specifically, decreasing the inter-battery distance within an optimal window suppresses charging fluctuations and accelerates the system's approach to a steady charged state. However, an excessively short distance amplifies environmental dissipation, thereby degrading the overall performance. Furthermore, while overly strong inter-battery coupling induces post-charging instability, moderate coupling achieves a favorable balance between maximum stored energy and stability. We also establish that the system-bath coupling and bath cutoff frequency predominantly govern the charging timescale, and that the planar architecture maintains its robustness against thermal fluctuations over a broad temperature range. These results highlight the critical role of geometry-controlled interactions in many-body quantum batteries, providing a theoretical foundation for the design and optimization of two-dimensional quantum energy-storage devices.

Surface oxides contribute to losses in superconducting transmon devices resulting in degraded performance. We explore the use of the damascene process to replace the sidewall native oxide of a device with a metal/substrate interface. We simulate sidewall oxidation by burying an oxide layer during fabrication. We observe a modest improvement between the two types of devices, which is suggestive of a reduction in the surface participation ratio.

We develop a minisuperspace formulation of the classical double copy for anisotropic Lifshitz spacetimes in arbitrary dimension. By imposing static symmetries at the level of the action, the gravitational system reduces to an effective one-dimensional radial problem with a universal structure, in which all theory dependence is captured by a single potential. Within this framework, we identify a radial operator that reproduces the Maxwell operator for the temporal component of the single-copy field directly from the reduced gravitational dynamics, without using the equations of motion. For non-relativistic Lifshitz backgrounds, this relation is modified by an additional contribution that encodes the deviation from maximal symmetry. We show that this term has a universal origin, determined by anisotropic scaling and horizon geometry, and that it vanishes smoothly in the relativistic limit. After imposing the Hamiltonian constraint, the matter sector generates the corresponding source term, reproducing known single-copy charge densities when a Kerr--Schild description exists and extending them beyond this setting. We further demonstrate that the same mechanism persists in higher-curvature theories, where the effective potential is replaced by its higher-order generalization while preserving the operator structure. Explicit Lifshitz black hole solutions illustrate how matter content and horizon topology enter the construction. As an additional check, we verify the consistency of the formalism in the relativistic limit using a charged AdS solution in Einstein--Gauss--Bonnet gravity.

Reducing orbital eccentricity in numerical relativity simulations of binary black holes is essential for producing astrophysically relevant gravitational wave models, as many of these systems are expected to be near-circular in nature. Standard eccentricity reduction procedures rely on iterative schemes, often requiring four or more trial simulations to achieve desired thresholds. This approach is computationally expensive because each trial simulation adds ~10% to the total simulation run time of multiple weeks to months. We introduce a data-driven approach that accelerates this process by learning the values of the initial orbital frequency, Omega_0, and radial velocity, adot_0, that yield an evolution with small eccentricity. This is done using a Gaussian Process Regression model trained on an archive of previously eccentricity-reduced numerical relativity simulations. For all configurations tested, using the trained model consistently reduces the number of required eccentricity reduction iterations to just zero or one, significantly lowering computational costs relative to post-Newtonian initial guesses. These results demonstrate the power of data-driven methods in accelerating expensive numerical relativity simulations.

We explore the performance of the time-convolutionless (TCL) projection operator technique using the Fano-Anderson model as a test case. Comparing the exact TCL master equation with an expansion in powers of the strength of the system-environment coupling, we analyze the transient dynamics as well as the steady-state behavior. For a Lorentzian spectral density we demonstrate that the dimensionless expansion parameter corresponds to the ratio of the environmental correlation time to the relaxation time of the system, and we derive the convergence radius for the TCL expansion, which is seen to depend on the ratio of detuning and width of the spectral density. We further study the quantum non-Markovianity of the model based on the evolution of the Bures distance between quantum states and how it is represented by the second and fourth order of the expansion. Our results highlight both the strengths and the limitations of the TCL formalism in capturing key features of open quantum systems and, in particular, the challenges of accurately describing strongly coupled systems and non-Markovian dynamics.

We develop a black hole response formulation of worldline quantum field theory (WQFT) adapted to the gravitational self-force expansion. Integrating out the worldline and graviton fluctuations yields connected graviton response functions: the "zeroth response" one-point function encodes the exact background metric, the "first response" two-point function describes the scattering of a gravitational wave off the black hole, and higher-point functions provide effective vertices for a systematic gravitational self-force (SF) expansion. As a first application, we consider a massless primary source, corresponding to the Aichelburg-Sexl shockwave or an ultra-boosted black hole. We show that the one-point response resums to the exact shockwave metric and that the resummed probe vertices reproduce the exact geodesics in this background. We then compute the off-shell graviton two-point response exactly in the gravitational coupling $G$ by resumming the full post-Minkowskian series. For on-shell external gravitons this yields the exact transfer matrix for gravitational-wave scattering off the shockwave, including recoil. The PM expansion exponentiates to a compact form in which the Born term is dressed by an overall phase which includes the Weinberg phase. Our results provide the basic WQFT ingredients for future 1SF computations of observables such as the impulse and waveform in the ultra high-energy regime.

Tidal dissipation in spinning compact binaries imprints characteristic corrections on the late-inspiral gravitational-wave signal. We develop a next-to-leading order post-Newtonian description of dissipative, electric-quadrupolar tides in spinning compact binaries, deriving the center-of-mass equations of motion, a generalized energy-balance law, and the corresponding Fourier-phase correction for quasi-circular orbits with spins aligned or anti-aligned with the orbital angular momentum. Using the most general, low-frequency, linear tidal response compatible with rotational symmetry, we show that spin-induced tidal dissipation enters the gravitational-wave phase at 2.5 post-Newtonian order and carries a logarithmic frequency dependence, so it is not degenerate with the coalescence phase. For binary black holes, our dissipative flux reproduces horizon absorption in the extreme-mass-ratio limit and points to a redshift-related correction in the comparable-mass case potentially not included in certain recent worldline effective field theory calculations. These results provide new waveform ingredients for precision modeling of spinning compact binaries in the high-signal-to-noise era.

We argue that kinetic mixing between topological flux sectors generates an effective shift of the QCD $\barθ$ angle, thereby inducing CP-violating effects. To demonstrate this mechanism, we analyze a $(1+1)$-dimensional $U(1)\times U(1)$ gauge theory as a controlled setting, where kinetic mixing leads to observable shifts in electric fluxes. We then extend the analysis to four dimensions using a three-form field description of QCD coupled to an additional $U(1)$ three-form gauge field. We find that hidden-sector fluxes, through kinetic mixing, shift the effective $θ$ parameter of QCD and induce a nonzero expectation value of $\langle G\tilde{G}\rangle$. We discuss the implications for the strong CP problem and clarify under which conditions standard solutions, such as axion or CP/parity-based mechanisms, are compromised or remain robust.

We study random entanglement percolation in heterogeneous quantum networks, where the singlet-conversion probabilities (SCPs) of the edges are drawn from a probability distribution rather than being fixed. After briefly recalling random classical and random quantum entanglement percolation, we focus on polarization-dependent loss (PDL) as a physical source of random edge entanglement in photonic networks. In this setting, polarization imbalance induces a simple map from the PDL magnitude to the edge SCP. We illustrate this map for representative PDL models and discuss the resulting implications for entanglement percolation.

A comprehensive analysis of cosmological perturbations and structure formation is presented for the Extended Proca-Nuevo (EPN) framework, a vector-tensor extension of General Relativity with a massive spin-1 field. In this scenario, the vector field modifies the background expansion through an algebraic constraint, leading to a characteristic Hubble evolution that interpolates between $Λ$CDM limits while introducing deviations at intermediate redshifts. Assuming minimal matter coupling, the linear perturbation equations are derived in gauge-invariant form and the resulting growth of matter inhomogeneities is analyzed. The EPN sector induces effective anisotropic stress and couples a single propagating scalar mode to the metric potentials, leaving the matter growth equation in its GR form but with a modified expansion history. The full scalar perturbation system is presented, stability conditions are discussed, and the results provide a foundation for testing the EPN framework against current and future cosmological observations.

New sources of charge-parity (CP) violation beyond those described in the Standard Model (SM) are required to explain the observed matter--antimatter asymmetry of the Universe. The Standard Model Effective Field Theory (SMEFT) provides a framework to introduce additional electroweak sources of CP-odd physics in a model-independent manner. However, these CP-violating signatures are mostly degenerate to CP-even SMEFT operators in polarisation-blind observables, distinguishable only in the SM-New Physics (NP) interference using the azimuthal decay angle. Using Quantum Tomography techniques, we present a new approach to constraining these NP effects. Reconstructing the spin density matrix (SDM) of a diboson system, we go beyond `interference resurrection' to exploit the full signature of the Beyond-SM physics, including the pure quadratic NP terms. We show that this approach provides superior simultaneous sensitivity to characteristic features of CP-even and CP-odd contributions, including effects not fully captured by traditional angular observables.

Systematic control errors remain a primary obstacle to realizing high-fidelity single-qubit gates. We introduce composite pulse sequences that implement X and Hadamard gates while simultaneously compensating amplitude (Rabi-frequency), detuning (frequency), and duration errors. Our construction uses two complementary strategies: (i) derivative-based cancellation of error terms in the full unitary (not just the transition probability), formulated via the Cayley-Klein parametrization, and (ii) direct minimization of the average gate infidelity over prescribed error ranges. We derive symmetric five-pulse solutions with closed-form phases that cancel all first-order terms (including the mixed derivative), and numerically optimize longer sequences -- up to 15 pulses -- to achieve higher-order suppression. We also show that standard ``universal'' five-pulse sequences (U5a/U5b) emerge as simple phase-shifted instances of our symmetric solutions, yielding broad robustness to both detuning and amplitude errors. Finally, we construct variable-area sequences for $R_x(π/2)$, which, up to virtual Z rotations, benchmark the Hadamard gate. Across all families we observe the expected trade-off between sequence length and robustness window, with substantial boosts in fidelity over large error domains.

There are two $R$-factors frequently used in the phenomenology of exclusive processes at small values of the Bjorken $x$ variable. One $R$-factor takes into account the effects of non-zero longitudinal momentum transfer, which is assumed to be zero in the dipole scattering amplitude. Another $R$-factor accounts for the real part of the elastic scattering amplitude which is often neglected, with the standard dipole scattering amplitude giving only the imaginary part of the elastic amplitude. In this work we present two new theoretical developments aimed at eliminating the need for the two $R$-factors. We argue that the $R$-factors can be replaced by (i) modifying the argument of the dipole scattering amplitude and by (ii) augmenting the initial conditions for its non-linear small-$x$ evolution. Specifically, we show that to account for the effects of non-zero skewness $ξ$, one has to replace the rapidity argument $Y = \ln (1/x)$ of the eikonal dipole amplitude $N$ and the odderon dipole amplitude $\cal O$ by $Y = \ln \min \left\{ 1/|x|, 1/|ξ|\right\}$. The prescription applies to the elastic scattering cross sections, as well as for calculations of the Generalized Parton Distributions and Generalized Transverse Momentum Dependent parton distributions at small $x$ and at small but non-zero skewness $ξ$. We also show that the real part of the scattering amplitude, proportional to Im~$N$, which is intimately connected to the signature factor of the amplitude, can be accounted for by a more careful evaluation of the initial condition for the evolution and by writing the non-linear evolution equation in an integral form. One can similarly construct Im~$\cal O$ for the odd-signature odderon amplitude. We hope that future implementation of our prescriptions presented here will eliminate the need for both phenomenological $R$-factors.

Nucleon decays into a lepton and two pseudoscalar mesons represent key channels for probing baryon number violation, complementing conventional two-body modes. In this Letter, we model-independently correlate two- and three-body processes within the framework of low-energy effective field theory, performing a global analysis that avoids single-operator-dominance assumption. We derive significantly improved bounds on 15 three-body modes with a lepton ($e^+,\,μ^+,\hatν=ν/\barν$) and two non-strange mesons ($π,η$). For charged-lepton modes, our lower limits on partial lifetimes ($Γ^{-1}$) surpass current Particle Data Group (PDG) values by more than three orders of magnitude. For 5 (anti)neutrino modes, we establish for the first time $Γ^{-1}\gtrsim 10^{34}\,\rm yr$. Additionally, our analysis improves constraints on two-body processes $n\to e^+π^-$, $n\to μ^+π^-$, and $p\to \hatνπ^+$ by approximately a factor of 2 compared to the PDG limits. These results highlight the importance of leveraging correlations among different processes to better probe new physics, enabling more stringent constraints on experimentally challenging processes from well-measured ones.

We investigate the dynamics and mass spectra of charged pseudoscalar ($π^+$) and vector ($ρ^+$) mesons in a background magnetic field at finite temperature using the two-flavor Nambu-Jona--Lasinio (NJL) model. By employing a quark propagator that isolates the Schwinger phase from its Landau level expansion, we formulate an exact non-commutative phase-space framework utilizing the Wigner-Weyl transform and the Moyal star product. This approach enables the algebraic diagonalization of the Bethe-Salpeter equations for composite states with asymmetric fractional constituent charges. For the pseudoscalar channel, we analytically verify the exact cancellation between the dynamical random phase approximation spatial sum rules and the vacuum gap equation. This identity preserves the generalized Goldstone theorem, causing the $π^+$ pole mass to strictly track the kinematic zero-point energy drift at order of $eB$. In the vector channel, our full phase-space evaluation reveals that the Zeeman spin-splitting emerges dynamically from microscopic threshold truncations governed by the chiral Dirac algebra. Notably, we find that the tachyonic instability of the spin-aligned $ρ^+$ state is quenched. The magnetic catalysis of the chiral condensate drives the continuum threshold ($2M$) upwards, overtaking the Zeeman attraction and preventing vector meson condensation within this mean-field framework. Furthermore, finite-temperature evaluations show a monotonic thermal suppression of the meson masses driven by Pauli blocking, yet all modes remain bound without undergoing Mott dissociation prior to chiral symmetry restoration.

Heavy quark (HQ) momentum ($κ$) and spatial diffusion ($D_s$) coefficients are computed in a non-perturbative thermal QCD medium in the presence of a background magnetic field of arbitrary strength. Both perturbative and non-perturbative effects are incorporated via the in-medium HQ potential, obtained from the resummed gluon propagator. We find that the momentum diffusion coefficients become anisotropic even in the static heavy quark limit, with the magnetic field direction defining the axis of anisotropy. This anisotropy originates from restrictions on longitudinal momentum diffusion in the gluon spectral function, and naturally leads to two spatial diffusion coefficients ($D_s^L$, $D_s^T$). Non-perturbative effects are found to be dominant at low temperatures. These results provide a more consistent input for Langevin based calculations of the heavy quark directed flow at RHIC and LHC energies.

A search for Higgs boson (H) production in association with two vector bosons (V = W, Z) via vector boson scattering (VBS) is presented using proton-proton collision data collected at $\sqrt{s}$ = 13 TeV by the CMS experiment, corresponding to an integrated luminosity of 138 fb$^{-1}$. Events containing two forward jets consistent with VBS, a large-radius jet from the decay of a boosted H to a pair of b quarks, and 0, 1, or 2 charged leptons coming from V decays are selected. The process is excluded at 95% CL for observed (expected) values of the VVHH coupling modifier $κ_\mathrm{VV}$ outside the interval 0.40 $\lt$ $κ_\mathrm{VV}$ $\lt$ 1.60 (0.34 $\lt$ $κ_\mathrm{VV}$ $\lt$ 1.66), assuming standard model values for all other couplings, thus establishing a novel probe of the VVHH interaction. Constraints are also set on the individual $κ_\mathrm{2W}$ and $κ_\mathrm{2Z}$ coupling modifiers, and on the allowed region in the $κ_\mathrm{2W}$-$κ_\mathrm{2Z}$ plane.

Both scalar and vector dark matter can be produced during a cosmological first order phase transition if the dark matter is coupled to the field undergoing the transition. Both kinds of particle are also produced by the plasma through the normal freeze out scenario. For different dark matter masses, we identify the regions of parameter space where there are significant deviations from the normal freeze out scenario and discover there are some rather general predictions. For dark matter particles in the traditional thermal relic GeV-TeV window, dark sector phase transitions around a GeV affect scalar dark matter and dark sector phase transitions around 10 MeV affect vector dark matter abundances (and therefore should take place in a dark sector). When the phase transitions are in the interesting temperature range, the normal range of dark matter masses are different to those predicted by thermal freeze out. We calculate the expected gravitational wave signal of these phase transitions.

We show that in critical loop models, torus 1-point functions can be expressed in terms of sphere 4-point functions at a different central charge. Unlike in the Moore--Seiberg formalism, crossing symmetry on the sphere therefore implies modular covariance on the torus. We systematically compute torus 1-point functions in critical loop models, using a numerical bootstrap approach. We focus on the 1-point functions of the 6 simplest primary fields, which give rise to 10 solutions of modular covariance equations. Such 1-point functions are infinite linear combinations of conformal blocks. The coefficients are products of double Gamma functions, times polynomial functions of loop weights. For each solution, we determine the first 6 to 12 polynomials.

We study the magnetic fields produced in the early Universe during the electroweak symmetry breaking by considering random configurations of an inhomogeneous Higgs field. By exploiting the inherent randomness of the initial configurations the spectrum of the produced magnetic field is essentially analytic, which bypasses the need for costly lattice simulations. On the numerical side, we devise a simulation framework which results in continuous fields capable of resolving the small-scale structure of the fields that was inaccessible for the lattice-based calculation. Finally, by revisiting the effects of statistical isotropy and causality on the spectrum, we define general correlation functions that are then fitted to the simulation data and compared to the analytic results.

In this paper, we present the production of double SM-like Higgs $ (h) $ at the future electron positron colliders within the context of $ S_{3} $ model with three Higgs doublets (S3-3H) and no CP violation to describe beyond the Standard Model Higgs Physics. We focus first on the numerically allowed parameter space of the model, taking into account theoretical bounds from perturbative unitarity and vacuum stability, as well as by data at the Large Hadron Collider (LHC) and the Tevatron. The double Higgs production in the S3-3H model can deviate from the SM predictions up to a few orders of magnitude in both the $ e^+e^- \rightarrow hhZ $ and the $ e^+ e^- \rightarrow h hν_{e} \barν_{e}$ channels. Thus, our findings indicate that the S3-3H model can lead to measurable deviations from the SM predictions of Higgs production at future $ e^+ e^- $ colliders. These results highlight the importance of studies at such colliders for searching physics beyond the Standard Model.

Through analyzing the quark propagator under the magnetic field, we establish that the axial anomaly originates from an additional Dirac structure in quark propagator induced by the magnetic field. This Dirac structure also allows one to connect the axial anomaly with the topological properties of the system by checking the axial vector Ward identity. For the tree level propagator, we reproduce the result of the anomalous axial current as in the Dirac Hamiltonian approach and kinetic theory. Particularly, we confirm that the chiral magnetic effect (CME) comes from the same term that is in charge of the axial anomaly, specifically, as the anomaly of the transversal axial vector Ward Identity. The identity guarantees that the CME conductivity $C_{\rm CME}$ is a constant as $C_{\rm CME}=\frac{1}{2π^2}$, and is robust against the temperature, chemical potential, magnetic field and also interaction. Finally, we verify this numerically by applying the full quark propagator under magnetic field calculated from the functional QCD methods.

Naturally tiny neutrino masses can be explained by the low scale seesaw with heavy neutral lepton $N$ coupling to the neutrinophilic Higgs doublet $Φ_ν$, which obtains a much smaller vacuum expectation value than the standard Higgs doublet $Φ$. Within this model, the neutrino masses originate from the new Yukawa interaction $y \overline{L}\tildeΦ_νN$. In this paper, we propose the novel same-sign tetralepton signature at the 2 TeV same-sign muon mode $μ^+μ^+$ of $μ$TRISTAN. We investigate two distinct channels of this signature, which are both generated by the Yukawa interaction $y \overline{L}\tildeΦ_νN$. One is from the pair production of charged Higgs $μ^+μ^+\to H^+ H^+\to μ^+N +μ^+ N\to μ^+ μ^+ jj + μ^+ μ^+ jj\to 4μ^+ + 4j$, and the other one is from the single production of charged Higgs $μ^+μ^+ \to μ^+ N H^+ \to μ^+N +μ^+ N\to μ^+ μ^+ jj + μ^+ μ^+ jj\to 4μ^+ + 4j$. We then perform a detailed simulation of this same-sign tetralepton signature, and obtain the promising region at $μ$TRISTAN.

The LHCb collaboration has recently observed the rare hyperon decay $Σ^+\to pμ^+μ^-$. It can also measure the corresponding antihyperon channel with comparable precision and is thus in a position to extract information on $CP$ violation in this mode. Interestingly, the long-distance contributions that dominate it within the standard model provide large absorptive phases that could drive substantial $CP$ violation through interference with potential new-physics contributions. Here we explore this possibility, finding that the decay rate asymmetry is currently allowed to be as high as tens of percent, which can be probed by LHCb in the near future. We additionally consider the same with regard to the dielectron mode $Σ^+\to pe^+e^-$ as well as the related radiative one $Σ^+\to pγ$.

We study the phenomenology of heavy neutral leptons (HNLs) in $B$-meson decays as probes of physics beyond the Standard Model. Focusing on the leptonic channels $B \to μN$ and $B \to τN$, we constrain the allowed regions in the $M_N$--$|U_{\ell N}|^2$ plane using current experimental data. Using these constraints, we investigate lepton-number violating ($ΔL=2$) processes mediated by on-shell HNLs, including $B_{(c)}^- \to π^+ μ^- μ^-$ and $B_c^- \to J/ψ\, π^+ μ^- μ^-$. For benchmark values $|U_{μN}|^2 = 10^{-6}$ and $M_N = 2$-- $3\,\mathrm{GeV}$, the predicted branching ratios lie in the range $\mathcal{O}(10^{-13})$--$\mathcal{O}(10^{-8})$. Among the channels, $B_c^- \to π^+ μ^- μ^-$ shows the largest enhancement, while $B_c^- \to J/ψ\, π^+ μ^- μ^-$ is strongly suppressed. These results indicate a clear channel dependence, with $B_c$ modes providing enhanced sensitivity to HNL effects and offering promising avenues for future searches of lepton number violation.

We study nonlinear effective field theories (EFTs) with factorially growing perturbative expansions, focusing on a class in which the relative entropy encodes an infinite tower of higher-dimensional operators. Using the resummed relative entropy, we derive bounds on EFT coefficients: the non-negativity of the resummed relative entropy fixes the sign of their asymptotic growth, while its violation signals instabilities. In fermionic QED, analytic continuation from Euclidean to Minkowski spacetime yields a concrete example: the Schwinger effect, a nonperturbative instability captured by the resummed relative entropy.

