arXiv Daily Digest - 2026-01-15

Vision-Language-Action (VLA) tasks require reasoning over complex visual scenes and executing adaptive actions in dynamic environments. While recent studies on reasoning VLAs show that explicit chain-of-thought (CoT) can improve generalization, they suffer from high inference latency due to lengthy reasoning traces. We propose Fast-ThinkAct, an efficient reasoning framework that achieves compact yet performant planning through verbalizable latent reasoning. Fast-ThinkAct learns to reason efficiently with latent CoTs by distilling from a teacher, driven by a preference-guided objective to align manipulation trajectories that transfers both linguistic and visual planning capabilities for embodied control. This enables reasoning-enhanced policy learning that effectively connects compact reasoning to action execution. Extensive experiments across diverse embodied manipulation and reasoning benchmarks demonstrate that Fast-ThinkAct achieves strong performance with up to 89.3\% reduced inference latency over state-of-the-art reasoning VLAs, while maintaining effective long-horizon planning, few-shot adaptation, and failure recovery.

Transformer-based language models often achieve strong results on mathematical reasoning benchmarks while remaining fragile on basic numerical understanding and arithmetic operations. A central limitation is that numbers are processed as symbolic tokens whose embeddings do not explicitly encode numerical value, leading to systematic errors. We introduce a value-aware numerical representation that augments standard tokenized inputs with a dedicated prefix token whose embedding is explicitly conditioned on the underlying numerical value. This mechanism injects magnitude information directly into the model's input space while remaining compatible with existing tokenizers and decoder-only Transformer architectures. Evaluation on arithmetic tasks shows that the proposed approach outperforms baselines across numerical formats, tasks, and operand lengths. These results indicate that explicitly encoding numerical value is an effective and efficient way to improve fundamental numerical robustness in language models.

Code generation tasks aim to automate the conversion of user requirements into executable code, significantly reducing manual development efforts and enhancing software productivity. The emergence of large language models (LLMs) has significantly advanced code generation, though their efficiency is still impacted by certain inherent architectural constraints. Each token generation necessitates a complete inference pass, requiring persistent retention of contextual information in memory and escalating resource consumption. While existing research prioritizes inference-phase optimizations such as prompt compression and model quantization, the generation phase remains underexplored. To tackle these challenges, we propose a knowledge-infused framework named ShortCoder, which optimizes code generation efficiency while preserving semantic equivalence and readability. In particular, we introduce: (1) ten syntax-level simplification rules for Python, derived from AST-preserving transformations, achieving 18.1% token reduction without functional compromise; (2) a hybrid data synthesis pipeline integrating rule-based rewriting with LLM-guided refinement, producing ShorterCodeBench, a corpus of validated tuples of original code and simplified code with semantic consistency; (3) a fine-tuning strategy that injects conciseness awareness into the base LLMs. Extensive experimental results demonstrate that ShortCoder consistently outperforms state-of-the-art methods on HumanEval, achieving an improvement of 18.1%-37.8% in generation efficiency over previous methods while ensuring the performance of code generation.

LLMs are increasingly being integrated into clinical workflows, yet they often lack clinical empathy, an essential aspect of effective doctor-patient communication. Existing NLP frameworks focus on reactively labeling empathy in doctors' responses but offer limited support for anticipatory modeling of empathy needs, especially in general health queries. We introduce the Empathy Applicability Framework (EAF), a theory-driven approach that classifies patient queries in terms of the applicability of emotional reactions and interpretations, based on clinical, contextual, and linguistic cues. We release a benchmark of real patient queries, dual-annotated by Humans and GPT-4o. In the subset with human consensus, we also observe substantial human-GPT alignment. To validate EAF, we train classifiers on human-labeled and GPT-only annotations to predict empathy applicability, achieving strong performance and outperforming the heuristic and zero-shot LLM baselines. Error analysis highlights persistent challenges: implicit distress, clinical-severity ambiguity, and contextual hardship, underscoring the need for multi-annotator modeling, clinician-in-the-loop calibration, and culturally diverse annotation. EAF provides a framework for identifying empathy needs before response generation, establishes a benchmark for anticipatory empathy modeling, and enables supporting empathetic communication in asynchronous healthcare.

The rapid evolution of Large Language Models (LLMs) has strongly impacted software engineering, leading to a growing number of studies on automated unit test generation. However, the standalone use of LLMs without post-processing has proven insufficient, often producing tests that fail to compile or achieve high coverage. Several techniques have been proposed to address these issues, reporting improvements in test compilation and coverage. While important, LLM-based test generation techniques have been evaluated against relatively weak baselines (for todays' standards), i.e., old LLM versions and relatively weak prompts, which may exacerbate the performance contribution of the approaches. In other words, stronger (newer) LLMs may obviate any advantage these techniques bring. We investigate this issue by replicating four state-of-the-art LLM-based test generation tools, HITS, SymPrompt, TestSpark, and CoverUp that include engineering components aimed at guiding the test generation process through compilation and execution feedback, and evaluate their relative effectiveness and efficiency over a plain LLM test generation method. We integrate current LLM versions in all approaches and run an experiment on 393 classes and 3,657 methods. Our results show that the plain LLM approach can outperform previous state-of-the-art approaches in all test effectiveness metrics we used: line coverage (by 17.72%), branch coverage (by 19.80%) and mutation score (by 20.92%), and it does so at a comparable cost (LLM queries). We also observe that the granularity at which the plain LLM is applied has a significant impact on the cost. We therefore propose targeting first the program classes, where test generation is more efficient, and then the uncovered methods to reduce the number of LLM requests. This strategy achieves comparable (slightly higher) effectiveness while requiring about 20% fewer LLM requests.

As Large Language Models (LLMs) continue to scale, post-training pruning has emerged as a promising approach to reduce computational costs while preserving performance. Existing methods such as SparseGPT and Wanda achieve high sparsity through layer-wise weight reconstruction or activation-aware magnitude pruning, but rely on uniform or hand-crafted heuristics to determine per-layer sparsity ratios. Moreover, recent work has shown that pruned LLMs suffer from severe factual knowledge degradation, with structured pruning methods experiencing near-total collapse in factual question-answering capabilities. We introduce agent-guided pruning, where a foundation model acts as an adaptive pruning agent to intelligently select which layers to prune at each iteration while preserving critical knowledge pathways. Our method constructs layer-wise sensitivity profiles by combining Wanda-inspired weight-activation metrics with gradient importance scores, normalized as z-scores for model-agnostic comparison. These statistics are processed by an LLM agent equipped with self-reflection capabilities, enabling it to learn from previous pruning outcomes and iteratively refine its strategy. A checkpoint rollback mechanism maintains model quality by reverting when perplexity degradation exceeds a threshold. We evaluate our approach on Qwen3 models (4B and 8B parameters) at approximately 45% sparsity, demonstrating substantial improvements over structured pruning baselines: 56% relative improvement in MMLU accuracy, 19x better factual knowledge retention on FreebaseQA, and 69% lower perplexity degradation. Notably, our framework requires no retraining, operates in a model-agnostic manner, and exhibits effective self-correction with only 2-4 rollbacks across 21-40 iterations, demonstrating that foundation models can effectively guide the compression of other foundation models.

Structure-based and ligand-based computational drug design have traditionally relied on disjoint data sources and modeling assumptions, limiting their joint use at scale. In this work, we introduce Contrastive Geometric Learning for Unified Computational Drug Design (ConGLUDe), a single contrastive geometric model that unifies structure- and ligand-based training. ConGLUDe couples a geometric protein encoder that produces whole-protein representations and implicit embeddings of predicted binding sites with a fast ligand encoder, removing the need for pre-defined pockets. By aligning ligands with both global protein representations and multiple candidate binding sites through contrastive learning, ConGLUDe supports ligand-conditioned pocket prediction in addition to virtual screening and target fishing, while being trained jointly on protein-ligand complexes and large-scale bioactivity data. Across diverse benchmarks, ConGLUDe achieves state-of-the-art zero-shot virtual screening performance in settings where no binding pocket information is provided as input, substantially outperforms existing methods on a challenging target fishing task, and demonstrates competitive ligand-conditioned pocket selection. These results highlight the advantages of unified structure-ligand training and position ConGLUDe as a step toward general-purpose foundation models for drug discovery.

Large Language Model (LLM) routers dynamically select optimal models for given inputs. Existing approaches typically assume access to ground-truth labeled data, which is often unavailable in practice, especially when user request distributions are heterogeneous and unknown. We introduce Routing with Generated Data (RGD), a challenging setting in which routers are trained exclusively on generated queries and answers produced from high-level task descriptions by generator LLMs. We evaluate query-answer routers (using both queries and labels) and query-only routers across four diverse benchmarks and 12 models, finding that query-answer routers degrade faster than query-only routers as generator quality decreases. Our analysis reveals two crucial characteristics of effective generators: they must accurately respond to their own questions, and their questions must produce sufficient performance differentiation among the model pool. We then show how filtering for these characteristics can improve the quality of generated data. We further propose CASCAL, a novel query-only router that estimates model correctness through consensus voting and identifies model-specific skill niches via hierarchical clustering. CASCAL is substantially more robust to generator quality, outperforming the best query-answer router by 4.6% absolute accuracy when trained on weak generator data.

Deep research systems are widely used for multi-step web research, analysis, and cross-source synthesis, yet their evaluation remains challenging. Existing benchmarks often require annotation-intensive task construction, rely on static evaluation dimensions, or fail to reliably verify facts when citations are missing. To bridge these gaps, we introduce DeepResearchEval, an automated framework for deep research task construction and agentic evaluation. For task construction, we propose a persona-driven pipeline generating realistic, complex research tasks anchored in diverse user profiles, applying a two-stage filter Task Qualification and Search Necessity to retain only tasks requiring multi-source evidence integration and external retrieval. For evaluation, we propose an agentic pipeline with two components: an Adaptive Point-wise Quality Evaluation that dynamically derives task-specific evaluation dimensions, criteria, and weights conditioned on each generated task, and an Active Fact-Checking that autonomously extracts and verifies report statements via web search, even when citations are missing.

Multi-Task Learning (MTL) combined with Low-Rank Adaptation (LoRA) has emerged as a promising direction for parameter-efficient deployment of Large Language Models (LLMs). By sharing a single adapter across multiple tasks, one can significantly reduce storage overhead. However, this approach suffers from negative transfer, where conflicting gradient updates from distinct tasks degrade the performance of individual tasks compared to single-task fine-tuning. This problem is exacerbated in LoRA due to the low-rank constraint, which limits the optimization landscape's capacity to accommodate diverse task requirements. In this paper, we propose Ortho-LoRA, a gradient projection method specifically tailored for the bipartite structure of LoRA. Ortho-LoRA dynamically projects conflicting task gradients onto the orthogonal complement of each other within the intrinsic LoRA subspace. Extensive experiments on the GLUE benchmark demonstrate that Ortho-LoRA effectively mitigates task interference, outperforming standard joint training and recovering 95\% of the performance gap between multi-task and single-task baselines with negligible computational overhead.

Modern supply chains are increasingly exposed to disruptions from geopolitical events, demand shocks, trade restrictions, to natural disasters. While many of these disruptions originate deep in the supply network, most companies still lack visibility beyond Tier-1 suppliers, leaving upstream vulnerabilities undetected until the impact cascades downstream. To overcome this blind-spot and move from reactive recovery to proactive resilience, we introduce a minimally supervised agentic AI framework that autonomously monitors, analyses, and responds to disruptions across extended supply networks. The architecture comprises seven specialised agents powered by large language models and deterministic tools that jointly detect disruption signals from unstructured news, map them to multi-tier supplier networks, evaluate exposure based on network structure, and recommend mitigations such as alternative sourcing options. \rev{We evaluate the framework across 30 synthesised scenarios covering three automotive manufacturers and five disruption classes. The system achieves high accuracy across core tasks, with F1 scores between 0.962 and 0.991, and performs full end-to-end analyses in a mean of 3.83 minutes at a cost of \$0.0836 per disruption. Relative to industry benchmarks of multi-day, analyst-driven assessments, this represents a reduction of more than three orders of magnitude in response time. A real-world case study of the 2022 Russia-Ukraine conflict further demonstrates operational applicability. This work establishes a foundational step toward building resilient, proactive, and autonomous supply chains capable of managing disruptions across deep-tier networks.

Multi-agent systems have evolved into practical LLM-driven collaborators for many applications, gaining robustness from diversity and cross-checking. However, multi-agent RL (MARL) training is resource-intensive and unstable: co-adapting teammates induce non-stationarity, and rewards are often sparse and high-variance. Therefore, we introduce \textbf{Multi-Agent Test-Time Reinforcement Learning (MATTRL)}, a framework that injects structured textual experience into multi-agent deliberation at inference time. MATTRL forms a multi-expert team of specialists for multi-turn discussions, retrieves and integrates test-time experiences, and reaches consensus for final decision-making. We also study credit assignment for constructing a turn-level experience pool, then reinjecting it into the dialogue. Across challenging benchmarks in medicine, math, and education, MATTRL improves accuracy by an average of 3.67\% over a multi-agent baseline, and by 8.67\% over comparable single-agent baselines. Ablation studies examine different credit-assignment schemes and provide a detailed comparison of how they affect training outcomes. MATTRL offers a stable, effective and efficient path to distribution-shift-robust multi-agent reasoning without tuning.

Tabular Foundation Models (TFMs) have recently shown strong in-context learning capabilities on structured data, achieving zero-shot performance comparable to traditional machine learning methods. We find that zero-shot TFMs already achieve strong performance, while the benefits of fine-tuning are highly model and data-dependent. Meta-learning and PEFT provide moderate gains under specific conditions, whereas full supervised fine-tuning (SFT) often reduces accuracy or calibration quality. This work presents the first comprehensive study of fine-tuning in TFMs across benchmarks including TALENT, OpenML-CC18, and TabZilla. We compare Zero-Shot, Meta-Learning, Supervised (SFT), and parameter-efficient (PEFT) approaches, analyzing how dataset factors such as imbalance, size, and dimensionality affect outcomes. Our findings cover performance, calibration, and fairness, offering practical guidelines on when fine-tuning is most beneficial and its limitations.

Word Sense Disambiguation (WSD) has been widely evaluated using the semantic frameworks of WordNet, BabelNet, and the Oxford Dictionary of English. However, for the UCREL Semantic Analysis System (USAS) framework, no open extensive evaluation has been performed beyond lexical coverage or single language evaluation. In this work, we perform the largest semantic tagging evaluation of the rule based system that uses the lexical resources in the USAS framework covering five different languages using four existing datasets and one novel Chinese dataset. We create a new silver labelled English dataset, to overcome the lack of manually tagged training data, that we train and evaluate various mono and multilingual neural models in both mono and cross-lingual evaluation setups with comparisons to their rule based counterparts, and show how a rule based system can be enhanced with a neural network model. The resulting neural network models, including the data they were trained on, the Chinese evaluation dataset, and all of the code have been released as open resources.

Text-to-image (T2I) models are increasingly popular, producing a large share of AI-generated images online. To compare model quality, voting-based leaderboards have become the standard, relying on anonymized model outputs for fairness. In this work, we show that such anonymity can be easily broken. We find that generations from each T2I model form distinctive clusters in the image embedding space, enabling accurate deanonymization without prompt control or training data. Using 22 models and 280 prompts (150K images), our centroid-based method achieves high accuracy and reveals systematic model-specific signatures. We further introduce a prompt-level distinguishability metric and conduct large-scale analyses showing how certain prompts can lead to near-perfect distinguishability. Our findings expose fundamental security flaws in T2I leaderboards and motivate stronger anonymization defenses.

While GUI agents have shown strong performance under explicit and completion instructions, real-world deployment requires aligning with users' more complex implicit intents. In this work, we highlight Hierarchical Implicit Intent Alignment for Personalized GUI Agent (PersonalAlign), a new agent task that requires agents to leverage long-term user records as persistent context to resolve omitted preferences in vague instructions and anticipate latent routines by user state for proactive assistance. To facilitate this study, we introduce AndroidIntent, a benchmark designed to evaluate agents' ability in resolving vague instructions and providing proactive suggestions through reasoning over long-term user records. We annotated 775 user-specific preferences and 215 routines from 20k long-term records across different users for evaluation. Furthermore, we introduce Hierarchical Intent Memory Agent (HIM-Agent), which maintains a continuously updating personal memory and hierarchically organizes user preferences and routines for personalization. Finally, we evaluate a range of GUI agents on AndroidIntent, including GPT-5, Qwen3-VL, and UI-TARS, further results show that HIM-Agent significantly improves both execution and proactive performance by 15.7% and 7.3%.

Large-scale optimization is a key backbone of modern business decision-making. However, building these models is often labor-intensive and time-consuming. We address this by proposing LEAN-LLM-OPT, a LightwEight AgeNtic workflow construction framework for LLM-assisted large-scale OPTimization auto-formulation. LEAN-LLM-OPT takes as input a problem description together with associated datasets and orchestrates a team of LLM agents to produce an optimization formulation. Specifically, upon receiving a query, two upstream LLM agents dynamically construct a workflow that specifies, step-by-step, how optimization models for similar problems can be formulated. A downstream LLM agent then follows this workflow to generate the final output. Leveraging LLMs' text-processing capabilities and common modeling practices, the workflow decomposes the modeling task into a sequence of structured sub-tasks and offloads mechanical data-handling operations to auxiliary tools. This design alleviates the downstream agent's burden related to planning and data handling, allowing it to focus on the most challenging components that cannot be readily standardized. Extensive simulations show that LEAN-LLM-OPT, instantiated with GPT-4.1 and the open source gpt-oss-20B, achieves strong performance on large-scale optimization modeling tasks and is competitive with state-of-the-art approaches. In addition, in a Singapore Airlines choice-based revenue management use case, LEAN-LLM-OPT demonstrates practical value by achieving leading performance across a range of scenarios. Along the way, we introduce Large-Scale-OR and Air-NRM, the first comprehensive benchmarks for large-scale optimization auto-formulation. The code and data of this work is available at https://github.com/CoraLiang01/lean-llm-opt.

Taxonomies form the backbone of structured knowledge representation across diverse domains, enabling applications such as e-commerce catalogs, semantic search, and biomedical discovery. Yet, manual taxonomy expansion is labor-intensive and cannot keep pace with the emergence of new concepts. Existing automated methods rely on point-based vector embeddings, which model symmetric similarity and thus struggle with the asymmetric "is-a" relationships that are fundamental to taxonomies. Box embeddings offer a promising alternative by enabling containment and disjointness, but they face key issues: (i) unstable gradients at the intersection boundaries, (ii) no notion of semantic uncertainty, and (iii) limited capacity to represent polysemy or ambiguity. We address these shortcomings with TaxoBell, a Gaussian box embedding framework that translates between box geometries and multivariate Gaussian distributions, where means encode semantic location and covariances encode uncertainty. Energy-based optimization yields stable optimization, robust modeling of ambiguous concepts, and interpretable hierarchical reasoning. Extensive experimentation on five benchmark datasets demonstrates that TaxoBell significantly outperforms eight state-of-the-art taxonomy expansion baselines by 19% in MRR and around 25% in Recall@k. We further demonstrate the advantages and pitfalls of TaxoBell with error analysis and ablation studies.

Large Language Models (LLMs), despite their remarkable capabilities across NLP tasks, struggle with phonologically-grounded phenomena like rhyme detection and generation. This is even more evident in lower-resource languages such as Modern Greek. In this paper, we present a hybrid system that combines LLMs with deterministic phonological algorithms to achieve accurate rhyme identification/analysis and generation. Our approach implements a comprehensive taxonomy of Greek rhyme types, including Pure, Rich, Imperfect, Mosaic, and Identical Pre-rhyme Vowel (IDV) patterns, and employs an agentic generation pipeline with phonological verification. We evaluate multiple prompting strategies (zero-shot, few-shot, Chain-of-Thought, and RAG-augmented) across several LLMs including Claude 3.7 and 4.5, GPT-4o, Gemini 2.0 and open-weight models like Llama 3.1 8B and 70B and Mistral Large. Results reveal a significant "Reasoning Gap": while native-like models (Claude 3.7) perform intuitively (40\% accuracy in identification), reasoning-heavy models (Claude 4.5) achieve state-of-the-art performance (54\%) only when prompted with Chain-of-Thought. Most critically, pure LLM generation fails catastrophically (under 4\% valid poems), while our hybrid verification loop restores performance to 73.1\%. We release our system and a crucial, rigorously cleaned corpus of 40,000+ rhymes, derived from the Anemoskala and Interwar Poetry corpora, to support future research.

Efficiently optimizing battery charging protocols is challenging because each evaluation is slow, costly, and non-differentiable. Many existing approaches address this difficulty by heavily constraining the protocol search space, which limits the diversity of protocols that can be explored, preventing the discovery of higher-performing solutions. We introduce two gradient-free, LLM-driven closed-loop methods: Prompt-to-Optimizer (P2O), which uses an LLM to propose the code for small neural-network-based protocols, which are then trained by an inner loop, and Prompt-to-Protocol (P2P), which simply writes an explicit function for the current and its scalar parameters. Across our case studies, LLM-guided P2O outperforms neural networks designed by Bayesian optimization, evolutionary algorithms, and random search. In a realistic fast charging scenario, both P2O and P2P yield around a 4.2 percent improvement in state of health (capacity retention based health metric under fast charging cycling) over a state-of-the-art multi-step constant current (CC) baseline, with P2P achieving this under matched evaluation budgets (same number of protocol evaluations). These results demonstrate that LLMs can expand the space of protocol functional forms, incorporate language-based constraints, and enable efficient optimization in high cost experimental settings.

The rapid adoption of large language model (LLM)-based systems -- from chatbots to autonomous agents capable of executing code and financial transactions -- has created a new attack surface that existing security frameworks inadequately address. The dominant framing of these threats as "prompt injection" -- a catch-all phrase for security failures in LLM-based systems -- obscures a more complex reality: Attacks on LLM-based systems increasingly involve multi-step sequences that mirror traditional malware campaigns. In this paper, we propose that attacks targeting LLM-based applications constitute a distinct class of malware, which we term \textit{promptware}, and introduce a five-step kill chain model for analyzing these threats. The framework comprises Initial Access (prompt injection), Privilege Escalation (jailbreaking), Persistence (memory and retrieval poisoning), Lateral Movement (cross-system and cross-user propagation), and Actions on Objective (ranging from data exfiltration to unauthorized transactions). By mapping recent attacks to this structure, we demonstrate that LLM-related attacks follow systematic sequences analogous to traditional malware campaigns. The promptware kill chain offers security practitioners a structured methodology for threat modeling and provides a common vocabulary for researchers across AI safety and cybersecurity to address a rapidly evolving threat landscape.

Machine unlearning is becoming essential for building trustworthy and compliant language models. Yet unlearning success varies considerably across individual samples: some are reliably erased, while others persist despite the same procedure. We argue that this disparity is not only a data-side phenomenon, but also reflects model-internal mechanisms that encode and protect memorized information. We study this problem from a mechanistic perspective based on model circuits--structured interaction pathways that govern how predictions are formed. We propose Circuit-guided Unlearning Difficulty (CUD), a {\em pre-unlearning} metric that assigns each sample a continuous difficulty score using circuit-level signals. Extensive experiments demonstrate that CUD reliably separates intrinsically easy and hard samples, and remains stable across unlearning methods. We identify key circuit-level patterns that reveal a mechanistic signature of difficulty: easy-to-unlearn samples are associated with shorter, shallower interactions concentrated in earlier-to-intermediate parts of the original model, whereas hard samples rely on longer and deeper pathways closer to late-stage computation. Compared to existing qualitative studies, CUD takes a first step toward a principled, fine-grained, and interpretable analysis of unlearning difficulty; and motivates the development of unlearning methods grounded in model mechanisms.

As artificial intelligence (AI) is increasingly integrated into news production, calls for transparency about the use of AI have gained considerable traction. Recent studies suggest that AI disclosures can lead to a ``transparency dilemma'', where disclosure reduces readers' trust. However, little is known about how the \textit{level of detail} in AI disclosures influences trust and contributes to this dilemma within the news context. In this 3$\times$2$\times$2 mixed factorial study with 40 participants, we investigate how three levels of AI disclosures (none, one-line, detailed) across two types of news (politics and lifestyle) and two levels of AI involvement (low and high) affect news readers' trust. We measured trust using the News Media Trust questionnaire, along with two decision behaviors: source-checking and subscription decisions. Questionnaire responses and subscription rates showed a decline in trust only for detailed AI disclosures, whereas source-checking behavior increased for both one-line and detailed disclosures, with the effect being more pronounced for detailed disclosures. Insights from semi-structured interviews suggest that source-checking behavior was primarily driven by interest in the topic, followed by trust, whereas trust was the main factor influencing subscription decisions. Around two-thirds of participants expressed a preference for detailed disclosures, while most participants who preferred one-line indicated a need for detail-on-demand disclosure formats. Our findings show that not all AI disclosures lead to a transparency dilemma, but instead reflect a trade-off between readers' desire for more transparency and their trust in AI-assisted news content.

Reproducing system-level concurrency bugs requires both input data and the precise interleaving order of system calls. This process is challenging because such bugs are non-deterministic, and bug reports often lack the detailed information needed. Additionally, the unstructured nature of reports written in natural language makes it difficult to extract necessary details. Existing tools are inadequate to reproduce these bugs due to their inability to manage the specific interleaving at the system call level. To address these challenges, we propose SysPro, a novel approach that automatically extracts relevant system call names from bug reports and identifies their locations in the source code. It generates input data by utilizing information retrieval, regular expression matching, and the category-partition method. This extracted input and interleaving data are then used to reproduce bugs through dynamic source code instrumentation. Our empirical study on real-world benchmarks demonstrates that SysPro is both effective and efficient at localizing and reproducing system-level concurrency bugs from bug reports.

Accurate and early perception of potential intrusion targets is essential for ensuring the safety of railway transportation systems. However, most existing systems focus narrowly on object classification within fixed visual scopes and apply rule-based heuristics to determine intrusion status, often overlooking targets that pose latent intrusion risks. Anticipating such risks requires the cognition of spatial context and temporal dynamics for the object of interest (OOI), which presents challenges for conventional visual models. To facilitate deep intrusion perception, we introduce a novel benchmark, CogRail, which integrates curated open-source datasets with cognitively driven question-answer annotations to support spatio-temporal reasoning and prediction. Building upon this benchmark, we conduct a systematic evaluation of state-of-the-art visual-language models (VLMs) using multimodal prompts to identify their strengths and limitations in this domain. Furthermore, we fine-tune VLMs for better performance and propose a joint fine-tuning framework that integrates three core tasks, position perception, movement prediction, and threat analysis, facilitating effective adaptation of general-purpose foundation models into specialized models tailored for cognitive intrusion perception. Extensive experiments reveal that current large-scale multimodal models struggle with the complex spatial-temporal reasoning required by the cognitive intrusion perception task, underscoring the limitations of existing foundation models in this safety-critical domain. In contrast, our proposed joint fine-tuning framework significantly enhances model performance by enabling targeted adaptation to domain-specific reasoning demands, highlighting the advantages of structured multi-task learning in improving both accuracy and interpretability. Code will be available at https://github.com/Hub-Tian/CogRail.

Scientific Workflow Management Systems (SWfMSs) such as Nextflow have become essential software frameworks for conducting reproducible, scalable, and portable computational analyses in data-intensive fields like genomics, transcriptomics, and proteomics. Building on Nextflow, the nf-core community curates standardized, peer-reviewed pipelines that follow strict testing, documentation, and governance guidelines. Despite its broad adoption, little is known about the challenges users face during the development and maintenance of these pipelines. This paper presents an empirical study of 25,173 issues and pull requests from these pipelines to uncover recurring challenges, management practices, and perceived difficulties. Using BERTopic modeling, we identify 13 key challenges, including pipeline development and integration, bug fixing, integrating genomic data, managing CI configurations, and handling version updates. We then examine issue resolution dynamics, showing that 89.38\% of issues and pull requests are eventually closed, with half resolved within three days. Statistical analysis reveals that the presence of labels (large effect, $δ$ = 0.94) and code snippets (medium effect, $δ$ = 0.50) significantly improve resolution likelihood. Further analysis reveals that tool development and repository maintenance poses the most significant challenges, followed by testing pipelines and CI configurations, and debugging containerized pipelines. Overall, this study provides actionable insights into the collaborative development and maintenance of nf-core pipelines, highlighting opportunities to enhance their usability, sustainability, and reproducibility.

Reinforcement learning (RL)-based enhancement of large language models (LLMs) often leads to reduced output diversity, undermining their utility in open-ended tasks like creative writing. Current methods lack explicit mechanisms for guiding diverse exploration and instead prioritize optimization efficiency and performance over diversity. This paper proposes an RL framework structured around a semi-structured long Chain-of-Thought (CoT), in which the generation process is decomposed into explicitly planned intermediate steps. We introduce a Diverse Planning Branching method that strategically introduces divergence at the planning phase based on diversity variation, alongside a group-aware diversity reward to encourage distinct trajectories. Experimental results on creative writing benchmarks demonstrate that our approach significantly improves output diversity without compromising generation quality, consistently outperforming existing baselines.

Vision-based policies for robot manipulation have achieved significant recent success, but are still brittle to distribution shifts such as camera viewpoint variations. Robot demonstration data is scarce and often lacks appropriate variation in camera viewpoints. Simulation offers a way to collect robot demonstrations at scale with comprehensive coverage of different viewpoints, but presents a visual sim2real challenge. To bridge this gap, we propose MANGO -- an unpaired image translation method with a novel segmentation-conditioned InfoNCE loss, a highly-regularized discriminator design, and a modified PatchNCE loss. We find that these elements are crucial for maintaining viewpoint consistency during sim2real translation. When training MANGO, we only require a small amount of fixed-camera data from the real world, but show that our method can generate diverse unseen viewpoints by translating simulated observations. In this domain, MANGO outperforms all other image translation methods we tested. Imitation-learning policies trained on data augmented by MANGO are able to achieve success rates as high as 60\% on views that the non-augmented policy fails completely on.

In recent years, foundation models have become very popular due to their exceptional performance, mainly in natural language (NLP) tasks where they were first introduced. These models usually consist of hundreds of millions, or even billions, of parameters, making them resource-intensive during training and in production systems, leading to increased costs. This paper focuses on the reduction of a foundation's model size when applied to music information retrieval (MIR) tasks. Our research combines the Branchformer architecture with SummaryMixing, which were first applied in speech recognition, along with a random quantization process. To facilitate reproducibility, we conduct pre-training on publicly available datasets, complemented by a proprietary dataset comparable in scale to other private datasets reported in the literature. We ensure robust evaluation by using a framework consisting of a variety of downstream MIR tasks. Our results show that our architecture achieves competitive performance when compared with other state-of-the-art models that use multi-head self-attention, while reducing the model size from 8.5% up to 12.3%.

Online information access (IA) platforms are targets of authoritarian capture. These concerns are particularly serious and urgent today in light of the rising levels of democratic erosion worldwide, the emerging capabilities of generative AI technologies such as AI persuasion, and the increasing concentration of economic and political power in the hands of Big Tech. This raises the question of what alternative IA infrastructure we must reimagine and build to mitigate the risks of authoritarian capture of our information ecosystems. We explore this question through the lens of Paulo Freire's theories of emancipatory pedagogy. Freire's theories provide a radically different lens for exploring IA's sociotechnical concerns relative to the current dominating frames of fairness, accountability, confidentiality, transparency, and safety. We make explicit, with the intention to challenge, the dichotomy of how we relate to technology as either technologists (who envision and build technology) and its users. We posit that this mirrors the teacher-student relationship in Freire's analysis. By extending Freire's analysis to IA, we challenge the notion that it is the burden of the (altruistic) technologists to come up with interventions to mitigate the risks that emerging technologies pose to marginalized communities. Instead, we advocate that the first task for the technologists is to pose these as problems to the marginalized communities, to encourage them to make and unmake the technology as part of their material struggle against oppression. Their second task is to redesign our online technology stacks to structurally expose spaces for community members to co-opt and co-construct the technology in aid of their emancipatory struggles. We operationalize Freire's theories to develop a problem-posing framework for envisioning emancipatory IA platforms of the future.

Experimental robot optimization often requires evaluating each candidate policy for seconds to minutes. The chosen evaluation time influences optimization because of a speed-accuracy tradeoff: shorter evaluations enable faster iteration, but are also more subject to noise. Here, we introduce a supplement to the CMA-ES optimization algorithm, named Adaptive Sampling CMA-ES (AS-CMA), which assigns sampling time to candidates based on predicted sorting difficulty, aiming to achieve consistent precision. We compared AS-CMA to CMA-ES and Bayesian optimization using a range of static sampling times in four simulated cost landscapes. AS-CMA converged on 98% of all runs without adjustment to its tunable parameter, and converged 24-65% faster and with 29-76% lower total cost than each landscape's best CMA-ES static sampling time. As compared to Bayesian optimization, AS-CMA converged more efficiently and reliably in complex landscapes, while in simpler landscapes, AS-CMA was less efficient but equally reliable. We deployed AS-CMA in an exoskeleton optimization experiment and found the optimizer's behavior was consistent with expectations. These results indicate that AS-CMA can improve optimization efficiency in the presence of noise while minimally affecting optimization setup complexity and tuning requirements.

Recent research suggests that looped Transformers have superior reasoning capabilities compared to standard deep architectures. Current approaches to training single-head looped architectures on benchmark tasks frequently fail or yield suboptimal performance due to a highly non-convex and irregular loss landscape. In these settings, optimization often stagnates in poor local minima and saddle points of the loss landscape, preventing the model from discovering the global minimum point. The internal mechanisms of these single-head looped transformer models remain poorly understood, and training them from scratch remains a significant challenge. In this paper, we propose a novel training framework that leverages Tsallis entropy and Hamiltonian dynamics to transform the geometry of the loss landscape. By treating the parameter updates as a physical flow, we successfully trained a single-head looped Transformer with model dimension $d = 8$ to solve induction head task with input sequence length of 1000 tokens. This success reveals the internal mechanism behind the superior reasoning capability.

Handwriting remains an essential skill, particularly in education. Therefore, providing visual feedback on handwritten documents is an important but understudied area. We outline the many challenges when going from an image of handwritten input to correctly placed informative error feedback. We empirically compare modular and end-to-end systems and find that both approaches currently do not achieve acceptable overall quality. We identify the major challenges and outline an agenda for future research.

Kernel-based methods are used in the context of Granger Causality to enable the identification of nonlinear causal relationships between time series variables. In this paper, we show that two state of the art kernel-based Granger Causality (GC) approaches can be theoretically unified under the framework of Kernel Principal Component Regression (KPCR), and introduce a method based on this unification, demonstrating that this approach can improve causal identification. Additionally, we introduce a Gaussian Process score-based model with Smooth Information Criterion penalisation on the marginal likelihood, and demonstrate improved performance over existing state of the art time-series nonlinear causal discovery methods. Furthermore, we propose a contemporaneous causal identification algorithm, fully based on GC, using the proposed score-based $GP_{SIC}$ method, and compare its performance to a state of the art contemporaneous time series causal discovery algorithm.

We study permutation (jumbled/Abelian) pattern matching over a general alphabet $Σ$. Given a pattern P of length m and a text T of length n, the classical task is to decide whether T contains a length-m substring whose Parikh vector equals that of P . While this existence problem admits a linear-time sliding-window solution, many practical applications require optimization and packing variants beyond mere detection. We present a unified sliding-window framework based on maintaining the Parikh-vector difference between P and the current window of T , enabling permutation matching in O(n + σ) time and O(σ) space, where σ = |Σ|. Building on this foundation, we introduce a combinatorial-optimization variant that we call Maximum Feasible Substring under Pattern Supply (MFSP): find the longest substring S of T whose symbol counts are component-wise bounded by those of P . We show that MFSP can also be solved in O(n + σ) time via a two-pointer feasibility maintenance algorithm, providing an exact packing interpretation of P as a resource budget. Finally, we address non-overlapping occurrence selection by modeling each permutation match as an equal-length interval and proving that a greedy earliest-finishing strategy yields a maximum-cardinality set of disjoint matches, computable in linear time once all matches are enumerated. Our results provide concise, provably correct algorithms with tight bounds, and connect frequency-based string matching to packing-style optimization primitives.

Autonomous systems conducting schema-grounded information-gathering dialogues face an instrumentation gap, lacking turn-level observables for monitoring acquisition efficiency and detecting when questioning becomes unproductive. We introduce Dialogue Telemetry (DT), a measurement framework that produces two model-agnostic signals after each question-answer exchange: (i) a Progress Estimator (PE) quantifying residual information potential per category (with a bits-based variant), and (ii) a Stalling Index (SI) detecting an observable failure signature characterized by repeated category probing with semantically similar, low-marginal-gain responses. SI flags this pattern without requiring causal diagnosis, supporting monitoring in settings where attributing degradation to specific causes may be impractical. We validate DT in controlled search-and-rescue (SAR)-inspired interviews using large language model (LLM)-based simulations, distinguishing efficient from stalled dialogue traces and illustrating downstream utility by integrating DT signals into a reinforcement learning (RL) policy. Across these settings, DT provides interpretable turn-level instrumentation that improves policy performance when stalling carries operational costs.

Large language models typically represent Chinese characters as discrete index-based tokens, largely ignoring their visual form. For logographic scripts, visual structure carries semantic and phonetic information, which may aid prediction. We investigate whether low-resolution visual inputs can serve as an alternative for character-level modeling. Instead of token IDs, our decoder receives grayscale images of individual characters, with resolutions as low as $8 \times 8$ pixels. Remarkably, these inputs achieve 39.2\% accuracy, comparable to the index-based baseline of 39.1\%. Such low-resource settings also exhibit a pronounced \emph{hot-start} effect: by 0.4\% of total training, accuracy reaches above 12\%, while index-based models lag at below 6\%. Overall, our results demonstrate that minimal visual structure can provide a robust and efficient signal for Chinese language modeling, offering an alternative perspective on character representation that complements traditional index-based approaches.