As the lightest isovector vector meson, the $ρ$ meson is an important object for investigating the structure of resonant states in strong interactions. Owing to its strong coupling to the $ππ$ channel and its large decay width, the conventional constituent quark model treatment, in which it is simply regarded as a pure $q\bar q$ bound state while the hadronic-channel coupling effects are neglected, is insufficient to fully characterize its physical properties. To this end, in the present work we establish a unified framework for studying the structure and resonant properties of the $ρ$ meson by combining the quark-gluon and hadronic degrees of freedom. At the quark-gluon level, we first determine the parameters of the chiral quark model by refitting a set of narrow mesons for which open OZI-allowed strong-decay channels are absent or strongly suppressed. With these parameters fixed, the bare mass of the $ρ$ meson is obtained and used as the input for the subsequent hadronic-level analysis. At the hadronic level, based on inverse scattering theory, we construct a model including the coupling between the bare state and the $ππ$ continuum, extract the $ρ_0-ππ$ interaction using the $P$-wave $ππ$ scattering phase-shift data, and further calculate the width of the $ρ$ meson as well as the bare-state component in the physical state. The present work also provides a generalizable analytical framework for further studies of other hadronic resonances with significant channel-coupling effects.

This contribution reviews recent progress in the low-lying scalar mesons and glueballs. We propose a new classification for the scalar nonet that includes $f_0(980)$ and $a_0(980)$ as the lowest states, while we identify $f_0(1500)$ as a primary glueball candidate. We demonstrate that the production yields of these states in heavy-ion collisions are mutually consistent across statistical, coalescence, and S-matrix frameworks. To investigate their internal structure, we move beyond standard phenomenology by describing glueballs as topological solitons. This approach yields an energy spectrum in excellent agreement with lattice QCD and experimental data, while interpreting $f_0(2470)$ as a tightly bound glueballonium to explain its anomalously long lifetime. This non-perturbative framework provides a predictive basis for the future experimental verification of exotic scalar states.

We report a detailed simulation study of the search for $s$-channel Higgs boson production in $e^+e^-$ collisions at a center-of-mass (c.m.) energy of $\sqrt{s}=125\,\mathrm{GeV}$ at the CERN Future Circular Collider (FCC-ee), as a means to constrain the electron Yukawa coupling, $y_e$. The process of interest is $e^+e^-\to H\to WW^*\to \ell^\pmν+ jj$ with four different $WW^*$ final states considered, involving both on- and off-shell $W$ bosons decaying either into dileptons ($\ell^\pm = e^\pm$ and $μ^\pm$, including those from $τ^\pm$ decays) or into dijets ($jj$). Signal and background events are discriminated through a multiclass gradient boosted decision tree exploiting a comprehensive set of kinematic and topological variables across the four final-state categories. Assuming a monochromatized c.m. energy spread of 4.1 MeV, yielding a $σ_{e^+e^-\to H} = 280\,\mathrm{ab}$ resonant cross section, and an integrated luminosity of $10\,\mathrm{ab}^{-1}$, the analysis achieves a combined statistical significance of 2.0 standard deviations. This corresponds to an upper limit on the coupling modifier $κ_e = y_e/y_e^{\rm SM} \lesssim 1.35$ at 95\% confidence level, and provides the most stringent constraint on the electron Yukawa coupling achieved in simulation-based studies to date.

Landau models serve as quantum mechanical systems for generating quantum matrix geometries. In this paper, we demonstrate that Howe duality provides the underlying structure of the super Landau model, reflecting a general feature of coset-type Landau models. The (super) Howe duality relates different Landau levels and accounts for the emergence of a dual fuzzy geometry. By employing super-spinor derivative operators, the supermonopole harmonics in both integer and half-integer Landau levels are explicitly constructed and the algebraic structure of the super-Hilbert space is revealed. We propose a consistent probabilistic interpretation for these wavefunctions defined on a supermanifold. Through a level projection method, we derive the matrix coordinates of fuzzy supersphere geometries for arbitrary Landau levels, along with a precise determination of the non-commutative scale factor. It is shown that the theta correspondence of Howe duality induces a geometric transformation between fuzzy objects. Finally, we point out that Howe duality realizes an internal-external space duality and underlies quantum matrix geometries, suggesting that it may play a fundamental role in understanding Matrix model geometries.

Using the gauge-gravity duality, we construct the anisotropic D3/D7 approach as a three-dimensional QCD-like theory, then investigate systematically the hadronic mass spectra, the dragging terms and the lowest hadronic interactions in the presence of the anisotropy in holography. Our derivation illustrates the dragging terms in the effective action are very necessary for an anisotropic theory since they are the key roles to affect the transport properties of the dual theory. And the numerical results in addition imply the hadronic system will become unstable if the anisotropy is much larger than its confinement energy scale. It agrees with that the confining phase in this approach becomes unstable if the anisotropy is sufficiently large in our previous works with this model. Therefore, this work is constructive to understand the anisotropy in the gauge field theory.

DUNE (Deep Underground Neutrino Experiment) is a long-baseline neutrino oscillation experiment currently under construction, whose far detectors will be the largest liquid argon time projection chambers ever built. This detector design calls for custom-built cryogenic front-end electronics to attain the required detector performance. This paper describes the charge readout electronics that will be used in the DUNE horizontal drift (HD) far detector and presents performance results using data from the ProtoDUNE-HD detector, a 770 ton liquid argon time projection chamber operated at the CERN Neutrino Platform in 2024 that served as the final prototype of the DUNE HD design.

Low-Gain Avalanche Diodes (LGADs) are fast silicon sensors with internal charge multiplication and are key candidates for precision timing layers in future high-energy hadron colliders. Their operation in harsh radiation environments, however, is limited by acceptor removal in the gain layer, which reduces the active acceptor concentration and degrades the internal electric field required for avalanche multiplication. Improving the radiation tolerance of the gain layer is therefore essential for future 4D tracking applications. In this work, we investigated several LGAD prototypes produced in collaboration with Hamamatsu Photonics K.K. (HPK), featuring modified gain-layer designs, including oxygen-modified, carbon-implanted, and boron--phosphorus compensated structures. The sensors were studied after proton and reactor-neutron irradiation. Radiation tolerance was characterized using the acceptor-removal coefficient extracted from IV measurements and the operation voltage required to recover the timing performance after irradiation. The results show that carbon implantation is the only approach among those studied here that provides a clear improvement in radiation tolerance. In contrast, neither oxygen-related modification, including the Partially Activated Boron (PAB) approach, nor gain-layer compensation alone yields a significant improvement, and the compensated carbon-implanted structure shows no clear advantage over the carbon-only case. In addition, the acceptor-removal coefficient is found to depend on the irradiation particle type and energy.

High-energy-resolution GaN $α$-particle detectors have significant potential for space radiation, nuclear instrumentation, and harsh-environment applications. However, existing GaN $α$-particle detectors still face several key challenges, including reducing the dead-layer thickness, suppressing leakage current under high reverse bias, improving energy resolution, and clarifying the physical mechanism underlying the low-energy tail phenomenon. This study presents a vertical homoepitaxial GaN $α$-particle detector integrating a 20-nm ultrathin dead layer and a guard-ring structure. The detector exhibits an ultralow leakage current of 2.195 nA at -200 V and an intrinsic energy resolution of 2.69% with a charge collection efficiency (CCE) of 95.9% at -260 V. More importantly, this work demonstrates for the first time through Geant4 simulations that depletion-width nonuniformity is the dominant source of partial energy leakage, leading to an extended low-energy tail in the energy spectrum. We establish a depletion-width nonuniformity model and observe good agreement between simulation and experiment. This finding provides practical guidance for the design and optimization of high-performance GaN-based radiation detectors.

We report on an experimental demonstration of efficient neutron generation based on direct laser acceleration in microwire-array targets irradiated by ultrashort (tens of femtoseconds) laser pulses. The optimal array period was identified, at which the maximum proton energy and the number of protons with energies exceeding $1~\mathrm{MeV}$ were significantly increased. Using a $1~\mathrm{PW}$, $\sim25~\mathrm{fs}$ laser at a moderate intensity of $\sim10^{20}~\mathrm{W/cm^2}$, a high neutron yield of up to $(8.33\pm0.84)\times10^{6}~\mathrm{n/sr/J}$ was detected from the LiD converter via $^7\mathrm{Li}(p,n)$ and $\mathrm{D}(p,n)^3\mathrm{He}$ nuclear reactions. Self-consistent integrated simulations reproduced the experimental results and predicted that with a Be converter, a forward pulsed neutron source with an unprecedented yield per joule of $3.67\times10^{7}~\mathrm{n/sr/J}$ can be obtained under identical laser conditions. This type of neutron source is favorable for applications that require a high repetition rate utilizing compact and economical laser systems.

Ultralight bosons sourced by macroscopic objects can generate long-range spin-independent and spin-dependent potentials that are accessible to precision interferometry. Such potentials induce phase shifts in Josephson junctions, detectable through precision current measurements. We propose three experimental scenarios to probe photophilic scalar interactions, Lorentz-violating scalar-mediated interactions, and axion-mediated monopole-dipole interactions, depending on the nature (unpolarized or polarized) of the source. The proposed setups provide sensitivities to novel mixed couplings that are largely unconstrained by existing bounds and enables the exploration of new forces at centimeter to micrometer length scales.

We show that the Weyl (trace) anomaly gives rise to a new non-dissipative vector current in accelerated relativistic fluids. The anomaly uniquely fixes the second-order transport coefficient governing the coupling between the electromagnetic field and the fluid acceleration. We derive this result by extending hydrodynamic anomaly matching to include the trace anomaly, and independently reproduce it in boundary quantum field theory by treating the Rindler horizon of an accelerated observer as an effective boundary. From the boundary perspective, the electric- and magnetic-field sectors correspond to screening and vacuum magnetization effects near the boundary. In the local rest frame, the electric-field contribution induces an additional charge density, while the magnetic-field contribution generates a transverse current with a Nernst-like, more generally thermomagnetic Hall-like, tensor structure. Our results reveal a new class of anomaly-induced transport governed by the trace anomaly.

We study the thermodynamics and interior structures of dynamical end of the world (EoW) branes in the rotating BTZ black hole. By mapping the induced metric of the branes to an effective Jackiw-Teitelboim (JT) system, we derive the first law of thermodynamics for the boundary conformal field theory (BCFT), incorporating boundary degrees of freedom. To construct this thermodynamics, we adopt two frameworks of black hole chemistry and the duality between the JT black hole and the Sachdev-Ye-Kitaev (SYK) model. In addition, we verify that the shadow entropy is equivalent to the boundary entropy via Hubeny-Ryu-Takayanagi (HRT) surfaces. Furthermore, we explore the possible interior configurations of the EoW brane inside the horizon. Two representative configurations, namely single and double-joint EoW branes, are considered. These disparate configurations share the same exterior brane configuration outside the horizon. By comparing their energies, we show that a transition could occur within the horizon.

Our recent study discussed the prospects for measuring single positrons produced in electron-laser collisions via the nonlinear Breit-Wheeler deep-tunneling process in the SLAC Experiment 320 at the FACET-II RF LINAC. In this work, we demonstrate how a tracking detector, that is a scaled-down version of the one discussed in the prospective simulation study, enables the measurement. This prototype detector, installed in Aug 2024, is built out of five layers of single ALPIDE chips. The data are taken from several standalone runs completed in Nov 2024 and Feb 2025. We use positrons generated through conversion of Bremsstrahlung photons as a proxy to the nonlinear Breit-Wheeler process. These positrons are produced by the beam electrons in a thin Beryllium foil close to the experiment's interaction point. The tracking approach used in this initial work is based on a Hough-Transform seeding algorithm followed by a straight line fit confined to the detector volume. Even with this relatively simple approach, we are able to measure a signal rate of $(1.20\pm0.06_{stat.}\pm0.56_{syst.})\times10^{-1}$ positrons per shot. This signal rate is comparable to the nonlinear Breit-Wheeler rate expected in the main experiment. Notably, the measurement is achieved under an extreme, unprecedented background hit density of ~1.7/mm$^2$, unlike the main experiment, where at least a twice lower density is expected. This large background is mostly due to secondary particles produced when the large flux of Bremsstrahlung photons interacts with the material of the beamline elements. When the foil is retracted, the false-positive signal rate is shown to be four orders of magnitude smaller than the signal rate. We further show that the high spatial tracking resolution of ~5 micron allows to characterize the positrons' spectra. The results are compared to simulations, which are found to be compatible with the data.

Beyond $B$ physics, charm and $τ$ physics constitute a central part of the Belle and Belle II physics programs. Here, we present recent results on charm baryon decays, including first measurements and observations of several previously unmeasured modes, together with new studies in $τ$ physics. Particular emphasis is placed on searches for CP violation in $τ$ decays and lepton-flavor-violating processes. In this context, a search for CP violation in $τ\to πK_{S} ν_τ$ decays is reported for the first time.

We study the kaon quasi-distribution amplitude (quasi-DA) and distribution amplitude (DA) within the large-momentum effective theory (LaMET) combined with the first-principles functional QCD. Using quark correlation functions and the kaon Bethe-Salpeter amplitude in the Euclidean space from the 2+1 flavour functional QCD [1] as inputs, we obtain the kaon quasi DA in the large longitudinal momentum region with the contour deformation method [2] in the complex plane of momentum. By performing $1/P_z^2$ and $1/P_z^4$ order extrapolations of the kaon quasi-DA for the choices of the maximal longitudinal momentum $P_z^{\max}\in[2,2.5]$ GeV, we obtain a single-peaked and asymmetric kaon DA with the uncertainties arising from the extrapolation interval and ansatz. We find the first and second order moments of the kaon DA, $\langle ξ\rangle_K = 0.020(3)$ and $\langle ξ^2 \rangle_K = 0.253(12)$, respectively.

We review recent developments on the QCD phase diagram at small chemical potentials and increasing temperature. There are three regimes/phases in QCD which differ by symmetries, degrees of freedom and N_c scaling: the hadron gas below the chiral restoration temperature T_ch, the stringy fluid between T_ch and the deconfinement temperature T_d and the quark-gluon plasma above T_d.

In this paper, we propose a universal diagrammatic interpretation of hidden zeros and $2$-splits of tree-level amplitudes. Originally developed for ${\rm Tr}(φ^3)$ amplitudes in our previous work, this interpretation is now extended to tree-level amplitudes in Nonlinear sigma model (NLSM) and Yang-Mills (YM) theories. The interpretation is based on a certain factorization behavior of Feynman diagrams under specific kinematic constraints, which we term shuffle factorization along a specific line (SFASL). This mechanism allows us to separate Feynman diagrams along specific lines after summing over shuffle permutations. When applied to NLSM and YM amplitudes, we perform proper extensions of the SFASL used in the ${\rm Tr}(φ^3)$ case. Through the SFASL, the interpretation for the hidden zeros and $2$-splits of tree amplitudes of ${\rm Tr}(φ^3)$, NLSM, and YM can be unified as: the hidden zeros are ascribed to the on-shell condition $k_j^2=0$ of a massless particle, while the $2$-splits are caused by separating each Feynman diagram along two lines, akin to unzipping two zippers.

Let $A:=\mathbb{C}[z_+,z_-]\otimes Λ(θ_1,θ_2,θ_3)$, with $z_\pm$ even and $θ_1,θ_2,θ_3$ odd. For a reductive Lie algebra $\mathfrak g$, let $\mathfrak g[A]:=\mathfrak g\otimes A$ be the corresponding current Lie superalgebra. Motivated by the Chang--Yin description of weak-coupling $1/16$-BPS cohomology in $\mathcal N=4$ super-Yang--Mills, we study the relative Lie algebra cohomology $H^\bullet(\mathfrak g[A],\mathfrak g;\mathbb{C})$. We isolate three finite-rank phenomena. First, the natural $3|2$ super-commuting restriction map, viewed as a super analogue of Chevalley restriction and its commuting-scheme variants, already fails to be an isomorphism for $\mathfrak g=\mathfrak{so}_7$; the obstruction is a non-Cartan class. Second, the same algebra produces explicit fortuitous classes for $\mathfrak{sl}_2$ and $\mathfrak{so}_7$, giving concrete counterexamples to naive stable-image expectations suggested by the type-A Loday--Quillen--Tsygan theorem and its current-algebra refinements. Third, the classical relative cohomologies for the Langlands-dual pair $(\mathfrak{so}_7,\mathfrak{sp}_6)$ are not isomorphic. We then record the conjectural quantum deformation of the differential expected to restore duality, together with first-order evidence pairing the fortuitous and non-Cartan $\mathfrak{so}_7$ classes.

We suggest a non-minimal renormalization scheme based on dimensional regularization that naturally incorporates threshold effects of heavy particles. By renormalizing couplings and masses to subtract all poles in $d \geq 4$, the resulting scheme is mass-dependent and circumvents shortcomings of mass-independent schemes like minimal subtraction. At the same time, many advantages of minimal subtraction such as gauge independence are retained. Through explicit one-loop computations in QCD, we demonstrate that this scheme reduces to minimal subtraction at high energies while providing smooth transitions at particle thresholds and implementing the Appelquist-Carazzone theorem. Potential future applications and extensions are discussed.

A search is presented for massive narrow-width resonances in the mass range of $1\text{-}4.5\,\text{TeV}$ decaying into pairs of Higgs bosons (HH), using proton-proton collision data at a center-of-mass energy of $13\,\text{TeV}$ collected with the CMS detector at the LHC during the $2016\text{-}2018$ data-taking. The data correspond to an integrated luminosity of $138~\mathrm{fb}^{-1}$. The analysis targets final states where one Higgs boson decays into a pair of bottom quarks and the other into a pair of tau leptons, $\mathrm{X}\rightarrow\mathrm{HH}\rightarrow \mathrm{b}\bar{\mathrm{b}}\,τ^{+}τ^{-}$. The observed data are found to be consistent with standard model background expectations. Upper limits at $95\%$ confidence level (CL) are set on the production cross section for resonant HH production for masses between $1$ and $4.5\,\text{TeV}$. This analysis sets the most sensitive LHC limits to date on $\mathrm{X}\rightarrow\mathrm{HH}\rightarrow \mathrm{b}\bar{\mathrm{b}}\,τ^{+}τ^{-}$ decays in the mass range of $1.4$ to $4.5\,\text{TeV}$.

Recent developments of Batalin-Vilkovisky (BV) formalism and related geometry are reviewed. Mathematical structures of BV formalism are summarized as a Q-manifold and a QP-manifold. Lie algebras, Lie algebroids and other higher algebroids are explained as typical examples of Q- and QP-manifolds. Finally, the BV action functionals are constructed by geometric structures of Lie algebroids and Courant algebroids.

The invariant mass spectrum of $e^+e^-$ pairs produced in high-energy Pb-emulsion collisions at 160 A GeV at CERN SPS exhibits a complex structure of many resonances resting on top of a broad enhancement at invariant masses below 50 MeV, with the prominent resonance at 19 $\pm$1 MeV providing independent support for the hypothetical X17 particle. We show that this complex structure may be coherently described as signatures for the neutral color-singlet $q\bar q$ quark matter in both its deconfined and confined phases. That is, the broad enhancement may arise from thermal annihilation of QED(U(1))-deconfined quarks and antiquarks into $e^+e^-$ pairs at the phase transition temperature $T_c$(QED), theoretically estimated to be 4.75 $\pm$ 1.2 MeV from the transitional equilibrium condition. The observed 3$\pm$1 and 7$\pm$1 MeV resonances may correspond to the QED(U(1))-deconfined $d\bar d$ and $u\bar u$ Coulomb bound states near their quark rest masses, respectively, whereas the observed 19 $\pm$ 1 MeV resonance may correspond to the QED(U(1))-confined isoscalar QED meson. The approximate agreement between the theoretical and the experimental spectrum suggests that both QED(U(1))-confined and QED(U(1))-deconfined neutral color-singlet $q\bar q$ quark matter may have been produced in these high-energy Pb-emulsion collisions. We propose future experiments to confirm or refute these findings.

For supersymmetric extensions of the Standard Model we construct some expressions that include Yukawa couplings for the third and second generations and receive relatively small quantum corrections. This implies that they slightly depend on scale and are therefore approximate renormalization group invariants. Using these invariants we try to analyse possible relations between the Yukawa couplings at the unification scale $M_X$ as well as the predictions for values of $\mbox{tg}\,β$ and $α(M_X)$. In particular, we suggest two variants of such relations and investigate whether they agree with the experimental values of elementary particle masses. It is demonstrated that the Yukawa unification for the third and second generations consistent with them can be achieved by adding exotic superfields forming 3 representations $5+\bar{5}$ of the group $SU(5)$ to the MSSM field content. We argue that this may indicate the possible underlying $E_6$ gauge symmetry.

We calculate the two-loop quarkonium Hamiltonian in the annihilation channel within the framework of potential-NRQCD effective field theory. The result agrees with the previous calculation of the corresponding four-quark operator in NRQCD for SU(N) color gauge group. We further obtain an expression with a more general color structure applicable to other gauge groups. Combined with the recently calculated two-loop Hamiltonian in the non-annihilation channel, this completes the full two-loop quarkonium Hamiltonian.

We investigate the three-body $DDK$ system with quantum numbers $I(J^P) = \frac{1}{2}(0^-)$ within a coupled-channel framework that incorporates both $DDK$ and $D^{*}D^{*}K$ configurations. The $D^{(*)}D^{(*)}$ interactions are described using the one-boson-exchange model constrained by the heavy-quark symmetry and fitted to the pole positions of $X(3872)$, $T_{cc}^+$, and $Z_c(3900)$. The $D^{(*)}K$ interaction is from the chiral effective theory, motivated by the molecular interpretation of $D_{s0}^*(2317)$, and is further constrained by lattice-QCD results for the $DK$ scattering lengths. The resulting three-body problem is solved using the Gaussian expansion method, while the complex scaling method is employed to search for possible resonant states. We find that coupled-channel effects from $D^{*}D^{*}K$ are negligible, and the $DDK$ system supports a deeply bound state across a wide range of parameters. Depending on the long-range behavior of the $DK$ interaction, an additional shallow state may emerge near the particle-dimer ($D$-$DK$) threshold. The deeply bound state exhibits a compact three-body structure, whereas the shallow state displays characteristic features of a three-body halo configuration. No clear resonance poles are identified within the explored parameter region. Similar results are obtained for the $D^{*}D^{*}K$ system. These findings may provide new insight into few-body dynamics in systems involving charmed mesons and kaons.

We expect black hole microstates to differ in their chaotic properties from states associated with other geometries. For supersymmetric black holes, ordinary level statistics cannot diagnose this distinction, since their energy levels are exactly degenerate. We propose that there is an intrinsic probe of chaos, encoded in the mixing of the microstates under changes in the couplings of the theory, as determined by the non-Abelian Berry curvature of the BPS states under certain deformations. For states dual to horizonless geometries in holographic systems, such as 1/2-BPS states in the D1/D5 CFT and 1/4-BPS states in $\mathcal{N}=4$ SYM, we find that the Berry curvature for marginal deformations is non-random and often exactly zero at generic couplings. By contrast, for states dual to supersymmetric black holes, we show through computations in $\mathcal{N}=2$ super-JT gravity and explicit numerics in the $\mathcal{N}=2$ SYK model that the Berry curvature resembles a random matrix. We also uncover interesting topological features of the $\mathcal{N}=2$ SYK moduli space, as probed by Chern numbers. These results suggest that the Berry curvature sharply distinguishes black hole microstates from smooth horizonless states and provides a robust diagnostic of chaos in supersymmetric sectors.

The enhancement of the baryon production relative to mesons in small-collision systems is considered a breakthrough result of the Large Hadron Collider since a similar effect in heavy-ion collisions is understood by invoking the formation of the strongly-interacting quark--gluon plasma. In this letter, a baryon enhancement is reported for $p_{\rm T}^{\rm ch,\, jet}>15$\,GeV/$c$ jets produced in pp collisions at $\sqrt{s}=13$\,TeV simulated with PYTHIA8. The effect can be explained as a transition between quark-initiated jets (low jet multiplicities) to gluon-initiated jets (high jet multiplicities). The present result challenges the interpretation about the multiplicity dependence of the baryon enhancement in terms of collective expansion of the medium and quark recombination.

Deep-inelastic $e/μ$ scattering data off targets ranging from deuterium to lead indicate that the nuclear modifications to the structure functions of bound nucleons are minimal in the kinematic region around the peak of the valence quark distributions. An analysis of world measurements of the isoscalar cross-section ratios $σ^A/σ^{{}^2\text{H}}$ in the region of $0.25 \leq x \leq 0.35$ shows a remarkable cancellation across all nuclei, with an average value of $0.9985 \pm 0.0022$. We discuss these results and possible interpretations in the context of a microscopic model of nuclear modifications of the structure functions.

We perform coupled channel analysis for $D \bar D^*$, $J/ψπ$ and related meson pairs for the $Z_c(3900)$ channel in an effective model of hadrons and quarks. The model incorporates meson exchange potential such as one pion and $D^{(*)}$ meson exchanges, and quark exchanges. It turns out that the meson exchange potential is small, while the off-diagonal interactions by the quark exchanges at short distances, particularly for transitions between $D\bar D^*$-$J/ψπ$ are strong, which plays a main role for the scattering amplitudes for the $Z_c(3900)$ channel, in consistent with the results of the lattice simulations of the HALQCD group.

Earth's matter affects the oscillation of atmospheric neutrinos and antineutrinos differently depending on the neutrino mass ordering (NMO). As more neutrinos than antineutrinos are expected to be detected in the IceCube detector, this matter effect can be used to probe the NMO. The fraction of energy transferred to the nucleon during a neutrino interaction, known as the inelasticity, has a different distribution for neutrinos and antineutrinos because of their opposite chirality. This can in theory be used to statistically separate neutrinos from antineutrinos, but hasn't been exploited in IceCube DeepCore analyses yet. To this end, two new inelasticity reconstructions were developed using a graph neural network and an ensemble of two-dimensional convolutional neural networks. This presentation discusses the development and performances of these reconstruction algorithms. The inelasticity is then used as a fourth observable, along with the particle energy, direction and flavor, to calculate new NMO sensitivities and determine the impact of adding the inelasticity in the measurement of the NMO with the IceCube DeepCore and upcoming IceCube Upgrade detectors.

The B-L Supersymmetric Standard Model with Inverse Seesaw (BLSSM-IS) extends the Minimal Supersymmetric Standard Model (MSSM) by incorporating a gauged B-L symmetry, right-handed neutrinos and an additional neutral gauge boson Z'. Searches at the Large Hadron Collider (LHC) constrain the mass of this gauge boson to be as low as only ~ 2.2 TeV in the BLSSM-IS, owing to interference effects with the SM. In this framework, mono-Z' events can arise from the associated production of a Z' boson and a singlet Higgs boson h', where h' subsequently decays into missing energy carried by a pair of the Lightest Supersymmetric Particle (LSP) - either a neutralino or a right-handed sneutrino - which serves as a Dark Matter (DM) candidate. Focusing on leptonic decays of the Z' (electrons and muons), we analyse the kinematic distributions of the final-state leptons and the missing transverse energy in order to extract a signal for this process which is independent of the nature of the BLSSM-IS DM.

High-fidelity simulation of particle-matter interactions provides the essential theoretical reference for diverse physics disciplines, yet generating synthetic datasets at the scale of current and future experiments has become prohibitive. Here, we introduce PHIN-GAN, a novel physics-informed generative adversarial network designed to address this challenge. We derive a set of analytical probability density functions, that effectively describe the ``straggling function'' identified with Landau. For the first time, this enables their continuous evaluation across the entire phase-space. These analytical forms are leveraged to enforce a parametric distribution-level learning objective. Rooted in first principles, PHIN-GAN offers a generalizable, interpretable and scalable proof-of-concept approach for a lossless generative model that maintains the high fidelity of the standard-bearer for simulating such interactions, namely GEANT4, at a fraction of the computational cost.

We construct an effective light-front mass-squared operator for quark-antiquark systems that incorporates quark dressing effects through a running quark mass. Starting from a Minkowski-space quark propagator constrained by lattice-QCD-inspired parametrization, we derive the disconnected light-front resolvent for a light quark-antiquark system using a generalized spectral representation of an individual quark propagator separating out the instantaneous contributions. By projecting the resolvent onto a constituent-quark helicity basis, we obtain an effective dressed mass-squared operator suitable for light-front Hamiltonian approaches. We introduce an effective light-front quark self-energy and analyze its momentum dependence. As an application, we study pion structure using representative light-front wave-function models and compute unpolarized transverse-momentum-dependent distributions, unpolarized parton distribution functions and distribution amplitudes. Our results show that quark dressing induces sizable infrared modifications while preserving controlled ultraviolet behavior, providing a framework to incorporate nonperturbative QCD dynamics into light-front descriptions of hadrons.

A formalism for radiative correction calculations in exclusive $η$ electroproduction on the proton is presented, extending the treatment developed for the pion channel. The EXCLURAD code is used in the radiative correction procedure with EtaMAID-2023 multipole amplitudes. The cross-section correction factor $δ$ varies by up to ${\sim}\,30\%$ across the resonance region $W = 1.49$-$2.0$ GeV at $E_{\rm beam} = 6.535$ GeV, with a local maximum near $W \simeq 1.66$ GeV driven by the $S_{11}(1535)$ and $S_{11}(1650)$ resonances. The beam-spin asymmetry is suppressed by 15-25% at the same kinematics. Numerical results covering $Q^2 = 0.3$-$4.0$ GeV$^2$ and the full angular range are provided for kinematics relevant to CLAS12 experiments at Jefferson Lab.

In this work we analyze the classical confinement of relativistic and supersymmetric spinning particles in higher-codimension braneworlds. Considering warped backgrounds generated by string-like $ n=2$ and global scalar defects $ n \geq 3$, we derive the effective radial dynamics from a Polyakov-type action. For spinless particles, the effective potential is monotonically decreasing, leading to repulsive behavior and the absence of confinement. In contrast, for $N=1,2 $ spinning particles, spin-curvature coupling modifies the potential, allowing the emergence of stable equilibrium points. Depending on the coupling parameter, particles may be confined on the membrane or in nearby regions, exhibiting bounded or satellite-like motion. These results emphasize the role of spin in localization mechanisms.

I briefly summarize a recent research program aiming to probe the landscape of low-energy phases of string theory from a global perspective. Borrowing conceptual lessons from the swampland program, I will discuss how the effective theories of gravity produced by low-energy string theory are far from generic; rather, their infrared data is connected by universal scaling relations which become non-trivial in species limits. In particular, a worldsheet analysis reveals that higher-derivative Wilson coefficients and the vacuum energy exhibit UV/IR relations which are invisible from the viewpoint of effective field theory, leading to parametric inequalities which take the form of holographic bounds. This lends a more solid theoretical support to swampland-motivated phenomenological scenarios, and to the broader hopes of extracting less direct, but empirically accessible, signatures of string theory from cosmological observations.

M-theory on ${\mathbf{S}}^1\vee{\mathbf{S}}^1$ has recently been proposed to yield, via quotients, ten-dimensional non-supersymmetric string theories. We revisit the construction that leads to the heterotic theories, finding a consistent set of rules that reproduces the light spectra and gauge groups, including indications of their global structure. Our approach uses the gauge enhancement mechanism of type I' string theory, applied to a setting in which the type I' interval is curled onto itself, with its two boundaries separated by a branch cut. Using these tools, we reproduce the ten-dimensional heterotic theories and provide some evidence for junctions among them.

Since their rediscovery in the 1990s, multiple zeta values have become ubiquitous in many areas of mathematics and physics. Their standard integral and sum representations can usually be traced back to a single source, namely the iterated integrals on the Riemann sphere with three punctures. We refer to such representations as the \emph{linear} geometry of multiple zeta values, since the denominators of the corresponding integrands factor completely into linear terms. However, there also exist equally important and entirely distinct integral representations for multiple zeta values arising in mathematics and physics, in which matrix determinants appear in the denominator of the integrand. We call this the \emph{non-linear} geometry of multiple zeta values. These lectures trace the origins of this non-linear geometry and provide an introductory journey through a range of topics including tropical geometry, the moduli spaces of tropical curves, Feynman integrals in quantum field theory, the general linear group of integer matrices, and the reduction theory of quadratic forms. In doing so, we propose a geometric framework for multiple zeta values based on such non-linear, determinantal representations and set out a number of open questions for future research.

We present an amplitude analysis of high-energy polarized photoproduction of $π^-Δ^{++}$ within a Regge exchange framework. A Regge amplitude model incorporating $π$, $ρ$, $b_1$, and $a_2$ trajectory exchanges is fit simultaneously to spin density matrix elements measured by the GlueX experiment at photon energies of $E_γ= 8.2$--$8.8$ GeV and differential cross section data from SLAC. By including SDME data, the fit constrains not only the magnitudes but also the relative phases of the helicity amplitudes. The results confirm the dominance of pion exchange at small momentum transfer, while natural parity exchanges become significant at larger $t$. We analytically continue the $s$-channel amplitude to the $t$-channel, taking care of the kinematical singularities, and isolate the dynamical residues at the meson poles. The extracted $πNΔ$ coupling constant is found to be consistent with the value obtained from the decay width of the $Δ(1232)$. For the $ρNΔ$, $b_1 NΔ$, and $a_2 NΔ$ vertices, first extractions of the relevant coupling constants are provided.

We use geometric Landau analysis to determine the singularity structure of four-point, one-cycle negative geometries in $\mathcal{N}=4$ super-Yang-Mills theory, which represent certain contributions to the logarithm of the four-point amplitude or equivalently the normalized quadrangular Wilson loop with a Lagrangian insertion. By analyzing the relevant Landau diagrams recursively, we prove that this quantity has singularities only at $z=-1,0$ and $\infty$ to all loop orders. This represents a first step towards obtaining a non-perturbative resummation for this quantity at next-to-leading order in the expansion over cycles.

A nonzero neutrino magnetic moment arises already in the minimally extended Standard Model with right-handed massive Dirac neutrinos. The well-known consequence of the neutrino magnetic moment is the phenomenon of neutrino spin oscillations in a magnetic field. It can manifest itself not only as a lack in the flux of active cosmic neutrinos arriving on Earth but also as characteristic features in low-energy neutrino elastic scattering processes. Following our approach developed earlier, in this work we study the influence of arbitrary spin-flavor state of incoming neutrino on low-energy neutrino scattering off different particles in a detector. We demonstrate that superposition of left- and right-handed helicity neutrino states gives rise to an azimuthal asymmetry in the angular distribution of the recoil momenta. We present numerical calculations for elastic neutrino scattering on electrons, protons and 40Ar and 132Xe nuclei, demonstrating the azimuthal-asymmetry effect. Our results indicate the unique potential of directional neutrino detection to observe the neutrino spin oscillations.

Sati and Schreiber [arXiv:2402.18473, arXiv:2512.12431] have proposed that charge quantisation in quantum field theory and string theory is governed by a homotopy type $\mathcal A$. We provide a refinement of this postulate, incorporating other currents including matter, connecting it to adjustments in higher gauge theory and providing a prescription for determining $\mathcal A$, and show that, while the homotopy groups of $\mathcal A$ classify the possible brane charges, the homology groups of $\mathcal A$ classify the invertible higher-form symmetries. Furthermore, we show that the charge-quantisation postulate implies a number of non-trivial constraints on quantum field theories similar to those implied by swampland conjectures; in particular, it rules out noncompact gauge groups and one-form field strengths that form a non-nilpotent Lie algebra. Finally, we argue that for theories of quantum gravity the space $\mathcal A$ must be contractible, in accordance with the swampland conjectures on the absence of global generalised symmetries and the completeness of the spectrum of charges, and explain how this explicitly arises in the case of Type I string theory.

We use the framework of generalized entanglement wedges to revisit the connected wedge theorem (CWT). This construction identifies an entanglement wedge associated for any bulk region and allows us to rephrase the CWT in terms of the entanglement entropies of bulk regions. We establish new upper and lower bounds on the mutual information of boundary decision regions in terms of the entropies of certain bulk regions associated with a scattering configuration. We then define new bulk decision regions for which we show that a non-empty scattering configuration implies a connected entanglement wedge. This generalization of the CWT extends to asymptotically flat spacetimes.

Despite the remarkable success of the Standard Model in describing fundamental interactions, unresolved phenomena such as dark matter, dark energy, and matter-antimatter asymmetry strongly suggest the existence of physics beyond the Standard Model. The absence of new particle discoveries at the LHC indicates that such New Physics may be significantly heavier than the electroweak scale. In this context, Effective Field Theories offer a powerful framework for studying the indirect effects of heavy New Physics. This contribution reviews some of the recent advancements, computational tools, and phenomenology of Effective Field Theories, with a particular focus on the Standard Model Effective Field Theory.

A natural approach for constructing a concrete example of flat space holography is to take the flat space limit of a well-understood example of AdS/CFT, such as the one relating M-theory in AdS$_4$ times an orbifolded 7-sphere to a certain three dimensional superconformal Chern-Simons-matter theory known as the ABJM theory living in the boundary of AdS$_4$. In particular, taking the flat space limit of the bulk corresponds to taking the speed of light $c$ to zero in the boundary, giving rise to a Carrollian superconformal theory. This limit is subtle to implement for fermions, however, since the Dirac algebra is sensitive to the spacetime metric and therefore takes a different form in Carrollian spacetimes than it does in Minkowski space. In fact, we show that there are four possible ways of realising Carrollian fermions, one of which arises at leading order in the $c\rightarrow0$ limit of relativistic fermions. In three dimensions, there is an additional complication that the minimal realisation of the Carrollian Dirac algebra requires $4\times4$ matrices rather than $2\times 2$ matrices familiar from the relativistic case. Nevertheless, we show that the $c\rightarrow0$ limit of the ABJM theory can be recast in terms of Carrollian Dirac matrices and enjoys and infinite-dimensional Carrollian superconformal symmetry whose bosonic subsector is the extended BMS$_4$ algebra encoding the asymptotic symmetries of four dimensional Minkowski space. This provides a concrete starting point for constructing a Carrollian gauge theory dual to M-theory in flat space.

The physics program of the Higgs factory will focus on measurements of the 125 GeV Higgs boson, with the Higgs-strahlung process being the dominant production channel at 250 GeV. However, similar production of exotic light scalars, in a scalar-strahlug process, is still not excluded by the existing experimental data, provided their coupling to the SM gauge bosons is sufficiently suppressed. This was selected as one of the focus topics of the ECFA Higgs/Top/EW factory study. Presented are the expected cross section limits from the search in the $b\,\bar{b}$ decay channel, based on a full simulation of the International Large Detector (ILD), as well as the expected sensitivity in $τ^+τ^-$ and invisible decay channels, based on the fast simulation in the Delphes framework, assuming 250 GeV ILC running scenario.

High Energy Physics (HEP) experiments like BESIII produce petabyte-scale data. Extracting physics results requires complex workflows (simulation, reconstruction, statistical analysis, etc.) that traditionally take experts months or years. Current manual methods are labor-intensive, prone to bias, and limit large-scale systematic scans. As data grows, this paradigm slows discovery. Large Language Models (LLMs) offer a solution. Their natural language understanding and code generation capabilities allow them to interpret scientific tasks and integrate with HEP tools (e.g., ROOT, BOSS) to act as an "AI partner" for autonomous analysis. We present Dr.Sai, an LLM-powered multi-agent system that translates natural language into rigorous physics workflows. As validation, Dr.Sai performed large-scale re-measurements of ten J/psi decay branching fractions - without manual coding. It successfully navigated the real BESIII computing environment and produced results matching established benchmarks. The article details Dr.Sai's architecture, the validation results, and performance evaluation. This work provides a blueprint for autonomous discovery, with relevance to other data-intensive fields like astronomy and genomics.

We calculate second- and selected fourth-order conserved-charge fluctuations in $(2+1)$-flavor QCD using Möbius domain-wall fermions (MDWF) along a line of constant physics. Gauge ensembles were generated for two light-to-strange quark-mass ratios, $m_l/m_s=1/10$ and $1/27.4$, corresponding to heavier-than-physical and physical pion masses, respectively. For $m_l/m_s=1/10$, calculations were carried out on lattices with temporal extents $N_τ=12$ and $16$, enabling an assessment of lattice-spacing effects at heavier pion mass. For $m_l/m_s=1/27.4$, calculations were performed at $N_τ=12$, allowing us to study the light-quark-mass dependence down to the physical point. Below the pseudocritical temperature, second-order electric-charge, strangeness, and off-diagonal conserved-charge fluctuations are consistent with QMHRG2020 hadron resonance gas calculations. Across the crossover region, these observables rise rapidly and tend toward their Stefan--Boltzmann limits. Selected fourth-order cumulants were also computed at the physical pion mass. Although these observables are statistically more demanding, several channels with controlled uncertainties permit a first comparison with hadron resonance gas calculations.

Searches for neutrino electromagnetic interactions at reactor sites require an unusual combination of ultra-low thresholds and a stable low-background environment. It is shown here that cryogenic undoped cesium iodide (CsI) naturally satisfies these conditions in a way prior detectors have not. Although suppression of nuclear recoil ionization efficiency at low energies limits the use of this scintillator for coherent elastic neutrino-nucleus scattering, that same property renders the detector effectively blind to those nuclear recoils from MeV-scale reactor antineutrinos. This leaves the low-energy regime free to expose neutrino-electron ($\barν_{e} -e^{-}$) scattering as the dominant observable channel and converts cryogenic CsI into a targeted probe of electromagnetic couplings. This work presents a conceptual design based on pure CsI crystals immersed in an active xenon-doped liquid argon veto evaluated under realistic intrinsic and environmental backgrounds. Under present detector capabilities, order-of-magnitude improvements over current reactor limits on the neutrino magnetic moment and millicharge are achievable. Cryogenic pure CsI therefore offers a distinctive and scalable route to leading studies of $\barν_{e} -e^{-}$ physics.

We consider a general reducible gauge theory deformed by mass or/and interaction terms violating gauge invariance. It is shown that in the Abelian case, by using the Stueckelberg-type procedure, this theory with broken gauge symmetry can be converted into exactly gauge-invariant theory which under a suitable choice of gauge conditions can be treated within the formalism of minimal wave operators manageable by the covariant Schwinger-DeWitt technique. We carry out quantization of such a theory in general terms when the initial generators of gauge transformations are of the first and second stages of reducibility and derive its partition function in terms of the functional integral with all corresponding ghost fields. This method is applied to quantization of massive fermionic totally antisymmetric tensor field models in $AdS$ space. One-loop quantum effective action for these models is derived in the form of the functional determinants of special Dirac-type differential operators in various dimensions.

On $T^6/(Z_3 \times Z_3)$ heterotic orbifolds where three quark generations arise from $Z_3$ fixed-point triplication, we prove that the leading-order tree-level Yukawa amplitude -- the three-point coupling among massless string states -- has an exact column texture: $Y_{\rm lead}(i,j) = c\,\varepsilon^{q_R[j]}$, with the $O(1)$ coefficient $c$ universal across all left-handed generations $i$. Five independent lines of evidence are given: (1) the worldsheet instanton geometry on the $SU(3)$ root lattice gives identical areas for all non-degenerate triangles, making the geometric $O(1)$ coefficient exactly $1$; (2) the generation direction necessarily has trivial Wilson line, rendering all three generations gauge-identical, as verified across all 77 MSSM-like models in the Mini-Landscape classification; (3) an extension to two-Wilson-line models, verified on the complete Parr-Vaudrevange-Wimmer classification of 3,337 $Z_3 \times Z_3$ MSSM models, confirms that no Wilson line configuration can break gauge blindness; (4) the Kähler metric is generation-universal by $Δ(54)$ representation theory; (5) the full Froggatt-Nielsen chain computation with 534 trilinear superpotential couplings and vacuum-aligned singlet VEVs produces left-circulant Yukawa matrices whose eigenstructure is generation-universal. The Froggatt-Nielsen column texture is therefore not an approximation but an exact property of the leading-order string amplitude. Non-trivial $O(1)$ coefficients, which are required for CKM mixing angles beyond the Wolfenstein hierarchy, must originate from beyond-leading-order contributions: integrated-out heavy messenger propagators (tree-level in the low-energy effective theory), vacuum-alignment effects, multi-instanton corrections, or loop corrections.

Solutions of the quark gap equation and the corresponding vacuum pressure are investigated within a modified Nambu-Jona-Lasinio model, which is a basic issue for studying the QCD equation of state (EOS) and the properties of hypothetical non-strange quark stars. In this study, the coupling strength $G$ is modified as $G=G_1+G_2\langle\barψψ\rangle$ to highlight the feedback effect of the quark condensate on the gluon propagator. Our analysis reveals that the influence of the vacuum pressure on EOS stiffness critically depends on whether the chiral phase transition is a first-order transition or a smooth crossover. A small ratio $G_1/G$ $(0.74\sim0.75)$ leads to a low vacuum pressure and a first-order chiral phase transition, a scenario favored by the existence of massive pulsars. Conversely, a large $G_1/G$ $(>0.96)$ leads to a high vacuum pressure and a crossover, but the corresponding EOS is ruled out by recent pulsar mass-radius observations. The model parameter space, restricted by four constraints, indicates the current quark mass is in the range $4.08\leq m\leq4.13$ MeV, with the quark condensate feedback contribution accounting for approximately 25\%. Furthermore, it is argued that the merging compact binary in GW170817 could be non-strange quark stars, and the tidal deformability is constrained to $Λ(1.4)\leq646$.