This paper presents SiliconHealth, a comprehensive blockchain-based healthcare infrastructure designed for resource-constrained regions, particularly sub-Saharan Africa. We demonstrate that obsolete Bitcoin mining Application-Specific Integrated Circuits (ASICs) can be repurposed to create a secure, low-cost, and energy-efficient medical records system. The proposed architecture employs a four-tier hierarchical network: regional hospitals using Antminer S19 Pro (90+ TH/s), urban health centers with Antminer S9 (14 TH/s), rural clinics equipped with Lucky Miner LV06 (500 GH/s, 13W), and mobile health points with portable ASIC devices. We introduce the Deterministic Hardware Fingerprinting (DHF) paradigm, which repurposes SHA-256 mining ASICs as cryptographic proof generators, achieving 100% verification rate across 23 test proofs during 300-second validation sessions. The system incorporates Reed-Solomon LSB watermarking for medical image authentication with 30-40% damage tolerance, semantic Retrieval-Augmented Generation (RAG) for intelligent medical record queries, and offline synchronization protocols for intermittent connectivity. Economic analysis demonstrates 96% cost reduction compared to GPU-based alternatives, with total deployment cost of $847 per rural clinic including 5-year solar power infrastructure. Validation experiments on Lucky Miner LV06 (BM1366 chip, 5nm) achieve 2.93 MH/W efficiency and confirm hardware universality. This work establishes a practical framework for deploying verifiable, tamper-proof electronic health records in regions where traditional healthcare IT infrastructure is economically unfeasible, potentially benefiting over 600 million people lacking access to basic health information systems.

Microscaling Floating-Point (MXFP) has emerged as a promising low-precision format for large language models (LLMs). Despite various post-training quantization (PTQ) algorithms being proposed, they mostly focus on integer quantization, while their applicability and behavior under MXFP formats remain largely unexplored. To address this gap, this work conducts a systematic investigation of PTQ under MXFP formats, encompassing over 7 PTQ algorithms, 15 evaluation benchmarks, and 3 LLM families. The key findings include: 1) MXFP8 consistently achieves near-lossless performance, while MXFP4 introduces substantial accuracy degradation and remains challenging; 2) PTQ effectiveness under MXFP depends strongly on format compatibility, with some algorithmic paradigms being consistently more effective than others; 3) PTQ performance exhibits highly consistent trends across model families and modalities, in particular, quantization sensitivity is dominated by the language model rather than the vision encoder in multimodal LLMs; 4) The scaling factor of quantization is a critical error source in MXFP4, and a simple pre-scale optimization strategy can significantly mitigate its impact. Together, these results provide practical guidance on adapting existing PTQ methods to MXFP quantization.

Multimodal Large Language Models (MLLMs) are making significant progress in multimodal reasoning. Early approaches focus on pure text-based reasoning. More recent studies have incorporated multimodal information into the reasoning steps; however, they often follow a single task-specific reasoning pattern, which limits their generalizability across various multimodal tasks. In fact, there are numerous multimodal tasks requiring diverse reasoning skills, such as zooming in on a specific region or marking an object within an image. To address this, we propose unified generative multimodal reasoning, which unifies diverse multimodal reasoning skills by generating intermediate images during the reasoning process. We instantiate this paradigm with Omni-R1, a two-stage SFT+RL framework featuring perception alignment loss and perception reward, thereby enabling functional image generation. Additionally, we introduce Omni-R1-Zero, which eliminates the need for multimodal annotations by bootstrapping step-wise visualizations from text-only reasoning data. Empirical results show that Omni-R1 achieves unified generative reasoning across a wide range of multimodal tasks, and Omni-R1-Zero can match or even surpass Omni-R1 on average, suggesting a promising direction for generative multimodal reasoning.

The energy transition challenges operational tasks based on simulations and optimisation. These computations need to be fast and flexible as the grid is ever-expanding, and renewables' uncertainty requires a flexible operational environment. Learned approximations, proxies or surrogates -- we refer to them as Neural Solvers -- excel in terms of evaluation speed, but are inflexible with respect to adjusting to changing tasks. Hence, neural solvers are usually applicable to highly specific tasks, which limits their usefulness in practice; a widely reusable, foundational neural solver is required. Therefore, this work proposes the Residual Power Flow (RPF) formulation. RPF formulates residual functions based on Kirchhoff's laws to quantify the infeasibility of an operating condition. The minimisation of the residuals determines the voltage solution; an additional slack variable is needed to achieve AC-feasibility. RPF forms a natural, foundational subtask of tasks subject to power flow constraints. We propose to learn RPF with neural solvers to exploit their speed. Furthermore, RPF improves learning performance compared to common power flow formulations. To solve operational tasks, we integrate the neural solver in a Predict-then-Optimise (PO) approach to combine speed and flexibility. The case study investigates the IEEE 9-bus system and three tasks (AC Optimal Power Flow (OPF), power-flow and quasi-steady state power flow) solved by PO. The results demonstrate the accuracy and flexibility of learning with RPF.

SMEs increasingly seek alternatives to cloud LLM APIs, which raise data privacy concerns. Dedicated cloud GPU instances offer improved privacy but with limited guarantees and ongoing costs, while professional on-premise hardware (A100, H100) remains prohibitively expensive. We present a systematic evaluation of NVIDIA's Blackwell consumer GPUs (RTX 5060 Ti, 5070 Ti, 5090) for production LLM inference, benchmarking four open-weight models (Qwen3-8B, Gemma3-12B, Gemma3-27B, GPT-OSS-20B) across 79 configurations spanning quantization formats (BF16, W4A16, NVFP4, MXFP4), context lengths (8k-64k), and three workloads: RAG, multi-LoRA agentic serving, and high-concurrency APIs. The RTX 5090 delivers 3.5-4.6x higher throughput than the 5060 Ti with 21x lower latency for RAG, but budget GPUs achieve the highest throughput-per-dollar for API workloads with sub-second latency. NVFP4 quantization provides 1.6x throughput over BF16 with 41% energy reduction and only 2-4% quality loss. Self-hosted inference costs $0.001-0.04 per million tokens (electricity only), which is 40-200x cheaper than budget-tier cloud APIs, with hardware breaking even in under four months at moderate volume (30M tokens/day). Our results show that consumer GPUs can reliably replace cloud inference for most SME workloads, except latency-critical long-context RAG, where high-end GPUs remain essential. We provide deployment guidance and release all benchmark data for reproducible SME-scale deployments.

Deep neural networks have achieved remarkable success across a variety of tasks, yet they often suffer from unreliable probability estimates. As a result, they can be overconfident in their predictions. Conformal Prediction (CP) offers a principled framework for uncertainty quantification, yielding prediction sets with rigorous coverage guarantees. Existing conformal training methods optimize for overall set size, but shaping the prediction sets in a class-conditional manner is not straightforward and typically requires prior knowledge of the data distribution. In this work, we introduce Class Adaptive Conformal Training (CaCT), which formulates conformal training as an augmented Lagrangian optimization problem that adaptively learns to shape prediction sets class-conditionally without making any distributional assumptions. Experiments on multiple benchmark datasets, including standard and long-tailed image recognition as well as text classification, demonstrate that CaCT consistently outperforms prior conformal training methods, producing significantly smaller and more informative prediction sets while maintaining the desired coverage guarantees.

Deep learning models define the state-of-the-art in Automatic Drum Transcription (ADT), yet their performance is contingent upon large-scale, paired audio-MIDI datasets, which are scarce. Existing workarounds that use synthetic data often introduce a significant domain gap, as they typically rely on low-fidelity SoundFont libraries that lack acoustic diversity. While high-quality one-shot samples offer a better alternative, they are not available in a standardized, large-scale format suitable for training. This paper introduces a new paradigm for ADT that circumvents the need for paired audio-MIDI training data. Our primary contribution is a semi-supervised method to automatically curate a large and diverse corpus of one-shot drum samples from unlabeled audio sources. We then use this corpus to synthesize a high-quality dataset from MIDI files alone, which we use to train a sequence-to-sequence transcription model. We evaluate our model on the ENST and MDB test sets, where it achieves new state-of-the-art results, significantly outperforming both fully supervised methods and previous synthetic-data approaches. The code for reproducing our experiments is publicly available at https://github.com/pier-maker92/ADT_STR

Enabling humanoid robots to physically interact with humans is a critical frontier, but progress is hindered by the scarcity of high-quality Human-Humanoid Interaction (HHoI) data. While leveraging abundant Human-Human Interaction (HHI) data presents a scalable alternative, we first demonstrate that standard retargeting fails by breaking the essential contacts. We address this with PAIR (Physics-Aware Interaction Retargeting), a contact-centric, two-stage pipeline that preserves contact semantics across morphology differences to generate physically consistent HHoI data. This high-quality data, however, exposes a second failure: conventional imitation learning policies merely mimic trajectories and lack interactive understanding. We therefore introduce D-STAR (Decoupled Spatio-Temporal Action Reasoner), a hierarchical policy that disentangles when to act from where to act. In D-STAR, Phase Attention (when) and a Multi-Scale Spatial module (where) are fused by the diffusion head to produce synchronized whole-body behaviors beyond mimicry. By decoupling these reasoning streams, our model learns robust temporal phases without being distracted by spatial noise, leading to responsive, synchronized collaboration. We validate our framework through extensive and rigorous simulations, demonstrating significant performance gains over baseline approaches and a complete, effective pipeline for learning complex whole-body interactions from HHI data.

Due to the dynamically evolving nature of real-world query streams, relevance models struggle to generalize to practical search scenarios. A sophisticated solution is self-evolution techniques. However, in large-scale industrial settings with massive query streams, this technique faces two challenges: (1) informative samples are often sparse and difficult to identify, and (2) pseudo-labels generated by the current model could be unreliable. To address these challenges, in this work, we propose a Self-Evolving Relevance Model approach (SERM), which comprises two complementary multi-agent modules: a multi-agent sample miner, designed to detect distributional shifts and identify informative training samples, and a multi-agent relevance annotator, which provides reliable labels through a two-level agreement framework. We evaluate SERM in a large-scale industrial setting, which serves billions of user requests daily. Experimental results demonstrate that SERM can achieve significant performance gains through iterative self-evolution, as validated by extensive offline multilingual evaluations and online testing.

To teach robots complex manipulation tasks, it is now a common practice to fine-tune a pre-trained vision-language-action model (VLA) on task-specific data. However, since this recipe updates existing representations, it is unsuitable for long-term operation in the real world, where robots must continually adapt to new tasks and environments while retaining the knowledge they have already acquired. Existing continual learning methods for robotics commonly require storing previous data (exemplars), struggle with long task sequences, or rely on task identifiers for deployment. To address these limitations, we propose CLARE, a general, parameter-efficient framework for exemplar-free continual learning with VLAs. CLARE introduces lightweight modular adapters into selected feedforward layers and autonomously expands the model only where necessary when learning a new task, guided by layer-wise feature similarity. During deployment, an autoencoder-based routing mechanism dynamically activates the most relevant adapters without requiring task labels. Through extensive experiments on the LIBERO benchmark, we show that CLARE achieves high performance on new tasks without catastrophic forgetting of earlier tasks, significantly outperforming even exemplar-based methods. Code and data are available at https://tum-lsy.github.io/clare.

Scientific surveys require not only summarizing large bodies of literature, but also organizing them into clear and coherent conceptual structures. Existing automatic survey generation methods typically focus on linear text generation and struggle to explicitly model hierarchical relations among research topics and structured methodological comparisons, resulting in gaps in structural organization compared to expert-written surveys. We propose MVSS, a multi-view structured survey generation framework that jointly generates and aligns citation-grounded hierarchical trees, structured comparison tables, and survey text. MVSS follows a structure-first paradigm: it first constructs a conceptual tree of the research domain, then generates comparison tables constrained by the tree, and finally uses both as structural constraints for text generation. This enables complementary multi-view representations across structure, comparison, and narrative. We introduce an evaluation framework assessing structural quality, comparative completeness, and citation fidelity. Experiments on 76 computer science topics show MVSS outperforms existing methods in organization and evidence grounding, achieving performance comparable to expert surveys.

Large language model (LLM) agents have demonstrated remarkable capabilities in complex decision-making and tool-use tasks, yet their ability to generalize across varying environments remains a under-examined concern. Current evaluation paradigms predominantly rely on trajectory-based metrics that measure task success, while failing to assess whether agents possess a grounded, transferable model of the environment. To address this gap, we propose Task-to-Quiz (T2Q), a deterministic and automated evaluation paradigm designed to decouple task execution from world-state understanding. We instantiate this paradigm in T2QBench, a suite comprising 30 environments and 1,967 grounded QA pairs across multiple difficulty levels. Our extensive experiments reveal that task success is often a poor proxy for environment understanding, and that current memory machanism can not effectively help agents acquire a grounded model of the environment. These findings identify proactive exploration and fine-grained state representation as primary bottlenecks, offering a robust foundation for developing more generalizable autonomous agents.

Object detectors often perform well in-distribution, yet degrade sharply on a different benchmark. We study cross-dataset object detection (CD-OD) through a lens of setting specificity. We group benchmarks into setting-agnostic datasets with diverse everyday scenes and setting-specific datasets tied to a narrow environment, and evaluate a standard detector family across all train--test pairs. This reveals a clear structure in CD-OD: transfer within the same setting type is relatively stable, while transfer across setting types drops substantially and is often asymmetric. The most severe breakdowns occur when transferring from specific sources to agnostic targets, and persist after open-label alignment, indicating that domain shift dominates in the hardest regimes. To disentangle domain shift from label mismatch, we compare closed-label transfer with an open-label protocol that maps predicted classes to the nearest target label using CLIP similarity. Open-label evaluation yields consistent but bounded gains, and many corrected cases correspond to semantic near-misses supported by the image evidence. Overall, we provide a principled characterization of CD-OD under setting specificity and practical guidance for evaluating detectors under distribution shift. Code will be released at \href{[https://github.com/Ritabrata04/cdod-icpr.git}{https://github.com/Ritabrata04/cdod-icpr}.

With the emergence of massively parallel processing units, parallelization has become a desirable property for new sequence models. The ability to parallelize the processing of sequences with respect to the sequence length during training is one of the main factors behind the uprising of the Transformer architecture. However, Transformers lack efficiency at sequence generation, as they need to reprocess all past timesteps at every generation step. Recently, state-space models (SSMs) emerged as a more efficient alternative. These new kinds of recurrent neural networks (RNNs) keep the efficient update of the RNNs while gaining parallelization by getting rid of nonlinear dynamics (or recurrence). SSMs can reach state-of-the art performance through the efficient training of potentially very large networks, but still suffer from limited representation capabilities. In particular, SSMs cannot exhibit persistent memory, or the capacity of retaining information for an infinite duration, because of their monostability. In this paper, we introduce a new family of RNNs, the memory recurrent units (MRUs), that combine the persistent memory capabilities of nonlinear RNNs with the parallelizable computations of SSMs. These units leverage multistability as a source of persistent memory, while getting rid of transient dynamics for efficient computations. We then derive a specific implementation as proof-of-concept: the bistable memory recurrent unit (BMRU). This new RNN is compatible with the parallel scan algorithm. We show that BMRU achieves good results in tasks with long-term dependencies, and can be combined with state-space models to create hybrid networks that are parallelizable and have transient dynamics as well as persistent memory.

Deep Operator Networks are emerging as fundamental tools among various neural network types to learn mappings between function spaces, and have recently gained attention due to their ability to approximate nonlinear operators. In particular, DeepONets offer a natural formulation for PDE solving, since the solution of a partial differential equation can be interpreted as an operator mapping an initial condition to its corresponding solution field. In this work, we applied DeepONets in the context of process modeling for adsorption technologies, to assess their feasibility as surrogates for cyclic adsorption process simulation and optimization. The goal is to accelerate convergence of cyclic processes such as Temperature-Vacuum Swing Adsorption (TVSA), which require repeated solution of transient PDEs, which are computationally expensive. Since each step of a cyclic adsorption process starts from the final state of the preceding step, effective surrogate modeling requires generalization across a wide range of initial conditions. The governing equations exhibit steep traveling fronts, providing a demanding benchmark for operator learning. To evaluate functional generalization under these conditions, we construct a mixed training dataset composed of heterogeneous initial conditions and train DeepONets to approximate the corresponding solution operators. The trained models are then tested on initial conditions outside the parameter ranges used during training, as well as on completely unseen functional forms. The results demonstrate accurate predictions both within and beyond the training distribution, highlighting DeepONets as potential efficient surrogates for accelerating cyclic adsorption simulations and optimization workflows.

The rapid evolution of Large Language Models (LLMs) has fostered diverse paradigms for automated slide generation, ranging from code-driven layouts to image-centric synthesis. However, evaluating these heterogeneous systems remains challenging, as existing protocols often struggle to provide comparable scores across architectures or rely on uncalibrated judgments. In this paper, we introduce SlidesGen-Bench, a benchmark designed to evaluate slide generation through a lens of three core principles: universality, quantification, and reliability. First, to establish a unified evaluation framework, we ground our analysis in the visual domain, treating terminal outputs as renderings to remain agnostic to the underlying generation method. Second, we propose a computational approach that quantitatively assesses slides across three distinct dimensions - Content, Aesthetics, and Editability - offering reproducible metrics where prior works relied on subjective or reference-dependent proxies. Finally, to ensure high correlation with human preference, we construct the Slides-Align1.5k dataset, a human preference aligned dataset covering slides from nine mainstream generation systems across seven scenarios. Our experiments demonstrate that SlidesGen-Bench achieves a higher degree of alignment with human judgment than existing evaluation pipelines. Our code and data are available at https://github.com/YunqiaoYang/SlidesGen-Bench.

Semantic understanding of popularity bias is a crucial yet underexplored challenge in recommender systems, where popular items are often favored at the expense of niche content. Most existing debiasing methods treat the semantic understanding of popularity bias as a matter of diversity enhancement or long-tail coverage, neglecting the deeper semantic layer that embodies the causal origins of the bias itself. Consequently, such shallow interpretations limit both their debiasing effectiveness and recommendation accuracy. In this paper, we propose FairLRM, a novel framework that bridges the gap in the semantic understanding of popularity bias with Recommendation via Large Language Model (RecLLM). FairLRM decomposes popularity bias into item-side and user-side components, using structured instruction-based prompts to enhance the model's comprehension of both global item distributions and individual user preferences. Unlike traditional methods that rely on surface-level features such as "diversity" or "debiasing", FairLRM improves the model's ability to semantically interpret and address the underlying bias. Through empirical evaluation, we show that FairLRM significantly enhances both fairness and recommendation accuracy, providing a more semantically aware and trustworthy approach to enhance the semantic understanding of popularity bias. The implementation is available at https://github.com/LuoRenqiang/FairLRM.

We address the problem of sampling from terminally constrained distributions with pre-trained flow-based generative models through an optimal control formulation. Theoretically, we characterize the value function by a Hamilton-Jacobi-Bellman equation and derive the optimal feedback control as the minimizer of the associated Hamiltonian. We show that as the control penalty increases, the controlled process recovers the reference distribution, while as the penalty vanishes, the terminal law converges to a generalized Wasserstein projection onto the constraint manifold. Algorithmically, we introduce Terminal Optimal Control with Flow-based models (TOCFlow), a geometry-aware sampling-time guidance method for pre-trained flows. Solving the control problem in a terminal co-moving frame that tracks reference trajectories yields a closed-form scalar damping factor along the Riemannian gradient, capturing second-order curvature effects without matrix inversions. TOCFlow therefore matches the geometric consistency of Gauss-Newton updates at the computational cost of standard gradient guidance. We evaluate TOCFlow on three high-dimensional scientific tasks spanning equality, inequality, and global statistical constraints, namely Darcy flow, constrained trajectory planning, and turbulence snapshot generation with Kolmogorov spectral scaling. Across all settings, TOCFlow improves constraint satisfaction over Euclidean guidance and projection baselines while preserving the reference model's generative quality.

Model merging enables multiple large language models (LLMs) to be combined into a single model while preserving performance. This makes it a valuable tool in LLM development, offering a competitive alternative to multi-task training. However, merging can be difficult at scale, as successful merging requires choosing the right merge operator, selecting the right models, and merging them in the right order. This often leads researchers to run expensive merge-and-evaluate searches to select the best merge. In this work, we provide an alternative by introducing \simmerge{}, \emph{a predictive merge-selection method} that selects the best merge using inexpensive, task-agnostic similarity signals between models. From a small set of unlabeled probes, we compute functional and structural features and use them to predict the performance of a given 2-way merge. Using these predictions, \simmerge{} selects the best merge operator, the subset of models to merge, and the merge order, eliminating the expensive merge-and-evaluate loop. We demonstrate that we surpass standard merge-operator performance on 2-way merges of 7B-parameter LLMs, and that \simmerge{} generalizes to multi-way merges and 111B-parameter LLM merges without retraining. Additionally, we present a bandit variant that supports adding new tasks, models, and operators on the fly. Our results suggest that learning how to merge is a practical route to scalable model composition when checkpoint catalogs are large and evaluation budgets are tight.

Multiple external representations (MERs) and personalized feedback support physics learning, yet evidence on how personalized feedback can effectively integrate MERs remains limited. This question is particularly timely given the emergence of multimodal large language models. We conducted a 16-24 week observational study in high school physics (N=661) using a computer-based platform that provided verification and optional elaborated feedback in verbal, graphical and mathematical forms. Linear mixed-effects models and strategy-cluster analyses (ANCOVA-adjusted comparisons) tested associations between feedback use and post-test performance and moderation by representational competence. Elaborated multirepresentational feedback showed a small but consistent positive association with post-test scores independent of prior knowledge and confidence. Learners adopted distinct representation-selection strategies; among students with lower representational competence, using a diverse set of representations related to higher learning, whereas this advantage diminished as competence increased. These findings motivate adaptive feedback designs and inform intelligent tutoring systems capable of tailoring feedback elaboration and representational format to learner profiles, advancing personalized instruction in physics education.

Graph unlearning has emerged as a critical mechanism for supporting sustainable and privacy-preserving social networks, enabling models to remove the influence of deleted nodes and thereby better safeguard user information. However, we observe that existing graph unlearning techniques insufficiently protect sensitive attributes, often leading to degraded algorithmic fairness compared with traditional graph learning methods. To address this gap, we introduce FairGU, a fairness-aware graph unlearning framework designed to preserve both utility and fairness during the unlearning process. FairGU integrates a dedicated fairness-aware module with effective data protection strategies, ensuring that sensitive attributes are neither inadvertently amplified nor structurally exposed when nodes are removed. Through extensive experiments on multiple real-world datasets, we demonstrate that FairGU consistently outperforms state-of-the-art graph unlearning methods and fairness-enhanced graph learning baselines in terms of both accuracy and fairness metrics. Our findings highlight a previously overlooked risk in current unlearning practices and establish FairGU as a robust and equitable solution for the next generation of socially sustainable networked systems. The codes are available at https://github.com/LuoRenqiang/FairGU.

Topographic models are essential for characterizing planetary surfaces and for inferring underlying geological processes. Nevertheless, meter-scale topographic data remain limited, which constrains detailed planetary investigations, even for the Moon, where extensive high-resolution orbital images are available. Recent advances in deep learning (DL) exploit single-view imagery, constrained by low-resolution topography, for fast and flexible reconstruction of fine-scale topography. However, their robustness and general applicability across diverse lunar landforms and illumination conditions remain insufficiently explored. In this study, we build upon our previously proposed DL framework by incorporating a more robust scale recovery scheme and extending the model to polar regions under low solar illumination conditions. We demonstrate that, compared with single-view shape-from-shading methods, the proposed DL approach exhibits greater robustness to varying illumination and achieves more consistent and accurate topographic reconstructions. Furthermore, it reliably reconstructs topography across lunar features of diverse scales, morphologies, and geological ages. High-quality topographic models are also produced for the lunar south polar areas, including permanently shadowed regions, demonstrating the method's capability in reconstructing complex and low-illumination terrain. These findings suggest that DL-based approaches have the potential to leverage extensive lunar datasets to support advanced exploration missions and enable investigations of the Moon at unprecedented topographic resolution.

Accurate global medium-range weather forecasting is fundamental to Earth system science. Most existing Transformer-based forecasting models adopt vision-centric architectures that neglect the Earth's spherical geometry and zonal periodicity. In addition, conventional autoregressive training is computationally expensive and limits forecast horizons due to error accumulation. To address these challenges, we propose the Shifted Earth Transformer (Searth Transformer), a physics-informed architecture that incorporates zonal periodicity and meridional boundaries into window-based self-attention for physically consistent global information exchange. We further introduce a Relay Autoregressive (RAR) fine-tuning strategy that enables learning long-range atmospheric evolution under constrained memory and computational budgets. Based on these methods, we develop YanTian, a global medium-range weather forecasting model. YanTian achieves higher accuracy than the high-resolution forecast of the European Centre for Medium-Range Weather Forecasts and performs competitively with state-of-the-art AI models at one-degree resolution, while requiring roughly 200 times lower computational cost than standard autoregressive fine-tuning. Furthermore, YanTian attains a longer skillful forecast lead time for Z500 (10.3 days) than HRES (9 days). Beyond weather forecasting, this work establishes a robust algorithmic foundation for predictive modeling of complex global-scale geophysical circulation systems, offering new pathways for Earth system science.

While LLM-based agents have shown promise for deep research, most existing approaches rely on fixed workflows that struggle to adapt to real-world, open-ended queries. Recent work therefore explores self-evolution by allowing agents to rewrite their own code or prompts to improve problem-solving ability, but unconstrained optimization often triggers instability, hallucinations, and instruction drift. We propose EvoFSM, a structured self-evolving framework that achieves both adaptability and control by evolving an explicit Finite State Machine (FSM) instead of relying on free-form rewriting. EvoFSM decouples the optimization space into macroscopic Flow (state-transition logic) and microscopic Skill (state-specific behaviors), enabling targeted improvements under clear behavioral boundaries. Guided by a critic mechanism, EvoFSM refines the FSM through a small set of constrained operations, and further incorporates a self-evolving memory that distills successful trajectories as reusable priors and failure patterns as constraints for future queries. Extensive evaluations on five multi-hop QA benchmarks demonstrate the effectiveness of EvoFSM. In particular, EvoFSM reaches 58.0% accuracy on the DeepSearch benchmark. Additional results on interactive decision-making tasks further validate its generalization.

In collaborative learning (CL), multiple parties jointly train a machine learning model on their private datasets. However, data can not be shared directly due to privacy concerns. To ensure input confidentiality, cryptographic techniques, e.g., multi-party computation (MPC), enable training on encrypted data. Yet, even securely trained models are vulnerable to inference attacks aiming to extract memorized data from model outputs. To ensure output privacy and mitigate inference attacks, differential privacy (DP) injects calibrated noise during training. While cryptography and DP offer complementary guarantees, combining them efficiently for cryptographic and differentially private CL (CPCL) is challenging. Cryptography incurs performance overheads, while DP degrades accuracy, creating a privacy-accuracy-performance trade-off that needs careful design considerations. This work systematizes the CPCL landscape. We introduce a unified framework that generalizes common phases across CPCL paradigms, and identify secure noise sampling as the foundational phase to achieve CPCL. We analyze trade-offs of different secure noise sampling techniques, noise types, and DP mechanisms discussing their implementation challenges and evaluating their accuracy and cryptographic overhead across CPCL paradigms. Additionally, we implement identified secure noise sampling options in MPC and evaluate their computation and communication costs in WAN and LAN. Finally, we propose future research directions based on identified key observations, gaps and possible enhancements in the literature.

Judicial reasoning in copyright damage awards poses a core challenge for computational legal analysis. Although federal courts follow the 1976 Copyright Act, their interpretations and factor weightings vary widely across jurisdictions. This inconsistency creates unpredictability for litigants and obscures the empirical basis of legal decisions. This research introduces a novel discourse-based Large Language Model (LLM) methodology that integrates Rhetorical Structure Theory (RST) with an agentic workflow to extract and quantify previously opaque reasoning patterns from judicial opinions. Our framework addresses a major gap in empirical legal scholarship by parsing opinions into hierarchical discourse structures and using a three-stage pipeline, i.e., Dataset Construction, Discourse Analysis, and Agentic Feature Extraction. This pipeline identifies reasoning components and extract feature labels with corresponding discourse subtrees. In analyzing copyright damage rulings, we show that discourse-augmented LLM analysis outperforms traditional methods while uncovering unquantified variations in factor weighting across circuits. These findings offer both methodological advances in computational legal analysis and practical insights into judicial reasoning, with implications for legal practitioners seeking predictive tools, scholars studying legal principle application, and policymakers confronting inconsistencies in copyright law.

Domain-specific metadata schemas are essential to improve the findability and reusability of research software and to follow the FAIR4RS principles. However, many domains, including energy research, lack established metadata schemas. To address this gap, we developed a metadata schema for energy research software based on a requirement analysis and evaluated it through user testing. Our results show that the schema balances the need for formalization and interoperability, while also meeting the specific needs of energy researchers. Meanwhile, the testing showed that a good presentation of the required information is key to enable researchers to create the required metadata. This paper provides insights into the challenges and opportunities of designing a metadata schema for energy research software.

Providing clear explanations to the choices of machine learning models is essential for these models to be deployed in crucial applications. Counterfactual and semi-factual explanations have emerged as two mechanisms for providing users with insights into the outputs of their models. We provide an overview of the computational complexity results in the literature for generating these explanations, finding that in many cases, generating explanations is computationally hard. We strengthen the argument for this considerably by further contributing our own inapproximability results showing that not only are explanations often hard to generate, but under certain assumptions, they are also hard to approximate. We discuss the implications of these complexity results for the XAI community and for policymakers seeking to regulate explanations in AI.

We propose Quamba-SE, a soft-edge quantizer for State Space Model (SSM) activation quantization. Unlike existing methods, using standard INT8 operation, Quamba-SE employs three adaptive scales: high-precision for small values, standard scale for normal values, and low-precision for outliers. This preserves outlier information instead of hard clipping, while maintaining precision for other values. We evaluate on Mamba- 130M across 6 zero-shot benchmarks. Results show that Quamba- SE consistently outperforms Quamba, achieving up to +2.68% on individual benchmarks and up to +0.83% improvement in the average accuracy of 6 datasets.

Conventional audio equalization is a static process that requires manual and cumbersome adjustments to adapt to changing listening contexts (e.g., mood, location, or social setting). In this paper, we introduce a Large Language Model (LLM)-based alternative that maps natural language text prompts to equalization settings. This enables a conversational approach to sound system control. By utilizing data collected from a controlled listening experiment, our models exploit in-context learning and parameter-efficient fine-tuning techniques to reliably align with population-preferred equalization settings. Our evaluation methods, which leverage distributional metrics that capture users' varied preferences, show statistically significant improvements in distributional alignment over random sampling and static preset baselines. These results indicate that LLMs could function as "artificial equalizers," contributing to the development of more accessible, context-aware, and expert-level audio tuning methods.

The use of formal language for deductive logical reasoning aligns well with language models (LMs), where translating natural language (NL) into first-order logic (FOL) and employing an external solver results in a verifiable and therefore reliable reasoning system. However, smaller LMs often struggle with this translation task, frequently producing incorrect symbolic outputs due to formatting and translation errors. Existing approaches typically rely on self-iteration to correct these errors, but such methods depend heavily on the capabilities of the underlying model. To address this, we first categorize common errors and fine-tune smaller LMs using data synthesized by large language models. The evaluation is performed using the defined error categories. We introduce incremental inference, which divides inference into two stages, predicate generation and FOL translation, providing greater control over model behavior and enhancing generation quality as measured by predicate metrics. This decomposition framework also enables the use of a verification module that targets predicate-arity errors to further improve performance. Our study evaluates three families of models across four logical-reasoning datasets. The comprehensive fine-tuning, incremental inference, and verification modules reduce error rates, increase predicate coverage, and improve reasoning performance for smaller LMs, moving us closer to developing reliable and accessible symbolic-reasoning systems.

In language models (LMs), intra-memory knowledge conflict largely arises when inconsistent information about the same event is encoded within the model's parametric knowledge. While prior work has primarily focused on resolving conflicts between a model's internal knowledge and external resources through approaches such as fine-tuning or knowledge editing, the problem of localizing conflicts that originate during pre-training within the model's internal representations remain unexplored. In this work, we design a framework based on mechanistic interpretability methods to identify where and how conflicting knowledge from the pre-training data is encoded within LMs. Our findings contribute to a growing body of evidence that specific internal components of a language model are responsible for encoding conflicting knowledge from pre-training, and we demonstrate how mechanistic interpretability methods can be leveraged to causally intervene in and control conflicting knowledge at inference time.

Deep learning libraries like Transformers and Megatron are now widely adopted in modern AI programs. However, when these libraries introduce defects, ranging from silent computation errors to subtle performance regressions, it is often challenging for downstream users to assess whether their own programs are affected. Such impact analysis requires not only understanding the defect semantics but also checking whether the client code satisfies complex triggering conditions involving configuration flags, runtime environments, and indirect API usage. We present DepRadar, an agent coordination framework for fine grained defect and impact analysis in DL library updates. DepRadar coordinates four specialized agents across three steps: 1. the PR Miner and Code Diff Analyzer extract structured defect semantics from commits or pull requests, 2. the Orchestrator Agent synthesizes these signals into a unified defect pattern with trigger conditions, and 3. the Impact Analyzer checks downstream programs to determine whether the defect can be triggered. To improve accuracy and explainability, DepRadar integrates static analysis with DL-specific domain rules for defect reasoning and client side tracing. We evaluate DepRadar on 157 PRs and 70 commits across two representative DL libraries. It achieves 90% precision in defect identification and generates high quality structured fields (average field score 1.6). On 122 client programs, DepRadar identifies affected cases with 90% recall and 80% precision, substantially outperforming other baselines.

In optoacoustic imaging, recovering the absorption coefficients of tissue by inverting the light transport remains a challenging problem. Improvements in solving this problem can greatly benefit the clinical value of optoacoustic imaging. Existing variational inversion methods require an accurate and differentiable model of this light transport. As neural surrogate models allow fast and differentiable simulations of complex physical processes, they are considered promising candidates to be used in solving such inverse problems. However, there are in general no guarantees that the derivatives of these surrogate models accurately match those of the underlying physical operator. As accurate derivatives are central to solving inverse problems, errors in the model derivative can considerably hinder high fidelity reconstructions. To overcome this limitation, we present a surrogate model for light transport in tissue that uses Sobolev training to improve the accuracy of the model derivatives. Additionally, the form of Sobolev training we used is suitable for high-dimensional models in general. Our results demonstrate that Sobolev training for a light transport surrogate model not only improves derivative accuracy but also reduces generalization error for in-distribution and out-of-distribution samples. These improvements promise to considerably enhance the utility of the surrogate model in downstream tasks, especially in solving inverse problems.

Large Language Model (LLM)-based Multi-Agent Systems (MAS) enhance complex problem solving through multi-agent collaboration, but often incur substantially higher costs than single-agent systems. Recent MAS routing methods aim to balance performance and overhead by dynamically selecting agent roles and language models. However, these approaches typically rely on a homogeneous collaboration mode, where all agents follow the same interaction pattern, limiting collaboration flexibility across different roles. Motivated by Social Capital Theory, which emphasizes that different roles benefit from distinct forms of collaboration, we propose SC-MAS, a framework for constructing heterogeneous and cost-efficient multi-agent systems. SC-MAS models MAS as directed graphs, where edges explicitly represent pairwise collaboration strategies, allowing different agent pairs to interact through tailored communication patterns. Given an input query, a unified controller progressively constructs an executable MAS by selecting task-relevant agent roles, assigning edge-level collaboration strategies, and allocating appropriate LLM backbones to individual agents. Experiments on multiple benchmarks demonstrate the effectiveness of SC-MAS. In particular, SC-MAS improves accuracy by 3.35% on MMLU while reducing inference cost by 15.38%, and achieves a 3.53% accuracy gain with a 12.13% cost reduction on MBPP. These results validate the feasibility of SC-MAS and highlight the effectiveness of heterogeneous collaboration in multi-agent systems.

Automated analysis of ancient coins has the potential to help researchers extract more historical insights from large collections of coins and to help collectors understand what they are buying or selling. Recent research in this area has shown promise in focusing on identification of semantic elements as they are commonly depicted on ancient coins, by using convolutional neural networks (CNNs). This paper is the first to apply the recently proposed Vision Transformer (ViT) deep learning architecture to the task of identification of semantic elements on coins, using fully automatic learning from multi-modal data (images and unstructured text). This article summarises previous research in the area, discusses the training and implementation of ViT and CNN models for ancient coins analysis and provides an evaluation of their performance. The ViT models were found to outperform the newly trained CNN models in accuracy.

We introduce a new method of generating Computer Aided Design (CAD) profiles via a sequence of simple geometric constructions including curve offsetting, rotations and intersections. These sequences start with geometry provided by a designer and build up the points and curves of the final profile step by step. We demonstrate that adding construction steps between the designer's input geometry and the final profile improves generation quality in a similar way to the introduction of a chain of thought in language models. Similar to the constraints in a parametric CAD model, the construction sequences reduce the degrees of freedom in the modeled shape to a small set of parameter values which can be adjusted by the designer, allowing parametric editing with the constructed geometry evaluated to floating point precision. In addition we show that applying reinforcement learning to the construction sequences gives further improvements over a wide range of metrics, including some which were not explicitly optimized.

Pre-trained language models (LMs) have, over the last few years, grown substantially in both societal adoption and training costs. This rapid growth in size has constrained progress in understanding and mitigating their biases. Since re-training LMs is prohibitively expensive, most debiasing work has focused on post-hoc or masking-based strategies, which often fail to address the underlying causes of bias. In this work, we seek to democratise pre-model debiasing research by using low-cost proxy models. Specifically, we investigate BabyLMs, compact BERT-like models trained on small and mutable corpora that can approximate bias acquisition and learning dynamics of larger models. We show that BabyLMs display closely aligned patterns of intrinsic bias formation and performance development compared to standard BERT models, despite their drastically reduced size. Furthermore, correlations between BabyLMs and BERT hold across multiple intra-model and post-model debiasing methods. Leveraging these similarities, we conduct pre-model debiasing experiments with BabyLMs, replicating prior findings and presenting new insights regarding the influence of gender imbalance and toxicity on bias formation. Our results demonstrate that BabyLMs can serve as an effective sandbox for large-scale LMs, reducing pre-training costs from over 500 GPU-hours to under 30 GPU-hours. This provides a way to democratise pre-model debiasing research and enables faster, more accessible exploration of methods for building fairer LMs.