Within the framework of the one-boson-exchange model, we systematically investigate the interaction between the vector meson $K^{*}$ and the baryon $Σ^{*}$ with the aim of exploring the possibility of forming hadronic molecular states. The $K^{*}Σ^{*}$ interaction potential is constructed from $ρ$, $ω$, and $π$ meson exchanges, and the nonrelativistic Schrödinger equation is solved using the Gaussian expansion method. The binding energies are calculated for different total angular momenta $J^{P}$ and isospin channels $I=1/2$ and $I=3/2$. Our results show that $S$--$D$ wave mixed $K^{*}Σ^{*}$ molecular states with $J^{P}=1/2^{-}$ can be formed only in the $I=3/2$ channel, while no bound state appears in the $I=1/2$ channel due to destructive interference of the interaction potentials in isospin space. In addition, the $S$--$D$ wave mixed states with $J^{P}=3/2^{-}$ and $J^{P}=5/2^{-}$ are also found to support bound-state solutions. For higher partial-wave states, the binding mechanism is governed by the interplay of partial-wave mixing, tensor forces, and spin--orbit interactions. In particular, the $J^{P}=1/2^{+}$ channel does not support a bound state because the meson-exchange interaction is predominantly repulsive. Our analysis further supports the interpretation of the experimentally observed $N(2250)$ and $Δ(2200)$ states as $K^{*}Σ^{*}$ molecular states, corresponding to $I=1/2,\ J^{P}=9/2^{-}$ and $I=3/2,\ J^{P}=7/2^{-}$, respectively.

The equation of state (EoS) of strongly interacting matter at finite temperature and chemical potentials (baryon, charge, and strangeness) is a crucial input for hydrodynamic simulations of relativistic heavy-ion collisions. We construct a four-dimensional EoS using a deep-learning-assisted quasi-particle model (DLQPM) within a physics-informed neural network (PINN) framework, in which the masses of light quarks, strange quarks, and gluons are parameterized as functions of temperature and chemical potentials ($T, μ_B, μ_Q, μ_S$). The model is constrained by lattice QCD data at vanishing chemical potentials and provides a thermodynamically consistent extrapolation to finite $μ_{B,Q,S}$. The DLQPM accurately reproduces the lattice-calculated cumulants $χ^{B,Q,S}_{i,j,k}$ at $μ_{B,Q,S}=0$, and its predicted EoS at various chemical potentials agrees well with results from the generalized $T'$-expansion method in lattice QCD. Furthermore, the calculated baryon-strangeness correlation $C_{BS}$ is consistent, within uncertainties, with preliminary STAR data. This work offers a reliable EoS for exploring the QCD phase structure in the beam energy scan region.

We perform a global analysis of HERA total inclusive cross section and charm quark production data to extract the non-perturbative initial condition for the next-to-leading order Balitsky-Kovchegov (BK) equation. We extend our previous analyses to full next-to-leading order + next-to-leading logarithm (NLO+NLL) accuracy by combining the NLO DIS impact factors with the NLO BK equation that includes the resummation of large double and single logarithms. The developed Bayesian setup extracts the posterior that describes the distribution of the parameters that best fit the data. The posterior distribution allows for estimates of the theoretical uncertainty of the dipole amplitude and, hence, offers a streamlined method for propagating uncertainties in the BK initial condition in NLO CGC calculations.

We construct an $SU(5)$ Grand Unified Theory on domain walls in the five-dimensional space-time. In this setup, we introduce an adjoint scalar field and a singlet that together form a set of five domain-wall solutions, realizing a dynamical brane-world. The same scalar fields also localize chiral fermion zero modes around the walls via the Jackiw-Rebbi mechanism, break $SU(5)$ down to the Standard Model gauge group via geometric Higgs mechanism and simultaneously trap gauge fields through a field-dependent gauge kinetic term. Furthermore, they enable localization of the Higgs field, providing a novel solution to the doublet-triplet splitting problem. As a result, all essential ingredients of the model are realized by a single adjoint scalar field and a singlet, making the construction very economical. We propose two realizations of the Higgs sector, derive the four-dimensional effective theory, and demonstrate that the Standard Model Yukawa couplings at the weak scale can be reproduced from the five-dimensional Yukawa couplings by the renormalization group analysis with a suitable choice of parameters.

We consider off-diagonal asymptotic series for integral kernels of functions of Laplace-type operators on curved backgrounds. These expansions are obtained by applying integral transforms to the DeWitt series for the heat kernel of the corresponding operator and thus represent a DeWitt-type series in the heat kernel coefficients with the coefficients of this expansion (which we call basis kernels) being some hypergeometric-type functions of the Synge world function. Basis kernels of a certain class of operator functions were found previously in terms of $N$-fold Mellin-Barnes integrals. In this paper we study series representations of the corresponding Mellin-Barnes integrals in both non-resonant and resonant cases and suggest a physical interpretation for the emerging series, which is related to the UV and IR properties of operator functions.

Recent anomalies in nuclear transitions observed by the ATOMKI collaboration suggest the existence of a new boson with a mass of $\sim 17$ MeV. A theoretically consistent interpretation requires a framework that not only matches the kinematics but also reproduces the observed decay rates while satisfying stringent experimental constraints. Among various possibilities, an axial-vector or mixed vector-axial-vector mediator $Z'$ emerges as the most viable candidate. However, getting such couplings for a light $Z'$ gauge boson is highly non trivial task. In this work, we construct a gauged chiral, flavor specific $U(1)_X$ extensions of the Standard Model where the associated $Z'$ boson acts as the $17$ MeV particle. By employing a two Higgs doublet framework, we generate the necessary non-vanishing axial-vector couplings while ensuring gauge anomaly cancellation and consistent fermion mass generation. Focusing on the $^8\mathrm{Be}$ and $^4\mathrm{He}$ signals, we show that in this model the viable parameter space to resolve the ATOMKI anomalies is also consistent with a diverse set of experimental constraints, including atomic parity violation, beam dump experiments, meson decays, and neutrino nucleus and neutrino electron scatterings. Our results demonstrate that this framework offers a theoretically sound and phenomenologically robust solution to the ATOMKI anomaly.

We quantify the internal structure of near-threshold bound, virtual, and resonance states in systems where Coulomb and short-range interactions coexist by evaluating the compositeness. Using the Coulomb-modified effective range expansion, we derive an expression for the compositeness in terms of the eigenenergy and Coulomb effective range in the weak-binding limit. We then apply the formulation to several near-threshold states in hadronic and nuclear systems, including $pp$, $αα$, $Ω^{-}Ω^{-}$, $Ω_{ccc}^{++}Ω_{ccc}^{++}$, $Ξ^{-}α$, and $Ω^{-}p$.

We investigate the vacuum expectation value of a complex scalar field on a two-dimensional torus with quantized magnetic flux $M$. A characteristic feature of this system is the emergence of a critical area: when the area of the torus exceeds this critical value, the vacuum expectation value becomes nonvanishing. Furthermore, any nonzero vacuum expectation value necessarily exhibits nontrivial dependence on the coordinates of the torus. Employing the lowest-mode approximation, we find a single vacuum configuration for $M=1$, whereas two and six degenerate vacuum configurations arise for $M=2$ and $M=3$, respectively. We then analyze the symmetry properties of these vacuum configurations and determine whether they preserve or spontaneously break the symmetry of the underlying system.

We present results for the inclusive semileptonic $D_s \to X_s \ell\barν$ decay rate from lattice QCD. Chiral and continuum extrapolations are performed using gauge ensembles generated with 2+1 flavours of Möbius domain-wall fermions. Systematic errors are fully addressed including those from the integral over all possible final states. Our results are in agreement with currently available experimental data, with an error at the few-percent level.

In this article, we study loop-induced Higgs decays into gauge bosons within the framework of Radiative Natural Supersymmetry. We reproduce the one-loop MSSM calculations for the Higgs partial decay widths into a Z boson-photon pair, two photons, and two gluons, providing the corresponding analytical expressions for the scattering amplitudes. We focus on the region of parameter space that maximizes the rare decay width of the process $h \to Zγ$, and analyze the correlated predictions for the remaining Higgs decay channels. In the selected region of parameter space, the $h \to Zγ$ decay width is enhanced, reaching a maximum value of $\simeq 7.5~\mathrm{keV}$, while remaining compatible with the current ATLAS measurement. At the same time, this region satisfies current Higgs constraints from the $h \to γγ$ and $h \to gg$ channels. The diphoton mode remains close to the Standard Model expectation, with deviations at the level of $\lesssim 5\%$. The gluon channel exhibits a stronger sensitivity to the considered region of parameter space, leading to a moderate suppression of about $12\%$ in the corresponding partial width.

The Belle and Belle~II experiments have accumulated a combined data set of $1.2~\mathrm{ab}^{-1}$ of $e^+e^- \to B\bar{B}$ collisions at the $Υ(4S)$ resonance. Owing to the clean event environment and the well-constrained initial-state kinematics, these data are particularly well suited for studying channels involving missing energy from neutrinos. This includes the reinterpretation of $B^+\to K^+ν\barν$ and the first inclusive measurement of $B\to X_sν\barν$ with the missing energy directly from $B$ decays, as well as searches involving missing energy from $τ$ decays, such as $B^0\to K^{*0}τ^+τ^-$, $B^0\to K_S^{0}τ^+τ^-$, and $B^+\to K^{+}τ^+τ^-$.

We outline the analysis of three-particle interactions and decay processes for leading-order modifications in a Lorentz-violating quantum field theory.

This thesis discusses exotic 7-spheres, i.e. manifolds that are homeomorphic but not diffeomorphic to the ordinary 7-sphere, using a set of analytical and computational tools from theoretical physics. The theory of fibre bundles and instantons, together with their relation to Yang-Mills theory, are reviewed, before presenting a generalisation of self-duality to twisted self-duality. The formalism required to derive and geometrically interpret some solutions to twisted-self-duality is relevant to the main subject of this thesis: investigating the geometry of the Gromoll-Meyer sphere. Through a Kaluza-Klein ansatz, motivated by bundle-theoretic arguments, an analytic expression for a family of Riemannian metrics on the Gromoll-Meyer sphere is derived. After a detailed study of its geometric constituents, recast as quaternionic-valued objects, the metric with maximal isometry is identified. Its curvature properties are also studied and the associated energy conditions are assessed. Then, an up-to-date and broader overview on the current work concerning exotic spheres and exotic manifolds in general is offered, before focusing again on the Gromoll-Meyer sphere, but this time under the lens of differential topology. Some explicit realisations of the homeomorphism between an exotic 7-sphere and an ordinary one are discussed, together with their possible interpretations in the context of general relativity. Finally, a numerical algorithm for finding Riemannian Einstein metrics on arbitrary manifolds is presented; it is based on machine learning, and highly generalisable in many directions. The current work on implementing its application to exotic spheres is also discussed. The thesis ends with an ample discussion of possible future directions.

Solar-reflected dark matter (SRDM) consists of dark-matter particles up-scattered and accelerated by energetic electrons in the solar interior, producing a high-velocity tail that can enhance signals in direct-detection experiments, especially for MeV-scale masses. We consider an inelastic dark matter (iDM) model, in which solar scattering populates the excited state; subsequent de-excitation in terrestrial detectors releases the mass-splitting energy, substantially helping the energy release of the collision to be larger than the detector threshold. Using detailed Monte Carlo simulations, we generate the velocity and energy distributions of solar-reflected iDM over a range of dark-matter masses $m_χ$ and mass splittings $Δ$. We then compute event rates and energy depositions for current xenon and semiconductor experiments. Our results show that these experiments can place new constraints on the parameter space of MeV-scale iDM.

This paper presents the first measurement of the forward-backward asymmetry, $A_{\rm FB}$, and the flat term, $F_{H}$, that parameterise the angular distribution of the $B^+\toπ^+μ^+μ^-$ decay. The proton-proton collision dataset used in the analysis corresponds to an integrated luminosity of 9 fb$^{-1}$, collected with the LHCb experiment between 2011 and 2018. The analysis is performed in two intervals of dimuon mass squared, one above and one below the region containing the $J\mskip -3mu/\mskip -2muψ$ and $ψ(2S)$ narrow charmonium resonances. The Standard Model predictions lie within the obtained $68\%$ confidence level interval in the high-mass and within the $99\%$ interval in the low-mass region.

We study self-dual instantons of topological charge $Q=r/N$, for any natural $r$, in $SU(N)$ Yang-Mills theory on a four torus with 't Hooft twists, by embedding them into worldvolume theories of $D$-branes. To study their moduli, we construct the wrapped intersecting brane configurations dual to general constant field strength instanton backgrounds. We show that, locally, the moduli space is identified with the Higgs branch of an $N=2$ supersymmetric theory. This parameterization of the moduli space is equivalent to one recently found in field theory, but is obtained with significantly less effort and has manifest hyper-K\" ahler structure. Our hope is that combining different perspectives on instantons on the twisted torus will help understand the still unknown global structure of the moduli space for general solutions with $Q=r/N$ as well as the nature of instantons with all moduli turned on -- when some $Q<1$ and all $Q \ge 1$ instantons become space-time dependent. For integer $Q$, these are expected to match the ADHM solution in an appropriately taken infinite volume limit.

We investigate theories of flavour based on genus $g=2$ modular invariance and analyze how fermion mass hierarchies can be generated in this context, in the vicinity of invariant points or regions in moduli space where a residual symmetry is preserved. We apply this mechanism of modular proximity-induced hierarchies to the quark sector, with the vacuum expectation values of the moduli being the only sources of spontaneous breaking of the flavour and CP symmetries. We present a benchmark model where quark mass hierarchies and CP violation are explained, with mass ratios vanishing in the symmetric limit, and quark mixing is reproduced. In this model, the values of the moduli turn out to be close to special values such as $τ_1 \simeq ω$ and $τ_2 \simeq ω, i$.

We study energy-energy correlators (EECs) in many-body quantum states, focusing on the matter produced in the aftermath of heavy-ion collisions. We analyze the angular structure of EECs in the collinear limit and identify a sequence of dynamical regimes. At the largest angular separations within the small-angle regime, the observable is dominated by disconnected contributions, leading to a classical scaling determined by the collective flow of the medium. We explicitly construct this contribution for hadrons produced from a hydrodynamic medium described by boost-invariant Gubser flow, obtaining the angular dependence of the EEC analytically. We further consider azimuthal perturbations to this flow, illustrating how EECs can be used to probe anisotropies in the initial state. At smaller angular separations, connected contributions become increasingly important. We argue that in this regime the EEC is controlled by collective hydrodynamic modes. The resulting angular behavior is similar to the one identified in the EECs of heavy and large-charge states of conformal field theories. At even smaller angles, this regime is expected to match onto the structure determined by the light-ray operator product expansion, before eventually crossing over to the smallest-angle behavior characteristic of dilute hadronic matter. Altogether, these results provide a unified picture of the angular structure of EECs in many-body QCD states and suggest new observables sensitive to the properties of matter in heavy-ion collisions.

At finite temperature, the decoupling of heavy Matsubara modes allows a four-dimensional quantum field theory to be matched onto a purely spatial, three-dimensional effective field theory (EFT). This dimensional reduction is a crucial prerequisite for the precise computation of thermal observables, most prominently those related to cosmological phase transitions. In this work, we present Matchotter -- a dedicated finite-temperature module natively integrated into the Matchete package -- which automates this matching process up to one-loop order for generic Lagrangians. By adapting modern functional matching techniques to the finite-temperature formalism, Matchotter efficiently extracts the low-energy EFT directly from the thermal path integral. Furthermore, the module fully automates supersoft matching, where the temporal gauge bosons, which acquire a Debye mass during the dimensional reduction process, are integrated out. We outline the underlying architecture of the program and demonstrate its capabilities across a range of models, including the Standard Model Effective Field Theory (SMEFT).

Novel target materials with anisotropic response will play a key role in detecting low-mass dark matter in upcoming experiments. Bilayer graphene is one such material that has been proposed for the detection of sub-MeV mass dark matter particles via electronic excitations. In this work, we calculate scattering rate via a massive mediator in bilayer graphene. With an exposure as small as $\sim$ 0.5 mg-year, bilayer graphene can probe new regions of the parameter space. The anisotropic response function of bilayer graphene leads to a sidereal-day modulation in the scattering rate, depending on its orientation with respect to the Galactic dark matter wind. We find significant modulation in the scattering rate for sub-MeV mass dark matter, demonstrating bilayer graphene's promise for a future experiment. We hope that our work will motivate the community to investigate bilayer graphene as a novel target material, and that it may lead us to discover the particle nature of dark matter.

Neutrino oscillations provide crucial insights into fundamental particle physics, with two-flavor approximations effectively describing reactor and atmospheric phenomena. This paper explores the use of Physics-Informed Neural Networks (PINNs) to solve the governing differential equations for neutrino evolution in both vacuum and matter environments. We review the theoretical framework, including vacuum mixing and the Mikheyev-Smirnov-Wolfenstein (MSW) effect in matter, and demonstrate PINN implementations for vacuum and constant-density profiles. This Machine learning based approach for reactor (low-energy) and atmospheric (high-energy) neutrinos shows high precision similar to analytical solutions, with mean squared errors of the order of \(10^{-3}-10^{-4)\). Although this work focuses on solving the evolution equation in general cases, we discuss the significant potential advantages of PINNs over traditional solvers, without any mesh requirements, high-dimensionality reduction, and applicability to complex geometries, along with future extensions to three-flavor effects.

A recently developed model-independent approach to measuring the Weyl potential has shown some tensions with $Λ$CDM (Lambda Cold Dark Matter) in DES (Dark Energy Survey) Y3 data. We apply this framework to Kilo-Degree Survey (KiDS-1000) weak lensing and BOSS/2dFLenS galaxy clustering using the KiDS Cosmology Analysis Pipeline (KCAP) in two redshift bins. Without external CMB priors, both the Weyl potential and $E_G$ measurements are consistent with late time constraints. After imposing Planck18 priors, the low redshift bin remains compatible with General Relativity, whereas the high redshift bin shows a weaker Weyl potential and lower $E_G$, corresponding to a mild $1.52σ$ deviation from the $Λ$CDM cosmology. Furthermore, phenomenological modified gravity models show a mild preference for a suppressed high redshift Weyl potential, but late time data alone remain consistent with $Λ$CDM. Our results also suggest that this deviation is primarily driven by the specific high redshift CMASS sample, reflecting the well-known ``Lensing is low'' problem.

Cosmological analyses using the latest set of type Ia SNe data weakly favor an evolving dark energy (EDE) model without strongly disfavoring the standard LCDM paradigm. Nonparametric reconstructions of the expansion history may reveal signal features potentially missed by a parametric LCDM model without laying out a specific functional form for the evolution of dark energy. Information field theory (IFT) is a Bayesian framework for optimal, nonparametric reconstruction algorithms. In this work, we present charm2, the successor to charm1, a previous IFT-based code to reconstruct the cosmic energy density's redshift evolution from SNe Ia. We apply our reconstruction algorithm to the Union2.1, Pantheon+, DESY5 and DESY5-Dovekie data sets to investigate the agreement between the nonparametric reconstruction and the signal suggested by a parametric, flat LCDM model. To enable an accurate Gaussian approximation, we employ geometric variational inference, which finds a coordinate transformation through which a curved posterior gets "flattened". The redshift evolution of the energy density can then be traced on a double-logarithmic scale, which, after de-trending, is well described by a stationary Gaussian process. The nonparametric charm2 reconstructions using the Union2.1 and Pantheon+ data sets are consistent with flat LCDM signal fields. The DESY5 and DESY5-Dovekie reconstructions deviate from flat LCDM comparison fields and are compatible with an evolving dark energy signal. However, using the evidence lower bound (ELBO) measure for model selection, we find no conclusive evidence supporting a preference for non-flat-LCDM features in any of the data sets. We note that at current DESY5 noise levels, the ELBO tends to favor flat LCDM over our nonparametric model although the latter better recovers the ground truth in synthetic EDE data; a trend reversing only at ~7x lower noise covariance.

The details of the spiral structure of the Milky Way are still debated due to large uncertainties in the distance estimates obtained through the most common tracers. X-ray dust scattering rings produced by short extragalactic X-ray transients provide instead a direct method to measure the 3D distribution of interstellar clouds up to the edges of our Galaxy with a few percent precision. We report on the analysis of all the available XMM-Newton and Chandra follow-up observations of three low-latitude gamma-ray bursts: GRB 031203 ($l \sim 255°$, $b \sim -5°$), GRB 160623A ($l \sim 84°$, $b \sim -3°$), and GRB 221009A ($l \sim 53°$, $b \sim 4°$). The previous detection of X-ray rings in these observations, produced by dust clouds located beyond 5 kpc, can be associated with dust in the Perseus, Outer, and Outer Scutum-Centaurus arms, thus providing direct distance measurements to these structures along three distinct lines of sight. We have identified two additional rings in the direction of GRB 160623A, produced by dusty clouds at $6.91\, \pm\,0.06$ kpc and $9.9\,\pm\,0.6$ kpc, and confirmed -- through a second XMM-Newton observation -- the presence of one cloud at $9.7\,\pm\,0.4$ kpc toward GRB 031203. We also accurately measured the distance of dusty clouds up to $19.0\,\pm\,0.2$ kpc owing to the analysis of one Chandra and four XMM-Newton observations of GRB 221009A. The small statistical and systematic uncertainties of these measurements place tight constraints on the geometry of the outer Milky Way and reveal significant deviations from current models, which critically depend on spectroscopy-based Galactic rotation curves at large distances.

Vision-language models (VLMs) are increasingly proposed as general-purpose tools for scientific data interpretation, yet their reliability on real astronomical observations across diverse modalities remains untested. We present AstroVLBench, a comprehensive benchmark comprising over 4,100 expert-verified instances across five tasks spanning optical imaging, radio interferometry, multi-wavelength photometry, time-domain light curves, and optical spectroscopy. Evaluating six frontier models, we find that performance is strongly modality-dependent: while one model (Gemini 3 Pro) emerges as the most consistently capable across tasks, task-specific strengths vary, and all models substantially underperform domain-specialized methods. Mechanistic ablations reveal that performance depends not only on directing attention to salient visual features but also on grounding those features in physical knowledge. Phenomenological prompts describing what to look for improve accuracy by sharpening model focus, but physical prompts explaining why those features matter perform better overall and yield more balanced classifications with reduced class-specific bias. Consistent with this picture, presenting the underlying one-dimensional measurements directly as numerical tables instead of rendered plots yields up to 13 percentage points improvement. Reasoning quality analysis further demonstrates that, without explicit physical grounding, models may reach correct predictions from phenomenologically plausible cues while providing physically imprecise justifications, establishing that accuracy alone is insufficient for trustworthy scientific deployment. These findings provide the first systematic, multi-modal baselines for VLMs in observational astronomy and identify the specific representation, grounding, and reasoning bottlenecks where current models fail.