Osteosarcoma (OS) is an aggressive primary bone malignancy. Accurate histopathological assessment of viable versus non-viable tumor regions after neoadjuvant chemotherapy is critical for prognosis and treatment planning, yet manual evaluation remains labor-intensive, subjective, and prone to inter-observer variability. Recent advances in digital pathology have enabled automated necrosis quantification. Evaluating on test data, independently sampled on patient-level, revealed that the deep learning model performance dropped significantly from the tile-level generalization ability reported in previous studies. First, this work proposes the use of radiomic features as additional input in model training. We show that, despite that they are derived from the images, such a multimodal input effectively improved the classification performance, in addition to its added benefits in interpretability. Second, this work proposes to optimize two binary classification tasks with hierarchical classes (i.e. tumor-vs-non-tumor and viable-vs-non-viable), as opposed to the alternative ``flat'' three-class classification task (i.e. non-tumor, non-viable tumor, viable tumor), thereby enabling a hierarchical loss. We show that such a hierarchical loss, with trainable weightings between the two tasks, the per-class performance can be improved significantly. Using the TCIA OS Tumor Assessment dataset, we experimentally demonstrate the benefits from each of the proposed new approaches and their combination, setting a what we consider new state-of-the-art performance on this open dataset for this application. Code and trained models: https://github.com/YaxiiC/RadiomicsOS.git.

We introduce a voice-agentic framework that learns one critical omni-understanding skill: knowing when to trust itself versus when to consult external audio perception. Our work is motivated by a crucial yet counterintuitive finding: naively fine-tuning an omni-model on both speech recognition and external sound understanding tasks often degrades performance, as the model can be easily misled by noisy hypotheses. To address this, our framework, Speech-Hands, recasts the problem as an explicit self-reflection decision. This learnable reflection primitive proves effective in preventing the model from being derailed by flawed external candidates. We show that this agentic action mechanism generalizes naturally from speech recognition to complex, multiple-choice audio reasoning. Across the OpenASR leaderboard, Speech-Hands consistently outperforms strong baselines by 12.1% WER on seven benchmarks. The model also achieves 77.37% accuracy and high F1 on audio QA decisions, showing robust generalization and reliability across diverse audio question answering datasets. By unifying perception and decision-making, our work offers a practical path toward more reliable and resilient audio intelligence.

Retrieval-Augmented Generation (RAG) enhances Large Language Models (LLMs) by incorporating external knowledge. Recently, some works have incorporated iterative knowledge accumulation processes into RAG models to progressively accumulate and refine query-related knowledge, thereby constructing more comprehensive knowledge representations. However, these iterative processes often lack a coherent organizational structure, which limits the construction of more comprehensive and cohesive knowledge representations. To address this, we propose PAGER, a page-driven autonomous knowledge representation framework for RAG. PAGER first prompts an LLM to construct a structured cognitive outline for a given question, which consists of multiple slots representing a distinct knowledge aspect. Then, PAGER iteratively retrieves and refines relevant documents to populate each slot, ultimately constructing a coherent page that serves as contextual input for guiding answer generation. Experiments on multiple knowledge-intensive benchmarks and backbone models show that PAGER consistently outperforms all RAG baselines. Further analyses demonstrate that PAGER constructs higher-quality and information-dense knowledge representations, better mitigates knowledge conflicts, and enables LLMs to leverage external knowledge more effectively. All code is available at https://github.com/OpenBMB/PAGER.

This paper introduces ACS2HER, a novel integration of the Anticipatory Classifier System (ACS2) with the Hindsight Experience Replay (HER) mechanism. While ACS2 is highly effective at building cognitive maps through latent learning, its performance often stagnates in environments characterized by sparse rewards. We propose a specific architectural variant that triggers hindsight learning when the agent fails to reach its primary goal, re-labeling visited states as virtual goals to densify the learning signal. The proposed model was evaluated on two benchmarks: the deterministic \texttt{Maze 6} and the stochastic \texttt{FrozenLake}. The results demonstrate that ACS2HER significantly accelerates knowledge acquisition and environmental mastery compared to the standard ACS2. However, this efficiency gain is accompanied by increased computational overhead and a substantial expansion in classifier numerosity. This work provides the first analysis of combining anticipatory mechanisms with retrospective goal-relabeling in Learning Classifier Systems.

Large language models can be continually pre-trained or fine-tuned to improve performance in specific domains, languages, or skills, but this specialization often degrades other capabilities and may cause catastrophic forgetting. We investigate how abilities are distributed within LLM parameters by analyzing module activations under domain- and language-specific inputs for closely related models. Across layers and modules, we find that ability-related activations are highly concentrated in a small set of channels (typically <5\%), and these channels are largely disentangled with good sufficiency and stability. Building on these observations, we propose ACT (Activation-Guided Channel-wise Ability Transfer), which localizes ability-relevant channels via activation differences and selectively transfers only the corresponding parameters, followed by lightweight fine-tuning for compatibility. Experiments on multilingual mathematical and scientific reasoning show that ACT can recover forgotten abilities while preserving retained skills. It can also merge multiple specialized models to integrate several abilities into a single model with minimal interference. Our code and data will be publicly released.

Graph Transformers (GTs) are increasingly applied to social network analysis, yet their deployment is often constrained by fairness concerns. This issue is particularly critical in incomplete social networks, where sensitive attributes are frequently missing due to privacy and ethical restrictions. Existing solutions commonly generate these incomplete attributes, which may introduce additional biases and further compromise user privacy. To address this challenge, FairGE (Fair Graph Encoding) is introduced as a fairness-aware framework for GTs in incomplete social networks. Instead of generating sensitive attributes, FairGE encodes fairness directly through spectral graph theory. By leveraging the principal eigenvector to represent structural information and padding incomplete sensitive attributes with zeros to maintain independence, FairGE ensures fairness without data reconstruction. Theoretical analysis demonstrates that the method suppresses the influence of non-principal spectral components, thereby enhancing fairness. Extensive experiments on seven real-world social network datasets confirm that FairGE achieves at least a 16% improvement in both statistical parity and equality of opportunity compared with state-of-the-art baselines. The source code is shown in https://github.com/LuoRenqiang/FairGE.

The transition from Cloud-Native to AI-Native architectures is fundamentally reshaping software engineering, replacing deterministic microservices with probabilistic agentic services. However, this shift renders traditional black-box evaluation paradigms insufficient: existing benchmarks measure raw model capabilities while remaining blind to system-level execution dynamics. To bridge this gap, we introduce AI-NativeBench, the first application-centric and white-box AI-Native benchmark suite grounded in Model Context Protocol (MCP) and Agent-to-Agent (A2A) standards. By treating agentic spans as first-class citizens within distributed traces, our methodology enables granular analysis of engineering characteristics beyond simple capabilities. Leveraging this benchmark across 21 system variants, we uncover critical engineering realities invisible to traditional metrics: a parameter paradox where lightweight models often surpass flagships in protocol adherence, a pervasive inference dominance that renders protocol overhead secondary, and an expensive failure pattern where self-healing mechanisms paradoxically act as cost multipliers on unviable workflows. This work provides the first systematic evidence to guide the transition from measuring model capability to engineering reliable AI-Native systems. To facilitate reproducibility and further research, we have open-sourced the benchmark and dataset.

The recent surge in open-source Multimodal Large Language Models (MLLM) frameworks, such as LLaVA, provides a convenient kickoff for artificial intelligence developers and researchers. However, most of the MLLM frameworks take vision as the main input modality, and provide limited in-depth support for the modality of speech, audio, and music. This situation hinders the development of audio-language models, and forces researchers to spend a lot of effort on code writing and hyperparameter tuning. We present SLAM-LLM, an open-source deep learning framework designed to train customized MLLMs, focused on speech, language, audio, and music processing. SLAM-LLM provides a modular configuration of different encoders, projectors, LLMs, and parameter-efficient fine-tuning plugins. SLAM-LLM also includes detailed training and inference recipes for mainstream tasks, along with high-performance checkpoints like LLM-based Automatic Speech Recognition (ASR), Automated Audio Captioning (AAC), and Music Captioning (MC). Some of these recipes have already reached or are nearing state-of-the-art performance, and some relevant techniques have also been accepted by academic papers. We hope SLAM-LLM will accelerate iteration, development, data engineering, and model training for researchers. We are committed to continually pushing forward audio-based MLLMs through this open-source framework, and call on the community to contribute to the LLM-based speech, audio and music processing.

Current large language model agents predominantly operate under a reactive paradigm, responding only to immediate user queries within short-term sessions. This limitation hinders their ability to maintain long-term user's intents and dynamically adapt to evolving external environments. In this paper, we propose a novel interaction paradigm for proactive Task-oriented Agents capable of bridging the gap between relatively static user's needs and a dynamic environment. We formalize proactivity through two key capabilities, (i) Intent-Conditioned Monitoring: The agent autonomously formulates trigger conditions based on dialog history; (ii) Event-Triggered Follow-up: The agent actively engages the user upon detecting useful environmental updates. We introduce a high-quality data synthesis pipeline to construct complex, multi-turn dialog data in a dynamic environment. Furthermore, we attempt to address the lack of evaluation criteria of task-oriented interaction in a dynamic environment by proposing a new benchmark, namely ChronosBench. We evaluated some leading close-source and open-source models at present and revealed their flaws in long-term task-oriented interaction. Furthermore, our fine-tuned model trained using synthetic data for supervised learning achieves a task completion rate of 85.19% for complex tasks including shifts in user intent, outperforming other models under test. And the result validated the effectiveness of our data-driven strategy.

In this paper, I discuss two logics for weighted finite structures: first-order logic with summation (FO(SUM)) and its recursive extension IFP(SUM). These logics originate from foundational work by Grädel, Gurevich, and Meer in the 1990s. In recent joint work with Standke, Steegmans, and Van den Bussche, we have investigated these logics as query languages for machine learning models, specifically neural networks, which are naturally represented as weighted graphs. I present illustrative examples of queries to neural networks that can be expressed in these logics and discuss fundamental results on their expressiveness and computational complexity.

In fault-tolerant quantum computing, a large number of physical qubits are required to construct a single logical qubit, and a single quantum node may be able to hold only a small number of logical qubits. In such a case, the idea of distributed fault-tolerant quantum computing (DFTQC) is important to demonstrate large-scale quantum computation using small-scale nodes. However, the design of distributed systems on small-scale nodes, where each node can store only one or a few logical qubits for computation, has not been explored well yet. In this paper, we propose network-based quantum computation (NBQC) to efficiently realize distributed fault-tolerant quantum computation using many small-scale nodes. A key idea of NBQC is to let computational data continuously move throughout the network while maintaining the connectivity to other nodes. We numerically show that, for practical benchmark tasks, our method achieves shorter execution times than circuit-based strategies and more node-efficient constructions than measurement-based quantum computing. Also, if we are allowed to specialize the network to the structure of quantum programs, such as peak access frequencies, the number of nodes can be significantly reduced. Thus, our methods provide a foundation in designing DFTQC architecture exploiting the redundancy of many small fault-tolerant nodes.

Do Large Language Models (LLMs) genuinely grasp the compositional semantics of events, or do they rely on surface-level probabilistic heuristics? We investigate the Imperfective Paradox, a logical phenomenon where the past progressive aspect entails event realization for activities (e.g., running $\to$ ran) but not for accomplishments (e.g., building $\nrightarrow$ built). We introduce ImperfectiveNLI, a diagnostic dataset designed to probe this distinction across diverse semantic classes. Evaluating state-of-the-art open-weight models, we uncover a pervasive Teleological Bias: models systematically hallucinate completion for goal-oriented events, often overriding explicit textual negation. Representational analyses show that while internal embeddings often distinguish process from result, inference decisions are dominated by strong priors about goal attainment. We further find that prompting-based interventions reduce hallucinated completions but also increase incorrect rejections of valid entailments. Our findings suggest that current LLMs lack structural aspectual awareness, operating as predictive narrative engines rather than faithful logical reasoners.

Zero-Knowledge (ZK) proof systems are cryptographic protocols that can (with overwhelming probability) demonstrate that the pair $(X, W)$ is in a relation $R$ without revealing information about the private input $W$. This membership checking is captured by a complex arithmetic circuit: a set of polynomial equations over a finite field. ZK programming languages, like Noir, have been proposed to simplify the description of these circuits. A developer can write a Noir program using traditional high-level constructs that can be compiled into a lower-level ACIR (Abstract Circuit Intermediate Representation), which is essentially a high-level description of an arithmetic circuit. In this paper, we formalise some of the ACIR language using SMT-LIB and its extended theory of finite fields. We use this formalisation to create an open-source formal verifier for the Noir language using the SMT solver cvc5. Our verifier can be used to check whether Noir programs behave appropriately. For instance, it can be used to check whether a Noir program has been properly constrained, that is, the finite-field polynomial equations generated truly capture the intended relation. We evaluate our verifier over 4 distinct sets of Noir programs, demonstrating its practical applicability and identifying a hard-to-check constraint type that charts an improvement path for our verification framework.

The scarcity of annotated datasets for clinical information extraction in non-English languages hinders the evaluation of large language model (LLM)-based methods developed primarily in English. In this study, we present the first comprehensive bilingual evaluation of LLMs for the clinical Relation Extraction (RE) task in both English and Turkish. To facilitate this evaluation, we introduce the first English-Turkish parallel clinical RE dataset, derived and carefully curated from the 2010 i2b2/VA relation classification corpus. We systematically assess a diverse set of prompting strategies, including multiple in-context learning (ICL) and Chain-of-Thought (CoT) approaches, and compare their performance to fine-tuned baselines such as PURE. Furthermore, we propose Relation-Aware Retrieval (RAR), a novel in-context example selection method based on contrastive learning, that is specifically designed to capture both sentence-level and relation-level semantics. Our results show that prompting-based LLM approaches consistently outperform traditional fine-tuned models. Moreover, evaluations for English performed better than their Turkish counterparts across all evaluated LLMs and prompting techniques. Among ICL methods, RAR achieves the highest performance, with Gemini 1.5 Flash reaching a micro-F1 score of 0.906 in English and 0.888 in Turkish. Performance further improves to 0.918 F1 in English when RAR is combined with a structured reasoning prompt using the DeepSeek-V3 model. These findings highlight the importance of high-quality demonstration retrieval and underscore the potential of advanced retrieval and prompting techniques to bridge resource gaps in clinical natural language processing.

Common ground plays a critical role in situated spoken dialogues, where interlocutors must establish and maintain shared references to entities, events, and relations to sustain coherent interaction. For dialog systems, the ability to correctly ground conversational content in order to refer back to it later is particularly important. Prior studies have demonstrated that LLMs are capable of performing grounding acts such as requesting clarification or producing acknowledgments, yet relatively little work has investigated how common ground can be explicitly represented and stored for later use. Without such mechanisms, it remains unclear whether acknowledgment or clarification behaviors truly reflect a grounded understanding. In this work, we evaluate a model's ability to establish and exploit common ground through relational references to entities within the shared context in a situational dialogue. We test multiple methods for representing common ground in situated dialogues and further propose approaches to improve both the establishment of common ground and its subsequent use in the conversation.

Reinforcement Learning with Verifiable Rewards (RLVR) is crucial for advancing large-scale reasoning models. However, existing parameter-efficient methods, such as PiSSA and MiLoRA, are designed for Supervised Fine-Tuning (SFT) and do not account for the distinct optimization dynamics and geometric structures of RLVR. Applying these methods directly leads to spectral collapse and optimization instability, which severely limit model performance. Meanwhile, alternative approaches that leverage update sparsity encounter significant efficiency bottlenecks on modern hardware due to unstructured computations. To address these challenges, we propose GeoRA (Geometry-Aware Low-Rank Adaptation), which exploits the anisotropic and compressible nature of RL update subspaces. GeoRA initializes adapters by extracting principal directions via Singular Value Decomposition (SVD) within a geometrically constrained subspace while freezing the residual components. This method preserves the pre-trained geometric structure and enables efficient GPU computation through dense operators. Experiments on Qwen and Llama demonstrate that GeoRA mitigates optimization bottlenecks caused by geometric misalignment. It consistently outperforms established low-rank baselines on key mathematical benchmarks, achieving state-of-the-art (SOTA) results. Moreover, GeoRA shows superior generalization and resilience to catastrophic forgetting in out-of-domain tasks.

When some resources are to be distributed among a set of agents following egalitarian social welfare, the goal is to maximize the utility of the agent whose utility turns out to be minimal. In this context, agents can have an incentive to lie about their actual preferences, so that more valuable resources are assigned to them. In this paper we analyze this situation, and we present a practical study where genetic algorithms are used to assess the benefits of lying under different situations.

Lane-free traffic environments allow vehicles to better harness the lateral capacity of the road without being restricted to lane-keeping, thereby increasing the traffic flow rates. As such, we have a distinct and more challenging setting for autonomous driving. In this work, we consider a Monte-Carlo Tree Search (MCTS) planning approach for single-agent autonomous driving in lane-free traffic, where the associated Markov Decision Process we formulate is influenced from existing approaches tied to reinforcement learning frameworks. In addition, MCTS is equipped with a pre-trained neural network (NN) that guides the selection phase. This procedure incorporates the predictive capabilities of NNs for a more informed tree search process under computational constraints. In our experimental evaluation, we consider metrics that address both safety (through collision rates) and efficacy (through measured speed). Then, we examine: (a) the influence of isotropic state information for vehicles in a lane-free environment, resulting in nudging behaviour--vehicles' policy reacts due to the presence of faster tailing ones, (b) the acceleration of performance for the NN-guided variant of MCTS, and (c) the trade-off between computational resources and solution quality.

The integration of artificial intelligence (AI) into the industrial sector has not only driven innovation but also expanded the ethical landscape, necessitating a reevaluation of principles governing technology and its applications and awareness in research and development of industrial AI solutions. This chapter explores how AI-empowered industrial innovation inherently intersects with ethics, as advancements in AI introduce new challenges related to transparency, accountability, and fairness. In the chapter, we then examine the ethical aspects of several examples of AI manifestation in industrial use cases and associated factors such as ethical practices in the research and development process and data sharing. With the progress of ethical industrial AI solutions, we emphasize the importance of embedding ethical principles into industrial AI systems and its potential to inspire technological breakthroughs and foster trust among stakeholders. This chapter also offers actionable insights to guide industrial research and development toward a future where AI serves as an enabler for ethical and responsible industrial progress as well as a more inclusive industrial ecosystem.

This work proposes a contextualised detection framework for implicitly hateful speech, implemented as a multi-agent system comprising a central Moderator Agent and dynamically constructed Community Agents representing specific demographic groups. Our approach explicitly integrates socio-cultural context from publicly available knowledge sources, enabling identity-aware moderation that surpasses state-of-the-art prompting methods (zero-shot prompting, few-shot prompting, chain-of-thought prompting) and alternative approaches on a challenging ToxiGen dataset. We enhance the technical rigour of performance evaluation by incorporating balanced accuracy as a central metric of classification fairness that accounts for the trade-off between true positive and true negative rates. We demonstrate that our community-driven consultative framework significantly improves both classification accuracy and fairness across all target groups.

The widespread deployment of large-scale, compute-intensive applications such as high-performance computing, artificial intelligence, and big data is leading to convergence between cloud and high-performance computing infrastructures. Cloud providers are increasingly integrating high-performance computing capabilities in their infrastructures, such as hardware accelerators and high-speed interconnects, while researchers in the high-performance computing community are starting to explore cloud-native paradigms to improve scalability, elasticity, and resource utilization. In this context, serverless computing emerges as a promising execution model to efficiently handle highly dynamic, parallel, and distributed workloads. This paper presents a comprehensive systematic literature review of 122 research articles published between 2018 and early 2025, exploring the use of the serverless paradigm to develop, deploy, and orchestrate compute-intensive applications across cloud, high-performance computing, and hybrid environments. From these, a taxonomy comprising eight primary research directions and nine targeted use case domains is proposed, alongside an analysis of recent publication trends and collaboration networks among authors, highlighting the growing interest and interconnections within this emerging research field. Overall, this work aims to offer a valuable foundation for both new researchers and experienced practitioners, guiding the development of next-generation serverless solutions for parallel compute-intensive applications.

Language models perform well on grammatical agreement, but it is unclear whether this reflects rule-based generalization or memorization. We study this question for German definite singular articles, whose forms depend on gender and case. Using GRADIEND, a gradient-based interpretability method, we learn parameter update directions for gender-case specific article transitions. We find that updates learned for a specific gender-case article transition frequently affect unrelated gender-case settings, with substantial overlap among the most affected neurons across settings. These results argue against a strictly rule-based encoding of German definite articles, indicating that models at least partly rely on memorized associations rather than abstract grammatical rules.

On-device recommendation is critical for a number of real-world applications, especially in scenarios that have agreements on execution latency, user privacy, and robust functionality when internet connectivity is unstable or even impossible. While large language models (LLMs) can now provide exceptional capabilities that model user behavior for sequential recommendation tasks, their substantial memory footprint and computational overhead make the deployment on resource-constrained devices a high risk proposition. In this paper, we propose OD-LLM, the first task-adaptive compression framework explicitly designed to provide efficient and accurate on-device deployment of LLMs for sequential recommendation tasks. OD-LLM uniquely integrates two complementary compression strategies: a low-rank structural compression algorithm which uses Singular Value Decomposition (SVD) to significantly reduce parameter redundancy in the model, and a novel tokenization normalization technique that better complements the low-rank decomposition process being used. Additionally, to minimize any potential performance degradation when using higher compression ratios, a novel progressive alignment algorithm is used to iteratively refine the parameters required layerwise in the target model. Empirical evaluations conducted on sequential recommendation benchmarks show that OD-LLM exhibits no loss in effectiveness when compared to the original recommendation model, when the deployed model size is halved. These promising results demonstrate the efficacy and scalability of OD-LLM, making this novel solution a practical alternative for real-time, on-device solutions wishing to replace expensive, remotely executed LLMs.

Federated Learning (FL) enables distributed learning across multiple clients without sharing raw data. When statistical heterogeneity across clients is severe, Clustered Federated Learning (CFL) can improve performance by grouping similar clients and training cluster-wise models. However, most CFL approaches rely on multiple communication rounds for cluster estimation and model updates, which limits their practicality under tight constraints on communication rounds. We propose Data Collaboration-based Clustered Federated Learning (DC-CFL), a single-round framework that completes both client clustering and cluster-wise learning, using only the information shared in DC analysis. DC-CFL quantifies inter-client similarity via total variation distance between label distributions, estimates clusters using hierarchical clustering, and performs cluster-wise learning via DC analysis. Experiments on multiple open datasets under representative non-IID conditions show that DC-CFL achieves accuracy comparable to multi-round baselines while requiring only one communication round. These results indicate that DC-CFL is a practical alternative for collaborative AI model development when multiple communication rounds are impractical.

Large Language Model (LLM) agents deployed in complex real-world scenarios typically operate as spatially distributed entities. However, this physical dispersion constrains agents to limited local perception and finite temporal horizons. We characterize this bottleneck as spatiotemporal partial observability. Given such fragmented awareness, distributed agents struggle to coordinate efficiently. To bridge this gap, we introduce MACRO-LLM, LLM-empowered multi-agent collaborative reasoning under spatiotemporal partial observability. The architecture addresses spatiotemporal constraints via three modules: (1) the CoProposer mitigates temporal uncertainty by verifying candidate actions via predictive rollouts; (2) the Negotiator overcomes spatial myopia by resolving conflicts through mean-field statistical aggregation; and (3) the Introspector ensures continuous adaptation by analyzing historical experience to refine strategies via semantic gradient descent. Extensive evaluations on two complex long-horizon tasks, cooperative adaptive cruise control and pandemic control, demonstrate that our framework effectively mitigates spatiotemporal partial observability through spatial and temporal strategies, enabling robust coordination.

We present a novel framework for solving Dynamic Job Shop Scheduling Problems under uncertainty, addressing the challenges introduced by stochastic job arrivals and unexpected machine breakdowns. Our approach follows a model-based paradigm, using Coloured Timed Petri Nets to represent the scheduling environment, and Maskable Proximal Policy Optimization to enable dynamic decision-making while restricting the agent to feasible actions at each decision point. To simulate realistic industrial conditions, dynamic job arrivals are modeled using a Gamma distribution, which captures complex temporal patterns such as bursts, clustering, and fluctuating workloads. Machine failures are modeled using a Weibull distribution to represent age-dependent degradation and wear-out dynamics. These stochastic models enable the framework to reflect real-world manufacturing scenarios better. In addition, we study two action-masking strategies: a non-gradient approach that overrides the probabilities of invalid actions, and a gradient-based approach that assigns negative gradients to invalid actions within the policy network. We conduct extensive experiments on dynamic JSSP benchmarks, demonstrating that our method consistently outperforms traditional heuristic and rule-based approaches in terms of makespan minimization. The results highlight the strength of combining interpretable Petri-net-based models with adaptive reinforcement learning policies, yielding a resilient, scalable, and explainable framework for real-time scheduling in dynamic and uncertain manufacturing environments.

We present an experimental evaluation that assesses the robustness of four open source LLMs claiming function-calling capabilities against three different attacks, and we measure the effectiveness of eight different defences. Our results show how these models are not safe by default, and how the defences are not yet employable in real-world scenarios.

The IEC 61850 Generic Object-Oriented Substation Event (GOOSE) protocol plays a critical role in real-time protection and automation of digital substations, yet its lack of native security mechanisms can expose power systems to sophisticated cyberattacks. Traditional rule-based and supervised intrusion detection techniques struggle to detect protocol-compliant and zero-day attacks under significant class imbalance and limited availability of labeled data. This paper proposes an explainable, unsupervised multi-view anomaly detection framework for IEC 61850 GOOSE networks that explicitly separates semantic integrity and temporal availability. The approach employs asymmetric autoencoders trained only on real operational GOOSE traffic to learn distinct latent representations of sequence-based protocol semantics and timing-related transmission dynamics in normal traffic. Anomaly detection is implemented using reconstruction errors mixed with statistically grounded thresholds, enabling robust detection without specified attack types. Feature-level reconstruction analysis provides intrinsic explainability by directly linking detection outcomes to IEC 61850 protocol characteristics. The proposed framework is evaluated using real substation traffic for training and a public dataset containing normal traffic and message suppression, data manipulation, and denial-of-service attacks for testing. Experimental results show attack detection rates above 99% with false positives remaining below 5% of total traffic, demonstrating strong generalization across environments and effective operation under extreme class imbalance and interpretable anomaly attribution.

Collaborative Filtering (CF) remains the cornerstone of modern recommender systems, with dense embedding--based methods dominating current practice. However, these approaches suffer from a critical limitation: our theoretical analysis reveals a fundamental signal-to-noise ratio (SNR) ceiling when modeling unpopular items, where parameter-based dense models experience diminishing SNR under severe data sparsity. To overcome this bottleneck, we propose SaD (Sparse and Dense), a unified framework that integrates the semantic expressiveness of dense embeddings with the structural reliability of sparse interaction patterns. We theoretically show that aligning these dual views yields a strictly superior global SNR. Concretely, SaD introduces a lightweight bidirectional alignment mechanism: the dense view enriches the sparse view by injecting semantic correlations, while the sparse view regularizes the dense model through explicit structural signals. Extensive experiments demonstrate that, under this dual-view alignment, even a simple matrix factorization--style dense model can achieve state-of-the-art performance. Moreover, SaD is plug-and-play and can be seamlessly applied to a wide range of existing recommender models, highlighting the enduring power of collaborative filtering when leveraged from dual perspectives. Further evaluations on real-world benchmarks show that SaD consistently outperforms strong baselines, ranking first on the BarsMatch leaderboard. The code is publicly available at https://github.com/harris26-G/SaD.

Metal-organic frameworks (MOFs) are porous crystalline materials with broad applications such as carbon capture and drug delivery, yet accurately predicting their 3D structures remains a significant challenge. While Large Language Models (LLMs) have shown promise in generating crystals, their application to MOFs is hindered by MOFs' high atomic complexity. Inspired by the success of block-wise paradigms in deep generative models, we pioneer the use of LLMs in this domain by introducing MOF-LLM, the first LLM framework specifically adapted for block-level MOF structure prediction. To effectively harness LLMs for this modular assembly task, our training paradigm integrates spatial-aware continual pre-training (CPT), structural supervised fine-tuning (SFT), and matching-driven reinforcement learning (RL). By incorporating explicit spatial priors and optimizing structural stability via Soft Adaptive Policy Optimization (SAPO), our approach substantially enhances the spatial reasoning capability of a Qwen-3 8B model for accurate MOF structure prediction. Comprehensive experiments demonstrate that MOF-LLM outperforms state-of-the-art denoising-based and LLM-based methods while exhibiting superior sampling efficiency.

Cluster workload allocation often requires complex configurations, creating a usability gap. This paper introduces a semantic, intent-driven scheduling paradigm for cluster systems using Natural Language Processing. The system employs a Large Language Model (LLM) integrated via a Kubernetes scheduler extender to interpret natural language allocation hint annotations for soft affinity preferences. A prototype featuring a cluster state cache and an intent analyzer (using AWS Bedrock) was developed. Empirical evaluation demonstrated high LLM parsing accuracy (>95% Subset Accuracy on an evaluation ground-truth dataset) for top-tier models like Amazon Nova Pro/Premier and Mistral Pixtral Large, significantly outperforming a baseline engine. Scheduling quality tests across six scenarios showed the prototype achieved superior or equivalent placement compared to standard Kubernetes configurations, particularly excelling in complex and quantitative scenarios and handling conflicting soft preferences. The results validate using LLMs for accessible scheduling but highlight limitations like synchronous LLM latency, suggesting asynchronous processing for production readiness. This work confirms the viability of semantic soft affinity for simplifying workload orchestration.

Large Reasoning Models (LRMs) have advanced automated multi-step reasoning, but their ability to generate complex Chain-of-Thought (CoT) trajectories introduces severe privacy risks, as sensitive information may be deeply embedded throughout the reasoning process. Existing Large Language Models (LLMs) unlearning approaches that typically focus on modifying only final answers are insufficient for LRMs, as they fail to remove sensitive content from intermediate steps, leading to persistent privacy leakage and degraded security. To address these challenges, we propose Sensitive Trajectory Regulation (STaR), a parameter-free, inference-time unlearning framework that achieves robust privacy protection throughout the reasoning process. Specifically, we first identify sensitive content via semantic-aware detection. Then, we inject global safety constraints through secure prompt prefix. Next, we perform trajectory-aware suppression to dynamically block sensitive content across the entire reasoning chain. Finally, we apply token-level adaptive filtering to prevent both exact and paraphrased sensitive tokens during generation. Furthermore, to overcome the inadequacies of existing evaluation protocols, we introduce two metrics: Multi-Decoding Consistency Assessment (MCS), which measures the consistency of unlearning across diverse decoding strategies, and Multi-Granularity Membership Inference Attack (MIA) Evaluation, which quantifies privacy protection at both answer and reasoning-chain levels. Experiments on the R-TOFU benchmark demonstrate that STaR achieves comprehensive and stable unlearning with minimal utility loss, setting a new standard for privacy-preserving reasoning in LRMs.

Recent studies in medical question answering (Medical QA) have actively explored the integration of large language models (LLMs) with biomedical knowledge graphs (KGs) to improve factual accuracy. However, most existing approaches still rely on traversing the entire KG or performing large-scale retrieval, which introduces substantial noise and leads to unstable multi-hop reasoning. We argue that the core challenge lies not in expanding access to knowledge, but in identifying and reasoning over the appropriate subset of evidence for each query. ReGraM is a region-first knowledge graph reasoning framework that addresses this challenge by constructing a query-aligned subgraph and performing stepwise reasoning constrained to this localized region under multiple evidence aware modes. By focusing inference on only the most relevant portion of the KG, ReGraM departs from the assumption that all relations are equally useful an assumption that rarely holds in domain-specific medical settings. Experiments on seven medical QA benchmarks demonstrate that ReGraM consistently outperforms a strong baseline (KGARevion), achieving an 8.04% absolute accuracy gain on MCQ, a 4.50% gain on SAQ, and a 42.9% reduction in hallucination rate. Ablation and qualitative analyses further show that aligning region construction with hop-wise reasoning is the primary driver of these improvements. Overall, our results highlight region-first KG reasoning as an effective paradigm for improving factual accuracy and consistency in medical QA.

Recent advances in DeepResearch-style agents have demonstrated strong capabilities in autonomous information acquisition and synthesize from real-world web environments. However, existing approaches remain fundamentally limited to text modality. Extending autonomous information-seeking agents to multimodal settings introduces critical challenges: the specialization-generalization trade-off that emerges when training models for multimodal tool-use at scale, and the severe scarcity of training data capturing complex, multi-step multimodal search trajectories. To address these challenges, we propose M$^3$Searcher, a modular multimodal information-seeking agent that explicitly decouples information acquisition from answer derivation. M$^3$Searcher is optimized with a retrieval-oriented multi-objective reward that jointly encourages factual accuracy, reasoning soundness, and retrieval fidelity. In addition, we develop MMSearchVQA, a multimodal multi-hop dataset to support retrieval centric RL training. Experimental results demonstrate that M$^3$Searcher outperforms existing approaches, exhibits strong transfer adaptability and effective reasoning in complex multimodal tasks.

Scientific reasoning relies not only on logical inference but also on activating prior knowledge and experiential structures. Memory can efficiently reuse knowledge and enhance reasoning consistency and stability. However, existing benchmarks mainly evaluate final answers or step-by-step coherence, overlooking the \textit{memory-driven} mechanisms that underlie human reasoning, which involves activating anchors and attractors, then integrating them into multi-step inference. To address this gap, we propose $A^3$-Bench~ https://a3-bench.github.io, a benchmark designed to evaluate scientific reasoning through dual-scale memory-driven activation, grounded in Anchor and Attractor Activation. First, we annotate 2,198 science reasoning problems across domains using the SAPM process(subject, anchor & attractor, problem, and memory developing). Second, we introduce a dual-scale memory evaluation framework utilizing anchors and attractors, along with the AAUI(Anchor--Attractor Utilization Index) metric to measure memory activation rates. Finally, through experiments with various base models and paradigms, we validate $A^3$-Bench and analyze how memory activation impacts reasoning performance, providing insights into memory-driven scientific reasoning.

With the rapid advancement of Multimodal Large Language Models (MLLMs), their potential has garnered significant attention in Chinese Classical Studies (CCS). While existing research has primarily focused on text and visual modalities, the audio corpus within this domain remains largely underexplored. To bridge this gap, we propose the Multi-task Classical Chinese Literary Genre Audio Corpus (MCGA). It encompasses a diverse range of literary genres across six tasks: Automatic Speech Recognition (ASR), Speech-to-Text Translation (S2TT), Speech Emotion Captioning (SEC), Spoken Question Answering (SQA), Speech Understanding (SU), and Speech Reasoning (SR). Through the evaluation of ten MLLMs, our experimental results demonstrate that current models still face substantial challenges when processed on the MCGA test set. Furthermore, we introduce an evaluation metric for SEC and a metric to measure the consistency between the speech and text capabilities of MLLMs. We release MCGA and our code to the public to facilitate the development of MLLMs with more robust multidimensional audio capabilities in CCS. MCGA Corpus: https://github.com/yxduir/MCGA

Recent work on domain-specific reasoning with large language models (LLMs) often relies on training-intensive approaches that require parameter updates. While activation steering has emerged as a parameter efficient alternative, existing methods apply static, manual interventions that fail to adapt to the dynamic nature of complex reasoning. To address this limitation, we propose RISER (Router-based Intervention for Steerable Enhancement of Reasoning), a plug-and-play intervention framework that adaptively steers LLM reasoning in activation space. RISER constructs a library of reusable reasoning vectors and employs a lightweight Router to dynamically compose them for each input. The Router is optimized via reinforcement learning under task-level rewards, activating latent cognitive primitives in an emergent and compositional manner. Across seven diverse benchmarks, RISER yields 3.4-6.5% average zero-shot accuracy improvements over the base model while surpassing CoT-style reasoning with 2-3x higher token efficiency and robust accuracy gains. Further analysis shows that RISER autonomously combines multiple vectors into interpretable, precise control strategies, pointing toward more controllable and efficient LLM reasoning.

Effective pandemic control requires timely and coordinated policymaking across administrative regions that are intrinsically interdependent. However, human-driven responses are often fragmented and reactive, with policies formulated in isolation and adjusted only after outbreaks escalate, undermining proactive intervention and global pandemic mitigation. To address this challenge, here we propose a large language model (LLM) multi-agent policymaking framework that supports coordinated and proactive pandemic control across regions. Within our framework, each administrative region is assigned an LLM agent as an AI policymaking assistant. The agent reasons over region-specific epidemiological dynamics while communicating with other agents to account for cross-regional interdependencies. By integrating real-world data, a pandemic evolution simulator, and structured inter-agent communication, our framework enables agents to jointly explore counterfactual intervention scenarios and synthesize coordinated policy decisions through a closed-loop simulation process. We validate the proposed framework using state-level COVID-19 data from the United States between April and December 2020, together with real-world mobility records and observed policy interventions. Compared with real-world pandemic outcomes, our approach reduces cumulative infections and deaths by up to 63.7% and 40.1%, respectively, at the individual state level, and by 39.0% and 27.0%, respectively, when aggregated across states. These results demonstrate that LLM multi-agent systems can enable more effective pandemic control with coordinated policymaking...

Delineating wildfire affected areas using satellite imagery remains challenging due to irregular and spatially heterogeneous spectral changes across the electromagnetic spectrum. While recent deep learning approaches achieve high accuracy when high-resolution multispectral data are available, their applicability in operational settings, where a quick delineation of the burn scar shortly after a wildfire incident is required, is limited by the trade-off between spatial resolution and temporal revisit frequency of current satellite systems. To address this limitation, we propose a novel deep learning model, namely BAM-MRCD, which employs multi-resolution, multi-source satellite imagery (MODIS and Sentinel-2) for the timely production of detailed burnt area maps with high spatial and temporal resolution. Our model manages to detect even small scale wildfires with high accuracy, surpassing similar change detection models as well as solid baselines. All data and code are available in the GitHub repository: https://github.com/Orion-AI-Lab/BAM-MRCD.