Context. High-resolution X-ray spectroscopy offers a powerful tool to investigate the physical and chemical properties of dust grains, especially through the analysis of absorption edges of elements such as oxygen, magnesium, silicon, and iron, which are the main constituents of interstellar dust. In all previous X-ray studies, these absorption edges have been modeled assuming the MRN grain size distribution. This model successfully reproduces the average interstellar extinction curve. However, with the advent of new observations, it shows important limitations, indicating that more complex grain-size distributions are required to fully describe interstellar dust properties. Aims. We aim to constrain the composition and size distribution of interstellar dust along the line of sight to the bright low-mass X-ray binary GX 13+1. Methods. We analyzed high-resolution X-ray spectra obtained with the Chandra HETG instrument (MEG+1 and MEG-3) and simultaneously modeled the Si K and Mg K absorption edges. For the first time, we compared the classical Mathis et al. 1977, ApJ, 217, 425 grain size distribution with other grain size distributions, thus exploring different ISM densities. Results. Our analysis rules out scenarios of both very diffuse and very dense ISM, favoring grain size distributions associated with average Galactic conditions along this line of sight. The dust composition is found to be dominated by amorphous olivine and the crystallinity contribution is about 2%. The depletion patterns and elemental abundances derived are consistent with prior X-ray and infrared studies.

In recent years, several aromatic molecules (benzene-based rings) have been detected in the cold molecular cloud TMC-1, with its CN-derivative, benzonitrile (c-C$_6$H$_5$CN), also identified in other nearby cold sources. However, observed abundances differ significantly from chemical model predictions, indicating an incomplete understanding of its chemistry and motivating searches in distinct environments. We report new detections of benzonitrile in two warmer molecular clouds of the Central Molecular Zone (CMZ): G+0.693-0.027 and G+0.633-0.0604. Using Yebes 40m ultra-deep surveys in the 31--50 GHz range, we performed LTE and non-LTE analyses to derive the physical parameters of the emission. We obtain column densities of $N$=(7.4$\pm$0.5)$\times10^{12}$ and (2.60$\pm$0.13)$\times10^{12}$ cm$^{-2}$, corresponding to abundances relative to H$_2$ of (6$\pm$1)$\times10^{-11}$ and (4.3$\pm$0.9)$\times10^{-11}$, consistent with values in cold Galactic clouds. The HC$_7$N/benzonitrile ratio is lower (2.15-2.4) than in colder sources (4.5-30), suggesting environmental effects and a relative enhancement of aromatic chemistry in the CMZ. These results confirm that benzonitrile is widespread and can survive in harsher environments (e.g., high temperatures, shocks, enhanced cosmic-ray ionization) than those in Galactic cold clouds. This suggests that aromatics are stable and abundant species that can significantly contribute to the total budget of interstellar carbon in molecular clouds. A top-down formation scenario, involving fragmentation of larger carbonaceous species, is consistent with the nearly constant abundances observed with molecular size.

The origin and structure of the hot ($T\gtrsim10^6$ K) gaseous halo around Milky Way (MW)-mass galaxies provide a critical test for galaxy formation models. We perform a comprehensive comparison for a sample of MW analogues from the TNG50 cosmological simulation by generating synthetic soft X-ray emission and O VII/O VIII absorption lines. The simulated halos successfully reproduce the observed global soft X-ray luminosity and key inner-halo properties, including X-ray surface brightness, emission measure, and O VII absorption strength. However, two critical discrepancies are identified: (i) the simulated X-ray surface brightness profiles decline too steeply at large radii compared to extended eROSITA stacking emission, and (ii) the halos systematically underproduce O VIII absorption, indicating a deficit of hotter-phase gas at $T\sim(1.6-3.2)\times10^{6}$ K. These findings indicate that the simulated halos are spatially too compact and lack the hotter gas phase observed in real galaxies. This suggests that the feedback model in TNG50, while generating hot gas, may deposit its energy too centrally and too vigorously to sustain a gently extended, multi-phase corona.

Context. Convolutional neural networks (CNNs) are widely used for automated galaxy morphological classification in large surveys. However, projection effects, image artefacts, and intrinsic degeneracies limit reliable identification of detailed features, requiring large-scale visual validation. Aims. To visually inspect SDSS galaxies at 0.02 < z < 0.1 classified by a CNN as merging, irregular, edge-on, barred, ringed, or dust-lane galaxies; assess CNN completeness and failure modes; construct visually verified morphological catalogues; and determine nuclear activity types via BPT diagrams. Methods. We visually inspected all galaxies assigned by the CNN to six morphological classes: merging (2,574), irregular (9,432), edge-on (17,000), barred (6,000), ringed (13,882), and dust-lane (588), regardless of CNN probability. Refined samples were cross-matched with Galaxy Zoo 2; remaining galaxies were classified here for the first time. Nuclear activity was determined from SDSS DR17 spectra using Hααα, H\b{eta}β\b{eta}, [O III]λ$5007, and [N II] λ$6583 line ratios. Results. We present catalogues of 612 merging, 9,372 irregular, 16,822 edge-on, 575 dust-lane, 811 barred, and 2,150 ringed galaxies. CNN misclassifications stem primarily from projection effects, foreground stars, faint tidal features, and irregular star-forming structures. We characterise nuclear activity types for edge-on, barred, ringed, and dust-lane galaxies, finding systematic differences in LINER-like and composite fractions across subsamples. Five strong polar ring galaxy candidates were identified. Conclusions. Visual validation remains essential for refining CNN-based classifications. The resulting datasets support morphological studies, investigations of galaxy structure and secular evolution, and provide robust training samples for future machine learning models.

This study explores the density profile of the stellar disk, radially and azimuthally, based on approximately 8.4 million red clump stars selected from Gaia Bp/Rp spectra. After correcting for selection effects and distance uncertainties, we fit the vertical stellar density profile of the Galactic disk with a two-component model consisting of geometrically thin and thick disks. Our derived density profile shows several breaks radially: (1) a steep exponential inside R$\sim3$ kpc; (2) a nearly flat plateau from R$\sim3$ to $\sim7$ kpc; (3) an exponential decline beyond the solar radius to around 13 kpc; (4) a sharper exponential drop-off beyond R$\sim$13 kpc. The parameters of these four main components depend on $φ$ to some extent. Variation of the termination radius of the first component suggests an interaction with the bar/bulge. Besides the typical flaring at $R>6.4$ kpc, we find that the thin disk also exhibits a similar and smooth thickening/flaring feature toward the Galactic center at $R<6.4$ kpc. The observed inner flaring may indicate heating effects introduced by the Galactic bar, since $R=6.4$ kpc lies close to the co-rotation radius where the bar's dynamical influence becomes significant. Additionally, we identify a localized density bump in the region $5<R<7$ kpc and $-30^\circ<φ<15^\circ$, where a corresponding metallicity bump is also visible near the Galactic plane. This density/metallicity bump may be related to the recently reported bimodal distribution of the guiding radius of super metal-rich stars in the solar vicinity through radial migration.

The upcoming imaging survey of the Chinese Space-station Survey Telescope (CSST) will deliver high-resolution imaging of an unprecedented number of galaxies for galaxy studies. To understand CSST's capability, and to support the preparation of early-science programs, we generate 470,526 mock CSST images for 22,406 simulated galaxies with $M_*>10^9 M_\odot$, whose parameters are calibrated to match real HST observations spanning photometric redshift $0<z\lesssim7$, across seven CSST filters and three planned survey depths: wide, deep, and extreme. We then perform source detection and Sérsic fitting. For point sources, we found that the 95% completeness magnitude in the g band reaches 26.3, 27.4, and 28.5 mag for the wide, deep, and extreme fields, respectively. For extended galaxies, their spatial extent dilutes the surface brightness, leading to brighter 95% completeness magnitudes of 24.4, 25.9, and 27.1 mag. The detection completeness remains above 95% at $z\lesssim3-4$ in the extreme field, while the corresponding redshift limits are $z\approx1$ in the deep field and $z\approx0.5$ in the wide field. Using three fitting codes, GALFIT, AstroPhot, and SourceXtractor++, we quantify measurement biases and uncertainties in galaxy magnitude ($m$), effective radius ($R_e$), effective surface brightness ($μ_e$), Sérsic index ($n$), and axis ratio ($q$). On average, for fainter galaxies, the reduced signal-to-noise ratio leads to systematic overestimates in $m$, $R_e$, and $μ_e$, and underestimates in $n$ and $q$. These biases, as well as the associated scatter, become progressively smaller in deeper fields. Overall, our results provide quantitative constraints on sample selection and the robustness of morphological measurements in CSST early-science and legacy surveys.

Uncertainties in the radial distribution of galaxies, $\boldsymbol{n}(\boldsymbol{z})$, are one of the major contributions to the error budget of early Stage-IV galaxy survey analyses of weak gravitational lensing, galaxy clustering and galaxy-galaxy lensing (3$\times$2-pt). Based on ensembles of simulated $\boldsymbol{n}(\boldsymbol{z})$ including stochastic and systematic variations, we study the impact of four different $\boldsymbol{n}(\boldsymbol{z})$ uncertainty models: shifts, shifts & stretches, Gaussian processes (GP) and principal component analysis (PCA). Due to the high dimensionality of the latter models, we make use of state-of-the-art gradient-based inference methods as well as approximate analytical marginalisation schemes. Our results show that Stage-IV 3$\times$2-pt analyses must go beyond simple shift & stretch models. In particular, we advocate for the adoption of PCA models even in early Stage-IV surveys. Our results show that considering a five-parameters PCA model only degrades the constraint on the $S_{\rm 8}$ parameter by $5$ per cent with respect to the case when only a shift and a stretch parameter are included, while incurring half the bias in its constituents parameters, $Ω_{\rm m}$ and $σ_{\rm 8}$. We demonstrate that all models considered can be safely marginalised analytically, with speed-ups of up to a factor of 25 depending on the dimensionality of the model. This will allow Stage-IV analyses to safely include higher-dimensional $\boldsymbol{n}(\boldsymbol{z})$ uncertainty models in their analysis at negligible additional computational cost.

TIFR-ARIES Near-Infrared Spectrometer (TANSPEC) is a spectrograph-cum-imager operating over the wavelength range $0.55 - 2.5~μ$m. The instrument is mounted on the 3.6-m Devasthal Optical Telescope (3.6-m DOT). It offers two resolution modes: Low Resolution (LR) with $R\sim100-350$ and Cross-Dispersed (XD) via various slits of different widths (0.5", 0.75", 1.0", 1.5", 2.0" and 4.0"). The LR mode provides a resolving power ($R$) of $\sim 100-350$, while the XD mode achieves $R\sim2500$ using the 0.5" slit. The previous version of the data reduction pipeline supported only wavelength-calibrated XD mode spectra and was limited to two slits (S-0.5 and S-1.0). In this work, we present an upgraded version of pyTANSPEC. The upgraded pipeline not only improves the data extraction algorithm but also introduces several new features for users. It now enables the reduction of spectra from all available slits for both LR and XD modes. The upgraded version also implements a template-matching method for more precise wavelength calibration. Additionally, a step for flux calibration is also included. Alongside pyTANSPEC, we upgraded HxRGproc, a Python package for cleaning and generating slope images from Non-Destructive Readout (NDR) frames taken with H1RG and H2RG detectors. The package performs non-linearity correction, flags saturated pixels, removes pink noise, and eliminates cosmic ray events. HxRGproc is updated to work for the H2RG detector of TANSPEC and is set up on the TANSPEC server, ensuring users receive data that are pre-cleaned and non-linearity corrected.

The identification of open cluster (OC) members has been revolutionized by high-precision Gaia astrometry, yet traditional kinematic membership selections remain inherently conservative, often overlooking stars in tidal tails or those with perturbed velocities. This study investigates the reliability of these kinematic probabilities by searching for leaky cluster members -- stars that fail standard kinematic membership criteria ($P < 0.7$) but possess chemical signatures identical to their host clusters. Using high-resolution spectroscopic data from the Gaia-ESO Survey, we established a seven-element chemical fingerprint ([Fe/H], Li, Si, Ca, Ti, Co, and Ni) for 34 OCs. We identified a sample of 63 stars across 22 clusters that are chemically indistinguishable from their host populations despite being kinematically rejected by standard algorithms. By cross-referencing these targets with Jacobi radii (rJ) derived from modern Milky Way potential models, we find that 35% are located in extra-tidal regions (d > rJ), providing direct evidence of active cluster dissolution and tidal debris. The remaining 65% are located within the Jacobi radius, suggesting that their kinematic rejection is likely due to orbital motion in unresolved binary systems. These results demonstrate that chemical tagging is a critical tool for overcoming the spatial and kinematic biases of astrometric catalogues. By recovering these lost members, we provide a more complete census of cluster mass loss and underscore the necessity of a hybrid chemical-kinematic approach to map the transition of stars from bound systems to the Galactic field.

We describe the public release of $>2.3$ petabytes of data from the FLAMINGO cosmological simulations. The suite consists of hydrodynamical simulations that include radiative cooling, star formation, stellar mass loss and the resulting chemical enrichment, supernova feedback, and two implementations of AGN feedback. Neutrinos are simulated explicitly using particles. Data products include snapshots, halo and galaxy catalogues, HEALPix all-sky lightcone maps, particle data for lightcone maps, and power spectra. The FLAMINGO set includes 22 hydrodynamical simulations. In addition, there are 16 gravity-only simulations, including the $10080^3$ particles FLAMINGO-10k run, with initial conditions that match those of the corresponding hydrodynamical runs. The fiducial hydrodynamical simulations span three numerical resolutions that have each been calibrated to reproduce the present-day galaxy stellar mass function and gas fractions in low-redshift clusters. Other simulations systematically vary the galaxy stellar mass function, cluster gas fractions, cosmology (including neutrino masses), and/or the nature of dark matter, in volumes of 1Gpc$^3$. The release includes hitherto unpublished simulations that use extra dark matter particles. While we provide a facility for downloading complete simulation outputs, we recognise that for many users this will not be possible due to limited local storage or network bandwidth. We implement a web service that enables users to explore available outputs and selectively download datasets or parts of datasets.

The study of the prompt high-energy emission of Gamma-Ray Bursts (GRBs) with SVOM relies on the observations performed by ECLAIRs (4-150 keV) and the Gamma-Ray Monitor (GRM, 0.015-5 MeV), the two wide field-of-view instruments on board the satellite. In this article, we introduce the eclgrm pipelines running at the French Science Center of SVOM, which combine the ECLAIRs and GRM data to generate scientific data products describing the GRB broad-band temporal and spectral properties. The architecture of the pipelines is presented, as well as their activation following each onboard trigger, and their workflow. The statistical data analysis methods employed by the pipelines are described, along with the scientific data products that are created in real time or from the full event data. We also present the eclgrm-ui user interface which allows the scientists on shift to monitor the automated data processings in the pipelines, and to optimize the analysis results interactively.

The Chinese Ground Follow-up Telescope (C-GFT) is an optical facility upgraded to support the Space Variable Objects Monitor mission (\textit{SVOM}). Located at the Jilin Observation Station, it is capable of rapidly identifying and monitoring the optical counterparts of Gamma-Ray Bursts (GRBs). The 1.2-m telescope is equipped with two switchable focal-plane instruments: the prime-focus wide-field LATIOS camera and the Cassegrain-focus three-channel CATCH camera. In this paper, we present a system overview, including the observatory, the telescope, the instrumentation, the automated operational framework managed by the Operations Center, and the data processing pipelines. We also report the performance results obtained during over one year of \textit{SVOM}'s post-launch operations. The results demonstrate that the system meets its design specifications and delivers robust observational and operational performance.

The VT--VHF data comprise three types of onboard-processed data results generated from four sequential observational sequences and transmitted to the ground via a Very High Frequency (VHF) downlink. On the ground, these data are processed by three successive pipelines: the pre-processing pipeline, the VT--VHF data processing pipeline (VVPP), and the VT afterglow candidate pipeline (VTAC). These pipelines perform packet decoding, astrometric and photometric calibration, and afterglow candidate identification, respectively. This paper describes the architecture and operational implementation of the VT--VHF ground processing system and assesses its end-to-end performance using the first year of SVOM operations. These data enable rapid identification of GRB optical counterparts. Early detections, while the source is still optically bright, facilitate spectroscopic redshift measurements. Dual-band colors provide preliminary redshift constraints and help identify high-redshift candidates, whereas non-detections in both bands may indicate very high redshift, significant extinction, or intrinsically dark bursts. In-orbit operations show that the VT--VHF ground processing system successfully identifies optical afterglow candidates for a significant fraction of ECLAIRs triggers with available VT--VHF data, demonstrating its robustness and readiness.

The SVOM Visible Telescope (VT) is critical for the rapid identification of gamma-ray burst (GRB) optical counterparts, particularly for high-redshift candidates that require immediate infrared spectroscopic follow-up. To address the stringent bandwidth constraints of the VHF downlink while ensuring real-time data availability, we developed the VT Onboard Data Processing Pipeline (VOPP).This paper details the software architecture, algorithms, and hardware implementation of VOPP using an FPGA and a CPU. The pipeline performs essential real-time tasks, including image quality assessment, dark and flat-field correction, and optimized image stacking to mitigate cosmic ray contamination and variable background noise. Furthermore, it generates compact source catalogs and highly compressed 1-bit images to facilitate rapid downlink.In-flight performance analysis confirms the pipeline's robustness, demonstrating the availability of VT VHF data for 78 percent of promptly slewed SVOM GRBs, with 56 percent leading to the identification of optical counterparts, typically within 18 minutes post-trigger.

The Gamma-Ray Monitor (GRM) is a key payload of the Space-based multiband astronomical Variable Objects Monitor (SVOM) mission, which is designed to detect gamma ray bursts (GRBs) within the energy range of 15 keV to 5 MeV. The GRM Instrument Center (GRM\_IC) features real-time data processing through the X-band, enabling rapid response of high-energy GRB events. The system employs an event-driven architecture and distributed design, achieving efficient processing and real-time monitoring of massive observational data. Through comprehensive data production processes and scientific data product management, the system achieves efficient production of scientific data products of the L1B / C level through the submission of jobs to the task scheduling system. Through modular architecture design and automated processing workflow, the GRM data processing system realizes precise conversion and scientific analysis of GRB detection data, providing robust technical support for future system upgrades and cross-platform collaboration.

VT (the Visible Telescope) is an optical telescope onboard the SVOM (Space-based Multi-band Astronomical Variable Objects Monitor) mission, specifically designed to detect optical counterparts of gamma-ray bursts (GRBs), study their afterglows, and select high-redshift candidates. It performs rapid follow-up observations simultaneously in two channels either via autonomous platform slewing or Target of Opportunity (ToO) observations. The science images acquired by VT and transmitted via the X-band downlink system are designated as VT X-band data. This paper provides an overview of GRB optical afterglow identifications with VT and describes the ground-based processing pipeline for VT X-band data, including preprocessing, astrometric calibration, and photometry. Up to 2025 December 3, VT has followed up 111 GRBs triggered by SVOM or external missions. The overall detection rate of optical counterparts is approximately 75%. Specifically, for bursts detected by SVOM/ECLAIRs, the detection rate is 77% when observed by VT within 30 minutes after the burst. A slightly higher detection rate of 81% is achieved for GRBs triggered by external missions through rapid ToO observations with a mid-time of less than 3 hours.

The SVOM mission (Space-based multi-band astronomical Variable Objects Monitor) is a Franco-Chinese mission dedicated to the study of the most distant explosions of stars, the gamma-ray bursts. Here, we introduce the real-time response and collaboration system of SVOM, with the adoption of the BeiDou-3 short message communication service. We present the SVOM on-board and on-ground system designs and data flow, together with the collaboration mechanism with other missions. In the first year of the in-flight operation, SVOM has detected 172 gamma-ray bursts, including 147 by the GRM instrument and 62 by the ECLAIRs instrument. At the same time, SVOM has performed 1040 observations, including 122 ToO-EX(Target of Opportunity-Exceptional) observations, 48 ToO-MM(Target of Opportunity-Multi-messenger) observations and 870 ToO-NOM(Target of Opportunity-Nominal) observations. All these have increased the scientific output of the mission.

The Ground Wide Angle Camera Network (GWAC-N) is a robotic telescope network. It consists of ten wide-field core telescopes (GWAC-A) and two 60cm narrow-field rapid follow-up telescopes (GWAC-F60A/B). The primary scientific goal of GWAC-N is to detect optical counterparts of gamma-ray bursts (GRBs) discovered by the SVOM satellite. This is achieved through synchronized monitoring with the GWAC-A array. Upon receiving a GRB trigger alert, the network conducts rapid, multi-band follow-up observations using the GWAC-F60A/B telescopes. The two-stage observation process involves many telescopes, making manual control impractical. Automated operations are therefore essential. They are realized through an integrated alert processing chain and an automated observation scheduling and dispatching mechanism. To enable this, we employ the SVOM Follow-up Observation Coordinating Service (FOCS) and GWAC-N's Automatic Observation Management (AOM) system. This paper presents the integrated alert processing workflow. It also describes the formulation of observation strategies, and the scheduling and execution of observations enabled by FOCS and AOM.

On-board the SVOM mission, the Microchannel X-ray Telescope observes the soft X-ray band of the gamma-ray bursts afterglows. The so-called lobster-eye optics focuses X-rays to the camera subsystem that performs imaging and spectroscopy of a region of the sky 58x58 arcmin2 wide centered on the burst detected by the ECLAIRs instrument. The recorded photon positions are used by the on-board scientific software to rapidly localize the source, whereas spectral information is used on ground to model the properties of the gamma-ray bursts. The first months in orbit were intensively used to tune the parameter settings of the detector and the calibration method to provide high availability of the camera and accurate spectroscopy to the users. The paper presents the design of the camera validated by on-ground testing, the tuning phase in flight and the performance of the camera at the beginning of the mission. Perspectives are given concerning the evolution of the spectral response during the mission.

ECLAIRs is a hard X-ray coded-mask telescope onboard the SVOM space mission, designed to detect and localize high-energy transients, in particular gamma-ray bursts. Operating over the 4-150 keV energy range, ECLAIRs extends coded-mask imaging to an unusually low-energy threshold. Achieving sensitivity down to 4 keV while maintaining performance up to 150 keV motivated the development of a novel self-supporting coded mask. This design addresses both scientific and mechanical challenges through dedicated pattern-generation algorithms and an innovative stiffened sandwich structure. We present the rationale, development, and final implementation of the ECLAIRs coded mask.

As one of the key ground-based facilities of the Chinese-French SVOM mission, the main scientific objectives of the Ground-based Wide Angle Camera array (GWAC) are to detect prompt optical emission of gamma-ray bursts or other short duration astronomical transients on a second-scale temporal resolution. GWAC is located at Xinglong observatory, China, and consists of 10 mounts and 40 cameras, providing a joint field of view of about 3600 square degrees.The detection ability is 16 magnitude in 10 seconds of exposure time in the visual band under the condition of the new moon phase. Here, we give an overview of GWAC and introduce the science motivation of the project, as well as the performance of the hardware and the software. The observation strategies and the data processing are briefly presented. The early sciences in the last 5 years since the first light are summarized.

COLIBRI, the French Mexican Ground Followup Telescope (FM GFT) for SVOM, is a 1.3 meter rapid response optical facility specifically developed for prompt, multiband observations of GRB afterglows and for delivering subarcsecond localisations of optical counterparts for detailed followup studies. The telescope operates through a fully automated system that manages the entire workflow, from alert reception to counterpart identification. Commissioning results confirm that the telescope meets design specifications, and this paper presents a comprehensive performance assessment of the capabilities.

We present the in-orbit calibration of the Visible Telescope (VT), one of the key instruments aboard the Space Variable Objects Monitor (SVOM) mission for gamma-ray burst (GRB) studies. Using Gaia Data Release 3 (DR3) as a reference, the VT achieves an astrometric precision better than 0.03'' for bright stars, degrading to ~0.25'' for faint targets. Shortly after launch, contamination was detected, reducing system transmission by ~40%. An initial bake-out successfully restored performance, but gradual recontamination caused transmission to decline by ~20% over the following 100 days before stabilizing. Despite this effect, routine standard star observations maintain precise zero-point calibration, ensuring a photometric stability of 0.02 mag. Using synthetic stellar spectra, we derived photometric transformations to the Gaia, SDSS, and Johnson-Cousins systems with typical residuals of 0.03 mag. These results demonstrate the VT system's capability and reliability in calibrating GRBs and other transients.

The SVOM (Space-based Variable Objects Monitor) mission, launched into low Earth orbit on 22 June 2024, is a French-Chinese multi-wavelength observatory dedicated to the study of the transient sky. Inspired by the Neil Gehrels Swift Observatory, it consists of an autonomous rapid-slewing satellite, linked in real time to several ground-based telescopes. The space segment comprises two X-ray/gamma-ray wide-field instruments (ECLAIRs and GRM) with real-time triggering capabilities combined with two narrow-field telescopes in X-ray (MXT) and in visible (VT). In addition, the SVOM collaboration has also developed a unique visible and NIR ground-based follow-up system to promptly respond to the gamma-ray transients detected on board. The core program of SVOM will provide new insights into the Gamma-Ray Burst physics by providing a homogeneous dataset covering both the prompt and afterglow emissions, as well as better studying the low luminosity and soft Gamma-Ray Burst populations. As a versatile satellite platform with fast slewing capabilities, SVOM also comprises a Target of Opportunity program and a General Program consisting in pointed observations scheduled over the year that will both significantly contribute to the multi-messenger and time-domain astronomy.

This paper reports on the scientific pipeline for the analysis of the ECLAIRs data of the SVOM mission. We describe the overall procedure, the different steps and the main algorithms of the data analysis for this hard X-ray coded mask instrument. The pipeline runs in automatic mode at the science center generating standard products but can also be used off-line with specific choices of parameters. Generated data products and preliminary performances are illustrated along with perspectives for future improvements.

The French-Chinese SVOM satellite mission (Space-based multi-band astronomical Variable Objects Monitor) was launched in mid-2024, with science objectives focused on the detection and study of astrophysical transient events, primarily Gamma-Ray Bursts (GRBs). The onboard trigger of the hard X-ray wide-field coded-mask instrument ECLAIRs autonomously detects and localizes GRBs on SVOM and requests automatic spacecraft slews toward these sources, enabling follow-up observations by the onboard narrow field-of-view instruments MXT and VT. The trigger also transmits real-time alerts via the SVOM VHF network to the ground, allowing rapid follow-up campaigns by the broader community, including multiple space- and ground-based facilities. We present an overview of the ECLAIRs trigger, with emphasis on the two trigger algorithms running simultaneously onboard: the Count-Rate Trigger (CRT) and the Image Trigger (IMT), both of which issue alerts for GRBs localized on the sky. The trigger has already detected several notable events, including both classical GRBs and peculiar X-ray-rich GRBs, enabling numerous redshift measurements, including high-redshift bursts.

The detection plane of the high-energy transient camera ECLAIRs onboard SVOM is made of 200 XRDPIX detection modules, each consisting of a matrix of $8\times 4$ Schottky-type CdTe detectors hybridized with the low-noise and low-consumption ASIC IDeF-X. The 6400 detectors are operated at -20°C and reverse biased at -300 V, reaching a low energy threshold of 4 keV. The 20 us time resolution readout electronics works in photon counting mode classifying detected events as single or multiple events and measuring the deposited energy in real time. In this paper, we present the detection plane sub-systems and their main features. We also discuss its overall performances as measured both on ground and in-flight, showing compliance with the ECLAIRs science requirements.

The SVOM mission, a Sino-French collaboration dedicated to Gamma-Ray Bursts (GRBs) and transient sources, began scientific operations in 2025. This paper describes the ground computing infrastructure and user support tools for SVOM's three observing programs: the Core Program (CP), the General Program (GP), and Targets of Opportunity Program (ToO), the latter two being open to the broader scientific community, provided they collaborate with a mission Co-I. The mission adopts operational roles inspired by Swift, including on-duty scientists such as Burst Advocates (BAs), who validate GRB triggers and coordinate follow-up observations, and Instrument Scientists (IS), who calibrate and validate data for all programs. Users can access observation schedules, public data products, and support tools via the French and Chinese mission centers. The SVOM portal serves as the primary interface for accessing these resources, including a GRB public table, API, and user documentation. This paper serves as a guide for both newcomers and external researchers interested in SVOM scientific operations, focusing on aspects related to the CP.

The Chinese-French SVOM (Space-based Multi-band Astronomical Variable Objects Monitor) mission is dedicated to the study of gamma-ray bursts (GRBs) from the distant universe. A key component of the SVOM Chinese Ground Segment, the Science User Support Services (SUSS) provides comprehensive support for the mission's scientific operations. SUSS consists of two integral pillars: a suite of specialized software tools that automate key workflows, and a dedicated User Support Team that delivers expert-led, human services. These human-delivered services include operational coordination across telescope networks, direct technical assistance to astronomers, user training, and proactive problem-solving throughout the observation lifecycle. This paper focuses on the organization of SVOM scientific operations and the role of SUSS in facilitating these tasks. We provide a detailed description of the SUSS software architecture and its functionalities, encompassing the General Platform, the Burst Advocate (BA) support tools for GRB counterpart identification, the Target of Opportunity (ToO) support tools, and the General Program (GP) support tools. The structure and services provided by the user support team at the Chinese Science Center (CSC) are also elaborated. Furthermore, we evaluate the performance of SUSS during its first operational year, assessing its effectiveness in fulfilling user requirements. The evaluation offers valuable insights to guide future user support strategies and software enhancements, ultimately enabling better service for the SVOM scientific community.

The core instrument of the SVOM Gamma-ray burst mission launched in June 2024 is the 4-150 keV 2-D coded mask camera ECLAIRs responsible for the autonomous trigger and localization of transient events within its field of view. The flight model of ECLAIRs has been built by several French labs (IRAP, CEA, APC) under the supervision of the French Space Agency (CNES), while APC, LUPM and IAP built a suite of data reduction and analysis software. This paper outlines the main science goals of ECLAIRs and describes the different instrument sub-systems and their main characteristics. The paper then discusses the instrument configuration and operation as well as the main in-flight measured performances. Finally, the paper summarizes the science performance of ECLAIRs up to March 31, 2025.

The 44-cm Visible Telescope (VT) aboard the Space-based Variable Objects Monitor (SVOM) is a dual-band (400-650 nm and 650-1000 nm) instrument designed to detect and characterize the optical counterparts of gamma-ray bursts (GRBs) and other high-energy transients. This paper presents the VT's design, scientific objectives, observing strategies, and both space- and ground-based data processing pipelines, along with its first-year in-orbit performance. In-orbit commissioning tests confirm a sensitivity of 22.5 AB mag (300 s exposure), extendable to $\sim\!24$ AB mag through stacking. This performance enables the VT to monitor over 100 GRBs in its first year with an exceptional $\sim\!80\%$ detection rate for \textit{SVOM}/ECLAIRS-triggered bursts and ToO-observed bursts from other missions (e.g., \textit{Swift, Fermi, Einstein Probe (EP)}), outperforming \textit{Swift}/UVOT's $\sim\!40\%$ detection rate. Beyond its exceptional detection efficiency, the VT played a key role in identifying high-redshift GRBs-most notably GRB 250314A (z = 7.3). Its deep upper limits at long wavelengths (up to 1 $μ$m) were pivotal in guiding follow-up observations with large ground-based telescopes, enabling crucial near-infrared (NIR) detections. With its rapid response, deep sensitivity, and real-time processing capabilities, the VT is a key instrument for GRB research in \textit{SVOM}-era, enabling critical studies of GRB optical afterglows, circumburst environments, relativistic jet dynamics, and the origins of optically dark bursts.

The Space-based multi-band astronomical Variable Objects Monitor (SVOM) mission was launched in June 2024. It is a joint Sino-French collaboration designed to detect, localize, and study gammaray bursts (GRBs) and other high-energy transients. Among its onboard instruments, the Microchannel Xray Telescope (MXT) plays a central role by providing follow-up X-ray observations of GRB afterglows and other transient phenomena. To ensure timely and accurate scientific exploitation of MXT observations, a dedicated ground processing pipeline has been developed. This pipeline automatically ingests raw event lists, performs calibration, background and time filtering, corrects instrumental effects, and produces science-ready data products such as images and light curves and spectra of detected sources. In this paper, we describe the architecture and key components of the MXT data analysis pipeline, highlighting its modular design and integration within the broader SVOM ground segment. We also show results from real datasets, demonstrating the pipeline's ability to meet the performance requirements of the mission.

The Microchannel X-ray Telescope (MXT) is a compact and lightweight focusing X-ray telescope, which is part of the space payload of the SVOM mission. The main goal of the MXT instrument is to precisely localize and physically characterize the early phases of the X-ray afterglows detected by the SVOM ECLAIRs coded mask telescope after a satellite slew. The MXT is composed by a "Lobster-Eye" type optics, with a 58$\times$58 arcmin$^{2}$ field of view, based on micro-pores of 40 $μ$m side. This innovative type of optics is coupled to an X-ray camera, which implements at its focal plane a low-noise pnCCD. The MXT system is completed by an onboard calculator, able to command the whole telescope and to analyze in real time the MXT data stream and hence to localize the sources within the MXT field of view. In this paper, we present the MXT design and in-flight performance, as measured during the SVOM Commissioning and early science operation phase. In particular, we will focus on the optical and spectral performances, the in flight localization capabilities, and how these compare with the pre-flight ground measurements.

Galaxy clusters are located at the nodes of the filamentary network known as the cosmic web. A more comprehensive understanding of galaxy clusters can be achieved by considering their environment, in particular, the filamentary structures to which they are connected. In this work, we aim to assess the reliability of the cosmic web reconstruction from mock observational data. In particular, we aim to quantify the effects of the 2D projection relative to the underlying 3D network and the impact of using the Sunyaev-Zel'dovich (SZ) effect as a tracer of the cosmic web. We reconstruct the filamentary networks in the outskirts of The Three Hundred simulated clusters with the filament finder DisPerSE. First, we extract the networks from the 2D gas distribution and evaluate their purity and completeness with respect to the 3D networks projected along the line of sight. We also compute the distances between the corresponding skeletons. Moreover, we identify filaments from simulated Compton-$y$ maps of the clusters at redshift $z=0$, and we compare them with the 2D gas network. The skeletons extracted from 2D maps provide good representations of the underlying 3D ones, both in terms of critical points and filaments. We find a median distance between the spines of the 2D and projected 3D networks of approximately $0.22 \, h^{-1}$ Mpc, although the connectivity derived from the 2D networks is slightly underestimated. We observe a good spatial agreement between the gas and SZ networks, with a median distance of $\approx 0.24 \, h^{-1}$ Mpc. Finally, we show that gas outside galaxy clusters is preferentially located in filamentary structures, which contribute $\sim 80\%$ of the integrated Compton-$Y$ parameter of clusters' outskirts.

We analyze zenith-pointing $(δ=-26.7^{\circ})$ Murchison Widefield Array (MWA) $ν_c=154.2 \,{\rm MHz}$ drift scan observations covering $349.0^{\circ} \le α\le 70.0^{\circ}$ with 163 pointing centers (PCs) spaced by $0.5^{\circ}$. We measure $D_{\ell}$, the mean-squared angular brightness temperature fluctuations, as a function of $α$. A broad peak at $α\approx 50.0^{\circ}$ corresponds to the bright extended source Fornax~A in the main lobe of the primary beam. A smaller peak at $α\approx 5.0^{\circ}$ possibly corresponds to Fornax~A in the first sidelobe. For $α\leq 22.0^{\circ}$ and $\ell \ge 200$, we find $D_{\ell} \propto \ell^2$, which we interpret as Poisson fluctuations from point sources. We present $Δ^2(k)$, the mean-squared 21-cm brightness temperature fluctuations from the Epoch of Reionization, as a function of $α$. Fornax~A causes strong contamination near $α\approx 50.5^{\circ}$, elsewhere several PCs are consistent with noise. The range $358.5^{\circ} \leq α\leq 11.5^{\circ}$ is relatively foreground-free and best suited for EoR science. The PC at $α= 11.0^{\circ}$ yields the best $2σ$ upper limit $Δ^{2}_{\rm UL}(k) = (173.13)^{2}\,{\rm mK^{2}}$ at $k = 0.161\,{\rm Mpc^{-1}}$. We incoherently combine $23$ PCs to obtain $Δ_{\rm UL}^2(k)=(98.67)^{2}\,{\rm mK}^{2}$ at $k=0.156\,{\rm Mpc}^{-1}$. This is the tightest upper limit from the MWA, being $\approx3$ times lower than earlier MWA limits, but $\approx2$ and $\approx21$ times higher than the LOFAR and HERA limits, respectively, and $\approx3$ orders of magnitude above theoretical predictions.

We present a multiwavelength follow-up study of 621 winged radio galaxies (WRGs) recently identified from LoTSS DR2, constituting the largest statistically significant samples of X-shaped (XRGs) and Z-shaped (ZRGs) radio galaxies to date. Our results show that WRGs are predominantly strongly radio-dominated, with XRGs on average more radio-luminous than ZRGs. Their optical hosts are massive elliptical galaxies residing in moderate-density environments. For 270 of XRGs, we measure angular offsets between the radio wings and the optical major axis. While most XRGs show large misalignments consistent with hydrodynamic backflow along the host minor axis, a substantial fraction ($\sim$25\%) exhibits small offsets (<30°), indicating that additional processes, such as jet reorientation, may also play a role. ZRGs, in contrast, are characterized by strongly antisymmetric deformations of their radio lobes pointing toward a coherent mechanism affecting both jets, modulated by local environmental interactions at the lobe termini. Mid-infrared diagnostics indicate merger-related cold gas in many WRGs, particularly XRGs, which also more frequently host powerful AGN, while ZRGs are more often classified as low-excitation radio galaxies (LERGs). This is consistent with our previous results showing that, although most WRGs exhibit FR II morphologies, FR I sources are almost exclusively ZRGs, suggesting that Z-shaped structures are statistically associated with lower jet power and are therefore more susceptible to perturbations. Nevertheless, the physical processes responsible for shaping XRGs and ZRGs need not be fundamentally different. Instead, the final morphology likely reflects the interplay between jet power, jet stability, and the surrounding environment.

The `Triangal' galaxy evolution schema is used to assess whether the Tuning Fork (bar strength) or the van~den~Bergh Trident and ATLAS$^{3D}$ Comb (spiral strength) offer greater evolutionary insight. A new catalogue of quantitative bar strengths (measured by the bar-to-total luminosity ratio, $P$), refined galaxy morphologies, and dust bin classifications is presented. It contains 137 galaxies with spheroid stellar masses, obtained from multi-component decompositions, and directly measured black hole masses, $M_{\rm bh}$. By placing these galaxies within the $M_{\rm bh}$-($M_{\rm\star,sph}$, $M_{\rm\star,gal}$) parameter space, an evolutionary reference frame reflecting integrated growth is established. Galaxies with varying bar strengths, and double bars, are observed to not occupy preferred locations, highlighting that bars are products of secular evolution-and can be transient or recurrent phenomena-and that they track neither hierarchical mass assembly nor galaxy speciation. In contrast, three physically distinct formation channels for S0/a galaxies are identified: (primeval S0)-to-S transitions; faded spiral galaxies; and, most commonly, wet-major-merger-built dust-rich S0 galaxies (on the `green mountain'). Galaxies with particularly strong spirals appear on the right-hand side of the spiral galaxy distribution. Furthermore, a `Dust Attrition/Retention' sequence places S0 (and compact massive ES,b) galaxies with dusty nuclear discs between the dust-poor and dust-rich S0 galaxies, and a `Disc Down-sizing' sequence is revealed, in which E galaxies with dusty nuclear discs-potentially formed through `damp' mergers-bridge the ES,e (ellicular) galaxies with intermediate-scale stellar discs and the dust-poor pure E galaxies. Extensive historical context is provided, and, finally, suspected biases in precision cosmology stemming from neglected precision galaxy morphology are discussed.

We present a model-independent, sound-horizon-free measurement of the Hubble constant $H_0$ using baryon acoustic oscillation tracers from the Dark Energy Spectroscopic Instrument Data Release 2. The function reconstructions are performed using the artificial neural network method, which is a completely data-driven approach that avoids the mild $Λ$CDM prior dependence. Our approach is based on the distance duality relation and combines three complementary observational probes, such as Type Ia supernovae, cosmic chronometer, and DESI DR2 BAO -- without requiring any knowledge of the sound horizon scale $r_d$ or any assumption about the absolute luminosity of SNe Ia. We obtain a joint constraint of $H_0 = 71.5\pm2.2$ km s$^{-1}$ Mpc$^{-1}$ at 68\% confidence for 1000 bootstrap realisations and 4096 neurons, which is consistent with the TRGB result and the SH0ES measurement within $0.6σ$, consistent with the Planck 2020 result within $2σ$. Our results favor a higher value of $H_0$ compared to the Planck CMB inference, adding independent support for the reality of the Hubble tension.

Accurate astrometric measurements for star-forming regions located on the far side of the Milky Way remain scarce. In this work, we present the astrometric results for a 22\,GHz water maser associated with star-forming region G040.96+02.48 located on the far side of the Milky Way, using the East Asian VLBI Network. The target water maser's proper motion was determined to be ($μ_α\cosδ, μ_δ$) = ($-2.06_{-0.51}^{+0.53}$, $-2.95_{-0.44}^{+0.45}$)~mas~yr$^{-1}$. The derived three-dimensional kinematic distance to the star-forming region is 20.2$\pm$3.2\,kpc, placing it slightly outside the Outer Scutum$-$Centaurus Arm. The corresponding vertical height of 872$\pm$139\,pc indicates a significant warp of the outer Galactic disk, which is in good agreement with the latest precessing warp model. Moreover, the resulting peculiar motions reveal a complex kinematic pattern, characterized by a large outward radial velocity of $-32\pm$18\,km~s$^{-1}$. Our observations substantially expand the valuable sample of star-forming regions with accurate astrometric measurements in the Extreme Outer Galaxy.

Circularization of the stream material into a debris cloud during tidal disruption events (TDEs) was recently demonstrated in one of the most accurate long duration TDE simulations to-date. The cooling envelope model (CEM) provides a description of the circularized debris cloud and its emission over time well beyond circularization across different disruption parameters. In the CEM, sub-Eddington accretion rates occur early in TDEs and the debris has a shallow density profile of roughly $ρ\propto r^{-1}$, with Eddington accretion only being achieved after several months. To explore the late stages of the CEM, we perform general relativistic radiation magnetohydrodynamics (GRRMHD) simulations of magnetized tori adapted from the near Eddington phase of the CEM for a $1M_\odot$ star disrupted around a $10^7 M_\odot$ black hole (BH). We find that the disk becomes thermally unstable within 17.1-46.5 days depending on the spin of the BH. Thermal spectra show a soft X-ray excess prior to collapse, with a nearly two order of magnitude decline in X-ray luminosity upon disk collapse. Furthermore, the evolution of the blackbody radius and temperature of our models are correlated with the spin of the black hole. The spectral properties and soft X-ray luminosity in our models are similar to the TDE AT2021ehb, which is a non-jetted TDE with late X-rays and a state transition after $\approx 271$ days.

The environmental mechanisms driving unwinding spiral arms in cluster galaxies remain debated. While earlier studies attributed it mainly to gravitational interactions, recent works suggest that RPS alone can induce unwinding. We present a VLT/MUSE spatially resolved analysis to investigate the mechanisms responsible for spiral-arm unwinding in two galaxies, UG101 and UG103, drawn from a larger sample. They are selected as tidal and RPS-driven candidates, respectively, based on the proximity of close neighbors. We estimate the galactocentric radius at which tidal forces, from a companion or the cluster potential, become relevant ($R_{\mathrm{tid}}$). We examine gas and stellar kinematics, exploiting their different responses to gravitational and hydrodynamical perturbations. SINOPSIS is used to map stellar populations in age bins and constrain unwinding timescales. For UG101, we find $R_{\mathrm{tid}} \sim 1.5 R_e$, while the unwound features extend beyond this radius. UG101 shows irregular stellar and gas kinematics; its rotation curve indicates similar motions, although the gas is truncated on one side and extended on the other. For UG103, neither the closest companion nor the cluster appear capable of triggering unwinding. UG103 displays regular stellar but disturbed gas kinematics, with truncation on the disk side likely facing the ICM wind and gas extended in the opposite direction. Stellar population maps show the emergence and unwinding of the spiral arms in UG103 on timescales consistent with its cluster infall time ($\sim 1.6$ Gyr). We conclude that unwinding in UG101 and UG103 is primarily driven by tidal interactions and RPS, respectively, although a combined effect cannot be excluded for UG101. Our methodology provides a framework to identify the mechanisms driving unwinding in cluster galaxies from spatially resolved properties.