Learning under unobservable feedback reliability poses a distinct challenge beyond optimization robustness: a system must decide whether to learn from an experience, not only how to learn stably. We study this setting as Epistemic Identifiability under Unobservable Reliability (EIUR), where each experience has a latent credibility, reliable and unreliable feedback can be locally indistinguishable, and data are generated in a closed loop by the learner's own evolving beliefs and actions. In EIUR, standard robust learning can converge stably yet form high-confidence, systematically wrong beliefs. We propose metacognitive regulation as a practical response: a second, introspective control loop that infers experience credibility from endogenous evidence in the learner's internal dynamics. We formalize this as a modular Monitor-Trust-Regulator (MTR) decomposition and instantiate it with self-diagnosis, which maintains a slowly varying experience-trust variable that softly modulates learning updates, without exogenous reliability labels or an explicit corruption model. Empirically, in the EIUR regimes studied here, self-diagnosis is associated with improved epistemic identifiability. In reinforcement learning, it enables calibrated skepticism and recovery under systematically corrupted rewards. In supervised learning, it exposes a critical dissociation: performance recovery does not imply epistemic recovery. Accuracy can rebound while internal belief dynamics remain locked-in by early misleading data, a failure detectable only through introspective diagnostics. Together, MTR and self-diagnosis provide an organizing abstraction and a concrete design template for intrinsic reliability assessment in autonomous learning under unobservable reliability.

High-quality chain-of-thought has demonstrated strong potential for unlocking the reasoning capabilities of large language models. However, current paradigms typically treat the reasoning process as an indivisible sequence, lacking an intrinsic mechanism to quantify step-wise information gain. This granularity gap manifests in two limitations: inference inefficiency from redundant exploration without explicit guidance, and optimization difficulty due to sparse outcome supervision or costly external verifiers. In this work, we propose CoT-Flow, a framework that reconceptualizes discrete reasoning steps as a continuous probabilistic flow, quantifying the contribution of each step toward the ground-truth answer. Built on this formulation, CoT-Flow enables two complementary methodologies: flow-guided decoding, which employs a greedy flow-based decoding strategy to extract information-efficient reasoning paths, and flow-based reinforcement learning, which constructs a verifier-free dense reward function. Experiments on challenging benchmarks demonstrate that CoT-Flow achieves a superior balance between inference efficiency and reasoning performance.

Large Language Model (LLM) Agents exhibit inherent reasoning abilities through the collaboration of multiple tools. However, during agent inference, existing methods often suffer from (i) locally myopic generation, due to the absence of lookahead, and (ii) trajectory instability, where minor early errors can escalate into divergent reasoning paths. These issues make it difficult to balance global effectiveness and computational efficiency. To address these two issues, we propose meta-adaptive exploration with LLM agents https://github.com/exoskeletonzj/MAXS, a meta-adaptive reasoning framework based on LLM Agents that flexibly integrates tool execution and reasoning planning. MAXS employs a lookahead strategy to extend reasoning paths a few steps ahead, estimating the advantage value of tool usage, and combines step consistency variance and inter-step trend slopes to jointly select stable, consistent, and high-value reasoning steps. Additionally, we introduce a trajectory convergence mechanism that controls computational cost by halting further rollouts once path consistency is achieved, enabling a balance between resource efficiency and global effectiveness in multi-tool reasoning. We conduct extensive empirical studies across three base models (MiMo-VL-7B, Qwen2.5-VL-7B, Qwen2.5-VL-32B) and five datasets, demonstrating that MAXS consistently outperforms existing methods in both performance and inference efficiency. Further analysis confirms the effectiveness of our lookahead strategy and tool usage.

LLM inference latency critically determines user experience and operational costs, directly impacting throughput under SLO constraints. Even brief latency spikes degrade service quality despite acceptable average performance. However, distributed inference environments featuring diverse software frameworks and XPU architectures combined with dynamic workloads make latency analysis challenging. Constrained by intrusive designs that necessitate service restarts or even suspension, and by hardware-bound implementations that fail to adapt to heterogeneous inference environments, existing AI profiling methods are often inadequate for real-time production analysis. We present LatencyPrism, the first zero-intrusion multi-platform latency sculpting system. It aims to break down the inference latency across pipeline, proactively alert on inference latency anomalies, and guarantee adherence to SLOs, all without requiring code modifications or service restarts. LatencyPrism has been deployed across thousands of XPUs for over six months. It enables low-overhead real-time monitoring at batch level with alerts triggered in milliseconds. This approach distinguishes between workload-driven latency variations and anomalies indicating underlying issues with an F1-score of 0.98. We also conduct extensive experiments and investigations into root cause analysis to demonstrate LatencyPrism's capability.

While Supervised Fine-Tuning (SFT) and Rejection Sampling Fine-Tuning (RFT) are standard for LLM alignment, they either rely on costly expert data or discard valuable negative samples, leading to data inefficiency. To address this, we propose Reward Informed Fine-Tuning (RIFT), a simple yet effective framework that utilizes all self-generated samples. Unlike the hard thresholding of RFT, RIFT repurposes negative trajectories, reweighting the loss with scalar rewards to learn from both the positive and negative trajectories from the model outputs. To overcome the training collapse caused by naive reward integration, where direct multiplication yields an unbounded loss, we introduce a stabilized loss formulation that ensures numerical robustness and optimization efficiency. Extensive experiments on mathematical benchmarks across various base models show that RIFT consistently outperforms RFT. Our results demonstrate that RIFT is a robust and data-efficient alternative for alignment using mixed-quality, self-generated data.

Fluid-structure interaction (FSI) systems involve distinct physical domains, fluid and solid, governed by different partial differential equations and coupled at a dynamic interface. While learning-based solvers offer a promising alternative to costly numerical simulations, existing methods struggle to capture the heterogeneous dynamics of FSI within a unified framework. This challenge is further exacerbated by inconsistencies in response across domains due to interface coupling and by disparities in learning difficulty across fluid and solid regions, leading to instability during prediction. To address these challenges, we propose the Heterogeneous Graph Attention Solver (HGATSolver). HGATSolver encodes the system as a heterogeneous graph, embedding physical structure directly into the model via distinct node and edge types for fluid, solid, and interface regions. This enables specialized message-passing mechanisms tailored to each physical domain. To stabilize explicit time stepping, we introduce a novel physics-conditioned gating mechanism that serves as a learnable, adaptive relaxation factor. Furthermore, an Inter-domain Gradient-Balancing Loss dynamically balances the optimization objectives across domains based on predictive uncertainty. Extensive experiments on two constructed FSI benchmarks and a public dataset demonstrate that HGATSolver achieves state-of-the-art performance, establishing an effective framework for surrogate modeling of coupled multi-physics systems.

Large Language Models (LLMs) are increasingly used for toxicity assessment in online moderation systems, where fairness across demographic groups is essential for equitable treatment. However, LLMs often produce inconsistent toxicity judgements for subtle expressions, particularly those involving implicit hate speech, revealing underlying biases that are difficult to correct through standard training. This raises a key question that existing approaches often overlook: when should corrective mechanisms be invoked to ensure fair and reliable assessments? To address this, we propose FairToT, an inference-time framework that enhances LLM fairness through prompt-guided toxicity assessment. FairToT identifies cases where demographic-related variation is likely to occur and determines when additional assessment should be applied. In addition, we introduce two interpretable fairness indicators that detect such cases and improve inference consistency without modifying model parameters. Experiments on benchmark datasets show that FairToT reduces group-level disparities while maintaining stable and reliable toxicity predictions, demonstrating that inference-time refinement offers an effective and practical approach for fairness improvement in LLM-based toxicity assessment systems. The source code can be found at https://aisuko.github.io/fair-tot/.

Autonomous agents such as cars, robots and drones need to precisely localize themselves in diverse environments, including in GPS-denied indoor environments. One approach for precise localization is visual place recognition (VPR), which estimates the place of an image based on previously seen places. State-of-the-art VPR models require high amounts of memory, making them unwieldy for mobile deployment, while more compact models lack robustness and generalization capabilities. This work overcomes these limitations for robotics using a combination of event-based vision sensors and an event-based novel guided variational autoencoder (VAE). The encoder part of our model is based on a spiking neural network model which is compatible with power-efficient low latency neuromorphic hardware. The VAE successfully disentangles the visual features of 16 distinct places in our new indoor VPR dataset with a classification performance comparable to other state-of-the-art approaches while, showing robust performance also under various illumination conditions. When tested with novel visual inputs from unknown scenes, our model can distinguish between these places, which demonstrates a high generalization capability by learning the essential features of location. Our compact and robust guided VAE with generalization capabilities poses a promising model for visual place recognition that can significantly enhance mobile robot navigation in known and unknown indoor environments.

Standardized Student Evaluation of Teaching often suffer from low reliability, restricted response options, and response distortion. Existing machine learning methods that mine open-ended comments usually reduce feedback to binary sentiment, which overlooks concrete concerns such as content clarity, feedback timeliness, and instructor demeanor, and provides limited guidance for instructional improvement.We propose TeachPro, a multi-label learning framework that systematically assesses five key teaching dimensions: professional expertise, instructional behavior, pedagogical efficacy, classroom experience, and other performance metrics. We first propose a Dimension-Anchored Evidence Encoder, which integrates three core components: (i) a pre-trained text encoder that transforms qualitative feedback annotations into contextualized embeddings; (ii) a prompt module that represents five teaching dimensions as learnable semantic anchors; and (iii) a cross-attention mechanism that aligns evidence with pedagogical dimensions within a structured semantic space. We then propose a Cross-View Graph Synergy Network to represent student comments. This network comprises two components: (i) a Syntactic Branch that extracts explicit grammatical dependencies from parse trees, and (ii) a Semantic Branch that models latent conceptual relations derived from BERT-based similarity graphs. BiAffine fusion module aligns syntactic and semantic units, while a differential regularizer disentangles embeddings to encourage complementary representations. Finally, a cross-attention mechanism bridges the dimension-anchored evidence with the multi-view comment representations. We also contribute a novel benchmark dataset featuring expert qualitative annotations and multi-label scores. Extensive experiments demonstrate that TeachPro offers superior diagnostic granularity and robustness across diverse evaluation settings.

Knowledge Graph Retrieval-Augmented Generation (KG-RAG) extends the RAG paradigm by incorporating structured knowledge from knowledge graphs, enabling Large Language Models (LLMs) to perform more precise and explainable reasoning. While KG-RAG improves factual accuracy in complex tasks, existing KG-RAG models are often severely overconfident, producing high-confidence predictions even when retrieved sub-graphs are incomplete or unreliable, which raises concerns for deployment in high-stakes domains. To address this issue, we propose Ca2KG, a Causality-aware Calibration framework for KG-RAG. Ca2KG integrates counterfactual prompting, which exposes retrieval-dependent uncertainties in knowledge quality and reasoning reliability, with a panel-based re-scoring mechanism that stabilises predictions across interventions. Extensive experiments on two complex QA datasets demonstrate that Ca2KG consistently improves calibration while maintaining or even enhancing predictive accuracy.

Despite the prevalent assumption of uniform variable importance in long-term time series forecasting models, real world applications often exhibit asymmetric causal relationships and varying data acquisition costs. Specifically, cost-effective exogenous data (e.g., local weather) can unilaterally influence dynamics of endogenous variables, such as lake surface temperature. Exploiting these links enables more effective forecasts when exogenous inputs are readily available. Transformer-based models capture long-range dependencies but incur high computation and suffer from permutation invariance. Patch-based variants improve efficiency yet can miss local temporal patterns. To efficiently exploit informative signals across both the temporal dimension and relevant exogenous variables, this study proposes XLinear, a lightweight time series forecasting model built upon MultiLayer Perceptrons (MLPs). XLinear uses a global token derived from an endogenous variable as a pivotal hub for interacting with exogenous variables, and employs MLPs with sigmoid activation to extract both temporal patterns and variate-wise dependencies. Its prediction head then integrates these signals to forecast the endogenous series. We evaluate XLinear on seven standard benchmarks and five real-world datasets with exogenous inputs. Compared with state-of-the-art models, XLinear delivers superior accuracy and efficiency for both multivariate forecasts and univariate forecasts influenced by exogenous inputs.

Reward design remains a significant bottleneck in applying reinforcement learning (RL) to real-world problems. A popular alternative is reward learning, where reward functions are inferred from human feedback rather than manually specified. Recent work has proposed learning reward functions from human feedback in the form of ratings, rather than traditional binary preferences, enabling richer and potentially less cognitively demanding supervision. Building on this paradigm, we introduce a new rating-based RL method, Ranked Return Regression for RL (R4). At its core, R4 employs a novel ranking mean squared error (rMSE) loss, which treats teacher-provided ratings as ordinal targets. Our approach learns from a dataset of trajectory-rating pairs, where each trajectory is labeled with a discrete rating (e.g., "bad," "neutral," "good"). At each training step, we sample a set of trajectories, predict their returns, and rank them using a differentiable sorting operator (soft ranks). We then optimize a mean squared error loss between the resulting soft ranks and the teacher's ratings. Unlike prior rating-based approaches, R4 offers formal guarantees: its solution set is provably minimal and complete under mild assumptions. Empirically, using simulated human feedback, we demonstrate that R4 consistently matches or outperforms existing rating and preference-based RL methods on robotic locomotion benchmarks from OpenAI Gym and the DeepMind Control Suite, while requiring significantly less feedback.

The prevailing post-training paradigm for Large Reasoning Models (LRMs)--Supervised Fine-Tuning (SFT) followed by Reinforcement Learning (RL)--suffers from an intrinsic optimization mismatch: the rigid supervision inherent in SFT induces distributional collapse, thereby exhausting the exploration space necessary for subsequent RL. In this paper, we reformulate SFT within a unified post-training framework and propose Gibbs Initialization with Finite Temperature (GIFT). We characterize standard SFT as a degenerate zero-temperature limit that suppresses base priors. Conversely, GIFT incorporates supervision as a finite-temperature energy potential, establishing a distributional bridge that ensures objective consistency throughout the post-training pipeline. Our experiments demonstrate that GIFT significantly outperforms standard SFT and other competitive baselines when utilized for RL initialization, providing a mathematically principled pathway toward achieving global optimality in post-training. Our code is available at https://github.com/zzy1127/GIFT.

Marked Temporal Point Processes (MTPPs) arise naturally in medical, social, commercial, and financial domains. However, existing Transformer-based methods mostly inject temporal information only via positional encodings, relying on shared or parametric decay structures, which limits their ability to capture heterogeneous and type-specific temporal effects. Inspired by this observation, we derive a novel attention operator called Hawkes Attention from the multivariate Hawkes process theory for MTPP, using learnable per-type neural kernels to modulate query, key and value projections, thereby replacing the corresponding parts in the traditional attention. Benefited from the design, Hawkes Attention unifies event timing and content interaction, learning both the time-relevant behavior and type-specific excitation patterns from the data. The experimental results show that our method achieves better performance compared to the baselines. In addition to the general MTPP, our attention mechanism can also be easily applied to specific temporal structures, such as time series forecasting.

User simulators serve as the critical interactive environment for agent post-training, and an ideal user simulator generalizes across domains and proactively engages in negotiation by challenging or bargaining. However, current methods exhibit two issues. They rely on static and context-unaware profiles, necessitating extensive manual redesign for new scenarios, thus limiting generalizability. Moreover, they neglect human strategic thinking, leading to vulnerability to agent manipulation. To address these issues, we propose UserLM-R1, a novel user language model with reasoning capability. Specifically, we first construct comprehensive user profiles with both static roles and dynamic scenario-specific goals for adaptation to diverse scenarios. Then, we propose a goal-driven decision-making policy to generate high-quality rationales before producing responses, and further refine the reasoning and improve strategic capabilities with supervised fine-tuning and multi-reward reinforcement learning. Extensive experimental results demonstrate that UserLM-R1 outperforms competitive baselines, particularly on the more challenging adversarial set.

Reconstructing natural visual scenes from neural activity is a key challenge in neuroscience and computer vision. We present SpikeVAEDiff, a novel two-stage framework that combines a Very Deep Variational Autoencoder (VDVAE) and the Versatile Diffusion model to generate high-resolution and semantically meaningful image reconstructions from neural spike data. In the first stage, VDVAE produces low-resolution preliminary reconstructions by mapping neural spike signals to latent representations. In the second stage, regression models map neural spike signals to CLIP-Vision and CLIP-Text features, enabling Versatile Diffusion to refine the images via image-to-image generation. We evaluate our approach on the Allen Visual Coding-Neuropixels dataset and analyze different brain regions. Our results show that the VISI region exhibits the most prominent activation and plays a key role in reconstruction quality. We present both successful and unsuccessful reconstruction examples, reflecting the challenges of decoding neural activity. Compared with fMRI-based approaches, spike data provides superior temporal and spatial resolution. We further validate the effectiveness of the VDVAE model and conduct ablation studies demonstrating that data from specific brain regions significantly enhances reconstruction performance.

Despite significant progress in autoregressive image generation, inference remains slow due to the sequential nature of AR models and the ambiguity of image tokens, even when using speculative decoding. Recent works attempt to address this with relaxed speculative decoding but lack theoretical grounding. In this paper, we establish the theoretical basis of relaxed SD and propose COOL-SD, an annealed relaxation of speculative decoding built on two key insights. The first analyzes the total variation (TV) distance between the target model and relaxed speculative decoding and yields an optimal resampling distribution that minimizes an upper bound of the distance. The second uses perturbation analysis to reveal an annealing behaviour in relaxed speculative decoding, motivating our annealed design. Together, these insights enable COOL-SD to generate images faster with comparable quality, or achieve better quality at similar latency. Experiments validate the effectiveness of COOL-SD, showing consistent improvements over prior methods in speed-quality trade-offs.

Recent progress in large language models and multimodal interaction has made it possible to develop AI companions that can have fluent and emotionally expressive conversations. However, many of these systems have problems keeping users satisfied and engaged over long periods. This paper argues that these problems do not come mainly from weak models, but from poor character design and unclear definitions of the user-AI relationship. I present Mikasa, an emotional AI companion inspired by Japanese Oshi culture-specifically its emphasis on long-term, non-exclusive commitment to a stable character-as a case study of character-driven companion design. Mikasa does not work as a general-purpose assistant or a chatbot that changes roles. Instead, Mikasa is designed as a coherent character with a stable personality and a clearly defined relationship as a partner. This relationship does not force exclusivity or obligation. Rather, it works as a reference point that stabilizes interaction norms and reduces the work users must do to keep redefining the relationship. Through an exploratory evaluation, I see that users describe their preferences using surface-level qualities such as conversational naturalness, but they also value relationship control and imaginative engagement in ways they do not state directly. These results suggest that character coherence and relationship definition work as latent structural elements that shape how good the interaction feels, without users recognizing them as main features. The contribution of this work is to show that character design is a functional part of AI companion systems, not just decoration. Mikasa is one example based on a specific cultural context, but the design principles-commitment to a consistent personality and clear relationship definition-can be used for many emotionally grounded AI companions.

We introduce A.X K1, a 519B-parameter Mixture-of-Experts (MoE) language model trained from scratch. Our design leverages scaling laws to optimize training configurations and vocabulary size under fixed computational budgets. A.X K1 is pre-trained on a corpus of approximately 10T tokens, curated by a multi-stage data processing pipeline. Designed to bridge the gap between reasoning capability and inference efficiency, A.X K1 supports explicitly controllable reasoning to facilitate scalable deployment across diverse real-world scenarios. We propose a simple yet effective Think-Fusion training recipe, enabling user-controlled switching between thinking and non-thinking modes within a single unified model. Extensive evaluations demonstrate that A.X K1 achieves performance competitive with leading open-source models, while establishing a distinctive advantage in Korean-language benchmarks.

Supervised fine-tuning (SFT) is a fundamental post-training strategy to align Large Language Models (LLMs) with human intent. However, traditional SFT often ignores the one-to-many nature of language by forcing alignment with a single reference answer, leading to the model overfitting to non-core expressions. Although our empirical analysis suggests that introducing multiple reference answers can mitigate this issue, the prohibitive data and computational costs necessitate a strategic shift: prioritizing the mitigation of single-reference overfitting over the costly pursuit of answer diversity. To achieve this, we reveal the intrinsic connection between token probability and semantic importance: high-probability tokens carry the core logical framework, while low-probability tokens are mostly replaceable expressions. Based on this insight, we propose ProFit, which selectively masks low-probability tokens to prevent surface-level overfitting. Extensive experiments confirm that ProFit consistently outperforms traditional SFT baselines on general reasoning and mathematical benchmarks.

Large language models and text encoders increasingly power web-based information systems in the social sciences, including digital libraries, data catalogues, and search interfaces used by researchers, policymakers, and civil society. Full fine-tuning is often computationally and energy intensive, which can be prohibitive for smaller institutions and non-profit organizations in the Web4Good ecosystem. Parameter-Efficient Fine-Tuning (PEFT), especially Low-Rank Adaptation (LoRA), reduces this cost by updating only a small number of parameters. We show that the standard LoRA update $ΔW = BA^\top$ has geometric drawbacks: gauge freedom, scale ambiguity, and a tendency toward rank collapse. We introduce OrthoGeoLoRA, which enforces an SVD-like form $ΔW = BΣA^\top$ by constraining the low-rank factors to be orthogonal (Stiefel manifold). A geometric reparameterization implements this constraint while remaining compatible with standard optimizers such as Adam and existing fine-tuning pipelines. We also propose a benchmark for hierarchical concept retrieval over the European Language Social Science Thesaurus (ELSST), widely used to organize social science resources in digital repositories. Experiments with a multilingual sentence encoder show that OrthoGeoLoRA outperforms standard LoRA and several strong PEFT variants on ranking metrics under the same low-rank budget, offering a more compute- and parameter-efficient path to adapt foundation models in resource-constrained settings.

The parallel Byzantine Fault Tolerant (BFT) protocol is viewed as a promising solution to address the consensus scalability issue of the permissioned blockchain. One of the main challenges in parallel BFT is the view change process that happens when the leader node fails, which can lead to performance bottlenecks. Existing parallel BFT protocols typically rely on passive view change mechanisms with blind leader rotation. Such approaches frequently select unavailable or slow nodes as leaders, resulting in degraded performance. To address these challenges, we propose a View Change Optimization (VCO) model based on mixed integer programming that optimizes leader selection and follower reassignment across parallel committees by considering communication delays and failure scenarios. We applied a decomposition method with efficient subproblems and improved benders cuts to solve the VCO model. Leveraging the results of improved decomposition solution method, we propose an efficient iterative backup leader selection algorithm as views proceed. By performing experiments in Microsoft Azure cloud environments, we demonstrate that the VCO-driven parallel BFT outperforms existing configuration methods under both normal operation and faulty condition. The results show that the VCO model is effective as network size increases, making it a suitable solution for high-performance parallel BFT systems.

The rapid expansion of AI research has intensified the Reviewer Gap, threatening the peer-review sustainability and perpetuating a cycle of low-quality evaluations. This position paper critiques existing LLM approaches that automatically generate reviews and argues for a paradigm shift that positions LLMs as tools for assisting and educating human reviewers. We define the core principles of high-quality peer review and propose two complementary systems grounded in these foundations: (i) an LLM-assisted mentoring system that cultivates reviewers' long-term competencies, and (ii) an LLM-assisted feedback system that helps reviewers refine the quality of their reviews. This human-centered approach aims to strengthen reviewer expertise and contribute to building a more sustainable scholarly ecosystem.

Large language models (LLMs) face significant deployment challenges due to their massive computational demands. % While pruning offers a promising compression solution, existing methods suffer from two critical limitations: (1) They neglect activation distribution shifts between calibration data and test data, resulting in inaccurate error estimations; (2) They overlook the long-tail distribution characteristics of activations in the attention module. To address these limitations, this paper proposes a novel pruning method, $D^2Prune$. First, we propose a dual Taylor expansion-based method that jointly models weight and activation perturbations for precise error estimation, leading to precise pruning mask selection and weight updating and facilitating error minimization during pruning. % Second, we propose an attention-aware dynamic update strategy that preserves the long-tail attention pattern by jointly minimizing the KL divergence of attention distributions and the reconstruction error. Extensive experiments show that $D^2Prune$ consistently outperforms SOTA methods across various LLMs (e.g., OPT-125M, LLaMA2/3, and Qwen3). Moreover, the dynamic attention update mechanism also generalizes well to ViT-based vision models like DeiT, achieving superior accuracy on ImageNet-1K.

Analysis of learned representations has a blind spot: it focuses on $similarity$, measuring how closely embeddings align with external references, but similarity reveals only what is represented, not whether that structure is robust. We introduce $geometric$ $stability$, a distinct dimension that quantifies how reliably representational geometry holds under perturbation, and present $Shesha$, a framework for measuring it. Across 2,463 configurations in seven domains, we show that stability and similarity are empirically uncorrelated ($ρ\approx 0.01$) and mechanistically distinct: similarity metrics collapse after removing the top principal components, while stability retains sensitivity to fine-grained manifold structure. This distinction yields actionable insights: for safety monitoring, stability acts as a functional geometric canary, detecting structural drift nearly 2$\times$ more sensitively than CKA while filtering out the non-functional noise that triggers false alarms in rigid distance metrics; for controllability, supervised stability predicts linear steerability ($ρ= 0.89$-$0.96$); for model selection, stability dissociates from transferability, revealing a geometric tax that transfer optimization incurs. Beyond machine learning, stability predicts CRISPR perturbation coherence and neural-behavioral coupling. By quantifying $how$ $reliably$ systems maintain structure, geometric stability provides a necessary complement to similarity for auditing representations across biological and computational systems.

As Large Language Models (LLMs) increasingly shape online content, removing targeted information from well-trained LLMs (also known as LLM unlearning) has become critical for web governance. A key challenge lies in sample-wise imbalance within the forget set: different samples exhibit widely varying unlearning difficulty, leading to asynchronous forgetting where some knowledge remains insufficiently erased while others become over-forgotten. To address this, we propose BalDRO, a novel and efficient framework for balanced LLM unlearning. BalDRO formulates unlearning as a min-sup process: an inner step identifies a worst-case data distribution that emphasizes hard-to-unlearn samples, while an outer step updates model parameters under this distribution. We instantiate BalDRO via two efficient variants: BalDRO-G, a discrete GroupDRO-based approximation focusing on high-loss subsets, and BalDRO-DV, a continuous Donsker-Varadhan dual method enabling smooth adaptive weighting within standard training pipelines. Experiments on TOFU and MUSE show that BalDRO significantly improves both forgetting quality and model utility over existing methods, and we release code for reproducibility.

In this work, we present SafePlanner, a systematic testing framework for identifying safety-critical flaws in the Plan model of Automated Driving Systems (ADS). SafePlanner targets two core challenges: generating structurally meaningful test scenarios and detecting hazardous planning behaviors. To maximize coverage, SafePlanner performs a structural analysis of the Plan model implementation - specifically, its scene-transition logic and hierarchical control flow - and uses this insight to extract feasible scene transitions from code. It then composes test scenarios by combining these transitions with non-player vehicle (NPC) behaviors. Guided fuzzing is applied to explore the behavioral space of the Plan model under these scenarios. We evaluate SafePlanner on Baidu Apollo, a production-grade level 4 ADS. It generates 20635 test cases and detects 520 hazardous behaviors, grouped into 15 root causes through manual analysis. For four of these, we applied patches based on our analysis; the issues disappeared, and no apparent side effects were observed. SafePlanner achieves 83.63 percent function and 63.22 percent decision coverage on the Plan model, outperforming baselines in both bug discovery and efficiency.

In this work, we propose N EIoU YOLOv9, a lightweight detection framework based on a signal aware bounding box regression loss derived from non monotonic gradient focusing and geometric decoupling principles, referred to as N EIoU (Non monotonic Efficient Intersection over Union). The proposed loss reshapes localization gradients by combining non monotonic focusing with decoupled width and height optimization, thereby enhancing weak regression signals for hard samples with low overlap while reducing gradient interference. This design is particularly effective for small and low contrast targets commonly observed in agricultural disease imagery. The proposed N EIoU loss is integrated into a lightweight YOLOv9t architecture and evaluated on a self collected field dataset comprising 5908 rice leaf images across four disease categories and healthy leaves. Experimental results demonstrate consistent performance gains over the standard CIoU loss, achieving a mean Average Precision of 90.3 percent, corresponding to a 4.3 percent improvement over the baseline, with improved localization accuracy under stricter evaluation criteria. For practical validation, the optimized model is deployed on an Android device using TensorFlow Lite with Float16 quantization, achieving an average inference time of 156 milliseconds per frame while maintaining accuracy. These results confirm that the proposed approach effectively balances accuracy, optimization stability, and computational efficiency for edge based agricultural monitoring systems.

Differentially private federated learning (DP-FL) suffers from slow convergence under tight privacy budgets due to the overwhelming noise introduced to preserve privacy. While adaptive optimizers can accelerate convergence, existing second-order methods such as DP-FedNew require O(d^2) memory at each client to maintain local feature covariance matrices, making them impractical for high-dimensional models. We propose DP-FedSOFIM, a server-side second-order optimization framework that leverages the Fisher Information Matrix (FIM) as a natural gradient preconditioner while requiring only O(d) memory per client. By employing the Sherman-Morrison formula for efficient matrix inversion, DP-FedSOFIM achieves O(d) computational complexity per round while maintaining the convergence benefits of second-order methods. Our analysis proves that the server-side preconditioning preserves (epsilon, delta)-differential privacy through the post-processing theorem. Empirical evaluation on CIFAR-10 demonstrates that DP-FedSOFIM achieves superior test accuracy compared to first-order baselines across multiple privacy regimes.

Building on the probability-domain distillation framework of Sparse-KD, we develop an axiomatic, operator-theoretic framework for multi-teacher ensemble knowledge distillation. Rather than prescribing a specific aggregation formula, we define five core axioms governing valid knowledge aggregation operators, encompassing convexity, positivity, continuity, weight monotonicity, and temperature coherence. We prove the existence and non-uniqueness of operator families satisfying these axioms, establishing that multiple distinct aggregation mechanisms conform to the same foundational principles. Within this framework, we establish operator-agnostic guarantees showing that multi-teacher aggregation reduces both stochastic variance and systematic supervisory bias under heterogeneous teachers, while providing Jensen-type bounds, log-loss guarantees, and safety attenuation properties. For aggregation operators linear in teacher weights, we further establish classical ensemble variance-reduction results under standard independence assumptions, with extensions to correlated-error regimes. The framework provides theoretical grounding for multi-teacher distillation from diverse frontier models while admitting multiple valid implementation strategies.

We study the Bandit Clustering (BC) problem under the fixed confidence setting, where the objective is to group a collection of data sequences (arms) into clusters through sequential sampling from adaptively selected arms at each time step while ensuring a fixed error probability at the stopping time. We consider a setting where arms in a cluster may have different distributions. Unlike existing results in this setting, which assume Gaussian-distributed arms, we study a broader class of vector-parametric distributions that satisfy mild regularity conditions. Existing asymptotically optimal BC algorithms require solving an optimization problem as part of their sampling rule at each step, which is computationally costly. We propose an Efficient Bandit Clustering algorithm (EBC), which, instead of solving the full optimization problem, takes a single step toward the optimal value at each time step, making it computationally efficient while remaining asymptotically optimal. We also propose a heuristic variant of EBC, called EBC-H, which further simplifies the sampling rule, with arm selection based on quantities computed as part of the stopping rule. We highlight the computational efficiency of EBC and EBC-H by comparing their per-sample run time with that of existing algorithms. The asymptotic optimality of EBC is supported through simulations on the synthetic datasets. Through simulations on both synthetic and real-world datasets, we show the performance gain of EBC and EBC-H over existing approaches.

While much of the current research in deep learning-based vulnerability detection relies on disassembled binaries, this paper explores the feasibility of extracting features directly from raw x86-64 machine code. Although assembly language is more interpretable for humans, it requires more complex models to capture token-level context. In contrast, machine code may enable more efficient, lightweight models and preserve all information that might be lost in disassembly. This paper approaches the task of vulnerability detection through an exploratory study on two specific deep learning model architectures and aims to systematically evaluate their performance across three vulnerability types. The results demonstrate that graph-based models consistently outperform sequential models, emphasizing the importance of control flow relationships, and that machine code contains sufficient information for effective vulnerability discovery.

Using Artificial Intelligence to improve teaching and learning benefits greater adaptivity and scalability in education. Knowledge Tracing (KT) is recognized for student modeling task due to its superior performance and application potential in education. To this end, we conceptualize and investigate counterfactual explanation as the connection from XAI for KT to education. Counterfactual explanations offer actionable recourse, are inherently causal and local, and easy for educational stakeholders to understand who are often non-experts. We propose KTCF, a counterfactual explanation generation method for KT that accounts for knowledge concept relationships, and a post-processing scheme that converts a counterfactual explanation into a sequence of educational instructions. We experiment on a large-scale educational dataset and show our KTCF method achieves superior and robust performance over existing methods, with improvements ranging from 5.7% to 34% across metrics. Additionally, we provide a qualitative evaluation of our post-processing scheme, demonstrating that the resulting educational instructions help in reducing large study burden. We show that counterfactuals have the potential to advance the responsible and practical use of AI in education. Future works on XAI for KT may benefit from educationally grounded conceptualization and developing stakeholder-centered methods.

This paper introduces PRA, an AI-agent design for simulating how individual users form privacy concerns in response to real-world news. Moving beyond population-level sentiment analysis, PRA integrates privacy and cognitive theories to simulate user-specific privacy reasoning grounded in personal comment histories and contextual cues. The agent reconstructs each user's "privacy mind", dynamically activates relevant privacy memory through a contextual filter that emulates bounded rationality, and generates synthetic comments reflecting how that user would likely respond to new privacy scenarios. A complementary LLM-as-a-Judge evaluator, calibrated against an established privacy concern taxonomy, quantifies the faithfulness of generated reasoning. Experiments on real-world Hacker News discussions show that \PRA outperforms baseline agents in privacy concern prediction and captures transferable reasoning patterns across domains including AI, e-commerce, and healthcare.

Large Language Models (LLMs) demonstrate potential to estimate the probability of uncertain events, by leveraging their extensive knowledge and reasoning capabilities. This ability can be applied to support intelligent decision-making across diverse fields, such as financial forecasting and preventive healthcare. However, directly prompting LLMs for probability estimation faces significant challenges: their outputs are often noisy, and the underlying predicting process is opaque. In this paper, we propose PRISM: Probability Reconstruction via Shapley Measures, a framework that brings transparency and precision to LLM-based probability estimation. PRISM decomposes an LLM's prediction by quantifying the marginal contribution of each input factor using Shapley values. These factor-level contributions are then aggregated to reconstruct a calibrated final estimate. In our experiments, we demonstrate PRISM improves predictive accuracy over direct prompting and other baselines, across multiple domains including finance, healthcare, and agriculture. Beyond performance, PRISM provides a transparent prediction pipeline: our case studies visualize how individual factors shape the final estimate, helping build trust in LLM-based decision support systems.

Zero-Shot Anomaly Detection (ZSAD) leverages Vision-Language Models (VLMs) to enable supervision-free industrial inspection. However, existing ZSAD paradigms are constrained by single visual backbones, which struggle to balance global semantic generalization with fine-grained structural discriminability. To bridge this gap, we propose Synergistic Semantic-Visual Prompting (SSVP), that efficiently fuses diverse visual encodings to elevate model's fine-grained perception. Specifically, SSVP introduces the Hierarchical Semantic-Visual Synergy (HSVS) mechanism, which deeply integrates DINOv3's multi-scale structural priors into the CLIP semantic space. Subsequently, the Vision-Conditioned Prompt Generator (VCPG) employs cross-modal attention to guide dynamic prompt generation, enabling linguistic queries to precisely anchor to specific anomaly patterns. Furthermore, to address the discrepancy between global scoring and local evidence, the Visual-Text Anomaly Mapper (VTAM) establishes a dual-gated calibration paradigm. Extensive evaluations on seven industrial benchmarks validate the robustness of our method; SSVP achieves state-of-the-art performance with 93.0\% Image-AUROC and 92.2\% Pixel-AUROC on MVTec-AD, significantly outperforming existing zero-shot approaches.

We present a transaction-driven dynamic reconfiguration protocol in Modern payment systems based on Byzantine Consistent Broadcast which can achieve high performance by avoiding global transaction ordering. We demonstrate the fundamental paradigm of modern payment systems, which combines user nonce based transactions ordering with periodic system-wide consensus mechanisms. Building on this foundation, we design PDCC(Payment Dynamic Config Change), which can lead a smooth reconfiguration process without impacting the original system's performance.

Neural operator learning accelerates PDE solution by approximating operators as mappings between continuous function spaces. Yet in many engineering settings, varying geometry induces discrete structural changes, including topological changes, abrupt changes in boundary conditions or boundary types, and changes in the effective computational domain, which break the smooth-variation premise. Here we introduce Discrete Solution Operator Learning (DiSOL), a complementary paradigm that learns discrete solution procedures rather than continuous function-space operators. DiSOL factorizes the solver into learnable stages that mirror classical discretizations: local contribution encoding, multiscale assembly, and implicit solution reconstruction on an embedded grid, thereby preserving procedure-level consistency while adapting to geometry-dependent discrete structures. Across geometry-dependent Poisson, advection-diffusion, linear elasticity, as well as spatiotemporal heat-conduction problems, DiSOL produces stable and accurate predictions under both in-distribution and strongly out-of-distribution geometries, including discontinuous boundaries and topological changes. These results highlight the need for procedural operator representations in geometry-dominated regimes and position discrete solution operator learning as a distinct, complementary direction in scientific machine learning.

Detecting evasive answers in earnings calls is critical for financial transparency, yet progress is hindered by the lack of large-scale benchmarks. We introduce EvasionBench, comprising 30,000 training samples and 1,000 human-annotated test samples (Cohen's Kappa 0.835) across three evasion levels. Our key contribution is a multi-model annotation framework leveraging a core insight: disagreement between frontier LLMs signals hard examples most valuable for training. We mine boundary cases where two strong annotators conflict, using a judge to resolve labels. This approach outperforms single-model distillation by 2.4 percent, with judge-resolved samples improving generalization despite higher training loss (0.421 vs 0.393) - evidence that disagreement mining acts as implicit regularization. Our trained model Eva-4B (4B parameters) achieves 81.3 percent accuracy, outperforming its base by 25 percentage points and approaching frontier LLM performance at a fraction of inference cost.