Detecting possible electromagnetic precursors to the gravitational signal from merging compact objects is challenging, but it can reveal intricate physical properties of the merging stars through their gravitational and electromagnetic interactions. We demonstrate, using 3D Particle-In-Cell simulations, that a neutron star moving through the magnetosphere of a merging companion generates a complicated system of dissipative currents, a relativistic analogue of planetary Alfven wings. Generated electric currents carry a large fraction of the electromagnetic power intersected by the neutron star. These currents may lead to the generation of beamed, pulsar-like coherent radio and high-energy emission. Orbital modulation will produce a nearly periodic signal, an Alfven-winged pulsar.

We present a comprehensive study of the galaxy UV luminosity function (UVLF) at $z=6-9$ leveraging deep JWST observations from the GLIMPSE survey. Thanks to gravitational lensing, we probe the UVLF to an unprecedented depth of $M_{\text{UV}} = -12$ mag, approximately three magnitudes deeper than previous robust constraints. Our UVLF determination incorporates a rigorous end-to-end uncertainty framework, including statistical and systematic lensing uncertainties. We find that the $z \sim 7$ UVLF continues to rise steeply with a faint-end slope of $α= -1.98_{-0.05}^{+0.06}$. Crucially, our data show no clear evidence of a turnover down to \muv $= -12.3$. The persistence of this faint population provides stringent constraints on galaxy formation models and cosmological simulations that predict an early flattening of the luminosity function due to radiative feedback or star-formation thresholds. Furthermore, post-JWST models specifically calibrated to match the UV-bright excess at $z > 10$ generally fail to reproduce the observed evolution toward lower redshifts and fainter magnitudes, highlighting a significant tension in our current understanding of early galaxy assembly. We derive a comoving ionizing emissivity at $z=7$ of log($n_{\mathrm ion}$ / s$^{-1}$ Mpc$^{-3}$) $\approx 50.85$, which suggests that faint galaxies dominate the ionizing budget, providing enough photons to maintain reionization even in a highly clumped IGM ($C_{\text{HII}} = 5$). As our detection of faint galaxies effectively rules out a luminosity function truncation at $M_{\text{UV}} \geq -15$, these results emphasize the need to either accurately characterize the ionizing properties of the global, low-mass galaxy population at $z > 6$, or to refine physical models of intergalactic medium clumping and its redshift evolution to maintain consistency with the observed reionization timeline.

We present the first data release (DR1) of the Spectroscopic Classification of Astronomical Transients (SCAT) survey, covering the first $\approx 5$ years of observations (March 2018 - January 2023). DR1 includes 1810 spectra of 1330 transients, which we sort into broad spectroscopic classes including supernovae (SNe), transients originating in galactic nuclei, and stellar variability. We collect multi-filter light curves from imaging surveys and fit them with phenomenological models to estimate peak brightnesses and the time of explosion/first-light. Extragalactic transients are matched to candidate host galaxies, and we compare host-galaxy luminosities and projected offsets by SN type. SNe appear to be a reliable way to augment the redshift coverage of nearby ($z\lesssim 0.1$) galaxies in tandem with dedicated redshift surveys. We present new redshifts for roughly half of the SN host galaxies, most of which are low-luminosity dwarfs similar to the Magellanic Clouds ($M_r \gtrsim -18$ mag). This set of transient spectra, light curves, luminosities, redshifts, and host galaxies offers an excellent testbed for real-time photometric/light curve classification pipelines in the modern era of deep and large-area surveys. We conclude with a brief discussion of the provided data products and status of the SCAT survey.

The Legacy Survey of Space and Time (LSST) on the Vera C. Rubin Observatory will generate a massive collection of time series (light curves) of the measured flux of transient and variable astronomical objects. With each new flux observation, light curve classifiers need to generate updated probability distributions over candidate classes, which will then be shared with the global community for the purpose of identifying interesting targets for follow-up observations as well as less time-sensitive analysis applications. Using the synthetic light curves and classification results of participating classifiers from the Extended LSST Astronomical Time-series Classification Challenge (ELAsTiCC), we investigate a novel framework to enhance existing light curve classifications by incorporating their classification histories and the temporal evolution of these histories. To demonstrate the potential of this approach, we introduce a model that combines a recurrent neural network and an additive attention module, which shows improved classification accuracy and more balanced precision-recall performance compared to existing classifiers from the challenge. Furthermore, at this stage, most, if not all, of the existing classifiers are evaluated by their final classification results on complete light curves; we propose new metrics that evaluate the stability, accuracy, and early classification performance of a classifier's predictions when using limited data by considering the Wasserstein distance between the temporally evolving classification probability distributions. Our metrics offer a more comprehensive perspective for model assessment by supplementing classical methods such as the confusion matrix and precision-recall.

The advent of multi-messenger astronomy has provided abundant information for understanding the acceleration and particle-production mechanisms of cosmic rays. In this work, we present a unified study of cosmic gamma-ray and neutrino spectra within the Gluon Condensation (GC) model. Derived from Quantum Chromodynamics (QCD), the GC model predicts that, in high-energy hadronic processes, gluons may condense near a critical momentum, leading to a dramatic enhancement in secondary-pion production and imprinting a characteristic broken power-law feature on the gamma-ray spectrum. Within this framework, we first derive the neutrino spectrum corresponding to the GC scenario and then investigate three astrophysical sources with both gamma-ray observations and neutrino candidate signals: the active galactic nuclei TXS~0506+056 and NGC~1068, and the supernova remnant G54.1+0.3. Using the GC model, we fit the observed gamma-ray spectra of these sources and predict their corresponding neutrino spectra. Our results show that the gamma-ray spectra of TXS~0506+056 and NGC~1068 are well described by the GC model, and that the predicted neutrino spectra are consistent with IceCube observations within uncertainties; in particular, clear relations are found between their relative magnitudes. For SNR~G54.1+0.3, however, the GC-predicted neutrino spectrum exhibits continuous hardening after the break, deviating from the typical power-law behavior expected for cosmic-ray secondaries and thus disfavoring a common GC origin. This study represents the first systematic attempt to correlate gamma-ray and neutrino spectra within the GC framework, offering a new perspective on multi-messenger emission from high-energy astrophysical sources.

The physical properties of Milky Way Mgii-bearing gas remain poorly constrained due to the saturation of the near-UV doublet. We utilize the weaker Mgii $λλ$1239, 1240 doublet from 482 archival HST/COS extragalactic sightlines to probe this cool gas phase. We identify 43 low-velocity absorbers ($|v_{\rm LSR}|<40\ {\rm km\ s^{-1}}$), yielding a covering fraction ($C_f$) of $32\pm5\%$ for $\log N_{\rm MgII} > 15$. We find that $C_f$ follows an exponential decay relative to equivalent width thresholds, marking a transition from a diffuse medium to localized, dense structures (e.g., cold neutral medium cores). The steep decline of the distribution at high column densities likely reflects the saturation of the turbulent log-normal spectrum and dust depletion. By integrating stellar data, we derive a Mgii scale height $h_{\rm MgII} = 0.12\pm0.02\ \rm\ kpc$ and mid-plane density $n_{0,\rm MgII} = (3.9\pm0.4)\times 10^{-6}\ \rm cm^{-3}$. A pronounced north-south asymmetry exists, with the northern hemisphere displaying a significantly higher mid-plane density ($n_{0,n} \approx 4.7 \times 10^{-6}\ \rm cm^{-3}$) than the south ($3.2 \times 10^{-6}\ \rm cm^{-3}$). This discrepancy suggests that the northern interstellar medium is more spatially concentrated and clumpy, whereas the southern gas is more ubiquitously distributed with a lower average density. These results indicate that Mgii is tightly confined to the disk, governed by a unified depletion law and restricted vertical extent.

From the onset of observations of JWST we have discovered unexpectedly luminous galaxies at redshifts $z>10$ and as high as $z=14$. With their discovery, the question immediately followed as to where their progenitors are, since such progenitors should be within reach of existing surveys. However, the discovery of several bright candidates at $z>15$ may indicate further discrepancies between pre-JWST model predictions and current observations. Progenitors of the bright $z\sim 14$ galaxies should be visible at redshifts as high as $z\sim 20$--$30$, showing in the data as F356W and F277W dropouts. We identify two such candidates in the Bullet Cluster JWST data; however, subsequent NIRSpec follow-up data show spectra that can be well fit with ultra-cool Y dwarf templates with temperatures $T_{\rm eff} = 350^{+110}_{-80}\,\mbox{K}$ and $T_{\rm eff} = 410^{+110}_{-50}\,\mbox{K}$ and distances of $\sim 500\,\mbox{pc}$. The first is one of the lowest temperature brown dwarfs known spectroscopically. With additional NIRCam imaging taken $\sim 1$ year later, we also detect their proper motions of $(49 \pm 8)\,\mbox{mas/yr}$ and $(24 \pm 3)\,\mbox{mas/yr}$, further indicating that at least some F277W and F356W dropouts are sub-stellar cold Milky Way objects such as brown dwarfs. We find a sky density of 0.14 Y dwarfs per arcmin$^2$ and caution that the probability of detecting such objects may increase significantly in surveys at low galactic latitudes.

We present a high-precision joint gravitational-lensing and kinematic analysis of nine massive galaxy clusters from the CLASH and CLASH-VLT surveys to test chameleon screening gravity and its $f(R)$ sub-class at Mpc scales. We investigate the dependence on the assumed parametrization of the total cluster mass profile by adopting three models, namely Navarro--Frenk--White (NFW), Burkert, and Hernquist. When cuspy models (NFW or Hernquist) are assumed in the general chameleon framework, the combined constraints from the nine clusters are fully consistent with General Relativity (GR), excluding large regions of the modified-gravity parameter space (the coupling constant $\mathcal{Q}$ and the background chameleon field $ φ_\infty$), providing one of the tightest bounds on general chameleon models with clusters to date. In contrast, adopting a Burkert profile -- disfavored by lensing data -- leads to a mild ($\sim 2σ$) departure from the GR expectation in joint analysis. When considering the $f(R)$ sub-case, we obtain a bound on the background scalaron field of $|f_R| \lesssim \mathrm{2-5}\times 10^{-5}$ (95\% C.L.) for NFW and Hernquist models, in agreement with current constraints at cosmological scales, and an apparent deviation from standard gravity of $\log_{10}|f_R| = -4.7 \pm 1.2$ for the Burkert case. We investigate the impact of systematics in the kinematical analysis, showing that the tension is mitigated when clusters exhibiting clear dynamical disturbance are excluded from the sample. [...[ The upcoming generation of wide-field lensing surveys and spectroscopic follow-up programs will enable similar analyses on substantially larger samples, offering the prospect of tightening cluster-based constraints on gravity and the dark sector.

In many astrophysical transients, outflows drive shocks into the ambient medium, accelerating electrons to non-thermal energy distributions that produce broadband synchrotron emission. At late times, even initially collimated relativistic jets evolve into quasi-spherical Newtonian blastwaves. As the shock decelerates, the post-shock internal energy per particle decreases; below a critical velocity $β_{\rm DN} \approx 0.2$, only a fraction $ξ_e < 1$ of electrons are accelerated to relativistic energies, defining the deep Newtonian (DN) regime. We develop a unified analytic framework for synchrotron emission in this phase, applicable to both single-velocity and stratified ejecta. For gamma-ray burst afterglows in a uniform medium, the DN transition occurs at $t_{\rm DN} \approx 3.7\,E_{51}^{1/3} n_0^{-1/3}$~yr, yielding a shallower decay by $δα= 6(p-2)/5$ relative to standard Newtonian predictions. For kilonova remnants ($E_0 = 10^{50.5}$~erg, $M_{\rm ej} = 0.1\,M_\odot$), the DN phase begins prior to deceleration; neglecting it underestimates radio flux by factors of $\sim 3$--$5$ during coasting and even more thereafter. Magnetar-boosted remnants ($E \sim10^{52}$~erg) should reach $\sim$\,10\,--\,100\,$μ$Jy at 3~GHz at $\sim$\,40\;Mpc, though limits on GW170817 already disfavor a long-lived millisecond magnetar. In core-collapse supernovae in a wind medium ($ρ\!\propto\!r^{-k}$), the peak luminosity remains constant during coasting, while $ν_{\rm pk} \propto t^{-1}$; for SN~2023ixf, we find $k = 1.29 \pm 0.14$. The DN SED typically satisfies $ν_m\!<\!ν_{\rm sa}\!<\!ν_c$, peaking at sub-GHz frequencies where LOFAR and SKA-low are most sensitive. Even non-detections place robust constraints on ambient density and outflow energetics.

The evolution of neutron stars (NSs) embedded within massive stellar envelopes is a critical phase in binary stellar evolution, potentially leading to the formation of Thorne-Żytkow Objects (TŻOs) or catastrophic collapse. We present the first fully coupled general relativistic hydrodynamics (GRHD) simulations of hypercritical accretion onto NSs that simultaneously incorporate grey two-moment (M1) neutrino transport and an $α$-chain nuclear reaction network. By investigating four distinct progenitor evolutionary stages, we resolve the complex interplay between intense neutrino cooling, multidimensional fluid dynamics, and nuclear feedback. Our results show that while vigorous convection is triggered in the post-shock region, the global energy budget is primarily governed by neutrino cooling, which effectively balances the accretion power. Crucially, even though our M1 transport scheme captures neutrino absorption and localized heating, the efficient cooling sink and high ram pressure of the infalling envelope prevent the formation of any core-collapse supernova-like explosion. We find that all nucleosynthetically processed material ($T > 5$~GK) remains strictly gravitationally bound, challenging the assumption that these systems contribute significantly to galactic nucleosynthetic yields via convective dredge-up. The lack of sustained outflows and the persistent hypercritical accretion rates suggest that embedded NSs will rapidly exceed the Tolman-Oppenheimer-Volkoff mass limit on timescales of minutes to hours. We conclude that these systems are not stable TŻOs, but are rather transient precursors to catastrophic black hole formation and potential central engines for high-energy transients.

Recent late-Universe observations suggest an open Universe. If confirmed, such a departure from spatial flatness would carry profound implications for our understanding of cosmic inflation and the ultimate fate of the Universe. Motivated by this intriguing result and the release of new data, we revisit the question using baryon acoustic oscillation measurements from DESI DR2, multiple Type Ia supernova samples, refined strong gravitational lensing time-delay analyses, and the most up-to-date cosmic chronometer data. We find that within the $Λ$ cold dark matter ($Λ$CDM) paradigm, the combined data still prefer an open Universe with $Ω_K=0.049\pm0.037$. However, this preference vanishes in extensions to $Λ$CDM, where the data instead favor a flat Universe. The model comparison shows that for $Λ$CDM, introducing new physics is preferred over merely allowing spatial curvature, and flat $Λ$CDM extensions perform better than their curved counterparts. We therefore argue that the mild open-Universe signal is an artifact of limited model flexibility, rather than a genuine feature of late-Universe observations.

The galaxy luminosity and stellar mass function (LF, SMF), and halo mass function (HMF), are fundamental quantities in astrophysics and crucial inputs to a range of astrophysical and cosmological analyses. They are typically parametrised by fitting functions that have been chosen "by eye" to match observed or simulated data. We apply symbolic regression -- specifically the Exhaustive Symbolic Regression (ESR) algorithm -- to automate the search for optimal LF, SMF and HMF functional forms. ESR scores all functions up to a maximum complexity composed of a user-defined basis set of operators using the description length, an approximation to the Bayesian evidence that balances accuracy with complexity. We find many functions outperforming the Schechter and double Schechter functions for the LF and SMF, and that outperform the Press--Schechter and Warren/Tinker functions for the HMF. By additionally imposing "physicality checks" on functions' extrapolation and integration properties, we identify the optimal, low-complexity functional forms in terms of accuracy, simplicity and behaviour beyond the data range. As well as providing drop-in replacements for literature LF, SMF and HMF fitting functions, and identifying robust behaviour across well-fitting functions, we present a framework with which symbolic regression may be used to automate the discovery of optimal functions for any astrophysical dataset.

Cygnus X-3, hosting a Wolf-Rayet (WR) star whose dense wind produces various spectral lines due to photoionization by X-rays from a compact object, provides an ideal laboratory for studying wind dynamics and density structure. We measured the orbital modulations of the Fe, Ca, Ar, and S Ly$α$ lines observed with the X-ray microcalorimeter (Resolve) onboard the $XRISM$, taking account of both emission and absorption lines of the Ly$α$ complexes. The modulations of Doppler shifts of the Fe, Ca, Ar, and S Ly$α$ lines showed amplitudes of 500 km s$^{-1}$ and phase offsets of 0.04, 0.09, 0.11, and 0.17, respectively, in units of an orbital period (4.8 hours) relative to the orbital motion of the compact object. This result indicated that H-like Fe most closely follows the compact object's motion. The line widths ranged from 400 to 1000 km s$^{-1}$. The intensities of both emission and absorption lines reached their minima around orbital phase 0.0 and their maxima around phase 0.5. The absorption peaks, however, did not align exactly with phase 0.5, suggesting the inhomogeneous structures such as an accretion wake and/or a bow shock. We compared the observed modulations with calculations based on a stellar wind model, accelerated by ultraviolet radiation from the WR star. One of the calculations qualitatively reproduced the observed trend that H-like Fe ions were concentrated near the compact object, whereas H-like S was distributed across the binary system, with H-like Ca and Ar showing intermediate spatial distributions. From this comparison, we estimated that a mass-loss rate of the stellar wind was approximately $5 \times 10^{-6}$-$1 \times 10^{-5}\ M_{\odot}\ {\rm yr}^{-1}$.

Understanding how galaxy populations emerge and evolve from the growth of dark matter structure is a central challenge in galaxy formation theory. Semi-analytic models (SAMs) provide an efficient framework to address this problem, but exploring large ensembles of merger trees across broad parameter spaces remains computationally demanding. We develop a conditional graph neural network surrogate model that combines merger tree information with SAM parameters to predict galaxy properties across cosmic time. Using merger trees of dark matter halos from the Uchuu simulation and the Galacticus SAM, the model predicts stellar mass, luminosity, angular momentum, gas metal mass, and specific star formation rate across the wide redshift range of 0 <= z <= 5. For instance, the model can predict stellar mass at 0 <= z <= 3 with a scatter of 0.19-0.28 dex and coefficient of determination R^2 of 0.946-0.973 (R^2 close to 1 indicates prediction closely matching the truth). The results show that a single graph based model can reproduce these galaxy properties with good accuracy over multiple SAM realizations, merger trees and redshifts. This catalog-level model provides a practical route for accelerating SAM based studies of galaxy formation to enable a more detailed investigation of the model parameter space. The inference code, trained models, and example data products are publicly available at https://github.com/MutongCat/sam2galaxy-gnn.

We construct mean spectral energy distributions (SEDs) for a substantial sample of 56,969 Sloan Digital Sky Survey DR16 quasars with $z < 0.75$, utilizing multiwavelength data from the mid-infrared (MIR) to ultraviolet (UV). These SEDs are built on eigenvector 1 parameters -- the relative optical $\rm Fe~ II$ strength ($R_{\rm Fe~II}$) and the H$β$ line width ($\rm Hβ$) -- that capture the principal spectral variance of quasar spectra. From three $R_{\rm Fe~II}$-dependent mean SEDs we find that quasars with a larger $R_{\rm Fe~II}$ exhibit redder UV and optical and redder MIR and near-infrared (NIR) continua, indicating more dust emission. We also split our sample directly into Eddington ratio $L_{\rm Bol} /L_{\rm Edd}$ (or dimensionless accretion rate $\dot{\mathscr{M}}$) bins to construct different mean SEDs and find that the continua become increasingly red with increasing $L_{\rm Bol} /L_{\rm Edd}$ (or $\dot{\mathscr{M}}$) in the MIR, NIR, and UV bands. This demonstrates that the shapes of Type 1 AGN SEDs depend on the accretion rate. However, the optical continuum shows the opposite trend (becoming harder and bluer), indicating the complexity of the optical emission region. From $\rm FWHM_{Hβ}$-dependent mean SEDs we find that quasars with a larger $\rm FWHM_{Hβ}$ show redder optical and NIR continua and bluer UV and MIR continua. The bluer MIR continuum suggests that a larger angle between of the line of sight and the torus plane results in weaker torus emission in the MIR.

Using \textit{Hubble Space Telescope} images from the PHAT and PHAST surveys, we construct an updated catalogue of 747 OB cluster (OBC) candidates. We introduce a dimensionless structural metric, the trace coefficient of variation ($CV_{\rm tr}$), derived from the Hessian matrix in four \textit{HST} bands, to quantify the internal photometric substructure of partially resolved OBC candidates. Cross-matching with the subset of M31 clusters that have independent colour--magnitude diagram (CMD) age estimates yields 247 objects in common. We find statistically significant anti-correlations between $CV_{\rm tr}$ and age in the UV and blue bands, suggesting a progressive smoothing of the light distribution as clusters evolve. Bootstrap resampling confirms the robustness of these trends. Forward modelling of synthetic clusters analysed with the same pipeline recovers a monotonic $CV_{\rm tr}$--age relation under simplified but physically motivated assumptions. These results show that second-order photometric structure contains measurable evolutionary information within the CMD-calibrated regime ($\sim10$--300~Myr).

The origin of ultra-high-energy cosmic rays (UHECRs) remains a fundamental question in astroparticle physics. While localized 3 $σ$ correlations with active galactic nuclei and starburst galaxies have been reported using time-integrated analyses, we propose and implement a spatiotemporal multiplet search method utilizing a pre-defined fixed window of 3 degrees and 15 days, a kinematic filter designed to isolate high-rigidity particles and keep chance coincidences low. Applying this method to 16 years of Pierre Auger Observatory data, we identify 28 UHECR multiplets (doublets) above 32 EeV. We backtrack these trajectories using three Galactic magnetic field models across eight distinct nuclear species. Testing the backtracked directions against the ten nearest bright radio galaxies yields an overall post-trial significance of 5.8 $σ$, conditioned on the individual best-fit associations. Specifically, we find a 4.5 $σ$ conditional post-trial significance for the joint association of 8 of these multiplets with the Fornax A region alone. These results point to radio galaxies, with a strong contribution from Fornax A, as long-term accelerators of heavy (Z > 3) UHECRs, possibly within the mildly relativistic backflows of their extended radio lobes, detected at Earth primarily as independent secondary fragments above 32 EeV.