Large Language Model (LLM) outputs often vary across user sociodemographic attributes, leading to disparities in factual accuracy, utility, and safety, even for objective questions where demographic information is irrelevant. Unlike prior work on stereotypical or representational bias, this paper studies identity-dependent degradation of core response quality. We show empirically that such degradation arises from biased generation behavior, despite factual knowledge being robustly encoded across identities. Motivated by this mismatch, we propose a lightweight, training-free framework for identity-robust generation that selectively neutralizes non-critical identity information while preserving semantically essential attributes, thus maintaining output content integrity. Experiments across four benchmarks and 18 sociodemographic identities demonstrate an average 77% reduction in identity-dependent bias compared to vanilla prompting and a 45% reduction relative to prompt-based defenses. Our work addresses a critical gap in mitigating the impact of user identity cues in prompts on core generation quality.

General-purpose Large Vision-Language Models (LVLMs), despite their massive scale, often falter in dermatology due to "diffuse attention" - the inability to disentangle subtle pathological lesions from background noise. In this paper, we challenge the assumption that parameter scaling is the only path to medical precision. We introduce SkinFlow, a framework that treats diagnosis as an optimization of visual information transmission efficiency. Our approach utilizes a Virtual-Width Dynamic Vision Encoder (DVE) to "unfold" complex pathological manifolds without physical parameter expansion, coupled with a two-stage Reinforcement Learning strategy. This strategy sequentially aligns explicit medical descriptions (Stage I) and reconstructs implicit diagnostic textures (Stage II) within a constrained semantic space. Furthermore, we propose a clinically grounded evaluation protocol that prioritizes diagnostic safety and hierarchical relevance over rigid label matching. Empirical results are compelling: our 7B model establishes a new state-of-the-art on the Fitzpatrick17k benchmark, achieving a +12.06% gain in Top-1 accuracy and a +28.57% boost in Top-6 accuracy over the massive general-purpose models (e.g., Qwen3VL-235B and GPT-5.2). These findings demonstrate that optimizing geometric capacity and information flow yields superior diagnostic reasoning compared to raw parameter scaling.

Vision Transformers (ViTs) have gained rapid adoption in computational pathology for their ability to model long-range dependencies through self-attention, addressing the limitations of convolutional neural networks that excel at local pattern capture but struggle with global contextual reasoning. Recent pathology-specific foundation models have further advanced performance by leveraging large-scale pretraining. However, standard ViTs remain inherently non-equivariant to transformations such as rotations and reflections, which are ubiquitous variations in histopathology imaging. To address this limitation, we propose Equi-ViT, which integrates an equivariant convolution kernel into the patch embedding stage of a ViT architecture, imparting built-in rotational equivariance to learned representations. Equi-ViT achieves superior rotation-consistent patch embeddings and stable classification performance across image orientations. Our results on a public colorectal cancer dataset demonstrate that incorporating equivariant patch embedding enhances data efficiency and robustness, suggesting that equivariant transformers could potentially serve as more generalizable backbones for the application of ViT in histopathology, such as digital pathology foundation models.

Current patent claim generation systems face three fundamental limitations: poor cross-jurisdictional generalization, inadequate semantic relationship modeling between claims and prior art, and unreliable quality assessment. We introduce a novel three-stage framework that addresses these challenges through relationship-aware similarity analysis, domain-adaptive claim generation, and unified quality assessment. Our approach employs multi-head attention with eight specialized heads for explicit relationship modeling, integrates curriculum learning with dynamic LoRA adapter selection across five patent domains, and implements cross-attention mechanisms between evaluation aspects for comprehensive quality assessment. Extensive experiments on USPTO HUPD dataset, EPO patent collections, and Patent-CE benchmark demonstrate substantial improvements: 7.6-point ROUGE-L gain over GPT-4o, 8.3\% BERTScore enhancement over Llama-3.1-8B, and 0.847 correlation with human experts compared to 0.623 for separate evaluation models. Our method maintains 89.4\% cross-jurisdictional performance retention versus 76.2\% for baselines, establishing a comprehensive solution for automated patent prosecution workflows.

Fine-grained labor market analysis increasingly relies on mapping unstructured job advertisements to standardized skill taxonomies such as ESCO. This mapping is naturally formulated as an Extreme Multi-Label Classification (XMLC) problem, but supervised solutions are constrained by the scarcity and cost of large-scale, taxonomy-aligned annotations--especially in non-English settings where job-ad language diverges substantially from formal skill definitions. We propose a zero-shot skill extraction framework that eliminates the need for manually labeled job-ad training data. The framework uses a Large Language Model (LLM) to synthesize training instances from ESCO definitions, and introduces hierarchically constrained multi-skill generation based on ESCO Level-2 categories to improve semantic coherence in multi-label contexts. On top of the synthetic corpus, we train a contrastive bi-encoder that aligns job-ad sentences with ESCO skill descriptions in a shared embedding space; the encoder augments a BERT backbone with BiLSTM and attention pooling to better model long, information-dense requirement statements. An upstream RoBERTa-based binary filter removes non-skill sentences to improve end-to-end precision. Experiments show that (i) hierarchy-conditioned generation improves both fluency and discriminability relative to unconstrained pairing, and (ii) the resulting multi-label model transfers effectively to real-world Chinese job advertisements, achieving strong zero-shot retrieval performance (F1@5 = 0.72) and outperforming TF--IDF and standard BERT baselines. Overall, the proposed pipeline provides a scalable, data-efficient pathway for automated skill coding in labor economics and workforce analytics.

Generative AI systems increasingly enable the production of highly realistic synthetic media. Civitai, a popular community-driven platform for AI-generated content, operates a monetized feature called Bounties, which allows users to commission the generation of content in exchange for payment. To examine how this mechanism is used and what content it incentivizes, we conduct a longitudinal analysis of all publicly available bounty requests collected over a 14-month period following the platform's launch. We find that the bounty marketplace is dominated by tools that let users steer AI models toward content they were not trained to generate. At the same time, requests for content that is "Not Safe For Work" are widespread and have increased steadily over time, now comprising a majority of all bounties. Participation in bounty creation is uneven, with 20% of requesters accounting for roughly half of requests. Requests for "deepfake" - media depicting identifiable real individuals - exhibit a higher concentration than other types of bounties. A nontrivial subset of these requests involves explicit deepfakes despite platform policies prohibiting such content. These bounties disproportionately target female celebrities, revealing a pronounced gender asymmetry in social harm. Together, these findings show how monetized, community-driven generative AI platforms can produce gendered harms, raising questions about consent, governance, and enforcement.

Real-world License Plate Recognition (LPR) faces significant challenges from severe degradations such as motion blur, low resolution, and complex illumination. The prevailing "restoration-then-recognition" two-stage paradigm suffers from a fundamental flaw: the pixel-level optimization objectives of image restoration models are misaligned with the semantic goals of character recognition, leading to artifact interference and error accumulation. While Vision-Language Models (VLMs) have demonstrated powerful general capabilities, they lack explicit structural modeling for license plate character sequences (e.g., fixed length, specific order). To address this, we propose an end-to-end structure-aware multimodal reasoning framework based on Qwen3-VL. The core innovation lies in the Character-Aware Multimodal Reasoning Module (CMRM), which introduces a set of learnable Character Slot Queries. Through a cross-attention mechanism, these queries actively retrieve fine-grained evidence corresponding to character positions from visual features. Subsequently, we inject these character-aware representations back into the visual tokens via residual modulation, enabling the language model to perform autoregressive generation based on explicit structural priors. Furthermore, combined with the LoRA parameter-efficient fine-tuning strategy, the model achieves domain adaptation while retaining the generalization capabilities of the large model. Extensive experiments on both synthetic and real-world severely degraded datasets demonstrate that our method significantly outperforms existing restoration-recognition combinations and general VLMs, validating the superiority of incorporating structured reasoning into large models for low-quality text recognition tasks.

The GEneral Matrix Multiplication (GEMM) is one of the essential algorithms in scientific computing. Single-thread GEMM implementations are well-optimised with techniques like blocking and autotuning. However, due to the complexity of modern multi-core shared memory systems, it is challenging to determine the number of threads that minimises the multi-thread GEMM runtime. We present a proof-of-concept approach to building an Architecture and Data-Structure Aware Linear Algebra (ADSALA) software library that uses machine learning to optimise the runtime performance of BLAS routines. More specifically, our method uses a machine learning model on-the-fly to automatically select the optimal number of threads for a given GEMM task based on the collected training data. Test results on two different HPC node architectures, one based on a two-socket Intel Cascade Lake and the other on a two-socket AMD Zen 3, revealed a 25 to 40 per cent speedup compared to traditional GEMM implementations in BLAS when using GEMM of memory usage within 100 MB.

Memory plays a foundational role in augmenting the reasoning, adaptability, and contextual fidelity of modern Large Language Models and Multi-Modal LLMs. As these models transition from static predictors to interactive systems capable of continual learning and personalized inference, the incorporation of memory mechanisms has emerged as a central theme in their architectural and functional evolution. This survey presents a comprehensive and structured synthesis of memory in LLMs and MLLMs, organizing the literature into a cohesive taxonomy comprising implicit, explicit, and agentic memory paradigms. Specifically, the survey delineates three primary memory frameworks. Implicit memory refers to the knowledge embedded within the internal parameters of pre-trained transformers, encompassing their capacity for memorization, associative retrieval, and contextual reasoning. Recent work has explored methods to interpret, manipulate, and reconfigure this latent memory. Explicit memory involves external storage and retrieval components designed to augment model outputs with dynamic, queryable knowledge representations, such as textual corpora, dense vectors, and graph-based structures, thereby enabling scalable and updatable interaction with information sources. Agentic memory introduces persistent, temporally extended memory structures within autonomous agents, facilitating long-term planning, self-consistency, and collaborative behavior in multi-agent systems, with relevance to embodied and interactive AI. Extending beyond text, the survey examines the integration of memory within multi-modal settings, where coherence across vision, language, audio, and action modalities is essential. Key architectural advances, benchmark tasks, and open challenges are discussed, including issues related to memory capacity, alignment, factual consistency, and cross-system interoperability.

Civil aviation is a cornerstone of global transportation and commerce, and ensuring its safety, efficiency and customer satisfaction is paramount. Yet conventional Artificial Intelligence (AI) solutions in aviation remain siloed and narrow, focusing on isolated tasks or single modalities. They struggle to integrate heterogeneous data such as voice communications, radar tracks, sensor streams and textual reports, which limits situational awareness, adaptability, and real-time decision support. This paper introduces the vision of AviationLMM, a Large Multimodal foundation Model for civil aviation, designed to unify the heterogeneous data streams of civil aviation and enable understanding, reasoning, generation and agentic applications. We firstly identify the gaps between existing AI solutions and requirements. Secondly, we describe the model architecture that ingests multimodal inputs such as air-ground voice, surveillance, on-board telemetry, video and structured texts, and performs cross-modal alignment and fusion, and produces flexible outputs ranging from situation summaries and risk alerts to predictive diagnostics and multimodal incident reconstructions. In order to fully realize this vision, we identify key research opportunities to address, including data acquisition, alignment and fusion, pretraining, reasoning, trustworthiness, privacy, robustness to missing modalities, and synthetic scenario generation. By articulating the design and challenges of AviationLMM, we aim to boost the civil aviation foundation model progress and catalyze coordinated research efforts toward an integrated, trustworthy and privacy-preserving aviation AI ecosystem.

Imbalanced electrocardiogram (ECG) data hampers the efficacy and resilience of algorithms in the automated processing and interpretation of cardiovascular diagnostic information, which in turn impedes deep learning-based ECG classification. Notably, certain cardiac conditions that are infrequently encountered are disproportionately underrepresented in these datasets. Although algorithmic generation and oversampling of specific ECG signal types can mitigate class skew, there is a lack of consensus regarding the effectiveness of such techniques in ECG classification. Furthermore, the methodologies and scenarios of ECG acquisition introduce noise, further complicating the processing of ECG data. This paper presents a significantly enhanced ECG classifier that simultaneously addresses both class imbalance and noise-related challenges in ECG analysis, as observed in the CPSC 2018 dataset. Specifically, we propose the application of feature fusion based on the wavelet transform, with a focus on wavelet transform-based interclass fusion, to generate the training feature library and the test set feature library. Subsequently, the original training and test data are amalgamated with their respective feature databases, resulting in more balanced training and test datasets. Employing this approach, our ECG model achieves recognition accuracies of up to 99%, 98%, 97%, 98%, 96%, 92%, and 93% for Normal, AF, I-AVB, LBBB, RBBB, PAC, PVC, STD, and STE, respectively. Furthermore, the average recognition accuracy for these categories ranges between 92\% and 98\%. Notably, our proposed data fusion methodology surpasses any known algorithms in terms of ECG classification accuracy in the CPSC 2018 dataset.

Production scheduling is highly susceptible to dynamic disruptions, such as variations in processing times, machine availability, and unexpected task insertions. Conventional approaches typically rely on event-specific models and explicit analytical formulations, which limits their adaptability and generalization across previously unseen disturbances. To overcome these limitations, this paper proposes DScheLLM, a dynamic scheduling approach that leverages fine-tuned large language models within a dual-system (fast-slow) reasoning architecture to address disturbances of different scales. A unified large language model-based framework is constructed to handle dynamic events, where training datasets for both fast and slow reasoning modes are generated using exact schedules obtained from an operations research solver. The Huawei OpenPangu Embedded-7B model is subsequently fine-tuned under the hybrid reasoning paradigms using LoRA. Experimental evaluations on standard job shop scheduling benchmarks demonstrate that the fast-thinking mode can efficiently generate high-quality schedules and the slow-thinking mode can produce solver-compatible and well-formatted decision inputs. To the best of our knowledge, this work represents one of the earliest studies applying large language models to job shop scheduling in dynamic environments, highlighting their considerable potential for intelligent and adaptive scheduling optimization.

Multi-constraint planning involves identifying, evaluating, and refining candidate plans while satisfying multiple, potentially conflicting constraints. Existing large language model (LLM) approaches face fundamental limitations in this domain. Pure reasoning paradigms, which rely on long natural language chains, are prone to inconsistency, error accumulation, and prohibitive cost as constraints compound. Conversely, LLMs combined with coding- or solver-based strategies lack flexibility: they often generate problem-specific code from scratch or depend on fixed solvers, failing to capture generalizable logic across diverse problems. To address these challenges, we introduce the Scalable COde Planning Engine (SCOPE), a framework that disentangles query-specific reasoning from generic code execution. By separating reasoning from execution, SCOPE produces solver functions that are consistent, deterministic, and reusable across queries while requiring only minimal changes to input parameters. SCOPE achieves state-of-the-art performance while lowering cost and latency. For example, with GPT-4o, it reaches 93.1% success on TravelPlanner, a 61.6% gain over the best baseline (CoT) while cutting inference cost by 1.4x and time by ~4.67x. Code is available at https://github.com/DerrickGXD/SCOPE.

Concrete is the most widely used construction material worldwide; however, reliable prediction of compressive strength remains challenging due to material heterogeneity, variable mix proportions, and sensitivity to field and environmental conditions. Recent advances in artificial intelligence enable data-driven modeling frameworks capable of supporting automated decision-making in construction quality control. This study leverages an industry-scale dataset consisting of approximately 70,000 compressive strength test records to evaluate and compare multiple predictive approaches, including linear regression, decision trees, random forests, transformer-based neural networks, and embedding-based neural networks. The models incorporate key mixture design and placement variables such as water cement ratio, cementitious material content, slump, air content, temperature, and placement conditions. Results indicate that the embedding-based neural network consistently outperforms traditional machine learning and transformer-based models, achieving a mean 28-day prediction error of approximately 2.5%. This level of accuracy is comparable to routine laboratory testing variability, demonstrating the potential of embedding-based learning frameworks to enable automated, data-driven quality control and decision support in large-scale construction operations.

Large Language Models (LLMs) can enhance reasoning capabilities through test-time scaling by generating multiple traces. However, the combination of lengthy reasoning traces with multiple sampling introduces substantial computation and high end-to-end latency. Prior work on accelerating this process has relied on similarity-based or confidence-based pruning, but these signals do not reliably indicate trace quality. To address these limitations, we propose STEP: Step-level Trace Evaluation and Pruning, a novel pruning framework that evaluates reasoning steps using hidden states and dynamically prunes unpromising traces during generation. We train a lightweight step scorer to estimate trace quality, and design a GPU memory-aware pruning strategy that triggers pruning as the GPU memory is saturated by KV cache to reduce end-to-end latency. Experiments across challenging reasoning benchmarks demonstrate that STEP reduces end-to-end inference latency by 45%-70% on average compared to self-consistency while also improving reasoning accuracy. Our code is released at: https://github.com/Supercomputing-System-AI-Lab/STEP

Recent advancements in large language models (LLMs) have significantly enhanced their reasoning capabilities. However, they continue to struggle with basic character-level tasks, such as counting letters in words, a problem rooted in their tokenization process. While existing benchmarks have highlighted this weakness through basic character operations, such failures are often dismissed due to lacking practical relevance. Yet, many real-world applications, such as navigating text-based maps or interpreting structured tables, rely heavily on precise sub-token understanding. In this regard, we introduce SubTokenTest, a comprehensive benchmark that assesses sub-token understanding through practical, utility-driven tasks. Our benchmark includes ten tasks across four domains and isolates tokenization-related failures by decoupling performance from complex reasoning. We provide a comprehensive evaluation of nine advanced LLMs. Additionally, we investigate the impact of test-time scaling on sub-token reasoning and explore how character-level information is encoded within the hidden states.

In this report, we introduce DASD-4B-Thinking, a lightweight yet highly capable, fully open-source reasoning model. It achieves SOTA performance among open-source models of comparable scale across challenging benchmarks in mathematics, scientific reasoning, and code generation -- even outperforming several larger models. We begin by critically reexamining a widely adopted distillation paradigm in the community: SFT on teacher-generated responses, also known as sequence-level distillation. Although a series of recent works following this scheme have demonstrated remarkable efficiency and strong empirical performance, they are primarily grounded in the SFT perspective. Consequently, these approaches focus predominantly on designing heuristic rules for SFT data filtering, while largely overlooking the core principle of distillation itself -- enabling the student model to learn the teacher's full output distribution so as to inherit its generalization capability. Specifically, we identify three critical limitations in current practice: i) Inadequate representation of the teacher's sequence-level distribution; ii) Misalignment between the teacher's output distribution and the student's learning capacity; and iii) Exposure bias arising from teacher-forced training versus autoregressive inference. In summary, these shortcomings reflect a systemic absence of explicit teacher-student interaction throughout the distillation process, leaving the essence of distillation underexploited. To address these issues, we propose several methodological innovations that collectively form an enhanced sequence-level distillation training pipeline. Remarkably, DASD-4B-Thinking obtains competitive results using only 448K training samples -- an order of magnitude fewer than those employed by most existing open-source efforts. To support community research, we publicly release our models and the training dataset.

Group Relative Policy Optimization (GRPO) has become a standard approach for training mathematical reasoning models; however, its reliance on multiple completions per prompt makes training computationally expensive. Although recent work has reduced the number of training steps required to reach peak performance, the overall wall-clock training time often remains unchanged or even increases due to higher per-step cost. We propose MMR-GRPO, which integrates Maximal Marginal Relevance to reweigh rewards based on completion diversity. Our key insight is that semantically redundant completions contribute limited marginal learning signal; prioritizing diverse solutions yields more informative updates and accelerates convergence. Extensive evaluations across three model sizes (1.5B, 7B, 8B), three GRPO variants, and five mathematical reasoning benchmarks show that MMR-GRPO achieves comparable peak performance while requiring on average 47.9% fewer training steps and 70.2% less wall-clock time. These gains are consistent across models, methods, and benchmarks. We will release our code, trained models, and experimental protocols.

Human preference evaluations are widely used to compare generative models, yet it remains unclear how many judgments are required to reliably detect small improvements. We show that when preference signal is diffuse across prompts (i.e., all prompt types are similarly informative), proportional allocation is minimax-optimal: no allocation strategy substantially improves detectability. Empirical analysis of large-scale human preference datasets shows that most comparisons fall into this diffuse regime, exhibiting small preference margins that require far more judgments than typically collected, even in well-sampled comparisons. These limits persist across evaluation protocols and modalities, including chat, image generation, and code generation with execution feedback. In contrast, curated benchmarks that reduce prompt induced variability systematically induce larger margins and improve detectability through a $1.5\times$ reduction in prompt-level variance. Our results show that inconclusive or negative human evaluation outcomes frequently reflect underpowered evaluation rather than model equivalence, underscoring the need to account explicitly for effect size, budget, and protocol design.

We present Speculative Rollout with Tree-Structured Cache (SRT), a simple, model-free approach to accelerate on-policy reinforcement learning (RL) for language models without sacrificing distributional correctness. SRT exploits the empirical similarity of rollouts for the same prompt across training steps by storing previously generated continuations in a per-prompt tree-structured cache. During generation, the current policy uses this tree as the draft model for performing speculative decoding. To keep the cache fresh and improve draft model quality, SRT updates trees online from ongoing rollouts and proactively performs run-ahead generation during idle GPU bubbles. Integrated into standard RL pipelines (\textit{e.g.}, PPO, GRPO and DAPO) and multi-turn settings, SRT consistently reduces generation and step latency and lowers per-token inference cost, achieving up to 2.08x wall-clock time speedup during rollout.

Split Federated Learning (SFL) enables collaborative training between resource-constrained edge devices and a compute-rich server. Communication overhead is a central issue in SFL and can be mitigated with auxiliary networks. Yet, the fundamental client-side computation challenge remains, as back-propagation requires substantial memory and computation costs, severely limiting the scale of models that edge devices can support. To enable more resource-efficient client computation and reduce the client-server communication, we propose HERON-SFL, a novel hybrid optimization framework that integrates zeroth-order (ZO) optimization for local client training while retaining first-order (FO) optimization on the server. With the assistance of auxiliary networks, ZO updates enable clients to approximate local gradients using perturbed forward-only evaluations per step, eliminating memory-intensive activation caching and avoiding explicit gradient computation in the traditional training process. Leveraging the low effective rank assumption, we theoretically prove that HERON-SFL's convergence rate is independent of model dimensionality, addressing a key scalability concern common to ZO algorithms. Empirically, on ResNet training and language model (LM) fine-tuning tasks, HERON-SFL matches benchmark accuracy while reducing client peak memory by up to 64% and client-side compute cost by up to 33% per step, substantially expanding the range of models that can be trained or adapted on resource-limited devices.

Developing safe, effective, and practically useful clinical prediction models (CPMs) traditionally requires iterative collaboration between clinical experts, data scientists, and informaticists. This process refines the often small but critical details of the model building process, such as which features/patients to include and how clinical categories should be defined. However, this traditional collaboration process is extremely time- and resource-intensive, resulting in only a small fraction of CPMs reaching clinical practice. This challenge intensifies when teams attempt to incorporate unstructured clinical notes, which can contain an enormous number of concepts. To address this challenge, we introduce HACHI, an iterative human-in-the-loop framework that uses AI agents to accelerate the development of fully interpretable CPMs by enabling the exploration of concepts in clinical notes. HACHI alternates between (i) an AI agent rapidly exploring and evaluating candidate concepts in clinical notes and (ii) clinical and domain experts providing feedback to improve the CPM learning process. HACHI defines concepts as simple yes-no questions that are used in linear models, allowing the clinical AI team to transparently review, refine, and validate the CPM learned in each round. In two real-world prediction tasks (acute kidney injury and traumatic brain injury), HACHI outperforms existing approaches, surfaces new clinically relevant concepts not included in commonly-used CPMs, and improves model generalizability across clinical sites and time periods. Furthermore, HACHI reveals the critical role of the clinical AI team, such as directing the AI agent to explore concepts that it had not previously considered, adjusting the granularity of concepts it considers, changing the objective function to better align with the clinical objectives, and identifying issues of data bias and leakage.

The existence of multiple, equally accurate models for a given predictive task leads to predictive multiplicity, where a ``Rashomon set'' of models achieve similar accuracy but diverges in their individual predictions. This inconsistency undermines trust in high-stakes applications where we want consistent predictions. We propose three approaches to reduce inconsistency among predictions for the members of the Rashomon set. The first approach is \textbf{outlier correction}. An outlier has a label that none of the good models are capable of predicting correctly. Outliers can cause the Rashomon set to have high variance predictions in a local area, so fixing them can lower variance. Our second approach is local patching. In a local region around a test point, models may disagree with each other because some of them are biased. We can detect and fix such biases using a validation set, which also reduces multiplicity. Our third approach is pairwise reconciliation, where we find pairs of models that disagree on a region around the test point. We modify predictions that disagree, making them less biased. These three approaches can be used together or separately, and they each have distinct advantages. The reconciled predictions can then be distilled into a single interpretable model for real-world deployment. In experiments across multiple datasets, our methods reduce disagreement metrics while maintaining competitive accuracy.

Knowledge graphs (KGs) have commonly been constructed using predefined symbolic relation schemas, typically implemented as categorical relation labels. This design has notable shortcomings: real-world relations are often contextual, nuanced, and sometimes uncertain, and compressing it into discrete relation labels abstracts away critical semantic detail. Nevertheless, symbolic-relation KGs remain widely used because they have been operationally effective and broadly compatible with pre-LLM downstream models and algorithms, in which KG knowledge could be retrieved or encoded into quantified features and embeddings at scale. The emergence of LLMs has reshaped how knowledge is created and consumed. LLMs support scalable synthesis of domain facts directly in concise natural language, and prompting-based inference favors context-rich free-form text over quantified representations. This position paper argues that these changes call for rethinking the representation of relations themselves rather than merely using LLMs to populate conventional schemas more efficiently. We therefore advocate moving from symbolic to natural-language relation descriptions, and we propose hybrid design principles that preserve a minimal structural backbone while enabling more flexible and context-sensitive relational representations.

We introduce Mi:dm 2.0, a bilingual large language model (LLM) specifically engineered to advance Korea-centric AI. This model goes beyond Korean text processing by integrating the values, reasoning patterns, and commonsense knowledge inherent to Korean society, enabling nuanced understanding of cultural contexts, emotional subtleties, and real-world scenarios to generate reliable and culturally appropriate responses. To address limitations of existing LLMs, often caused by insufficient or low-quality Korean data and lack of cultural alignment, Mi:dm 2.0 emphasizes robust data quality through a comprehensive pipeline that includes proprietary data cleansing, high-quality synthetic data generation, strategic data mixing with curriculum learning, and a custom Korean-optimized tokenizer to improve efficiency and coverage. To realize this vision, we offer two complementary configurations: Mi:dm 2.0 Base (11.5B parameters), built with a depth-up scaling strategy for general-purpose use, and Mi:dm 2.0 Mini (2.3B parameters), optimized for resource-constrained environments and specialized tasks. Mi:dm 2.0 achieves state-of-the-art performance on Korean-specific benchmarks, with top-tier zero-shot results on KMMLU and strong internal evaluation results across language, humanities, and social science tasks. The Mi:dm 2.0 lineup is released under the MIT license to support extensive research and commercial use. By offering accessible and high-performance Korea-centric LLMs, KT aims to accelerate AI adoption across Korean industries, public services, and education, strengthen the Korean AI developer community, and lay the groundwork for the broader vision of K-intelligence. Our models are available at https://huggingface.co/K-intelligence. For technical inquiries, please contact midm-llm@kt.com.

Annotator disagreement is widespread in NLP, particularly for subjective and ambiguous tasks such as toxicity detection and stance analysis. While early approaches treated disagreement as noise to be removed, recent work increasingly models it as a meaningful signal reflecting variation in interpretation and perspective. This survey provides a unified view of disagreement-aware NLP methods. We first present a domain-agnostic taxonomy of the sources of disagreement spanning data, task, and annotator factors. We then synthesize modeling approaches using a common framework defined by prediction targets and pooling structure, highlighting a shift from consensus learning toward explicitly modeling disagreement, and toward capturing structured relationships among annotators. We review evaluation metrics for both predictive performance and annotator behavior, and noting that most fairness evaluations remain descriptive rather than normative. We conclude by identifying open challenges and future directions, including integrating multiple sources of variation, developing disagreement-aware interpretability frameworks, and grappling with the practical tradeoffs of perspectivist modeling.

This paper presents team Kl33n3x's multilingual dialogue summarization and question answering system developed for the NLPAI4Health 2025 shared task. The approach employs a three-stage pipeline: forward translation from Indic languages to English, multitask text generation using a 2.55B parameter distilled language model, and reverse translation back to source languages. By leveraging knowledge distillation techniques, this work demonstrates that compact models can achieve highly competitive performance across nine languages. The system achieved strong win rates across the competition's tasks, with particularly robust performance on Marathi (86.7% QnA), Tamil (86.7% QnA), and Hindi (80.0% QnA), demonstrating the effectiveness of translation-based approaches for low-resource language processing without task-specific fine-tuning.

Stylometry--the identification of an author through analysis of a text's style (i.e., authorship attribution)--serves many constructive purposes: it supports copyright and plagiarism investigations, aids detection of harmful content, offers exploratory cues for certain medical conditions (e.g., early signs of dementia or depression), provides historical context for literary works, and helps uncover misinformation and disinformation. In contrast, when stylometry is employed as a tool for authorship verification--confirming whether a text truly originates from a claimed author--it can also be weaponized for malicious purposes. Techniques such as de-anonymization, re-identification, tracking, profiling, and downstream effects like censorship illustrate the privacy threats that stylometric analysis can enable. Building on these concerns, this paper further explores how adversarial stylometry combined with steganography can counteract stylometric analysis. We first present enhancements to our adversarial attack, $\textit{TraceTarnish}$, providing stronger evidence of its capacity to confound stylometric systems and reduce their attribution and verification accuracy. Next, we examine how steganographic embedding can be fine-tuned to mask an author's stylistic fingerprint, quantifying the level of authorship obfuscation achievable as a function of the proportion of words altered with zero-width Unicode characters. Based on our findings, steganographic coverage of 33% or higher seemingly ensures authorship obfuscation. Finally, we reflect on the ways stylometry can be used to undermine privacy and argue for the necessity of defensive tools like $\textit{TraceTarnish}$.

Incomplete multi-view clustering (IMVC) aims to discover shared cluster structures from multi-view data with partial observations. The core challenges lie in accurately imputing missing views without introducing bias, while maintaining semantic consistency across views and compactness within clusters. To address these challenges, we propose DIMVC-HIA, a novel deep IMVC framework that integrates hierarchical imputation and alignment with four key components: (1) view-specific autoencoders for latent feature extraction, coupled with a view-shared clustering predictor to produce soft cluster assignments; (2) a hierarchical imputation module that first estimates missing cluster assignments based on cross-view contrastive similarity, and then reconstructs missing features using intra-view, intra-cluster statistics; (3) an energy-based semantic alignment module, which promotes intra-cluster compactness by minimizing energy variance around low-energy cluster anchors; and (4) a contrastive assignment alignment module, which enhances cross-view consistency and encourages confident, well-separated cluster predictions. Experiments on benchmarks demonstrate that our framework achieves superior performance under varying levels of missingness.

Tonal low-resource languages are widely spoken yet remain underserved by modern speech technology. A key challenge is learning representations that are robust to nuisance variation such as gender while remaining tone-aware for different lexical meanings. To address this, we propose SITA, a lightweight adaptation recipe that enforces Speaker-Invariance and Tone-Awareness for pretrained wav2vec-style encoders. SITA uses staged multi-objective training: (i) a cross-gender contrastive objective encourages lexical consistency across speakers, while a tone-repulsive loss prevents tone collapse by explicitly separating same-word different-tone realizations; and (ii) an auxiliary Connectionist Temporal Classification (CTC)-based ASR objective with distillation stabilizes recognition-relevant structure. We evaluate primarily on Hmong, a highly tonal and severely under-resourced language where off-the-shelf multilingual encoders fail to represent tone effectively. On a curated Hmong word corpus, SITA improves cross-gender lexical retrieval accuracy, while maintaining usable ASR accuracy relative to an ASR-adapted XLS-R teacher. We further observe similar gains when transferring the same recipe to Mandarin, suggesting SITA is a general, plug-in approach for adapting multilingual speech encoders to tonal languages.

While Large Language Models (LLMs) excel at factual retrieval, they often struggle with the "curse of two-hop reasoning" in compositional tasks. Recent research suggests that parameter-sharing transformers can bridge this gap by forming a "Generalization Circuit" during a prolonged "grokking" phase. A fundamental question arises: Is a grokked model superior to its non-grokked counterparts on downstream tasks? Furthermore, is the extensive computational cost of waiting for the grokking phase worthwhile? In this work, we conduct a mechanistic study to evaluate the Generalization Circuit's role in knowledge assimilation and transfer. We demonstrate that: (i) The inference paths established by non-grokked and grokked models for in-distribution compositional queries are identical. This suggests that the "Generalization Circuit" does not represent the sudden acquisition of a new reasoning paradigm. Instead, we argue that grokking is the process of integrating memorized atomic facts into an naturally established reasoning path. (ii) Achieving high accuracy on unseen cases after prolonged training and the formation of a certain reasoning path are not bound; they can occur independently under specific data regimes. (iii) Even a mature circuit exhibits limited transferability when integrating new knowledge, suggesting that "grokked" Transformers do not achieve a full mastery of compositional logic.

Horseshoe mixtures-of-experts (HS-MoE) models provide a Bayesian framework for sparse expert selection in mixture-of-experts architectures. We combine the horseshoe prior's adaptive global-local shrinkage with input-dependent gating, yielding data-adaptive sparsity in expert usage. Our primary methodological contribution is a particle learning algorithm for sequential inference, in which the filter is propagated forward in time while tracking only sufficient statistics. We also discuss how HS-MoE relates to modern mixture-of-experts layers in large language models, which are deployed under extreme sparsity constraints (e.g., activating a small number of experts per token out of a large pool).

This work addresses the fundamental problem of unbounded metric movement costs in bandit online convex optimization, by considering high-dimensional dynamic quadratic hitting costs and $\ell_2$-norm switching costs in a noisy bandit feedback model. For a general class of stochastic environments, we provide the first algorithm SCaLE that provably achieves a distribution-agnostic sub-linear dynamic regret, without the knowledge of hitting cost structure. En-route, we present a novel spectral regret analysis that separately quantifies eigenvalue-error driven regret and eigenbasis-perturbation driven regret. Extensive numerical experiments, against online-learning baselines, corroborate our claims, and highlight statistical consistency of our algorithm.

Large language models generate judgments that resemble those of humans. Yet the extent to which these models align with human judgments in interpreting figurative and socially grounded language remains uncertain. To investigate this, human participants and four instruction-tuned LLMs of different sizes (GPT-4, Gemma-2-9B, Llama-3.2, and Mistral-7B) rated 240 dialogue-based sentences representing six linguistic traits: conventionality, sarcasm, funny, emotional, idiomacy, and slang. Each of the 240 sentences was paired with 40 interpretive questions, and both humans and LLMs rated these sentences on a 10-point Likert scale. Results indicated that humans and LLMs aligned at the surface level with humans, but diverged significantly at the representational level, especially in interpreting figurative sentences involving idioms and Gen Z slang. GPT-4 most closely approximates human representational patterns, while all models struggle with context-dependent and socio-pragmatic expressions like sarcasm, slang, and idiomacy.

Probabilistic computers built from p-bits offer a promising path for combinatorial optimization, but the dense connectivity required by real-world problems scales poorly in hardware. Here, we address this through graph sparsification with auxiliary copy variables and demonstrate a fully on-chip parallel tempering solver on an FPGA. Targeting MIMO detection, a dense, NP-hard problem central to wireless communications, we fit 15 temperature replicas of a 128-node sparsified system (1,920 p-bits) entirely on-chip and achieve bit error rates significantly below conventional linear detectors. We report complete end-to-end solution times of 4.7 ms per instance, with all loading, sampling, readout, and verification overheads included. ASIC projections in 7 nm technology indicate about 90 MHz operation with less than 200 mW power dissipation, suggesting that massive parallelism across multiple chips could approach the throughput demands of next-generation wireless systems. However, sparsification introduces sensitivity to the copy-constraint strength. Employing Two-Dimensional Parallel Tempering (2D-PT), which exchanges replicas across both temperature and constraint dimensions, we demonstrate over 10X faster convergence without manual parameter tuning. These results establish an on-chip p-bit architecture and a scalable algorithmic framework for dense combinatorial optimization.

Scientific reasoning increasingly requires linking structured experimental data with the unstructured literature that explains it, yet most large language model (LLM) assistants cannot reason jointly across these modalities. We introduce SpectraQuery, a hybrid natural-language query framework that integrates a relational Raman spectroscopy database with a vector-indexed scientific literature corpus using a Structured and Unstructured Query Language (SUQL)-inspired design. By combining semantic parsing with retrieval-augmented generation, SpectraQuery translates open-ended questions into coordinated SQL and literature retrieval operations, producing cited answers that unify numerical evidence with mechanistic explanation. Across SQL correctness, answer groundedness, retrieval effectiveness, and expert evaluation, SpectraQuery demonstrates strong performance: approximately 80 percent of generated SQL queries are fully correct, synthesized answers reach 93-97 percent groundedness with 10-15 retrieved passages, and battery scientists rate responses highly across accuracy, relevance, grounding, and clarity (4.1-4.6/5). These results show that hybrid retrieval architectures can meaningfully support scientific workflows by bridging data and discourse for high-volume experimental datasets.

The parallel evolution of Large Language Models (LLMs) with advanced code-understanding capabilities and the increasing sophistication of malware presents a new frontier for cybersecurity research. This paper evaluates the efficacy of state-of-the-art LLMs in classifying executable code as either benign or malicious. We introduce an automated pipeline that first decompiles Windows executable into a C code using Ghidra disassembler and then leverages LLMs to perform the classification. Our evaluation reveals that while standard LLMs show promise, they are not yet robust enough to replace traditional anti-virus software. We demonstrate that a fine-tuned model, trained on curated malware and benign datasets, significantly outperforms its vanilla counterpart. However, the performance of even this specialized model degrades notably when encountering newer malware. This finding demonstrates the critical need for continuous fine-tuning with emerging threats to maintain model effectiveness against the changing coding patterns and behaviors of malicious software.

Jahn--Teller (JT) distortions are a key driver of physical properties in many correlated oxide materials. Cooperative JT distortions, in which long-range orbital order reduces the symmetry of the average structure macroscopically, are common in JT-distorted materials at low temperatures. This long-range order will often melt on heating, \textit{via} a transition to a high-temperature state without long-range orbital order. The nature of this transition has been observed to vary with different materials depending on crystal structure; in LaMnO$_3$ the transition has generally been interpreted as order-disorder, whereas in layered nickelates $A$NiO$_2$ ($A$=Li,Na) there is a displacive transition. Alternatively, recent theoretical work has suggested that previous attributions of order-disorder may in fact be a consequence of phonon anharmonicity, rather than persistence of JT distortions, which would suggest that the displacive transition may be more common than currently believed. In this work, we run Monte Carlo simulations with a simple Hamiltonian which is modified to include terms dependent on the JT amplitude $ρ$, which is allowed to vary within the simulation \textit{via} the Metropolis algorithm. Our simulations yield distributions of JT amplitudes consistent with displacive rather than order-disorder behaviour for both perovskites and layered nickelates, suggesting that displacive-like JT transitions may be more common than previously assumed in both perovskites and layered nickelates. We also find significant differences between the transition observed for perovskites compared with layered nickelates, which we attribute to differing extensivity of configurational entropy on the two lattices, showing the crucial role of lattice geometry in determining behaviour.

A recent STM experiment in 2D bilayer graphene [Y.-C. Tsui, et al., Nature 628, 287 (2024)], under a strong perpendicular magnetic field, has made a direct observation of the existence of three distinct filling-factor-dependent quantum phases in the lowest Landau level: the incompressible fractional quantum Hall liquid, a crystalline compressible hexagonal Wigner crystal with long-range order and rotational symmetry-breaking, and a random localized solid phase with no spatial order. We argue that the spatially random localized phase at low filling is the recently proposed disorder-dominated strongly localized amorphous "Anderson solid" phase [A. Babber, et al., arXiv:2601.03521], which appears generically at a sample-dependent filling factor.

We investigate heat transport in one-dimensional harmonic chains with mass disorder and weak bond disorder, coupled at both ends to oscillator heat baths through weak impedance mismatches. The model incorporates correlations in mass disorder, in bond disorder, and between the two. We find that the scaling of thermal conductivity $κ$ with system size $N$ is determined solely by either mass disorder or bond disorder. This indicates that cross-correlations between the two types of disorder play no important role in the scaling behavior of $κ$. Consequently, by tuning the self-correlations, it is possible to control how the thermal conductivity scales with the system size. Such control could have potential applications in thermoelectric devices and thermal insulation technologies.

We propose a mechanism for the inverse Faraday and the inverse Cotton--Mouton effects arising from quantum geometry, characterized by the quantum metric quadrupole and the weighted quantum metric. Within a semiclassical framework based on the Boltzmann transport theory, we establish a general formalism describing light-induced magnetization in electronic systems as a second-order response to the electric field of light. Using continuum and tight-binding models, we discuss the symmetry constraints on these effects and estimate the magnitudes of the resulting magnetizations. Our results highlight a direct manifestation of quantum-geometric quantities in nonlinear magneto-optical responses and suggest a viable pathway for experimental detection.

Ferroelectric materials exhibit a switchable, spontaneous polarization at the unit cell level--an attractive property utilized in many emerging technologies including, among others, high-density memory storage, low-power transistors, and high-speed fiber optic communication. Understanding the local polarization switching behavior, through domain nucleation and evolution, is critical to advancing these technologies and requires characterization of the local domain microstructure. However, in application-relevant polycrystalline materials exhibiting a distribution of grain orientations, a direct mapping of the polarization direction in three dimensions has remained inaccessible using conventional experimental approaches. Here, taking barium titanate single crystals and lead zirconium titanate polycrystals as our bulk model systems, we map the local polarization directions using a new electron backscatter diffraction indexing technique based on simulated pattern-matching. Through improved pre-processing techniques (including optimized pattern processing, a new pseudosymmetry-sensitive neighbor pattern averaging method, and DIC-based global sample-detector geometry calibration) and a new pseudosymmetry confidence index (which considers not only pattern similarity but pattern dissimilarity trends with other domain variant patterns), we successfully distinguish between the six polarization directions, despite the challengingly small unit cell aspect ratio of the selected materials. The methods developed in this work are not only applicable to ferroelectrics but any material which exhibits close crystallographic pseudosymmetries--extending the current capabilities of EBSD.

We introduce the genuine multipartite Rains entanglement (GMRE) as a measure of genuine multipartite entanglement that can be computed using semi-definite programming. Similar to the Rains relative entropy (its bipartite counterpart), the GMRE is monotone under selective quantum operations that completely preserve the positivity of the partial transpose, implying that it is a multipartite entanglement monotone. As a consequence, we show that the GMRE bounds both the one-shot standard and probabilistic approximate GHZ-distillable entanglement from above. We also develop a generalization of this quantity that incorporates other entropies, including quantum Renyi relative entropies.

Stiff forces, which bind objects together or otherwise confine motion, are found widely in soft-matter systems - colloids with short range attractions, ligand-receptor contacts, particles in optical traps, fibres that resist stretching, etc. To assess the long-term effect of these stiff forces on dynamics and structure, it is useful to consider the limit where they are treated as constraints, so the system evolves strictly within allowed configurations. Efforts to derive equations involving both constraints, and the stochastic motion appropriate at the scales of soft matter, began around 50 years ago, yet, we are still lacking a straightforward way to extract the projected equations and apply them in modern formulations of mesoscale dynamics. Here, we address this gap with two key contributions: (1) a practical summary of the constrained Brownian dynamics equations with ``soft'' constraints, i.e. constraints imposed by stiff forces, which is illustrated through several representative examples, taking care to highlight the nontrivial effects of the constraints; and (2) a novel derivation using singular perturbation theory, establishing the validity of these equations over timescales exceeding the relaxation of stiffly constrained degrees of freedom. We further extend our approach to ``soft soft'' constraints, where mobility varies on lengthscales comparable to the restraining forces - a scenario typical for particles in fluids experiencing hydrodynamic interactions. We hope our results will be useful for soft matter research, as a robust toolkit for studying tethered or confined systems.

Dynamical perturbations can modify the energy landscape of a physical system, such that unstable equilibrium configurations become stable when subject to an external drive. The magnetic analog of such dynamical stabilization corresponds to saturation of the magnetization against an external field. Here we report dynamical stabilization of the magnetization in thin film bismuth-substituted yttrium iron garnet by spin current injection from an adjacent Pt layer. Magneto-optical Kerr effect measurements demonstrate magnetization reversal against magnetic fields up to 3000 times larger than the film's coercivity once the spin injection surpasses a critical threshold associated with negative damping. Micromagnetic simulations reveal that this process is mediated by the excitation of a large population of incoherent magnons with non-zero wave vector, leading to a transient shortening and subsequent stabilization of the inverted magnetization. The elementary excitations of the high-energy inverted magnetization state are shown to be antimagnons, quasi-particles carrying opposite energy and spin relative to magnons. Our results further reveal how the system's size and minimization of nonlinear magnon scattering processes play a key role in dynamical stabilization, opening new avenues for magnetic state control beyond conventional magnetization switching. Dissipation-driven phase transitions in large-area magnetic systems provide a solid-state platform to study magnonic analogs of relativistic phenomena, such as Klein tunneling and black holes, as well as spin-wave amplification and lasing.

Many-body localization (MBL) appears to be a robust example of ergodicity breaking in many-body interacting systems. Here, we review different aspects of MBL, concentrating on various ways the disorder may be introduced into the system studied. In particular, we consider both the random and quasiperiodic diagonal (i.e., on-site) disorder as well as bond disorder as realized in randomly distributed atoms interacting via long-range interactions. We also review the quantum sun model, which seems to be the ideal, albeit artificial, model exhibiting MBL.

Error-mitigation methods for Ising machines are reexamined not merely as noise-suppression techniques but as a structural design problem of replica-coupled Ising models. Using simulated annealing as a hardware-noise-free testbed, we systematically compare the penalty-spin (PS) model, which couples replicas through a centralized auxiliary layer, with the stacked model, which couples adjacent replicas directly. Numerical experiments on the quadratic assignment problem reveal that the ferromagnetically coupled stacked model stably maintains constraint satisfaction and improves solution quality over a broad parameter range, exhibiting favorable scalability with both the number of replicas and problem size. In contrast, the PS model suffers from cooperation collapse at large parallelism: many-replica averaging in the PS layer washes out sparse solution information, preventing effective inter-replica coordination. These findings demonstrate that the topology of inter-replica couplings decisively influences search robustness, and provide practical guidelines for model selection and parameter tuning in constrained optimization.

1D quasicrystals such as the Fibonacci chain have been said to ``inherit" their topological properties from the 2D Quantum Hall problem. Yet, a direct way to see the connection was lacking until a common ancestor, the Fibonacci-Hall model, was introduced recently \cite{aj2025}. This 2D ancestor model relates the role of the external magnetic flux in the Hall problem and that of a geometric flux which describes the winding of the quasicrystal in 2D, in the cut-and-project method. Doing this enables us to extend the notion of Chern numbers as defined in 2D, to the energy bands of the 1D chain by adiabatic continuity. The older notion of gap labels in the 1D system are now seen to be derivable from the Chern numbers of the 2D bands. The Fibonacci-Hall model thus provides an important link between physics of two paradigmatic models, the Fibonacci quasicrystal and the quantum Hall insulator. The generalization to other 1D quasiperiodic models is expected to be relatively straightforward. The extension to 2D cut-and-project tilings is left for future studies.

We study composite topological excitations in ferromagnetic superconductors consisting of bound states of magnetic spin textures (skyrmions) and superconducting vortices. Using a Ginzburg--Landau framework with Zeeman coupling between the magnetization and superconducting magnetic field, we demonstrate that skyrmion-vortex pairs (SVPs) form energetically stable bound states. By analyzing their asymptotic interactions, we identify regimes in which SVPs exhibit both short-range repulsion and long-range attraction, leading to clustering phenomena. Our results provide a field-theoretical basis for understanding suggest pathways for controlling hybrid topological matter through long-range interactions.

The role of interface states and dielectric mismatch is studied in ultrathin P-doped silicon-on-insulator (SOI) films with thickness of the device layer ($H_{SOI}$) varying from 30 to 8 nm and dopant concentration ($n_{D}$) ranging from 10$^{18}$ to nearly 10$^{20}$ cm$^{-3}$. P concentration is determined by Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS). Sample resistivity ($ρ$), carrier concentration ($n_e$), and mobility ($μ_e$) are extracted by combining sheet resistance and Hall measurements in van der Pauw configuration. When $H_{SOI}$ = 30 nm, transport properties at room temperature are fully compatible with those of a similarly doped bulk Si. Progressive 2D confinement by reduction of $H_{SOI}$ below 30 nm results in a reduction of the carrier concentration and a concomitant degradation of $μ_e$. These effects, which are steadily enhanced decreasing $n_D$, are attributed to non-passivated interface states at the SiO$_2$/Si interface and can be significantly mitigated by high temperature rapid thermal oxidation (RTO). The effectiveness of this approach was verified by electron-paramagnetic resonance (EPR) spectra and capacitance-voltage (CV) measurements, which allowed the assessment of the quality of the RTO-SiO$_2$/Si interface and the correlation with observed electrical properties. After effective interface engineering, low temperature electrical characterization revealed a significant increase in P ionization energy in samples with $H_{SOI}$ <= 15 nm, a result directly related to the dielectric mismatch.

The autonomous motion of liquid crystal oil droplets in micellar media arises from spontaneous breaking of time reversal symmetry via nonlinear coupling between Marangoni stresses and surfactant transport. While this phenomenon has been widely studied, the influence of micellar solute structure remains unexplored. By modifying micellar architecture using a structure forming salt, we uncover a pronounced non monotonic dependence of droplet velocity on salt concentration. Increasing salt simultaneously raises the medium viscosity and drives a transition of micelles from spherical to rod-like or worm like morphologies. Using complementary experiments, we quantify the viscosity and micellar interaction lengthscale as functions of the salt to surfactant ratio and develop a theoretical model that consistently reproduces the measured propulsion speeds. Flow fields around the droplets are characterized by particle image velocimetry. Our results demonstrate that salt surfactant composition governs active droplet propulsion by jointly controlling micellar solute interaction lengthscales and medium viscosity.

In this paper, we establish a connection between the formalism of $\mathcal{R}$-transforms for non-Hermitian random matrices and the framework of spherical integrals, using the replica method. This connection was previously proved in the Hermitian setting and in the case of bi-invariant random matrices. We show that the $\mathcal{R}$-transforms used in the non-Hermitian context in fact originate from a single scalar function of two variables. This provides a new and transparent way to compute $\mathcal{R}$-transforms, which until now had been known only in restricted cases such as bi-invariant, Hermitian, or elliptic ensembles.

The Eigenstate Thermalization Hypothesis explains thermalization in isolated quantum systems through the statistical properties of observables in the energy eigenbasis. We investigate the crossover from integrability to chaos in the spin-$1/2$ XXZ chain, establishing a direct correspondence between the spectral correlations of the Hamiltonian and local observables expressed in the energy eigenbasis as a signature of ergodicity breaking. By introducing a local perturbation that drives the system from integrability to chaos, we track the standard ETH indicators and the eigenstate entanglement entropy. We introduce a submatrix-based framework for analyzing local observables in the energy eigenbasis. By extracting real-symmetric blocks along the diagonal of the local observables represented in eigenbasis, we show that these submatrices exhibit both the short-range and long-range spectral features of the Hamiltonian. Remarkably, this correspondence persists even in a partially ergodic regime, indicating that the emergence of chaos is already encoded locally within the observables' matrix structure and that small blocks are sufficient to capture the underlying spectral correlations.

We present a systematic study of the Ruderman-Kittel-Kasuya-Yosida (RKKY) interaction in MnBi$_2$Te$_4$ films under both dark and illuminated conditions. In the dark, the intrinsic magnetism of MnBi$_2$Te$_4$ is shown to yield a stronger anisotropic RKKY spin model compared to nonmagnetic topological insulators, providing a clear signature for differentiating these systems. Furthermore, key band properties -- such as energy gap, band degeneracy/splitting, and topological deformations of the Fermi surface -- imprint distinct signatures on the RKKY interaction, enabling clear discrimination between even- and odd-septuple-layer (SL) films. This discrimination manifests in multiple ways: through the Fermi-energy dependence or spatial oscillations of the interaction for impurities on the same surface, or via the presence versus absence of spin-frustrated terms for those on different surfaces. Under off-resonant circularly polarized light, we track photoinduced topological phase transitions and identify two characteristic signatures at the phase boundary: a sign reversal in spin-frustrated terms and a dip in collinear RKKY components. These serve as fingerprints for circular-polarization-chirality-dependent topological transitions in even- and odd-SL films, respectively. Overall, this work establishes RKKY interactions as a sensitive magnetic probe for revealing both distinctive band properties and light-driven phase transitions in MnBi$_2$Te$_4$ films, thereby complementing conventional electrical measurements while providing new insights into the influence of intrinsic magnetism on the surface-state band structure.

The time to first crossing for the Poisson counting process with respect to a linear moving barrier with offset is a classic problem, although key results remain scattered across the literature and their equivalence is often unclear. Here we present a unified and pedagogical treatment of two approaches: the direct time-domain approach based on path-decomposition techniques and the Laplace-domain approach based on the Pollaczek-Spitzer formula. Beyond streamlining existing derivations and establishing their consistency, we leverage the complementary nature of the two methods to obtain new exact analytical results. Specifically, we derive an explicit large deviation function for the first-passage time distribution in the subcritical regime and closed-form expressions for the conditional mean first-passage time for arbitrary offset. Despite its simplicity, this first crossing process exhibits non-trivial critical behavior and provides a rare example where all the main results of interest can be derived exactly.

Heterostructures combining graphene with 3d transition metal ferromagnets (FMs) enable various spin-based phenomena at ultrafast timescales. However, challenges such as the interfacial impedance mismatch, FM deposition-induced defect generation, and interface modification by interfacial coupling or hybridization can impede their functionalization for spin-orbitronics. In this work, we utilize insulating TiOx barrier layers (BLs) to modify the interfacial spin conductance structurally, disentangle spin pumping and magnetic proximity effects (MPE), and establish external control over ultrafast magnetization dynamics in single-layer graphene/TiOx/Co systems. All-optical time-resolved magneto-optical Kerr effect measurements of femtosecond to nanosecond spin dynamics reveal systematic tunability of ultrafast magnetic parameters via barrier engineering. The thickness-dependent damping modulation in Co indicates strong spin pumping, with interfacial spin transparency close to half its physical limit in the presence of an ultrathin BL, where MPE is eliminated. Our results show that appropriately chosen ultrathin BLs can prevent interfacial alterations from ferromagnetic metals, facilitating efficient spin detection in graphene and enhancing control over spin angular momentum dissipation in graphene/FM interfaces.

Choosing a representative element volume (REV) from finite cylindrical $μ$CT scans becomes ambiguous when a key field variable exhibits a slow axial trend, because estimated statistics can change systematically with subvolume size and position rather than converging under simple averaging. A practical workflow is presented to size an REV under such nonstationary conditions by first suppressing axial drift/trend to obtain a residual field suitable for second-order analysis, and then selecting the smallest analysis diameter for which low-wavenumber content stabilizes within a prescribed tolerance. The approach is demonstrated on \textit{Thalassinoides}-bearing rocks, whose branching, connected burrow networks impose heterogeneity on length scales comparable to typical laboratory core diameters, making imaging-based microstructural statistics and downstream digital-rock proxies highly sensitive to the chosen subvolume. From segmented data, a scalar ``burrowsity'' field--capturing burrow-related pore spaces and infills--is defined, and axial detrending (with optional normalization) is applied to mitigate acquisition drift and nonstationary trends. Representativeness is then posed as a diameter-convergence problem on nested inscribed cylinders: the two-point covariance and its isotropic spectral counterpart $\widehat{C}$ are estimated, and the smallest diameter at which the low-wavenumber plateau becomes stable is selected. Applied to a segmented \textit{Thalassinoides} core, the method identifies a minimum analysis cylinder of approximately $D_{\mathrm{REV}}\approx 93~\mathrm{mm}$ and $H_{\mathrm{REV}}\approx 83~\mathrm{mm}$, enabling reproducible correlation-scale reporting and connectivity-sensitive property estimation.

One-dimensional semiconductors are characterized by strongly bound excitons. Therefore, the Frenkel regime of excitons localized within a few unit cells is readily reached and traditional Wannier exciton models become inadequate. In the presence of strong electric fields, excitons are polarized and, in extreme cases, ionized. Such strong-field effects have previously been described analytically for Wannier excitons. In the present work, we show that analytical results can be extended to the more involved Frenkel case as well. Hence, by analytically solving the difference equation describing Frenkel excitons in electric fields, we derive close-form expressions for resonances providing Stark shifts and ionization rates. Moreover, closed-form results for exciton electroabsorption spectra and dynamic polarizability are obtained.

We present explicit matrix product operator (MPO) representations for the local conserved quantities of the spin-$1/2$ XYZ chain. Through these MPO representations, we simplify the coefficients appearing in the local conserved quantities originally derived by one of the authors, and reveal their combinatorial meaning: the coefficients prove to be a polynomial generalization of the Catalan numbers, defined via weighted monotonic lattice paths. Furthermore, we obtain a new simple $3 \times 3$ Lax operator for the XYZ chain that, unlike Baxter's R-matrix, does not involve elliptic functions.

In pursuit of a colloidal analogue to quantum density functional theory (DFT) predictions of atomic crystal structures, we report a new, classical DFT that predicts the relative thermodynamic stability of colloidal crystals of hard, convex particle shapes. In contrast to standard classical DFT approaches, our theory maps the hard particle system to an auxiliary system in which we treat the particles as fixed "nuclei" embedded in a fictitious, spatially varying density field that distributes throughout the auxiliary system. By minimizing the free energy of the auxiliary system, and through comparison with known equations of state and free energy calculations using thermodynamic integration, we show that the auxiliary system with the lowest free energy corresponds to the most probable crystal of hard shapes in the original system.

We propose a new method to describe the population dynamics of distinct configurational states based on a continuous-time description of state-to-state transitions. According to classical reaction dynamics theory, the probability density associated with a given state obeys the Liouville equation, the probability density associate d with a given state obeys the Li ouville equation, including influx from and efflux to neighboring states. By introducing a Markov approximation for the crossing of boundaries separating the states, tractable integral equations governing the state populations are derived. Once the time-dependent quantities appearing in these equations are evaluated, the population dynamics on long timescales can be obtained. Because these quantities depend only on a few states in the local neighborhood of a given state, they can be computed using a set of short-timescale molecular dynamics (MD) simulations. We apply the present method to the binding and unbinding kinetics of CH$_4$/CH$_4$, Na$^+$/Cl$^-$, and 18-crown-6-ether (crown ether)/K$^+$ in water. For both kinetics, the time constants estimated from the present method are almost comparable to those obtained from brute-force MD simulations. The required timescale of each MD trajectory in the present method is approximately two orders of magnitude shorter than that in the brute-force MD approach in the crown ether/K$^+$ system. This reduction in the trajectory timescale enables applications to complex binding and unbinding sy stems whose characteristic timescales a re far beyond those directly acce ssible by brute-force MD simulati ons.

We investigate vapor-liquid phase separation of an active near critical Lennard-Jones fluid confined within a cylindrical pore using molecular dynamics simulations. Activity is introduced via Vicsek-type alignment interactions, enabling a systematic study of how self-propulsion modifies domain morphology and coarsening kinetics under quasi-one-dimensional confinement. In the passive limit, the system undergoes early-time spinodal decomposition (diffusive growth characterized by the Lifshitz-Slyozov exponent $α= 1/3$), followed by the formation of periodically modulated, plug-like liquid domains along the pore axis. At late times, coarsening becomes kinetically arrested, and the system remains trapped in a metastable striped state. Introducing activity destabilizes this arrested morphology by enhancing collective domain transport, leading to frequent domain mergers and complete phase separation at sufficiently high activity. The late-stage coarsening then exhibits a crossover to faster, ballistic growth with an effective exponent $α= 2/3$, consistent with a cluster-coalescence mechanism. Analysis of two-point correlation functions and structure factors confirms dynamic scaling across all activity regimes. Our results demonstrate that alignment-induced activity can overcome confinement-driven kinetic arrest, providing new insight into phase separation in confined active fluids. The relevant growth laws are analyzed and interpreted using appropriate theoretical frameworks.

We experimentally demonstrate the generation of multiple three-magnon splitting processes in an in-plane magnetized submicron Bi-YIG disk using micro-focused Brillouin light scattering. The low magnetic damping and strong magneto-optical response of BiYIG enable the detection of nonlinear spin-wave interactions at low threshold powers. By tuning the in-plane static magnetic field, excitation frequency, and power, we observe the generation of three pairs of secondary modes symmetrically distributed around half the excitation frequency. Time-resolved BLS measurements present temporal dynamics and threshold behavior associated with the successive activation of three-magnon pairs.

Exciton-polaritons are light-matter hybrid quasi-particles that have emerged as a flexible platform for developing quantum technologies and engineering material properties. However, the fundamental mechanistic principles that govern their dynamics and relaxation remain elusive. In this work, we provide the microscopic mechanistic understanding of the exciton-polariton relaxation process that follows from an excitation in the upper polariton. Using both mixed quantum-classical simulations and analytical analysis, we reveal that phonon-induced upper-to-lower polariton relaxation proceeds via two steps: the first step is a vertical inter-band transition from the upper to the lower polariton, which is followed by a second step that is a phonon-induced Fröhlich scattering within the lower polariton. We find that in materials of finite thickness (which include filled cavities), phonon-induced polaritonic intraband Fröhlich scattering is significantly suppressed. We show that the microscopic origin of this suppression is phonon-fluctuations synchronization (or self-averaging) due to the polaritonic spatial delocalization in the quantization direction. Finally, we show that the same phonon fluctuation-synchronization effect plays a central role across polaritonic relaxation pathways, and we derive simple analytical expressions that relate a material's finite thickness to the corresponding relaxation rate constants.

The origin of the excess vibrational density of states (DOS) beyond Debye's theory in amorphous solids (often referred to as the Boson peak) has been attributed to the presence of quasi-localized vibrational modes in recent years. However, by dispersing the total DOS onto each degree of freedom (DOF), the results of this report provide evidence that \(99.9\%\) of DOFs, and hence almost all particles, contribute to the Boson peak (BP). These results challenge the prevailing opinion that BP is contributed by a minority of particles and highlight its long-neglected global and collective origin.

We present here a study of the bright soliton dynamics in an inhomogeneous fibre by means of variable coefficient Fokas-Lenells equation with time varying dispersion, nonlinearity and gain/loss parameter. At first, we propose our system that governs the propagation of ultrashort pulses in an inhomogeneous fibre. Secondly, under a suitable gauge transformation, we transform the system into a simplified form of variable coefficient Fokas-Lenells equation. The Lax integrability and conservation laws are exhibited. We also study the stability of the generalised plane wave against small amplitude perturbations. Thereafter, by using a nonstandard Hirota bilinearization method with the help of a suitable auxiliary function, we obtain the bright one soliton, two soliton and provide a scheme for obtaining N-bright soliton solutions. The elastic collision dynamics of the two solitons is studied using asymptotic analysis. We also investigate the soliton acceleration/retardation under a suitable choice of dispersion and nonlinearity coefficients. Finally, the dramatic effect of the nonlinear tunnelling of the bright one and two-soliton is also studied under some Gaussian dispersion or nonlinearity.

In the present work we describe a way to assess memory capability of real devices, while proposing to the engineering community what to pursue to create devices with deep associated memory capability. The study of the signal produced by a real memristor nano-device focused on the description in terms of the Landau φ4 theory for the critical phenomena in finite systems. This further allowed the utilization of the property of the anomalous enhancement of the autocorrelation function when a system is on the Spontaneous Symmetry Breaking (SSB), for improving the quantity of the demonstrated memory, while simultaneously maintaining a very good quality, as this is expressed by the stability of the autocorrelation function. In this proof-of-concept case, the morphology of the signal allowed us to impose the appropriate modifications on the signal so that we finally show how to get very close to the characteristics of the SSB and thus achieve our goal to get as close as possible to the ideal behavior of a Memristor that yields deep memory. Finally, we provide proof of the stability of memristor's operation by showing that solitons "follow" as a skeleton structure the experimentally derived time series.

We experimentally investigate the effects of finite-system size on the dynamics of weakly nonlinear random gravity-capillary surface waves. Experiments are conducted in rectangular tanks with varying aspect ratios, in which the fluid surface is perturbed locally and erratically by small, partially submerged magnets. Driven by an oscillating vertical electromagnetic field, these magnets generate a statistically homogeneous and isotropic random wave field. This setup enables us to probe finite-size effects without the dominant influence of global forcing present in horizontally oscillated tanks. Spatiotemporal measurements of the wave field reveal multiple branches in the wave-energy spectrum along the unconfined direction, corresponding to sloshing modes in the confined direction. We show that the spectral properties of these modes can be tuned by varying either the wave steepness or the confinement. Signatures of discrete wave turbulence in the confined direction and mesoscopic continuous wave turbulence in the unconfined direction are observed. As the confinement is gradually relaxed, we further demonstrate a smooth transition from discrete to continuous wave turbulence, consistent with the nonlinear-to-discreteness timescale ratio. Using high-order correlation analysis, we also show that finite-size effects alter wave dynamics by depleting two-dimensional three-wave resonant interactions along the confined direction.

In this paper the Mikhailov model is discretized by means of the Cauchy matrix approach. A pair of discrete Miura transformations are constructed. The discrete Mikhailov model is a coupled system, in which one equation comes from the compatibility of the two Miura transformations, the other is transformed from the discrete negative order Ablowitz-Kaup-Newell-Segur system by using the Miura transformations. Explicit solutions, including solitons and multiple-pole solutions, are presented via two Cauchy matrix schemes respectively, namely, the Ablowitz-Kaup-Newell-Segur type and the Kadomtsev-Petviashvili type. By straight continuum limits, semi-discrete and continuous Mikhailov models together with their Cauchy matrix structures and solutions are recovered.

We study the dynamics of the geodesics of pp-wave spacetimes with polynomial profiles of degrees $3\leq n\leq10$, which are dynamically equivalent to the motion of a classical particle in a two-dimensional harmonic polynomial potential. By analysing the escape basins associated with different asymptotic outcomes, we show that all basins exhibit the Wada property for every polynomial degree considered. We further compute the basin entropy $S_{b}$, finding that it increases monotonically with the polynomial degree, indicating enhanced unpredictability of the final state of the system. The boundary basin entropy $S_{bb}$ is also evaluated, and the $\ln(2)$ criterion confirms that the basin boundaries are fractal for $n>3$. We conjecture that the Wada property persists for polynomial degrees $n>10$.

Existing match classification models in the tournament design literature have two major limitations: a contestant is considered indifferent only if uncertain future results do never affect its prize, and competitive matches are not distinguished with respect to the incentives of the contestants. We propose a probabilistic framework to address both issues. For each match, our approach relies on simulating all other matches played simultaneously or later to compute the qualifying probabilities under the three main outcomes (win, draw, loss), which allows the classification of each match into six different categories. The suggested model is applied to the previous group stage and the new incomplete round-robin league, introduced in the 2024/25 season of UEFA club competitions. An incomplete round-robin tournament is found to contain fewer stakeless matches where both contestants are indifferent, and substantially more matches where both contestants should play offensively. However, the robustly higher proportion of potentially collusive matches can threaten with serious scandals.

In this work, we characterized the material properties of an animal model of the rotator cuff tendon using full volume datasets of both its intact and injured states by capturing internal strain behavior throughout the tendon. Our experimental setup, involving tension along the fiber direction, activated volumetric, tensile, and shear mechanisms due to the tendon's complex geometry. We implemented an approach to model inference that we refer to as variational system identification (VSI) to solve the weak form of the stress equilibrium equation using these full volume displacements. Three constitutive models were used for parameter inference: a neo-Hookean model, a modified Holzapfel-Gasser-Ogden (HGO) model with higher-order terms in the first and second invariants, and a reduced polynomial model consisting of terms based on the first, second, and fiber-related invariants. Inferred parameters were further refined using an adjoint-based partial differential equation (PDE)-constrained optimization framework. Our results show that the modified HGO model captures the tendon's deformation mechanisms with reasonable accuracy, while the neo-Hookean model fails to reproduce key internal features, particularly the shear behavior in the injured tendon. Surprisingly, the simplified polynomial model performed comparably to the modified HGO formulation using only three terms. These findings suggest that while current constitutive models do not fully replicate the complex internal mechanics of the tendon, they are capable of capturing key trends in both intact and damaged tissue, using a homogeneous modeling approach. Continued model development is needed to bridge this gap and enable clinical-grade, predictive simulations of tendon injury and repair.

Efficient Monte Carlo (MC) sampling of many-body systems with long-range electrostatics is often limited by the cost of per-move energy-difference evaluation under periodic boundary conditions. We present DMK-MC, an accelerated MC method that adapts the dual-space multilevel kernel-splitting (DMK) framework to single-particle Metropolis updates. DMK-MC computes the energy change and, upon acceptance, updates the stored incoming plane-wave fields with $O(1)$ work per tree level, yielding an overall $O(\log N)$ expected work per trial move for fixed accuracy. The method decomposes the Coulomb kernel into three components: a global, periodized smooth part; a multilevel sequence of smooth difference kernels whose interactions are restricted to same-level colleague boxes; and a singular residual kernel whose short-range interactions are evaluated directly. Benchmarks on uniform, highly nonuniform, and implicit-solvent electrolyte and colloidal configurations show that DMK-MC consistently outperforms a recent FMM-based $O(\log N)$ Monte Carlo method, delivering several-fold speedups at comparable tolerances.

There is little debate about the importance of the ancestral recombination graph in population genetics. An important theoretical tool, the main obstacle to its widespread usage is the computational cost required to match the ever-increasing scale of the data being analyzed. Many of these difficulties have been overcome in the past two decades, which have consequently seen the development of increasingly sophisticated ARG simulation and inference software. Nonetheless, challenges remain, especially in the area of ancestry inference. This paper is a comprehensive review of ARG samplers that have emerged in the past three decades to meet the need for scalable and flexible ancestry simulation and inference solutions. It specifically focuses on their performance, usability, and the biological realism of the underlying algorithm, and aims primarily to provide a technical overview of the field for researchers seeking to write their own coalescent-with-recombination sampler. As a complement to this article, we have compiled links to software, source code and documentation and made them available at https://www.patrickfournier.ca/arg-samplers-review/graph.

Sweepstakes reproduction may be generated by chance matching of reproduction with favorable environmental conditions. Gene genealogies generated by sweepstakes reproduction are in the domain of attraction of multiple-merger coalescents where a random number of lineages merges at such times. We consider population genetic models of sweepstakes reproduction for haploid panmictic populations of both constant ($N$), and varying population size, and evolving in a random environment. We construct our models so that we can recover the observed number of new mutations in a given sample without requiring strong assumptions regarding the population size or the mutation rate. Our main results are {\it (i)} continuous-time coalescents that are either the Kingman coalescent or specific families of Beta- or Poisson-Dirichlet coalescents; when combining the results the parameter $α$ of the Beta-coalescent ranges from 0 to 2, and the Beta-coalescents may be incomplete due to an upper bound on the number of potential offspring an arbitrary individual may produce; {\it (ii)} in large populations we measure time in units proportional to either $ N/\log N$ or $N$ generations; {\it (iii)} incorporating fluctuations in population size leads to time-changed multiple-merger coalescents where the time-change does not depend on $α$; {\it (iv)} using simulations we show that in some cases approximations of functionals of a given coalescent do not match the ones of the ancestral process in the domain of attraction of the given coalescent; {\it (v)} approximations of functionals obtained by conditioning on the population ancestry (the ancestral relations of all gene copies at all times) are broadly similar (for the models considered here) to the approximations obtained without conditioning on the population ancestry.

Cerebellar-like networks, in which input activity patterns are separated by projection to a much higher-dimensional space before classification, are a recurring neurobiological motif, present in the cerebellum, dentate gyrus, insect olfactory system, and electrosensory system of the electric fish. Their relatively well-understood design presents a promising test-case for probing principles of biological learning. The circuits' expansive projections have long been modelled as random, enabling effective general purpose pattern separation. However, electron-microscopy studies have discovered interesting hints of structure in both the fly mushroom body and mouse cerebellum. Recent numerical work suggested that this non-random connectivity enables the circuit to prioritise learning of some, presumably natural, tasks over others. Here, rather than numerical results, we present a robust mathematical link between the observed connectivity patterns and the cerebellar circuit's learning ability. In particular, we extend a simplified kernel regression model of the system and use recent machine learning theory results to relate connectivity to learning. We find that the reported structure in the projection weights shapes the network's inductive bias in intuitive ways: functions are easier to learn if they depend on inputs that are oversampled, or on collections of neurons that tend to connect to the same hidden layer neurons. Our approach is analytically tractable and pleasingly simple, and we hope it continues to serve as a model for understanding the functional implications of other processing motifs in cerebellar-like networks.

We introduce a category $\mathsf{qGph}$ of quantum graphs, whose definition is motivated entirely from noncommutative geometry. For all quantum graphs $G$ and $H$ in $\mathsf{qGph}$, we then construct a quantum graph $[G,H]$ of homomorphisms from $G$ to $H$, making $\mathsf{qGph}$ a closed symmetric monoidal category. We prove that for all finite graphs $G$ and $H$, the quantum graph $[G,H]$ is nonempty iff the $(G,H)$-homomorphism game has a winning quantum strategy, directly generalizing the classical case. The finite quantum graphs in $\mathsf{qGph}$ are tracial, real, and self-adjoint, and the morphisms between them are CP morphisms that are adjoint to a unital $*$-homomorphism. We show that Weaver's two notions of a CP morphism coincide in this context. We also show that every finite reflexive quantum graph is the confusability quantum graph of a quantum channel, answering a question of Daws.

GW231123 is an exceptional gravitational-wave event consistent with the merger of two massive, highly-spinning black holes. Reliable inference of the source properties is crucial for accurate interpretation of its astrophysical implications. However, characterization of GW231123 is challenging: only few signal cycles are observed and different signal models result in systematically different parameters. We investigate whether the interpretation of GW231123 is robust against model systematics and Gaussian detector noise. We show that the model systematics observed in GW231123 can be reproduced for a simulated signal based on the numerical-relativity surrogate model NRSur7dq4. Simulating data using the maximum-likelihood NRSur7dq4 waveform for GW231123 and no noise realization, we closely recover the systematics observed for the real signal. We then explore how the headline properties of GW231123 are impacted by Gaussian detector noise. Using the NRSur7dq4 maximum-likelihood waveform and different noise realizations, we consistently find support for large masses, high spin magnitudes (median $χ_1\geq 0.7$), and high spin precession (median $χ_\mathrm{p}\geq 0.68$). The spin in the direction of the angular momentum ($χ_\mathrm{eff}$) fluctuates more. Finally, again comparing to simulated signals, we show that any differences in the GW231123 inference based on each separate detector are not statistically significant. These results show that the properties of GW231123, and most importantly the high mass and high spin magnitudes inferred by NRSur7dq4, are robust.

The generation of non-Gaussian quantum states is a key requirement for universal continuous-variable quantum information processing. We report the experimental generation of large-amplitude squeezed coherent-state superpositions (squeezed cat states) on free-space optical pulses, reaching an amplitude of $α= 2.47$, which, to our knowledge, exceeds all previously reported values. Our protocol relies on the controlled mixing of the Fock states $|1\rangle$ and $|2\rangle$ through a tunable beam splitter, followed by heralding via homodyne detection. The resulting state displays three well-resolved negative regions in its Wigner function and achieves a fidelity of $0.53$ with the target state $\propto \hat{S}(z)(|α\rangle - |-α\rangle)$, with $α= 2.47$ and squeezing parameter $z = 0.56$. These results constitute a significant milestone for temporal breeding protocols and for the iterative generation of optical GKP states, opening new perspectives for scalable and fault-tolerant photonic quantum architectures.