A rare multiple open cluster system has been analyzed using Gaia DR3 astrometry and photometry data. Using Agglomerative Hierarchical clustering and Bayesian-HDBSCAN, we identify a compact core consisting of seven known open clusters and two additional components, including a new candidate, forming a nine-member association. Membership probabilities are refined through statistical modeling, combining GUMM, Bayesian-KDE, and Bayesian-XDGMM for tidal tails identification. Backward orbital integrations confirm coherent motions over 20-30 million years, suggesting a common origin from the same massive molecular cloud. This system offers a unique laboratory for investigating cluster multiplicity, dynamical evolution, and Galactic structure.

Lagrangian tracer particles have long been used to track the history of individual gas parcels in hydrodynamical codes. Particles advected by the cell-centered velocity carry no representation of underlying numerical diffusion, and thus exhibit systematic bias. The Monte-Carlo (MC) tracer resolves this with discrete probabilistic cell-to-cell, flux-based jumps, at the cost of trajectories that are discontinuous in time. We introduce the Itô tracer, a continuous-time Lagrangian particle with moments matched to the advection, diffusion, and dispersion of the gas. A subgrid-scale variant (SGS-Itô) replaces the numerical diffusion with a Smagorinsky--Lilly turbulent diffusivity, illustrating that the form of the diffusion matters less than its magnitude. We validate these methods with a 1D square-pulse advection test and 3D decaying turbulence at $σ_{\rm rms} = 15\,c_{\rm s}$. We compare the different tracer particle methods using several statistical tests. Itô tracers largely reproduce or improve upon MC tracers statistics across column-density maps, joint density histograms, log-density-ratio PDFs, and density power spectra. In the turbulence test, Itô tracers improve the correlation between tracers and gas over the MC tracers by >3\%, and reduce the width of the log-density ratio PDF by nearly 50\%. Relative to classical tracers, these improvements are $\gtrsim$30\% and 230\%, respectively. Because Itô tracers follow a stochastic differential equation, the method maps onto other continuous-trajectory Lagrangian processes (e.g. dust grains, charged particles, cosmic rays), admits variance-reduction techniques, higher-order integrators, and GPU-friendly implementations -- all of which are unavailable to discrete-jump schemes.

While there has been significant progress in our understanding of the origin and evolu-tion of planetary nebulae in the last 50 years, there remain several unsolved problems. These include the true 3D morphological structure of the nebulae, origin of multipolar nebulae, the dust and molecular distribution relative to the optical nebulosity, large-scale structures outside of the main nebulae, the relevance of binarity to planetary nebulae evolution, and a precise definition of the planetary nebula phenomenon. The long-standing problem of elemental abundance discrepancy still remains unsolved. In this paper, we summarize current observations related to these problems and present possible future directions to tackle them.

Real-time ranking of optical transient candidates during gravitational-wave (GW) and multimessenger follow-up is challenging when only sparse early-time, multi-band photometry is available.We present \texttt{KilonovaSCORER}, an open-source framework for scoring and ranking in this regime. It quantifies the consistency of each candidate with a physically motivated kilonova model grid in absolute magnitude space using two complementary per-observation metrics, $P_{\mathrm{tail},\mathrm{KNe}}$ and $P_{\mathrm{near},\mathrm{KNe}}$. These are aggregated into a cumulative ranking score via inverse-variance weighting in logit space, naturally accounting for heterogeneous observational uncertainties across bands and epochs. A sequential Approximate Bayesian Computation (ABC) diagnostic tracks photometric consistency across epochs, penalizing candidates whose temporal evolution is incompatible with kilonova expectations. We validate the framework on AT\,2017gfo and SN\,2025ulz, and test it against supernova simulations under a realistic Rubin/LSST Target-of-Opportunity strategy. The framework recovers kilonova candidates with high confidence while ruling out supernova contaminants within five days of the gravitational-wave trigger. In our LSST ToO simulations, median cumulative scores for thermonuclear and core-collapse supernova contaminants fall to zero by $3$--$4$\,d post-trigger, whereas kilonova medians remain $\gtrsim 0.4$. \texttt{KilonovaSCORER} supports real-time workflows for ToO teams and LSST alert brokers, integrates with follow-up coordination platforms such as the Tool for Rapid Object Vetting and Examination, and is publicly available at https://github.com/phelipedarc/KilonovaSCORER/tree/main.

The Solar Neighborhood is dominated by stars smaller, colder, and fainter than the Sun: the M dwarfs. If we are to understand the context in which the Sun formed and evolved, then we must investigate the system architectures of our low-mass neighbors. We have therefore carried out the Pervasive Overview of Kompanions of Every M Dwarf in Our Neighborhood (POKEMON) speckle survey of nearby M-dwarf primaries. We created the survey with the goal of observing a volume-limited (north of -30 degrees) sample of M-dwarf primaries through M9 out to 15 pc at diffraction-limited resolution. Pre-Gaia parallax measurements yielded a catalog of 454 nearby M-dwarf primaries. However, the precise astrometry from Gaia indicated that there are additional low-mass sources within 15 pc. Here we present the POKEMON-Distance Limited Catalog (POKEMON-DLC), a supplemental catalog that consists of speckle observations for the 66 additional M-dwarf primaries identified by Gaia, increasing the number of ultracool dwarf (later than M6.5) primaries in the POKEMON catalog by a factor of 1.6. In our observations we detect four likely bound companions. After carrying out a literature search for additional companions, we update the projected separation distribution and find a peak at 7.91 au (σlog(a) = 1.1, SElog(a) = 0.10). We also update the M-dwarf stellar multiplicity and companion rates, and find values of 22.7 p/m 1.8% and 27.5 p/m 2.0%, respectively. These results emphasize the utility of Gaia for identifying low-mass, nearby sources, and we find that ensuing characterization of these sources by SPHEREx will continue to clarify the nature of the Solar Neighborhood.

We present an analysis of the temperature structure of the Perseus cluster atmosphere using XRISM Resolve observations. The average temperature rises from 3.3 keV near the nucleus of NGC 1275 to 8 keV at 10 arcmin (210 kpc), which is consistent with Chandra and XMM measurements. The velocity and velocity dispersion profiles are broadly consistent with those in arXiv:2509.04421. While the gas at altitudes beyond $\sim60$ kpc can be modeled as a single temperature plasma, we find evidence for more than one gas phase in the inner $\sim60$ kpc. The hotter gas component, traced primarily by the Fe He$α$ line, has a velocity dispersion of $\lesssim140$ km s$^{-1}$. We detect a cooler, $\sim1.87-2.43$ keV, gas component with a velocity dispersion of $\sim300-400$ km s$^{-1}$ and a bulk velocity of $\sim 21-213$ km s$^{-1}$ with respect to the central galaxy. These ranges reflect large systematic uncertainties associated with modeling spatial-spectral mixing and the bright central point source. Potential low energy gain variations may add additional uncertainties. The cooler component is identified by broad wings in prominent emission lines, most notably S Ly$α$ and Fe He$α$. This cooler component's Mach number $\sim0.73-0.96$ and non-thermal pressure fraction of $\sim22.9-33.7\%$ are much higher than found for the hotter gas. The cooler gas may be associated with merging halos along the line of sight which formed the cool, sloshing spiral and/or cooling gas being disturbed by the radio jets and lobes.

Planetary systems can act as sensitive gravitational detectors for dark matter. We investigate the gravitational scattering of Oort cloud objects by primordial black holes (PBHs) as a potential component of the Galactic dark matter halo. Calculating the rates at which PBH encounters eject objects from the Oort cloud or inject them into Earth crossing orbits, we find a linear scaling $Γ\propto m_{\mathrm{PBH}}$ for $m_{\mathrm{PBH}} \gtrsim 10^{-10} M_\odot$. For $m_{\mathrm{PBH}} \sim 10^3 M_\odot$, PBHs constituting all local dark matter would eject $\sim1.3\times10^{12}$ objects over the Solar System's lifetime, comparable to the total Oort cloud population and inject $\sim2.6\times10^{10}$ objects into Earth-crossing orbits. Comparing these rates with observational constraints from long period comet fluxes and terrestrial impact records, we derive upper limits on the PBH dark matter fraction $f_{\mathrm{PBH}}$. Our most stringent constraints exclude $f_{\mathrm{PBH}}=1$ for $10^2 M_\odot \lesssim m_{\mathrm{PBH}} \lesssim 10^5 M_\odot$, with $f_{\mathrm{PBH}} \lesssim 0.002$ at $m_{\mathrm{PBH}} = 10^3 M_\odot$. For the asteroid mass window ($10^{17}$-$10^{23}$ g), scattering rates are far too low to produce observable effects. These Solar System-based constraints complement existing astrophysical probes and demonstrate that planetary systems can serve as sensitive gravitational detectors for compact dark matter.

We present results from a multi-epoch Very Long Baseline Array (VLBA) survey conducted as part of the DYNAMO-VLBA project, aimed at measuring the dynamical masses of young stellar systems in the Orion complex. Our observations include 19 radio sources associated with 15 binary or multiple young systems. For four visual binaries in which both components were detected, the derived Keplerian orbits yield model-independent stellar masses; in particular, Brun~656 and HD~294300 show excellent agreement between VLBA-based and spectral-energy-distribution-based estimates, providing valuable benchmarks for pre-main-sequence evolutionary models. The component NU Ori C is confirmed as an intermediate-mass ($\sim$7\,M$_\odot$) star with nonthermal radio emission, offering rare evidence of magnetic activity near the boundary with the high-mass regime. Several additional sources exhibit astrometric accelerations or periodic residuals, revealing unseen companions and extending dynamical constraints to systems with only one radio-emitting component. These results highlight the capability of very long baseline interferometry astrometry to obtain precise and model-independent masses of young binaries, providing critical empirical anchors for stellar evolution models and new insights into the origin of magnetism in intermediate-mass stars.

The physics driving type Ia supernovae (SNe~Ia) standardisation in cosmology remains poorly-understood. Recent advances however mean that it is now possible to systematically analyse the explosion properties of large numbers of cosmological SNe~Ia. To that end we use riddler, a machine learning based framework for rapidly modelling SNe~Ia based on realistic explosion simulations, to perform quantitative spectral modelling of the Zwicky Transient Facility SN~Ia DR2 sample and determine their best-fitting explosion mechanism(s). We find that approximately two thirds of our sample is best reproduced by sub-Chandrasekhar mass explosions. Analysing their light curve and host galaxy properties, we find that Chandrasekhar mass explosions are not favoured for the fastest-evolving SNe~Ia, while sub-Chandrasekhar mass explosions are favoured for the reddest SNe~Ia. Due to the differences in their environments, selecting SNe~Ia in massive, passive galaxies could produce a homogeneous sample of violent merger SNe~Ia. We show that standardising each explosion mechanism independently reduces scatter in distance estimates and previously claimed environmental and non-linear light curve shape corrections may be due to changes in the relative populations of different explosion mechanisms. Although a step forward towards understanding SNe~Ia physics in cosmology, we highlight a number of limitations affecting our conclusions, including sample biases and small numbers. We therefore cannot assess the statistical significance of our results and they should be treated with caution. Larger and more uniformly observed samples will be key to determining the significance of any trends hinted at here.

Observations of type Ia supernovae (SNe Ia) have led to suggestions of multiple progenitor and explosion scenarios. Distinguishing between scenarios and tying specific SNe Ia to individual scenarios however has so far been challenging. Constraints on the explosion physics are often achieved through empirical modelling of SNe Ia spectra and qualitative assessments of the level of agreement. While this approach has provided useful insights, it cannot be scaled up to large numbers of SNe Ia in a robust and systematic way. As a machine learning based framework for automated and quantitative fitting of SNe Ia, riddler is designed to overcome these limitations. Neural networks are used as radiative transfer emulators and, in conjunction with nested sampling, emulated spectra are fit to observations of SNe Ia to determine the best-fitting input parameters and explosion scenario. In this work, we present recent improvements to riddler, including a significantly expanded training dataset covering pure deflagrations, delayed detonations, double detonations, gravitationally confined detonations, and violent mergers. We show that despite the increased complexity and variety of our training data, riddler is able to accurately recover the input parameters and explosion scenario of spectra unseen during training. Using riddler, we fit observations of three SNe Ia covering different sub-classes: SN 2011fe, SN 2005hk, and SN 2018byg. We detail a number of limitations and assumptions that should be considered when applying similar approaches. Nevertheless, the benefits of this approach and riddler will result in automated fitting playing an increasingly important role in the coming years.

Numerical simulations have established that star clusters with densities comparable to the high redshift ($z>6$-$10$) James Webb Space Telescope (JWST) proto globular clusters may build up extremely massive (EMSs; $m_\mathrm{\star}>1000 M_\odot$) or even supermassive stars (SMSs; $m_\mathrm{\star}>10000 M_\odot$) and potentially intermediate mass black holes (IMBHs) through runaway stellar collisions. Using direct simulations of assembling star clusters including post-Newtonian black hole dynamics and stellar evolution, we demonstrate that in such dense environments ($Σ_\mathrm{h} \gtrsim 10^6 M_\odot$pc$^\mathrm{-2}$) stellar BHs ($m_\bullet \lesssim 60 M_\odot$), driven by rapid mass segregation and relaxation effects within the sphere of influence of the EMSs/SMSs, may strongly interact with the extremely massive stars and become embedded within their gaseous layers. We suggest that this quasi-star (QS) like embedded BH phase is a natural outcome of the runaway formation of EMSs/SMSs in the densest star clusters. The QS phase is orders of magnitude longer in duration than the lifetime of the SMS, enabling an extended growth period by stellar collisions, and allows the formation of embedded gravitational wave sources if the QS captures more than a single stellar BH. The star cluster assembly region sizes ($\sim100$ pc), QS masses ($\gtrsim 10^4 M_\odot$) and their proximity to young, massive blue star forming clumps are consistent with the faint population of multiple little red dots (LRDs) recently discovered by the JWST.

Ultra-fast outflows (UFOs) are mildly relativistic (outflow velocity $v_{out}>0.1c$) nuclear winds detected as blueshifted absorption lines from highly ionized, dense gas in the X-ray spectra of active galactic nuclei. The AGN feedback mechanism is believed to be powered by these outflows, which can inject a large amount of energy and momentum into the surrounding interstellar medium, shaping the coevolution of the AGNs and their host galaxies. We performed a systematic search and rigorous statistical assessment of the presence of UFOs in the 7-12 keV band, in a sample of bright local AGNs. This study also aims to understand whether the presence and characteristics of UFOs depend on the state of the sources, by studying the relations between the incidence of UFOs and the accretion properties of AGNs. We collected X-ray spectroscopic flux-limited XMM data of 33 observations of local (z<0.2) type 1 AGNs. We modeled their spectra in the 2-12 keV band using a combination of direct-continuum and reflection components and searched for absorption features. This represents the first systematic search for UFOs up to 12 keV. We performed Monte Carlo simulations to assess the statistical significance of the detected lines. We report strong detections of UFOs in six sources of the sample at the >95% confidence level via MC simulations, corresponding to a fraction of 18% in our sample. From the observed energies of each absorption line, we evaluated the respective wind velocities, which in some cases exceed 40\% of the speed of light. The velocity distribution found in this work is therefore shifted to higher energies than those found in previous searches for UFOs in local sources, which were limited to 10 keV. Moreover, our analysis shows no correlation between the accretion properties of the SMBHs and the presence of winds. Furthermore, our study highlights the temporal variability of UFOs.

Green valley (GV) galaxies are thought to represent a transitional population between star-forming and quiescent systems. However, their spatial distribution relative to galaxy systems remains unclear, particularly in relation to the large-scale environmental influence on galaxy quenching. We aim to determine whether GV galaxies preferentially inhabit specific environments within galaxy systems. We analyse the spatial distribution of GV galaxies using the cluster-galaxy cross-correlation function (CCF), based on the hydrodynamical simulation Illustris TNG300-1 (TNG) and observational data from the Sloan Digital Sky Survey (SDSS). Galaxy systems with $\log(M_{200}/M_{\odot}) \geq 13.5$ are used as cluster centres, while galaxies classified as blue, green, or red serve as tracers for the correlation analysis. In TNG, GV galaxies show an increasing relative fraction with cluster-centric distance, peaking in the outskirts, particularly for low-mass galaxies and haloes, and in some cases the GV fraction exceeds that of red galaxies. SDSS data reveal qualitatively similar trends, with the GV fraction remaining below that of red galaxies at all scales. Mock catalogues built from TNG and matched to SDSS selection functions reproduce the observational signal, indicating that projection effects drive the differences between datasets. GV galaxies preferentially reside in the outskirts of galaxy systems as satellites bound to the central halo, supporting a scenario in which they are transitioning objects influenced by environmental quenching.

Cosmological simulations of galaxy clusters are unable to resolve dwarf galaxies due to limited numerical resolution which drives the artificial disruption of dark matter substructures. We address these limitations by combining the results of the cosmological hydrodynamical simulation TNG50 in $Λ$CDM with an empirical model of tidal evolution of cluster galaxies calibrated using high-resolution idealized N-body simulations. Applied to the three most massive clusters in TNG50, our model allows us to study the stellar mass and radial distribution of dwarfs well below the formal resolution limit of the parent simulation. We find that, at $z=0$, clusters with virial mass $M_{200} \sim 10^{14}~\mathrm{M_\odot}$ host a vast population of dwarf galaxies within the virial radius, amounting to $2000$-$7000$ systems with $M_* > 100~\mathrm{M_\odot}$. Taken together, these satellites follow a radial distribution that matches the underlying dark matter profile of the host. However, applying a minimum mass or luminosity threshold for detection, as expected in observational studies, tends to exclude the most heavily-stripped objects, which tend to populate the inner regions. Future surveys targeting ultra-faint galaxies in group and cluster environments, such as those made possible by the Euclid, Rubin, or Roman telescopes, will be fundamental to refute or confirm this prediction.

Future weak lensing analyses with the Nancy Grace Roman Space Telescope will require highly realistic image simulations to control shear systematics at unprecedented precision. A key limitation of existing approaches is their reliance on analytic light-profile models, which cannot fully capture the complex, non-parametric morphologies revealed by high-resolution observations. We present a diffusion-based framework for generating realistic galaxy image simulations tailored to the weak lensing requirements of the Roman High Latitude Survey. We construct Roman-like galaxy images from multi-band JWST/NIRCam observations in the GOODS-S and GOODS-N fields, transforming them into the Roman observing regime through point-spread-function matching, pixel-scale conversion, and interloper masking that preserves correlated noise properties. These data are used to train a denoising diffusion probabilistic model to generate multi-band galaxy postage stamps in the Roman Y, J, and H filters. We validate the generated sample against an independent dataset using a consistent photometric pipeline, comparing key galaxy observables including magnitude, size, ellipticity, peak surface brightness, and three-band colors. The generated galaxies reproduce both the marginal distributions and the covariance structure of these properties, with only modest deviations in low-occupancy regions of parameter space. These results demonstrate that diffusion models provide a scalable and physically motivated alternative to analytic simulations, enabling high-fidelity galaxy populations for Roman weak lensing calibration and, more generally, for survey preparation in upcoming cosmological experiments.

The Lighthouse pulsar (PSR J1101$-$6101) sports a bright X-ray trail and filament. The synchrotron emission from both structures is expected to be polarized, with electric vector position angle (EVPA) perpendicular to the magnetic field direction and polarization degree (PD) indicating the local degree of magnetic turbulence. We present a 1 megasecond Imaging X-ray Polarimetry Explorer (IXPE) observation of the Lighthouse complex. At the 99% confidence level, we detect the filament polarization with PD $55 \pm 18\%$ and EVPA indicating a magnetic field parallel to the filament axis. The large PD implies a turbulent magnetic field weaker than the background field, in conflict with some existing models. We also detect polarization from the pulsar and trail. The trail's X-ray polarization is nearly orthogonal to the radio polarization, suggesting spatial separation between the X-ray- and radio-emitting leptons. The pulsar polarization is well-fit by the rotating vector model.

Stars embedded in the accretion disks of active galactic nuclei (AGN) can accrete rapidly from their surroundings, dramatically altering their structure and evolution. However, feedback from the release of gravitational potential energy and radiative enthalpy by accreting gas can limit accretion rates, as recently demonstrated in radiation hydrodynamics simulations. To determine the importance of these effects neglected in earlier stellar evolution calculations, we incorporate these feedback processes into a semi-analytical model of stellar structure and evolution and conduct a suite of calculations spanning a broad parameter space of AGN disk conditions drawn from $α$-disk models with central black hole masses $M_\bullet/M_\odot \in [10^6, 10^9]$. We find that accretion feedback limits stellar accretion rates below $\sim 10^{-1}\,M_\odot\,\mathrm{yr}^{-1}$, reducing the sensitivity of stellar evolution on disk properties. This suppression eliminates runaway accretion in models where it would otherwise occur, broadening the parameter space over which stars can reach long-lived ``immortal'' equilibria between accretion and mass loss. When gap opening is also accounted for, accretion feedback significantly alters stellar properties: it can reduce accretion and mass-loss rates by over an order of magnitude, reducing the strength of accretion shocks and thereby increasing equilibrium stellar masses and radii. These higher masses correspond to higher intrinsic luminosities, suggesting that neglecting accretion feedback may lead to an underestimate of disk chemical enrichment rates. Additionally, accretion feedback is important for predicting the properties of stellar populations within AGN disks, and associated transient phenomena.

Highly magnified stars at cosmological distances ($z \gtrsim 0.7$) become detectable thanks to microlensing by intracluster stars near the critical curves of galaxy clusters. Multi-epoch photometric campaigns targeting caustic crossing galaxies magnified by massive galaxy clusters enable the detection of these objects as transient events. Such stars provide unique opportunities to study stellar populations at early cosmic times, probe the nature of dark matter, reveal small-scale structure in the cluster, and improve lens models. To date, only a few dozen high-redshift stars have been reported, with a single lensed galaxy, the Dragon, holding the current record of 44 detections. These numbers, however, remain insufficient to exploit their full potential. In this paper, owing to the inclusion of new observations, we report the identification of more than 100 magnified stellar events in the Dragon, behind the massive galaxy cluster Abell 370. The relatively low redshift of the Dragon ($z\approx0.725$) facilitates the detection of its most massive stars. Using imaging data from three different cycles (2022--2024) with the James Webb Space Telescope, we apply a time-domain technique to identify flux variations associated with caustic-crossing events. From the spatial distribution of stellar events we constrain the high-end slope of the stellar luminosity function, finding $β=2.18^{+0.20}_{-0.30}$. Alternatively, assuming a fixed slope, we constrain the microlens surface mass density. In addition, we examine the parity asymmetry of the detected caustic-crossing events, a proposed probe of wave dark matter, and find that it remains present. We also use the events to trace the regions of highest magnification, offering an alternative way to map the system critical curves.