In this paper, we explore the inflationary dynamics of the $β$-exponential potential model, where a scalar field couples to quadratic $(R + R^2)$ gravity. In this model, the inflaton is the field that determines the size of the extra dimension. We employ the Palatini formalism to derive the resulting Einstein-frame generalized $k$-inflation effective theory, which we analyze under the assumption that the constant-roll condition is satisfied. We scan the parameter space for inflationary predictions, specifically the spectral index $n_s$ and the tensor-to-scalar ratio $r$, ensuring consistency with the results from ACT DR6. The compliant regions are depicted accordingly. For a suitable range of the model parameters, the values obtained for the inflationary observables align with the most recent observations by the Atacama Cosmology Telescope (ACT) collaboration and/or the Planck mission.

We present a closed-form surrogate for the equivalent diameter of the Kerr black-hole shadow, defined as the diameter of the circle with the same area as the shadow's critical curve. The construction enforces the exact face-on (polar) limit by explicitly separating an analytically computed polar contribution based on the spherical photon-orbit branch where the horizontal impact parameter vanishes. The remaining inclination dependence is captured by a compact 15-parameter polynomial placed inside an exponential correction. The coefficients are determined by ordinary least squares on a deterministic reference grid generated from the Kerr critical-curve area. Over the practical domain of dimensionless spin from 0 to 0.998 and inclination from just above 0 degrees up to 90 degrees (with the exactly polar point treated analytically), the surrogate achieves sub-percent accuracy. On the training grid the median absolute percent error is 0.0105 percent with a worst case of 0.782 percent, and on a denser out-of-sample validation set (including inclinations down to 0.5 degrees) the median, 95th-percentile, and worst-case errors are 0.023 percent, 0.471 percent, and 1.64 percent, respectively. The resulting expression provides fast evaluations of the shadow size without numerical ray tracing, making it convenient for repeated calls in parameter inference and rapid model comparisons.

Molecular spins are promising candidates for quantum information science, leveraging coherent electronic spin states for quantum sensing and computation. However, the practical application of these systems is hindered by electronic spin decoherence, driven by interactions with nuclear spins in the molecule and the surrounding environment at low temperatures. Predicting dephasing dynamics remains a formidable challenge due to the complexity of the spin bath. In this work, we develop a non-Markovian time-convolutionless master equation to treat an electronic spin coupled to a nuclear-spin bath. By relating ab initio electronic structure parameters directly to the decoherence dynamics, we provide a framework that accounts for pure dephasing in the low-temperature limit. We apply this method to a series of molecular qubit candidates and demonstrate good agreement with experimental relaxation trends. This approach offers a computationally efficient path for the prediction of low-temperature decoherence trends in molecular spin systems.

We present the electro-thermal characterization of transition-edge sensor (TES) detectors suspended on Si membranes fabricated using a silicon-on-insulator (SOI) wafer. The use of an all-silicon fabrication platform, in contrast to the more commonly used silicon nitride membranes, is compatible with monolithic fabrication of integrated TES and SQUID circuits. The all-silicon architecture additionally allows efficient use of focal plane area; the readout circuitry may be positioned out of the focal plane by bending a thinned portion of the chip. Compatibility with integrated fabrication and efficient use of focal plane area provide a path to an efficient soft X-ray spectrometer. This work is motivated by our goal to develop a 10,000-pixel TES spectrometer to overcome critical measurement limitations in catalysis research. The characterization of fragile, carbon-based intermediates via techniques like Resonant Inelastic X-ray Scattering (RIXS) is often precluded by the slow, high-flux nature of existing technologies. The new instrument will allow for fast RIXS measurements to be made without causing sample damage. We verify the detector models and measure the energy resolution using a pulsed optical laser, demonstrating the viability of this approach for the final instrument to be deployed at the National Synchrotron Light Source II (NSLS-II).

We explore the existence and stability of exceptional points (EPs) in finite waveguide arrays subject to single-site dissipation. We show that the EP landscape is dictated by a geometry-dependent parity effect, leading to strictly distinct spectral behaviors for arrays with even versus odd numbers of waveguides. Through analytical derivation and numerical analysis, we define the conditions under which these singularities emerge and evolve. Our findings clarify the mechanisms of symmetry breaking in finite non-Hermitian lattices, offering new guidelines for the design of robust optical structures that exploit or avoid exceptional points.

Quantum Field Theory (QFT) introduced us to the notion that a causal diamond in space-time corresponded to a subsystem of a quantum mechanical system defined on the global space-time. Work by Jacobson, Fischler and Susskind, and particularly Bousso suggested that, in the quantum theory of gravity, this subsystem should have a density matrix of finite entropy. These authors formalized older intuitive arguments based on black hole physics. Although mathematically, Type II von Neumann algebras admit finite entropy density matrices, the black hole arguments suggest that the number of physical states in these subsystems is finite. The conjecture that de Sitter (dS) space has a finite number of physical states was first made by Fischler and one of the present authors. Leutheusser and Liu showed that, in the $N = \infty$ limit, causal diamonds with finite area in AdS radius units had Type $III_1$ von Neumann sub-algebras of the full operator algebra. They claimed that this was true for finite values of the UV cutoff, and that the algebra was the algebra of bulk local fields in the diamond. We will argue that the second part of this conjecture is incorrect and that the bulk field algebra emerges only in a double scaled limit, where the boundary UV cutoff is taken to infinity as $N$ is taken to infinity. There is never a bulk field theory description that resolves distances smaller than the AdS radius.

The accurate computation of quasinormal modes from rotating black holes beyond general relativity is crucial for testing fundamental physics with gravitational waves. In this study, we assess the accuracy of the eikonal and post-Kerr approximations in predicting the quasinormal mode spectrum of a scalar field on a deformed Kerr spacetime. To obtain benchmark results and to analyze the ringdown dynamics from generic perturbations, we further employ a 2+1-dimensional numerical time-evolution framework. This approach enables a systematic quantification of theoretical uncertainties across multiple angular harmonics, a broad range of spin parameters, and progressively stronger deviations from the Kerr geometry. We then confront these modeling errors with simple projections of statistical uncertainties in quasinormal mode frequencies as a function of the signal-to-noise ratio, thereby exploring the domain of validity of approximate methods for prospective high-precision black-hole spectroscopy. We also report that near-horizon deformations can affect prograde and retrograde modes differently and provide a geometrical explanation.

Highly multipartite entangled states play an important role in various quantum computing tasks. We investigate the dissipative generation of a complex entanglement structure as in a cluster state through engineered Markovian dynamics in the spin systems coupled via Ising interactions. Using the Lindblad master equation, we design a projection based dissipative channel that drives the system toward a unique pure steady state corresponding to the desired cluster state. This is done by removing the contribution of the orthogonal states. By explicitly constructing the Liouvillian superoperator in the full $2^N$-dimensional Hilbert space, we compute the steady-state density matrix, the Liouvillian spectral gap, entanglement witness and the fidelity with respect to the ideal cluster state. The results demonstrate that the cluster state emerges as the steady state when the engineered Liouvillian dissipation dominates over the local Ising interaction between spins. Moreover, we find that the fidelity and Liouvillian spectral gap is relatively insensitive to the system size once the saturation dissipation has been achieved that scales linearly with the qubit number. This analysis illustrates a physically realizable path towards steady-state entanglement generation in the spin systems using engineered dissipation.

The pseudocomplex version of the FLRW model is presented within the framework of pseudocomplex General Relativity (pcGR). In this approach, dark energy arises as a geometric consequence of the pseudocomplex structure, leading to a time dependent Hubble parameter rather than a strictly constant H0. The relation between the tiderived and constrained using recent DESI BAO data. Fitting beta yields a best-fit value beta = 1.0426, corresponding to a deceleration parameter q = -0.9361 and a present day Hubble acceleration me derivative of the Hubble parameter and a single geometric parameter beta in the effective dark energy equation of state is derived and constrained using recent DESI BAO data. Fitting beta yields a best-fit value beta = 1.0426, corresponding to a deceleration parameter q = -0.9361 and a present day Hubble acceleration H0 sim 0.94 x10-17 (km/s2)/Mpc. Using the exact Sandage Loeb relation, the predicted redshift drift over 20 years for a source at z = 4 is Delta-v sim -11.1 cm/s, in close agreement with the Lambda CDM prediction. In pcGR, however, the non-vanishing H0 is a direct geometric prediction, providing a clear and testable target for future high-precision spectroscopic observations.

Compared to other stable isotopes of $\rm{Yb}^+$, $^{173}\rm{Yb}^+$ has a richer hyperfine structure, which leads to more favorable clock transitions, spectroscopic techniques for probing new physics, and more sophisticated quantum computing architectures. However, to date, its electronic spectrum remains poorly characterized. Here, we report on efficient laser cooling, state preparation, and detection of a single trapped $^{173}\rm{Yb}^+$ ion. The previously unobserved $^2\!S_{1/2} \rightarrow {}^2\!D_{3/2}$ electric quadrupole transition at 436 nm is coherently excited, and the isotope shift between $^{171}\rm{Yb}^+$ and $^{173}\rm{Yb}^+$ on this transition is determined with an uncertainty of 1.4 Hz. Using microwave spectroscopy, we resolve the hyperfine structure (HFS) of the ${}^2\!D_{3/2}$ state with a relative uncertainty below $10^{-8}$. From the HFS measurement data, we infer for ${}^{173}$Yb a nuclear magnetic octupole moment $Ω= -0.062(8)\,({\rm b} \times μ_N)$ with uncertainty reduced by more than 2 orders of magnitude compared to previous studies. The data also allow us to determine hyperfine anomalies for the ${}^2\!S_{1/2}$ and ${}^2\!D_{3/2}$ states.

We extend a recently developed "tangent fermion" method to discretize the Hamiltonian of a helical Luttinger liquid on a one-dimensional lattice, including two-particle backscattering processes that may open a gap in the spectrum. The fermion-doubling obstruction of the sine dispersion is avoided by working with a tangent dispersion, preserving the time-reversal symmetry of the Hamiltonian. The numerical results from a tensor network calculation on a finite lattice confirm the expectation from infinite-system analytics, that a gapped phase with spontaneously broken time-reversal symmetry emerges when the Fermi level is tuned to the Dirac point and the Luttinger parameter crosses a critical value.

The Berry phase, a fundamental geometric phase in quantum systems, has become a crucial tool for probing the topological properties of materials. Quantum oscillations, such as Shubnikov-de Haas (SdH) oscillations, are widely used to extract this phase, but its unambiguous determination remains challenging. This work highlights the inherent ambiguities in interpreting the oscillation phase solely from SdH data, primarily due to the influence of the spin factor $R_S$, which depends on the Landé $g$-factor and effective mass. While the Lifshitz-Kosevich (LK) theory provides a framework for analyzing oscillations, the unknown g-factor introduces significant uncertainty. For instance, a zero oscillation phase could arise either from a nontrivial Berry phase or a negative $R_S$. We demonstrate that neglecting $R_S$ in modern studies, especially for topological materials with strong spin-orbit coupling, can lead to doubtful conclusions. Through theoretical analysis and numerical examples, we show how the interplay between the Berry phase and Zeeman effect complicates phase determination. Additionally, we also discuss another underappreciated mechanism - the magnetic field dependence of the Fermi level. Our discussion underscores the need for complementary experimental techniques to resolve these ambiguities and calls for further research to refine the interpretation of quantum oscillations in topological systems.

We investigate the quantum properties of heavy-fermion pairs, such as $t\bar t$ or $τ^+τ^-$, produced in lepton-lepton collisions with polarised beams. Focusing on spin correlations, entanglement, Bell-inequality violation, and quantum-information--theoretic measures such as purity and magic, we analyse how beam polarisation shapes the structure of the spin-density matrix. We derive analytic expressions for a wide range of helicity configurations, including both Standard Model contributions and generic new-physics effects parametrised by scalar, vector, and tensor four-fermion operators within an effective field theory framework. We show that beam polarisation unlocks a substantially richer set of spin configurations and significantly enhances sensitivity to non-standard interactions. As a phenomenological application, we study $t\bar t$ production at a future linear collider and demonstrate that quantum observables provide a comprehensive and complementary probe of top-quark interactions and stronger constraints on the scale of new physics.

This book gives a geometry-first, hardware-aware route through quantum-information workflows, with one goal: connect states, circuits, and measurement to deterministic classical pipelines that make hybrid quantum systems run. Part 1 develops the backbone (essential linear algebra, the Bloch-sphere viewpoint, differential-geometric intuition, and quantum Fisher information geometry) so evolution can be read as motion on curved spaces and measurement as statistics. Part 2 reframes circuits as dataflow graphs: measurement outcomes are parsed, aggregated, and reduced to small linear-algebra updates that schedule the next pulses, highlighting why low-latency, low-jitter streaming matters. Part 3 treats multi-qubit structure and entanglement as geometry and computation, including teleportation, superdense coding, entanglement detection, and Shor's algorithm via quantum phase estimation. Part 4 focuses on topological error correction and real-time decoding (Track A): stabilizer codes, surface-code decoding as "topology -> graph -> algorithm", and Union-Find decoders down to microarchitectural/RTL constraints, with verification, fault injection, and host/control-stack integration under product metrics (bounded latency, p99 tails, fail-closed policies, observability). Optional Track C covers quantum cryptography and streaming post-processing (BB84/E91, QBER/abort rules, privacy amplification, and zero-knowledge/post-quantum themes), emphasizing FSMs, counters, and hash pipelines. Appendices provide visualization-driven iCEstick labs (switch-to-bit conditioning, fixed-point phase arithmetic, FSM sequencing, minimal control ISAs), bridging principles to implementable systems.

Modern understanding of symmetry in quantum field theory includes both invertible and non-invertible operations. Motivated by this, we extend Nielsen's geometric approach to quantum circuit complexity to incorporate non-invertible gates. These arise naturally from fusion of topological defects and allow transitions between superselection sectors. We realise fusion operations as completely positive, trace-preserving quantum channels. Including such gates makes the sector-changing optimisation problem discrete: it reduces to a weighted shortest-path problem on the fusion graph. Circuit complexity therefore combines continuous geometry within sectors with discrete sector jumps. We illustrate the framework in rational conformal field theories and briefly comment on an AdS$_3$ interpretation in which fusion-induced transitions correspond to geometry-changing boundary operations. A companion paper provides full derivations and extended examples.

Two of the most pressing challenges in cosmology are the persistent discrepancy in measurements of the Hubble constant, referred to as the Hubble tension, and the deficit of baryons in the local Universe, known as the missing baryon problem. Fast radio bursts (FRBs) encode the integrated electron column density along their lines of sight, offering a unique probe of both the cosmic expansion rate ($H_0$) and the baryon density ($Ω_{\rm b}$). However, constraints from FRBs alone suffer from a severe $H_0$-$Ω_{\rm b}$ degeneracy that prevents them from resolving either problem independently. We show that this degeneracy can be broken by combining FRBs with other emerging probes whose degeneracy directions differ in the $H_0$-$Ω_{\rm b}$ plane. Specifically, we quantify three multi-messenger approaches: FRBs paired with gravitational wave (GW) standard sirens, strong gravitational lensing (SGL) time delays, and 21 cm intensity mapping (IM). The combinations FRB+GW, FRB+SGL, and FRB+21 cm IM each deliver simultaneous constraints on $H_0$ and $Ω_{\rm b}$ better than ($1\%$, $1\%$) in the $Λ$CDM model, ($1.5\%$, $2\%$) in the $w$CDM model, and ($2\%$, $3.5\%$) in the CPL model. Moreover, in a model-independent framework, both FRB+GW and FRB+SGL constrain $H_0$ and $Ω_{\rm b}$ to better than ($1\%$, $2\%$) precision. These results demonstrate that the synergy between FRBs and other emerging probes holds great promise for resolving the Hubble tension and locating the missing baryons.

We present an analytically tractable theory of reservoir-engineered refrigeration of a superconducting microwave cavity and map it onto a realistic solid-state implementation based on gate-defined double-quantum-dot (DQD) spin qubits. Treating the DQD not as a spectroscopic element but as a tunable engineered reservoir, we show how gate control of populations, coherences, linewidths, and detuning defines an effective photon birth-death process with predictable detailed balance. This framework yields closed-form expressions for the cavity steady state, identifies cooling bounds and detuning-dependent refrigeration valleys, and clarifies when refrigeration can drive the cavity below both the bath temperature and the DQD setpoint. By distinguishing refreshed (collision-like) and persistent reservoir regimes, we show how memory effects, saturation, and dark-state formation constrain cooling in realistic devices, while collective bright-mode coupling in a two-dot configuration can enhance refrigeration subject to mismatch and dephasing, as confirmed by numerical Lindblad simulations demonstrating targeted millikelvin cavity cooling relevant for cryogenic circuit-QED architectures.

We investigated the spectral properties of squeezed light generated via parametric down-conversion in the high-gain regime, considering both unapodized and apodized dispersion-engineered waveguides. The gain-dependent evolution of these states is examined starting from the low-gain regime, which includes both highly correlated and nearly uncorrelated cases. For the unapodized configuration, we observe a monotonic increase in spectral purity with gain, whereas the apodized configuration exhibits a nonmonotonic dependence, initially decreasing and then recovering at higher gain. By combining Schmidt-mode analysis with a group-velocity-based interpretation, we explain why different dispersion conditions exhibit distinct gain-dependent behavior, specifically that rapid purification occurs when the pump group velocity lies between those of the signal and idler. Our study shows that the evolution of spectral purity is governed primarily by the underlying dispersion of the waveguide. These results demonstrate that dispersion engineering and parametric gain can be jointly exploited to tailor the spectral-mode structure of squeezed-light sources, enabling their optimization for a broad range of quantum applications.

We investigate tidal forces, geodesic deviation, and tidal disruption in the black hole spacetime described by the Kalb-Ramond-ModMax solution, where electromagnetic nonlinearity is governed by the parameter $γ$ and Lorentz symmetry violation by the parameter $l$. In the canonical sector ($α=1$), the radial tidal force exhibits a transition marked by a sign inversion between the horizons $r_{-}$ and $r_+$, signaling internal regimes of radial compression analogous to those of charged black holes; the parameter $l$ controls the strength and location of this transition, while $γ$ regulates the nonlinear electromagnetic contribution. The angular tidal force is predominantly compressive, $l$ shaping the effective geometry, and $γ$ acting as a damping factor. In the phantom sector ($α=-1$), tidal forces and geodesic deviation diverge, indicating a tidal instability, with $l$ and $γ$ affecting only the magnitude of the response. We further show that $l$ shifts the relation between the horizon radius $r_+$ and the tidal disruption radius $r_{\rm Roche}$, thereby modifying the critical (Hills) mass defined by $r_{\rm Roche}=r_+$. Tidal disruption of neutron stars occurs inside the horizon for supermassive black holes, whereas Sun-like stars are disrupted outside the horizon, with $γ$ becoming relevant only for ultramassive black holes with masses $\sim 10^{8}M_{\odot}$. Our results demonstrate that Kalb-Ramon-ModMax effects are largely suppressed for supermassive black holes, but may be relevant for intermediate-mass systems and observable tidal disruption events, offering an indirect probe of Lorentz violation and nonlinear electrodynamics in the strong-field regime.

The NANOGrav 15-year (NG15) data set provides evidence for a gravitational-wave background (GWB) signal at nanohertz frequencies, which is expected to originate either from a cosmic population of inspiraling supermassive black-hole binaries or new particle physics in the early Universe. A firm identification of the source of the NG15 signal requires an accurate reconstruction of its frequency spectrum. In this paper, we provide such a spectral characterization of the NG15 signal based on a piecewise power-law (PPL) ansatz that strikes a balance between existing alternatives in the literature. Our PPL reconstruction is more flexible than the standard constant-power-law model, which describes the GWB spectrum in terms of only two parameters: an amplitude A and a spectral index gamma. Concurrently, it better approximates physically realistic GWB spectra -- especially those of cosmological origin -- than the free spectral model, since the latter allows for arbitrary variations in the GWB amplitude from one frequency bin to the next. Our PPL reconstruction of the NG15 signal relies on individual PPL models with a fixed number of internal nodes (i.e., constant power law, broken power law, doubly broken power law, etc.) that are ultimately combined in a Bayesian model average. The data products resulting from our analysis provide the basis for fast refits of spectral GWB models.

In this work we show that results of Rayleigh-Schrödinger perturbation theory can be easily obtained using the recently proposed supersymmetric expansion algorithm. Our formalism avoids the sums over intermediate states and yield directly corrections to the energy and eigenstates in terms of integrals weighted by the probability densities for the edge states of the involved supersymmetric Hamiltonians.

Non-Gaussian mechanical states are a key resource for quantum-enhanced sensing and tests of macroscopic quantum physics. We propose a measurement-based protocol to herald delocalized, nonclassical states of a mechanical oscillator in cavity optomechanics by conditioning on Geiger photodetection of the optical output. We analyse under which conditions Stokes-induced optomechanical entanglement give rise to mechanical Wigner Function negativity upon detection. We develop and compare a blue-detuned pulsed scheme and a continuous-wave steady-state scheme employing temporal-mode filtering, and we quantify heralding rates and robustness to finite temperature under realistic detection efficiencies.

We here investigate the observational viability of Nash gravity as an alternative to the standard $Λ$CDM cosmology. Based on Nash's embedding theorem, the model introduces orthogonal perturbations via variations in the extrinsic curvature, generating scalar-type metric perturbations directly from geometry, without the need to introduce additional fields. We confront the model with current observational data, including Cosmic Microwave Background (CMB) measurements from Planck, Baryon Acoustic Oscillations (BAO) from DESI DR2, and recent Type Ia supernova (SN Ia) compilations. Our analysis shows that Nash gravity provides a good fit to the data, yielding a slightly higher value for the Hubble constant, $H_0 = 69.32 \pm 0.72$ km/s/Mpc, compared to the $Λ$CDM model, thus offering a potential alleviation of the $H_0$ tension. Furthermore, the model naturally predicts a suppressed growth of structure, with $S_8 \approx 0.76$ across various joint analyses, potentially alleviating the so-called $S_8$ tension, assuming that this discrepancy is not solely due to systematic effects in other independent measurements. In some cases, Nash gravity achieves a better fit to the data than the $Λ$CDM paradigm at the $2σ$ level.

The open quantum Rabi model is studied in this work, with the explicit $\mathbf{A}^{2}$ term incorporated as required by the Thomas-Reich-Kuhn sum rule. It is shown that anisotropy provides a generic and robust mechanism for overcoming the no-go theorem in dissipative quantum systems, thereby establishing a genuine platform for observing dissipative phase transitions. The inclusion of the $\mathbf{A}^{2}$ term yields a significantly richer and asymmetric steady-state phase diagram, consisting of normal, superradiant, and bistable phases that intersect at tricritical points, while isolated bistable phases also emerge and the number of tricritical points is reduced. Notably, it is near the intersection of the two critical-line branches enclosing the superradiant phases, rather than at the tricritical points, that the $\mathbf{A}^{2}$ term fundamentally alters the scaling of photon-number fluctuations. Given the inherent role of the $\mathbf{A}^{2}$ term in light-matter interactions, our findings open a realistic route toward the experimental investigation and dynamical control of nonequilibrium critical phenomena in practical open quantum platforms.

Holographic black holes exhibit a striking relation between thermal boundary one-point functions and bulk geodesic lengths. In the large conformal-dimension limit, the one-point function of a primary operator is given by the exponential of the geodesic length from its boundary insertion point to the horizon. We test the robustness of this relation under perturbations by considering an arbitrary radial deformation of an Euclidean BTZ black hole and working to first order in the perturbation. We find that the relation remains robust: the corrected one-point function at large conformal dimension is still governed by an exponent proportional to the modified boundary-to-horizon geodesic length. The result is established using WKB and saddle-point methods, with the validity of the WKB approximation justified by exact analyses.

The preparation of quantum states is one of the most fundamental tasks in quantum computing, and a key primitive in many quantum algorithms. Of particular interest to areas such as quantum simulation and linear-system solvers are sparse quantum states, which contain only a small number $s$ of non-zero computational basis states compared to a generic state. In this work, we present an approach that prepares $s$-sparse states on $n$ qubits, reducing the number of Toffoli gates required compared to prior art. We work in the established framework of first preparing a dense state on a $\lceil{\log(s)}\rceil$-qubit sub-register, and then mapping this state to the target state via an isometry, with the latter step dominating the cost of the full algorithm. The speed-up is achieved by designing an efficient algorithm for finding and implementing the isometry. The worst-case Toffoli cost of our isometry circuit, which may be viewed as a batched version of an approach by Malvetti et al., is essentially $2s$ for sufficiently large values of $n$, yielding roughly a $\log(s)/2$ improvement factor over the state-of-the-art. In numerical benchmarks on randomly chosen states, the cost is closer to $s$. With the improved isometry circuit, we examine the dense-state preparation step and present ways to optimize the joint cost of both steps, particularly in the case of target states with purely real coefficients, by outsourcing some sub-tasks from the dense-state preparation to the isometry.

One of the key considerations in the development of Quantum Machine Learning (QML) protocols is the encoding of classical data onto a quantum device. In this chapter we introduce the Matrix Product State representation of quantum systems and show how it may be used to construct circuits which encode a desired state. Putting this in the context of QML we show how this process may be modified to give a low depth approximate encoding and crucially that this encoding does not hinder classification accuracy and is indeed exhibits an increased robustness against classical adversarial attacks. This is illustrated by demonstrations of adversarially robust variational quantum classifiers for the MNIST and FMNIST dataset, as well as a small-scale experimental demonstration on a superconducting quantum device.

Coherent multidimensional spectroscopy enables detailed investigations of vibronic effects in molecular and solid-state systems. We present explicit analytical expressions for multidimensional nonlinear response functions in the presence of Herzberg-Teller (non-Condon) coupling, within the displaced harmonic oscillator model. The formulation applies to electronic systems with an arbitrary number N of electronic states and to response functions of arbitrary order M in the light-matter interaction. We show that Herzberg-Teller coupling introduces additional oscillatory factors in the time-domain response functions, leading, upon Fourier transformation, to replicas of the Franck-Condon multidimensional spectra shifted by integer multiples of the vibrational frequencies. The present results provide a general analytical framework for the interpretation of non-Condon effects in coherent multidimensional spectroscopies.

We study the CHSH inequality from an informational, timing-sensitive viewpoint using game-theoretic probability, which avoids assuming an underlying probability space. The locality loophole and the measurement-dependence (``freedom-of-choice'') loophole are reformulated as structural constraints in a sequential hidden-variable game between Scientists and Nature. We construct a loopholes-closed game with capital processes that test (i) convergence of empirical conditional frequencies to the CHSH correlations and (ii) the absence of systematic correlations between measurement settings and Nature's hidden-variable assignments, and prove that Nature cannot satisfy both simultaneously: at least one capital process must diverge. This yields an operational winning strategy for Scientists and a game-theoretic probabilistic interpretation of experimentally observed CHSH violations.

The continual production of long wavelength scalars and gravitons during inflation injects secular growth into loop corrections which would be constant in flat space. One typically finds that each additional factor of the loop counting parameter can induce up to a certain number of logarithms of the scale factor. Loop corrections that attain this number are known as ``leading logarithms''; those with fewer are sub-leading. Starobinsky's stochastic formalism has long been known to reproduce the leading logarithms of scalar potential models. We show that the first sub-leading logarithm is captured by applying the stochastic formalism to a certain part of the 1-loop effective potential. This is checked at 2-loops for a massless, minimally coupled scalar with a quartic self-interaction on de Sitter background.

The generalized quantum Arimoto--Blahut (QAB) algorithm is a powerful derivative-free iterative method in quantum information theory. A key obstacle to its broader use is that existing convergence guarantees typically rely on analytical conditions that are either overly restrictive or difficult to verify for concrete problems. We address this issue by introducing an a posteriori certification viewpoint: instead of requiring fully a priori verifiable assumptions, we provide convergence and error guarantees that can be validated directly from the iterates produced by the algorithm. Specifically, we prove a generalized global convergence theorem showing that, under convexity and a substantially weaker numerically verifiable condition, the QAB iteration converges to the global minimizer. This theorem yields a practical certification procedure: by checking explicit inequalities along the computed trajectory, one can certify global optimality and bound the suboptimality of the obtained value. As an application, we develop a certified iterative scheme for computing the quantum relative entropy of channels, a fundamental measure of distinguishability in quantum dynamics. This quantity is notoriously challenging to evaluate numerically: gradient-based methods are impeded by the complexity of matrix functions such as square roots and logarithms, while recent semidefinite programming approaches can become computationally and memory intensive at high precision. Our method avoids these bottlenecks by combining the QAB iteration with a posteriori certification, yielding an efficient and scalable algorithm. Numerical experiments demonstrate rapid convergence and improved scalability and adaptivity compared with SDP-based approaches.

We investigate the complex time evolution of a vacuum state with the insertion of a local primary operator in two-dimensional conformal field theories (2d CFTs). This complex time evolution can be considered as a composite process constructed from Lorentzian time evolution and a Euclidean evolution induced by a post-selected measurement. Our main finding is that in the spatially-compact system, this complex time evolution drives the state of the subsystems to those of the primary state with the same conformal dimensions of the inserted operator. Contrary to the compact system, the subsystems of the spatially non-compact system evolve to states that depend on the non-unitary process during a certain time regime. In holographic systems with a compact spatial direction, this process induced by a heavy local operator can correspond to the relaxation from a black hole with an inhomogeneous horizon to that with a uniform one, while in the ones with a non-compact spatial direction, it can correspond to the relaxation to that with a horizon depending on the non-unitary process.

We propose a geometric hybrid Poincaré sphere (GHPS) as a unified geometrical framework for describing structured photon states with independently controllable spin angular momentum (SAM) and orbital angular momentum (OAM). Unlike the conventional higher-order Poincaré sphere, in which the SAM and OAM are intrinsically coupled through fixed basis states, the GHPS is constructed by defining its poles as direct products of arbitrary orthogonal bases on the Poincaré sphere (PS) and orbital Poincaré sphere (OPS) and by superposing these pole states. Using numerical simulations, we analyze representative GHPS states and show that the GHPS spherical coordinates govern the amplitude ratio and relative phase between the pole bases. This framework enables spatially inhomogeneous polarization distributions and intensity patterns, including nonseparable structures in which polarization and intensity are intrinsically intertwined, and provides a systematic state-space description for the coherent geometrical control of advanced structured light fields.

We investigate exact cosmological solutions with a massive scalar field minimally coupled to the Einstein-Hilbert action in General Relativity. For an extended Higgs-like scalar self-interaction, we find that the resulting field equations belong to the damped Ermakov-Painlevé II class and construct novel analytical solutions within the framework of the Chiellini integrability condition. We analyze whether the expanding branch of the solutions can describe a late-time cosmic acceleration, using a combined statistical analysis of BAO, CMB, cosmic chronometer and Pantheon+SHOES supernova datasets. A crucial outcome of this exercise is the analytical emergence of a smooth phantom divide crossing in the dark energy equation of state, achieved without introducing any pathological instabilities. The reconstruction yields a present-day Hubble parameter $H_0 \gtrsim 70 \,\mathrm{km\,s^{-1}\,Mpc^{-1}}$, with a reduced tension relative to the $Λ$CDM cosmology. The results indicate that Chiellini-integrable scalar cosmologies are capable of providing a robust and analytically controlled framework for modeling late-time cosmic acceleration and phantom divide crossing, offering a viable alternative to phenomenological dark-energy parametrizations.

Scale invariance in quantum mechanics can be broken in several ways. A well-known example is the breakdown of continuous scale invariance to discrete scale invariance, whose typical realization is the Efimov effect of three-body problems. Here we discuss yet another discrete symmetry to which continuous scale invariance can be broken: discrete phase invariance. We first revisit the one-body problem on the half line in the presence of an inverse-square potential -- the simplest example of nontrivial scale-invariant quantum mechanics -- and show that continuous scale invariance can be broken to discrete phase invariance in a small window of coupling constant. We also show that discrete phase invariance manifests itself as circularly distributed simple poles on Riemann sheets of the S-matrix. We then present three examples of few-body problems that exhibit discrete phase invariance. These examples are the one-body Aharonov-Bohm problem, a two-body problem of nonidentical particles in two dimensions, and a three-body problem of nonidentical particles in one dimension, all of which contain a codimension-two ``magnetic'' flux in configuration spaces.

We study the compactification of higher-dimensional Lovelock gravity on compact irreducible symmetric spaces, focusing on conditions under which a physically healthy four-dimensional Minkowski vacuum exists. We show that when the internal dimension is five or less, or when the theory is restricted to the Einstein-Gauss-Bonnet sector, the four-dimensional graviton (tensor sector) is necessarily a ghost. Inclusion of the cubic Lovelock term removes this ghost instability; however, the resulting Minkowski vacuum is generically only metastable, being accompanied by energetically favored Anti-de Sitter vacua. While such metastability cannot be avoided for spherical internal spaces, we identify an infinite class of higher-rank symmetric spaces where the true vacuum can be pushed to infinity in moduli space, thereby realizing genuinely stable and ghost-free Minkowski vacua at the level of the four-dimensional effective theory. To support these conclusions, we explicitly compute Lovelock terms up to cubic order on these spaces, confirming a universal log-convexity among the linear, quadratic, and cubic invariants, which plays a central role in our analysis.

Nonlocality is a peculiar nature of quanta and it stands as an important quantum resource in application. Yet mere linear property of it, viz. the first order in moment, has been explored through various inequalities. Noticing the vast higher-order regime unexplored, in this study we investigate the higher-order quantum correlations in entangled hyperon-antihyperon system, which may be generated massively in charmonium decays. A new type of Clauser-Horne inequality for statistical cumulants and central moments is formulated. We find that a significant violation of the third-order constraint, indicating the existence of higher-order correlation, exists in hyperon-antihyperon system and can be observed in high energy physics experiments, like BESIII and Belle II. Notably, the violation manifests more in higher energy systems of the $Λ\barΛ$ pair against the kinematic contamination of timelike events.

We develop a unified many-body theory of two-photon dark-state laser cooling, the workhorse for preparing trapped ions close to their motional quantum ground state. For ions with a $Λ$ level structure, driven by Raman lasers, we identify an ion-number-dependent crossover between weak and strong coupling where both the cooling rate and final temperature are simultaneously optimized. We obtain simple analytic results in both extremes: In the weak coupling limit, we show a Lorentzian spin-absorption spectrum determines the cooling rate and final occupation of the motional state, which are both independent of the number of ions. We also highlight the benefit of including an additional spin dependent force in this case. In the strong coupling regime, our theory reveals the role of collective dynamics arising from phonon exchange between dark and bright states, allowing us to explain the enhancement of the cooling rate with increasing ion number. Our analytic results agree closely with exact numerical simulations and provide experimentally accessible guidelines for optimizing cooling in large ion crystals, a key step toward scalable, high-fidelity trapped-ion quantum technologies.

We systematically construct ghost-free scalar-tensor theories whose Lagrangian includes up to third-order derivatives of the scalar field. Using a spatially covariant action written in terms of the ADM variables, we impose degeneracy and consistency conditions that ensure the propagation of only one scalar and two tensor degrees of freedom. The resultant theories extend the generalized disformal Horndeski and U-DHOST theories. We discuss the transformation properties of these theories under generalized disformal transformations.

An initial value algorithm is utilized to examine the time dependent evolution of the electromagnetic fields arising from oblique scattering of bounded pulses from an infinite planar dielectric interface. Since the qubit lattice algorithm (QLA) is almost fully unitary, one finds excellent conservation of electromagnetic energy. Various Gaussian envelope pulses are considered in regimes where the incident angle is below that needed for total internal reflection. While the reflected pulse retains its overall Gaussian shape, the transmitted pulse exhibits a combination of a Gaussian envelope along with Huygen-like emitted wave fronts from the collision point of the initial pulse with the infinite dielectric interface. The strength of these Huygen wavefronts depends on the width of the incident pulse.

A quantum Latin square of order $n$ (denoted as QLS$(n)$) is an $n\times n$ array whose entries are unit column vectors from the $n$-dimensional Hilbert space $\mathcal{H}_n$, such that each row and column forms an orthonormal basis. Two unit vectors $|u\rangle, |v\rangle\in \mathcal{H}_n$ are regarded as identical if there exists a real number $θ$ such that $|u\rangle=e^{iθ}|v\rangle$; otherwise, they are considered distinct. The cardinality $c$ of a QLS$(n)$ is the number of distinct vectors in the array. In this note,we use sub-QLS$(6)$ to prove that for any integer $m\geq 2$ and any $c\in [6m,36m^2]\setminus \{6m+1\}$, there is a QLS$(6m)$ with cardinality $c$.

High-rate concatenated quantum codes offer a promising pathway toward fault-tolerant quantum computation, yet designing efficient decoders that fully exploit their error-correction capability remains a significant challenge. In this work, we introduce a hard-decision decoder for concatenated quantum Hamming codes with time complexity polynomial in the block length. This decoder overcomes the limitations of conventional local decoding by leveraging higher-level syndrome information to revise lower-level recovery decisions -- a strategy we refer to as bidirectional decoding. For the concatenated $[[15,7,3]]$ quantum Hamming code under independent bit-flip noise, the bidirectional decoder improves the threshold from approximately $1.56\%$ to $4.35\%$ compared with standard local decoding. Moreover, the decoder empirically preserves the full $3^{L}$ code-distance scaling for at least three levels of concatenation, resulting in substantially faster logical-error suppression than the $2^{L+1}$ scaling offered by local decoders. Our results can enhance the competitiveness of concatenated-code architectures for low-overhead fault-tolerant quantum computation.

In the noisy intermediate-scale quantum (NISQ) era, distributed quantum computation has garnered considerable interest, as it overcomes the physical limitations of single-device architectures and enables scalable quantum information processing. In this study, we focus on the challenge of achieving exact amplitude amplification for quantum states with arbitrary amplitude distributions and subsequently propose a Distributed Exact Quantum Amplitude Amplification Algorithm (DEQAAA). Specifically, (1) it supports partitioning across any number of nodes $t$ within the range $2 \leq t \leq n$; (2) the maximum qubit count required for any single node is expressed as $\max \left(n_0,n_1,\dots,n_{t-1} \right) $, where $n_j$ represents the number of qubits at the $j$-th node, with $\sum_{j=0}^{t-1} n_j =n$; (3) it can realize exact amplitude amplification for multiple targets of a quantum state with arbitrary amplitude distributions; (4) we verify the effectiveness of DEQAAA by resolving a specific exact amplitude amplification task involving two targets (8 and 14 in decimal) via MindSpore Quantum, a quantum simulation software, with tests conducted on 4-qubit, 6-qubit, 8-qubit and 10-qubit systems. Notably, through the decomposition of $C^{n-1}PS$ gates, DEQAAA demonstrates remarkable advantages in both quantum gate count and circuit depth as the qubit number scales, thereby boosting its noise resilience. In the 10-qubit scenario, for instance, it achieves a reduction of over $97\%$ in both indicators compared to QAAA and EQAAA, underscoring its outstanding resource-saving performance.

We present SASHMAG (Saturated Absorption Spectroscopy High-field MAGnetometer), an atomic sensor designed for precision magnetic-field measurements in the intermediate-to-high field regime ($>0.2\,\text{T}$) using Rubidium-87 ($^{87}Rb$). The sensor operates in the hyperfine Paschen-Back regime, where the hyperfine and Zeeman interactions decouple, and utilizes counter-propagating pump-probe configuration in Faraday geometry to resolve isolated, Doppler-free Zeeman transitions. To interpret the resulting spectra in this strongly field-dependent regime, we developed a comprehensive multilevel optical Bloch-equation model solved explicitly in the uncoupled $\ket{m_I, m_J}$ basis, capturing state mixing and nonlinear saturation dynamics. This model reproduces measured spectra at sub-Doppler resolution and is consistent with analytical expectations for power broadening and thermal Doppler scaling. Magnetic field estimation is performed using a physics-constrained optimization routine that infers the magnetic field by minimizing the residual between experimentally extracted line centers and calculated transition frequencies from the field-dependent Hamiltonian. We demonstrate magnetic field retrieval from $0.2\,\text{T}$ to $0.4\,\text{T}$ with a precision of $\pm 0.0017 \,\text{T}$). Furthermore, the validated simulation establishes a foundation for generating synthetic training datasets, paving the way for autonomous, Machine Learning-enhanced magnetometry in applications ranging from MRI to fusion reactors.

We develop a data-driven framework for identifying non-Markovian dynamical equations of motion for open quantum systems. Starting from the Nakajima--Zwanzig formalism, we vectorize the reduced density matrix into a four-dimensional state vector and cast the dynamics as a Volterra integro-differential equation with an operator-valued memory kernel. The learning task is then formulated as a constrained optimization problem over the admissible operator space, where correlation functions are approximated by rational functions using Padé approximants. We establish well-posedness of the learnin

We propose a displacement-squeeze receiver (DSR) for discriminating BPSK displaced squeezed vacuum states (S-BPSK). The receiver applies a displacement followed by a squeezing operation with the squeezing axis rotated by $\fracπ{2}$, and performs photon-number-resolving detection with a MAP threshold decision. This processing effectively increases the distinguishability of the input states by elongating their distance in phase space and reducing their population overlap in Fock basis. We show that for all signal energy N, $P_\text{err}^\text{DSR} \in \left[P_\text{HB}^\text{DSS}, 2P_\text{HB}^\text{DSS}\right]$, under equal priors and ideal condition. In the low-energy regime, DSR beats the S-BPSK SQL at $N \approx 0.3$ and drops below the coherent-state BPSK (C-BPSK) Helstrom bound at $N \approx 0.4$, reaching $P_\text{err}^\text{DSR} < 1\%$ near $N \approx 0.6$. Finally, we quantify performance under non-unit efficiency and dark counts, phase diffusion, and receiver thermal noise, with MAP threshold adaptation providing robustness across these nonidealities.

We present the results of a search for gravitational waves (GWs) from individual sources using high-cadence observations of PSR J1909\(-\)3744 obtained during an intensive observing campaign with the International Pulsar Timing Array second data release (IPTA-DR2) between July 2010 and November 2012. The observations, conducted at three different radio frequencies with the Nançay Radio Telescope (NRT) and Parkes Telescope (PKS) and five frequencies with the Green Bank Telescope (GBT), enabled precise corrections for dispersion measure effects and scattering variations. After these corrections, the timing residuals showed an unmodeled periodic noise component with an amplitude of 340 ns. Our analysis yields upper limits on the GW strain from individual sources, constraining it to be below \(1.9 \times 10^{-14}\) at 71 nHz and \(2.3 \times 10^{-13}\) at 1 \textmu Hz for average sky locations, while for optimal source locations the limits improve to \(6.2 \times 10^{-15}\) and \(8.9 \times 10^{-14}\) at the same frequencies, respectively. Our new limits are about a factor of 1.52 more stringent than those of Perera et al. based on an earlier EPTA data.

Recent work revealed a tension between the Gross-Mende analysis of the high-energy fixed-angle behavior of string amplitudes and the explicit numerical data. Motivated by this puzzle, we revisit the problem of classifying saddle-point geometries for the one-loop amplitude. We find an infinite family of complex saddles that dominate the high-energy regime. Using general constraints and matching to numerical data, we formulate a bootstrap problem that determines their multiplicities. This procedure yields a precise asymptotic expansion of the one-loop amplitude at high energies. The resulting oscillatory contributions lead to a much richer high-energy behavior than that predicted by the original Gross-Mende analysis.

We report on a broadband search for axion-like-particle (ALP) interactions using a radio-frequency-operated $^{87}\mathrm{Rb}$ atomic magnetometer. The instrument provides wide spectral coverage and sensitivity to an oscillating pseudomagnetic field that may be generated by the gradient coupling of the ALP field to the constituent fermions of atoms. We search for an ALP-gradient signature in the mass range $2.40\times10^{-10}\,\mathrm{eV}/c^{2}$--$2.11\times10^{-9}\,\mathrm{eV}/c^{2}$. No statistically significant signatures of an oscillating magnetic field are observed, and we derive upper limits on the corresponding ALP-proton, -neutron and -electron couplings, $g_{αpp}$, $g_{αnn}$ and $g_{αee}$, respectively. The result on $g_{αpp}$ improves over previous laboratory searches, while the limits on $g_{αnn}$ and $g_{αee}$ complement earlier laboratory searches and astrophysical bounds. The work extends searches for ALP-fermion interactions into a mass region largely unexplored in a dark-matter context, demonstrating the potential of our method for broadband axion-like particle searches targeting the Galactic dark-matter halo.

We investigate the effective potential and the string tension for the spatial Wilson loop (SWL) in hot dense QGP with two types of anisotropy, i.e. external magnetic field and spatial anisotropy, employing a holographic approach for the heavy quark model. In this approach, the string is extended in the 5th, holographic direction and has a turning point either on a dynamical wall (DW) configuration or on the horizon configuration in the 5th direction. We obtain the magnetic catalysis behavior for a phase transition between DW and horizon configuration of the string. The structure of the phase diagram does not depend on the boundary conditions choice for the dilaton field. Inclusion of the external magnetic field and spatial anisotropy enhance the string tension in the horizon configuration, namely drag force. For the spatially isotropic case $ν= 1$ at different magnetic field values the string tension is proportional to $T^2$ and is qualitatively consistent with lattice results. However, for the anisotropic case, $ν= 4.5$, it deviates from the quadratic term.

The calculation of precise predictions for Higgs decays is a necessary ingredient for determining Higgs properties at the LHC and future colliders. We compute all two- and three-body Higgs decays at next-to-leading order (NLO) in both QCD and electroweak interactions using the dimension-6 Standard Model Effective Field Theory (SMEFT). Results for four-body Higgs decays that are accurate to NLO QCD/electroweak order in the SMEFT are obtained using the narrow width approximation. Our results are contained in a flexible Monte Carlo program, NEWiSH, that is publicly available and we illustrate the impact of the NLO electroweak corrections for HL-LHC, Tera-Z, and Higgstrahlung projections.

We perform a general analysis of minimal modular fixed point models based on two right-handed neutrinos (2RHNs) and three modular fixed points, and find that the only viable possibilities are based on modular $S_4'$ and $A_5$ symmetry. Such models are highly predictive, with neutrino masses and the lepton mixing mixing matrix being fixed by three real parameters, as in the Littlest Seesaw Models. We perform an exhaustive scan over all possible models in this class and find many viable fixed points and modular form alignments, after confronting them with the latest neutrino oscillation global fits. The resulting models have the new feature that the two Dirac columns take more general forms than traditional Littlest Seesaw models, resulting in new sum rule relations between the solar and reactor angles, beyond those associated with TM1 (where the first column of the tri-bimaximal mixing matrix is preserved), which are compared to present and future projected JUNO results. We also compare the predictions of these models for the atmospheric angle and CP violating phase to current global fits and future T2HK and DUNE sensitivities.

We propose to discuss the modular flavor model and the stabilization of single modulus field in the Jordan frame supergravity with non-minimal scalar-curvature coupling of the form $Φ(τ,\barτ)R$. Modular invariance and positivity of the scale factor constrain stringently the form of the frame function, consequently the Kahler potential by the relation $Φ(τ,\barτ)=-3\exp[-K(τ,\barτ)/3]$. We discuss some general properties of scalar potentials after the scale transformation from the Jordan frame to the Einstein frame. We find that the shape of the resulting scalar potential in the Einstein frame is quite different from that of ordinary single modulus stabilization mechanism. The scalar potential could be stationary at the $i\infty$ fixed point, leading to a runaway type vacuum. We also discuss numerically the modulus stabilization for some simplified scenarios.

We perform for the first time a full study of spin correlations in inclusive WZ production at the LHC with leptonic decays in the presence of NLO QCD corrections and of effects from a dimension-six operator in the SMEFT modifying the electroweak triple-gauge coupling. We carry out the complete quantum-state tomography of the di-boson system and relate its results to common purity and spin-entanglement markers, highlighting the sizeable impact of both QCD corrections and SMEFT insertions. Additionally, we show how a naive truncation at dimension-six in the SMEFT expansion of the spin-density matrix can lead to a cumbersome spin interpretation of the quantum-tomography results.

We extend Nielsen's formulation of quantum circuit complexity to include intrinsically non-invertible operations. Such gates arise from fusion with topological defect operators and remove a basic limitation of symmetry-based circuits: the inability to change superselection sectors, or in two-dimensional CFTs, conformal families. We realise fusion operations as completely positive, trace-preserving quantum channels acting between sectors, with consistency ensured by the fusion and associator data of an underlying unitary modular tensor category. In contrast to standard Nielsen circuits, non-invertible circuits lead to an optimisation problem that is no longer governed by geodesics on a continuous group manifold but instead reduces to a discrete shortest-path problem on the fusion graph of superselection sectors. We illustrate the framework in representative rational conformal field theories. Finally, we interpret fusion-induced transitions as discrete changes in boundary stress-tensor data, corresponding to shock-like defects in AdS$_3$ gravity.

We report the recent progress from our group in extracting observables of both inclusive and exclusive semileptonic heavy-meson decays directly from lattice QCD four-point correlators. On the inclusive side, we illustrate how to estimate the systematic uncertainties from omitted higher-order terms and non-zero smearing of the kernel approximation, building on two important features of the Chebyshev expansion. On the exclusive side, we perform BCL parameterizations of the pseudoscalar to pseudoscalar form factors and compare the fitted coefficients with those from earlier results by HPQCD. We also perform a HQET-based parameterization of the P-wave form factors to shed new light on the 1/2-vs-3/2 puzzle. This work constitutes a step toward a unified lattice treatment of inclusive and exclusive semileptonic decays, relevant for the Vcb puzzle. In this study, we use lattice ensembles from the RBC/UKQCD collaboration for numerical investigations. Future developments from our group will focus on the control of other systematic effects for inclusive decays and investigations of other techniques with reduced statistical errors to extract exclusive contributions from lattice four-point correlators.

The ePIC collaboration is developing a multidetector system to explore the fundamental properties of the strong interaction at the future Electron-Ion Collider (EIC), to be built at Brookhaven National Laboratory. A key component of the ePIC detector is the Silicon Vertex Tracker (SVT), which provides high-precision tracking and microvertex reconstruction. The SVT consists of the Inner Barrel (IB), the Outer Barrel (OB), and the Forward/Backward Disks, all based on Monolithic Active Pixel Sensors (MAPS) that combine high granularity, low power consumption, and minimal material budget. This paper presents a concise overview of the SVT design and its development status.

Interference of the ultralight dark matter (ULDM) field generates time-varying gravitational potential fluctuations, which stochastically heat stellar systems embedded in ULDM halos. Small-sized stellar systems are therefore often used to set stringent constraints on ULDM. However, the evolution of systems with sizes well below the ULDM de Broglie wavelength remains poorly explored. Using numerical simulations, we show that the evolution of compact stellar systems in ULDM halos is governed by the interplay between internal stellar relaxation and ULDM-induced heating. We find the following main results. First, in sufficiently compact systems, relaxation-driven core collapse dominates, allowing the system to remain bound and dense, while ULDM-induced stripping of outer stars further accelerates the collapse. Second, in more extended systems, ULDM heating dominates and ultimately disrupts the system. Near the disruption threshold, we identify systems resembling ultra-faint dwarfs like Segue 1. Third, we further introduce a dimensionless parameter to quantify the relative importance of heating and relaxation and finally lead to an evolutionary phase diagram. Our results reveal the rich and nontrivial dynamics of compact stellar systems in ULDM halos, indicating that precise system modeling is essential for robust ULDM constraints.

In this letter, we construct an explicit rephasing transformation that converts an arbitrary unitary matrix into the Kobayashi--Maskawa (KM) parameterization and identify all independent CP phases in the mixing matrix as the arguments of its matrix elements. Furthermore, by applying this rephasing transformation to the fermion diagonalization matrices $U^{f}$, we show that the Majorana phases are represented by fermion-specific phases $δ^{ν, e}_{\rm KM}$ and their relative phases. In particular, by neglecting the 3-1 elements $U_{31}^{ν,e}$ of the diagonalization matrices for the two fermions, the KM phase $δ_{\rm KM}$ is concisely expressed by fermion-specific rephasing invariants involving two relative phases $δ_{\rm KM} = \arg \left [1 + ({U^{e * }_{21} U^ν_{21} / U^{e * }_{11} U^ν_{11} }) \right ] + \arg \left [ - { U_{32}^{e *} U_{32}^ν / U^{e * }_{22} U^ν_{22} } \right] $.

We study the near-threshold exclusive photoproduction of heavy vector mesons (quarkonia $J/ψ$ and $Υ$) off the proton within the framework of generalized parton distribution (GPD) factorization. The gluon GPDs are computed using a spectator model in which the proton emits a gluon and the remaining constituents are treated as a single spectator particle. Model parameters are determined by fitting the gluon unpolarized and helicity collinear parton distribution functions (PDFs) from global analyses. We compare our results with the latest near-threshold $J/ψ$ production data from the GlueX and $J/ψ$-007 experiments at Jefferson Laboratory, finding good agreement for both differential and total cross sections. Predictions are also provided for $Υ$ photoproduction, which can be tested at future Electron-Ion Colliders.

Sub-GeV dark matter is an appealing thermal target because it can still be produced via the standard freeze-out mechanism; at such low masses, achieving freeze-out naturally points to the presence of a light mediator, which shifts the most promising discovery avenues from the energy frontier to the intensity frontier. Realizing this picture is nonetheless challenging, since CMB observations tightly constrain energy injection from dark-matter annihilation at recombination and therefore strongly disfavor simple $s$-wave annihilation into visible Standard-Model final states. In this work, we propose a concrete neutrinophilic framework for sub-GeV thermal dark matter (''light WIMPs'') based on an additional gauge symmetry $\mathrm{U}(1)_{\mathrm{B}-\mathrm{L}+x\mathrm{Y}}$; for an appropriate choice of $x$, the new gauge boson couples predominantly to dark matter and neutrinos while its couplings to charged leptons are suppressed, so that sub-GeV dark matter annihilates almost exclusively into neutrinos, with hadronic modes kinematically closed. We map the parameter space in which the observed relic abundance is reproduced via standard thermal freeze-out in a conventional cosmological history, and show that sizable regions remain viable after imposing current cosmological, indirect-detection, and terrestrial constraints; in part of the allowed parameter space, the dark matter also exhibits sufficiently large self-interactions to potentially alleviate small-scale structure tensions.

We discuss in rather general terms the properties of space-like baryon transition form factors. In particular, we argue why these are necessarily complex-valued, what can be deduced from the respective phase motion and why dealing with real valued transition form factors in general leads to misleading results. For illustration the transition form factors for the Roper resonance as derived in the Jülich-Bonn-Washington framework are discussed.

Charged lepton flavor violation (CLFV) is forbidden in the Standard Model but predicted by many new physics models. We present searches for CLFV in charmonium decays using world-leading datasets collected by the BESIII detector. The processes $J/ψ\to eτ$, $J/ψ\to eμ$, and $ψ(3686)\to eμ$ are investigated using world-leading $J/ψ$ and $ψ(3686)$ dataset collected by BESIII. No significant signals are observed, and upper limits on branching fractions are set at $\mathcal{B}(J/ψ\to eτ)<7.5\times10^{-8}$, $\mathcal{B}(J/ψ\to eμ)<4.5\times10^{-9}$, and $\mathcal{B}(ψ(3686)\to eμ)<1.4\times10^{-8}$ at 90\% confidence level. These results provide constraints on Wilson coefficients in effective field theory and probe new physics at high energy scales.

We propose a generic axion--driven cogenesis scenario in which both the baryon asymmetry and dark matter abundance originate from the kinetic misalignment. The framework unifies the Peccei--Quinn (PQ) mechanism with a Type--I seesaw sector, where Hubble--induced masses and higher-dimensional PQ--violating operators drive early--time axion rotation. Working within the DFSZ axion model augmented by heavy neutrinos, we identify the parametric window of right-handed neutrino masses, determined by its decay rate, and the range of Hubble scales compatible with successful cogenesis, while maintaining the axion solution to the strong CP problem and satisfying current limits on axion isocurvature perturbations. Our results establish kinetic axion misalignment as a robust and predictive mechanism for axion cogenesis, independent of the inflationary microphysics.

We propose a new strategy to probe heavy neutrinos with non-universal fermion couplings at the Large Hadron Collider (LHC) using a novel production mechanism and machine-learning algorithms. Focusing on proton--proton collisions at $\sqrt{s} = 13.6~\mathrm{TeV}$, we investigate final states containing a charged lepton, missing transverse energy, and two jets. For heavy neutrino masses below $\mathcal{O}(1~\mathrm{TeV})$, production is dominated by the $s$ channel process. At higher masses, vector boson fusion becomes the dominant production mechanism, with cross sections that decrease slowly as the heavy neutrino mass increases. We simulate both signal and Standard Model background events and employ gradient-boosted decision trees to optimize event classification. Assuming an integrated luminosity of $3000~\mathrm{fb^{-1}}$, expected for the high-luminosity, and considering realistic statistical and systematic uncertainties, we find that heavy neutrinos in the mass range $50~\mathrm{GeV}$--$10~\mathrm{TeV}$ can be probed with sensitivity to the mixing parameter $|V_{\ell N}|^2$ spanning from $\mathcal{O}(10^{-5})$ to 1. This approach enhances the discovery potential for heavy neutrinos and provides a complementary pathway to existing search strategies.

We derive several new Bogomol'nyi (self-dual) equations in two-species $U(1)\times U(1)$ gauge theories governed by the Born--Infeld nonlinear electrodynamics. By identifying appropriate Born--Infeld type Higgs potentials, we show that the highly nonlinear energy functionals admit exact topological lower bounds saturated by coupled first-order equations. The resulting models accommodate both vortex-vortex and vortex-antivortex configurations and generalize previously known single-species Born--Infeld systems to interacting multi-component settings. Beyond the derivation of the Bogomol'nyi equations, we develop an exact thermodynamic theory for pinned multivortex configurations in both the full plane and compact doubly periodic domains. Owing to the linear dependence of the Bogomol'nyi energy spectrum on topological charges, we obtain closed-form expressions for the canonical partition function, internal energy, heat capacity, and magnetization. In compact domains, the Bradlow type geometric bounds constrain admissible vortex numbers and lead to qualitatively new high-temperature behavior. In particular, vortex-only systems exhibit spontaneous magnetization, while vortex-antivortex systems do not, reflecting the underlying symmetry between opposite topological charges. These results provide a rare analytically solvable framework for studying thermodynamics in nonlinear multi-component gauge theories regulated by the Born--Infeld electrodynamics.

It is known that a two-dimensional bosonic theory with a non-anomalous $\mathbb{Z}_2$ symmetry can be fermionized. Recent work shows that if the bosonic theory also has non-anomalous time-reversal symmetry, fermionization extends to non-orientable surfaces and yields a fermionic theory that depends on a $\mathrm{Pin}^-$ structure. Besides fermionization, one can define various topological manipulations, such as gauging and stacking invertible phases, which together generate a web of dualities. We prove that their group structure is the dihedral group $D_8$ of order 16. Furthermore, we systematically investigate the web from two perspectives: Symmetry TFT and actions on sectors of the $S^1$ Hilbert space.

On 2025 January 10, a thermonuclear (Type I) X-ray burst from the neutron star low-mass X-ray binary \textit{4U~0614+091} was detected with the ECLAIRs instrument on board the \textit{SVOM} mission. We present here a time-resolved spectroscopic analysis of the burst, along with the detection of burst oscillations within a 51-second interval during the decay phase. The oscillation frequency is measured to be $ν= 413.674 \pm 0.002\,\mathrm{Hz}$, consistent with previous reports. However, we detect a significant downward frequency drift over the burst duration, characterized by $\dotν = (-4.7 \pm 0.3) \times 10^{-3}\,\mathrm{Hz\,s^{-1}}$. This frequency evolution is atypical compared to those observed in similar burst oscillation sources. We tentatively attribute the observed drift to a Doppler shift induced by orbital motion. Under this interpretation, the inferred orbital period must be shorter than 20 minutes, placing \textit{4U~0614+091} among the most compact known low-mass X-ray binaries.

We utilize Gaia DR3 data to report the discovery of a new star cluster, Nasser 1, located 32 arcmin from Czernik 38 at coordinates $α= 282.11 \pm 0.05$ and $δ= 4.56 \pm 0.05$. Using a variable membership probability threshold technique with pyUPMASK, we confirm Nasser 1 as a genuine open cluster. It exhibits a distinct King profile and a well-defined CMD with an age of $125.0 \pm 12.30$ Myr, a distance modulus of $12.87 \pm 0.21$ mag, and a color excess of $2.42 \pm 0.09$. Nasser 1 shares consistent physical parameters (age, distance, kinematics, and reddening) with Czernik 38, suggesting they constitute a young primordial binary system in the Carina-Sagittarius spiral arm. Both clusters display elongations indicative of differential rotation tides; Nasser 1 is additionally perturbed by the spiral arm's gravitational field. Gaussian mass function analysis suggests the two were formerly a single cluster, violently torn apart by differential rotation.

Recent studies report a mild discrepancy between baryon acoustic oscillation (BAO) and cosmic microwave background (CMB) measurements within the $Λ$CDM framework. This discrepancy could be explained if the optical depth $τ$ inferred from the CMB large-scale E-mode polarization is underestimated, which may be biased by foreground-subtraction or instrumental systematics. In this work, we present a determination of $τ$ independent of the large-scale E-mode polarization, using the latest measurements of the redshift evolution of the neutral hydrogen fraction $x_\mathrm{HI}(z)$, which is constrained by Lyman-$α$ forest and damping-wing absorption measurements at $z\sim5$-$14$, based on ground-based optical and JWST observations. Combining $x_\mathrm{HI}(z)$ with the Planck CMB power spectra excluding the large-scale E-mode polarization, we obtain $τ=0.0552^{+0.0019}_{-0.0026}$, a stringent constraint consistent with the previous CMB results including the large-scale E-mode. We also evaluate a potential systematic error in our method associated with absorption modeling, obtaining $τ=0.0552^{+0.0075}_{-0.0049}$. Using this constraint on $τ$, we resolve the degeneracy in the $τ$-$Ω_m$ plane and find a $2.4σ$ tension with the DESI DR2 BAO results, thereby confirming the claimed mild discrepancy suggestive of physics beyond $Λ$CDM. Finally, we derive an upper limit on the sum of neutrino masses, $Σm_ν<0.0550\,(0.0717)$ eV at the 95% (99%) confidence level. This limit favors the normal mass ordering and, when combined with the lower limits from neutrino oscillation experiments, yields a further constraint, $Σm_ν=0.0594_{-0.0007}^{+0.0113}$ eV. However, the cosmological upper limit and the oscillation-based lower limit show a mild $2.2σ$ tension, providing an independent indication of possible physics beyond $Λ$CDM.

We investigate plasma wake generation via Compton scattering from photon bursts, a non-ponderomotive process relevant when the photon wavelength is smaller than the interparticle distance but larger than the Compton wavelength. In this regime, electrons can reach relativistic velocities. We extend linear theory to the nonlinear regime, showing that plasma waves can reach the wave-breaking limit. Perfectly collimated drivers produce wakes propagating at the speed of light, allowing electron phase-locking (limited by driver depletion). Non-collimated drivers induce subluminal phase velocities, limiting acceleration via dephasing. Two-dimensional simulations reveal unique transverse fields compared to laser wakefields, with a DC magnetic field leading to consistent focusing. The work considers observational prospects in laboratory and astrophysical scenarios such as around highly luminous compact objects (e.g., pulsars, gamma-ray bursts) interacting with tenuous interstellar or intergalactic plasmas, where conditions favor Comptondominated wakefield acceleration.

We present observations of jets within 2000 au of three deeply embedded protostars using 2.9-27 micron observations with JWST. These observations show the morphologies and kinematics of the collimated jets from three protostars, the low-mass Class 0 protostars B335 and HOPS 153, and the intermediate-mass protostar HOPS 370. These jets are traced by shock-ionized fine-structure line emission observed with the JWST NIRSpec and MIRI IFUs. We find that [Fe II] emission traces the full extent of the inner 1000 to 2000 au of the jets, depending on distance to the protostar, while other ions mostly trace isolated shocked knots. The jets show evidence of wiggling motion in the plane of the sky as well as asymmetries between blue and red-shifted lobes. The widths of the jets increase non-monotonically with distance from the central protostar, with opening angles ranging from 2.1 degrees to < 10.1 degrees for the three protostars in the sample. The jets have total velocities ranging from 147 to 184 km/s after correcting for disk inclination. For B335, an 8-month gap between NIRSpec and MIRI MRS observations enabled measurement of the tangential velocity of a shocked knot; in combination with the radial velocity, this shows that the jet has a different inclination than the outflow cavity. We find multiple knots before and during a recent outburst in B335, although the knots were more frequent during the burst. The asymmetries between blue- and red-shifted lobes strongly suggest complex interactions between the circumstellar disks and magnetic fields.

Formation of a rapidly spinning, strongly magnetized neutron star (NS) may occur in various classes of core-collapse events. If the NS injects an amount of energy comparable to the explosion energy of the accompanying supernova (SN) before the SN ejecta becomes transparent, the nascent NS wind bubble can overtake the outer ejecta and undergo a blowout driven by hydrodynamic instabilities. Based on multidimensional numerical studies, we construct a minimal semi-analytic framework to follow the post-blowout dynamics and radiative evolution, map the blowout conditions by scanning the ejecta and NS parameters, and compute survey-ready multi-band light curves. For stripped-envelope SNe with an ejecta mass of $M_\mathrm{ej} \sim 10\,M_\odot$ and an explosion energy of $E_\mathrm{sn} \sim 10^{51}\,\mathrm{erg}$, blowout occurs for NSs with magnetic field strengths of $B_{\mathrm{dip}} \gtrsim 10^{13}\,\mathrm{G}$ and spin periods of $P_\mathrm{NS} \lesssim \mathrm{a\ few}\,\mathrm{ms}$. Relatively weak-field cases with $B_\mathrm{dip} \lesssim 10^{14}\,\mathrm{G}$ produce luminous double-peaked UV/optical light curves, as observed in the superluminous SN LSQ14bdq, while stronger-field cases with $B_\mathrm{dip} \gtrsim 10^{14}\,\mathrm{G}$ result in hypernovae preceded by X-ray blowout precursors. We also examine weaker and lower-mass SN explosions representing ultra-stripped SNe and accretion- or merger-induced collapse events, in which blowout is more readily achieved over a broader range of NS parameters, producing fast X-ray transients with durations of $ 10^{2\mbox{--}4}\,\mathrm{s}$ and peak luminosities of $10^{42\mbox{--}48}\,\mathrm{erg\,s^{-1}}$. Our results encourage coordinated UV, optical, and X-ray observations which constrain the formation of the most energetic NSs in the universe.

We employ the MIT bag model with density-dependent bag constant for the equation of state (EOS) to estimate the gravitational mass and Keplerian frequency of rapidly rotating strange quark stars (SQS). In a companion paper we discuss the structural parameters of such rotating stars under the influence of strong magnetic fields. We use the LORENE library to compute the structural parameters at different rotational frequencies in the range of 1100-1300~Hz for a non-magnetized SQS. While there is no minimum limit for the mass of slowly rotating self-bound stars, by computing the maximum rotational frequency, known as the mass-shedding limit, we show that SQS must have a minimum mass to sustain high rotational frequencies. The mass-shedding frequency in our EOS model is lower than that estimated from the MIT bag model EOS with a fixed bag constant. The Keplerian frequency in our model depends linearly on the gravitational mass at the mass-shedding limit (and similarly on the minimum mass) with the slope of 0.08~${\rm kHz}/M_\odot$. We obtain mass limits aligned with the observational data for both the heaviest and the lightest observed pulsars.

We investigate the structural parameters of strange quark stars (SQS) under the influence of strong magnetic fields and varying rotational frequencies. The equation of state is computed using the MIT bag model with a density-dependent bag constant and considering the Landau quantization effect regarding the strong magnetic fields up to $5\times10^{17}\,$G in the interior of SQS. Employing the LORENE library, we calculate the structural parameters under different magnetic field strengths and rotational frequencies. Our models are compared in terms of maximum gravitational mass, deformation parameter, binding energy, and compactness. Our equation of state model demonstrates that the gravitational masses are higher than those computed using a MIT bag model with a fixed bag constant. We find the gravitational masses beyond $2.3 \,M_\odot$, which are compatible with the masses of observed compact objects, such as the ``black widow'' pulsar \emph{PSR J0952-0607}, and the \emph{GW190814} event detected by the LIGO/Virgo collaboration. The deformation parameter and maximum gravitational mass of SQS are characterized by fitted functions accounting for variations in both magnetic field strength and rotational frequency. We find the maximum deformation parameter of 1.55 and the maximum gravitational mass of $2.8\, M_\odot$ in the fast-rotating strongly magnetized model.

Ultraluminous X-ray sources (ULXs) are off-nuclear compact objects with apparent luminosities above 10^39 erg/s, often exceeding the Eddington limit for stellar-mass black holes. Beaming is a commonly invoked mechanism to explain their extreme brightness, and the dependence of the beaming factor on accretion rate is a critical parameter. In this work, we investigate how different beaming prescriptions affect the predicted properties of ULX populations. Using binary population synthesis, we construct synthetic X-ray luminosity functions (XLFs) for both classical and log-modified beaming models at solar and sub-solar metallicities. The classical model predicts a larger intrinsic number of bright ULXs, but strong beaming reduces their observable fraction, resulting in fewer visible ULXs compared to the log-modified model. The log-modified prescription yields a shallower slope at high-luminosity, aligning better with observed XLFs, and increases the fraction of observable neutron star ULXs above 10^39 erg/s. These results underscore the significant role of the beaming law in shaping ULX statistical distributions and assessing neutron star contributions to the population.

We present a catalog of 4539 fast radio bursts (FRBs) observed with the Canadian Hydrogen Intensity Mapping Experiment (CHIME) telescope between 25 July 2018 and 15 September 2023. These bursts originate from 3641 unique sources, including 981 bursts from 83 known repeating sources. For each FRB, the catalog provides a $O(10')$ estimate of sky location along with corresponding measurements of cumulative exposure time and survey sensitivity over the observing period. It includes a total-intensity dynamic spectrum between 400 and 800 MHz at 0.983 ms resolution. From this spectrum, we constrain a model of the burst morphology and measure key parameters such as arrival time, intrinsic temporal width, dispersion measure, scattering time, and flux density. This second catalog includes all FRBs from the first catalog, with every event reprocessed using a uniform and improved analysis framework. We show that previously published inferences remain valid under the updated measurements. We assess consistency of the detection rate across observational parameters, present initial distributions of burst properties, and outline ongoing and future studies that will use this catalog to investigate the nature of FRBs and their utility as astrophysical and cosmological probes.

In the interstellar medium, cosmic rays (CRs) generate a field of ultraviolet (UV) photons via the excitation and subsequent radiative decay of $\rm H_2$ molecules. This UV field is a major agent of ionization and dissociation in the inner regions of molecular clouds that are shielded from the effects of the interstellar radiation field. In particular, the dissociation of $\rm H_2$, by far the most abundant molecule in interstellar clouds, leads to the production of atomic hydrogen which then takes part in the production of a multitude of molecules, in particular complex organics on the surfaces of interstellar dust grains. Precise knowledge of the rates of CR-induced dissociation processes is thus crucial for constructing reliable chemical models. For the present paper, we have derived a new value of $k_{\rm diss, CR}(\mbox{$\rm H_2$})=0.831ζ$ for the rate of $\rm H_2$ dissociation, where $ζ$ is the CR ionization rate of $\rm H_2$. This prediction contrasts a previous value from the Leiden database which overestimated the rate due to an inconsistent treatment of the $\rm H_2$ abundances and photodissociation cross sections. By running a series of chemical models, we show that the overestimated dissociation rate has a large effect on the results of chemical simulations, with the abundance of methanol being overestimated by over one order of magnitude. Hence, we strongly recommend the adoption of our new estimate $k_{\rm diss, CR}(\mbox{$\rm H_2$})=0.831ζ$ in all chemical models that include this process. Our newly derived value corresponds to $\rm H_2$ being purely in the para form ($J^{\prime\prime} = 0$). However, in the interiors of molecular clouds the $\rm H_2$ ortho-to-para ratio is low and using the rate for para-$\rm H_2$ is an adequate approximation.

One of the key challenges in strong gravitational lensing cosmography is the accurate measurement of time delays between multiple lensed images, which are essential for constraining the Hubble constant ($H_0$). We investigate how lens mass-profile assumptions affect time delays. Specifically, we implement a Broken Power Law (BPL) mass model within the Lenstronomy framework (Birrer & Amara 2018), which introduces additional flexibility in the radial mass distribution and can phenomenologically capture deviations from a single power-law profile. This model is combined with a numerical approach to compute time delays at the image positions. We validate the BPL implementation using simulated lenses and compare the results with those obtained from the elliptical power-law (EPL) model. We then apply both model families to the quadruply imaged quasar WGD~2038-4008. Both models fit the imaging and kinematic data comparably well, yet the greater radial freedom in the BPL model shifts the inferred time-delay distance -- and thus $H_0$ -- by an amount comparable to the current discrepancy between early- and late-universe $H_0$ estimates. In a flat $Λ$CDM cosmology, the $H_0$ inferred using the BPL lens model is $75^{+23.1}_{-16.3} \ \mathrm{km \ s^{-1} \ Mpc^{-1}},$ while the EPL model gives $H_0 = 61^{+19.2}_{-13.2} \ \mathrm{km \ s^{-1} \ Mpc^{-1}}.$ This difference is largely due to uncertainties in the inner mass profile ($θ<0.2''$), a region where point spread function (PSF) reconstruction is a critical factor -- a finding consistent with results reported in Shajib et al. (2022). This highlights how time-delay cosmography remains sensitive to assumptions about the lens mass profile. With current precision, this difference does not favor one cosmological scenario over another, but rather underscores the importance of flexible mass modeling and PSF modeling.

Extragalactic jets are broadly classified into two categories based on radio observations: core-brightened jets, known as Fanaroff-Riley Type I (FR I), and edge-brightened jets, classified as Type II (FR II). This FR dichotomy may arise due to variation in the ambient medium and/or the properties of the jet itself, such as injection speed, temperature, composition, magnetization, etc. To investigate this, we perform large-scale three-dimensional magnetohydrodynamic (3D-MHD) simulations of low-power, supersonic jets extending to kiloparsec scales. We inject a jet beam carrying an initially toroidal magnetic field into a denser, unmagnetized, and stratified ambient medium through a cylindrical nozzle. Our simulations explore jets with varying injection parameters to investigate their impact on morphology and emission properties. Furthermore, we examine jets with significantly different plasma compositions, such as hadronic and mixed electron-positron-proton configurations, to study the conditions that may drive transitions between FR I and FR II morphologies. We find that, under the same injection parameters, mixed plasma composition jets tend to evolve into FR I structures. In contrast, electron-proton jets exhibit a transition between FR I and FR II morphologies at different stages of their evolution.

This paper presents Owl-z, a Bayesian code aiming at identifying z \geq 7 quasars in wide field optical and near-infrared surveys. By construction,the code can also be used to select objects that contaminate the high-z quasar population, i.e. brown dwarfs and early-type galaxies at intermediate redshifts. The code can be adapted for the selection of high-z galaxies, and although it has been tuned to the Euclid Wide Survey, it can be easily adapted to other photometric surveys. The code input data are the object's photometric data and its galactic longitude and latitude, and the code output data are the probabilities of the modelled populations of high-z quasars, brown dwarfs and early-type galaxies at intermediate redshift. As part of the validation, Owl-z could re-identify all spectroscopically confirmed quasars at z \geq 7, demonstrating the code's versatility in applying to different photometric catalogues. The performance of Owl-z, based on a metric combining completeness and purity called F-measure, is analysed in the case of Euclid using simulated data in a wide range of redshifts (7 \leq z \leq 12) and H-band Euclid magnitudes (18 \leq H_E \leq 24.5). The results show that Owl-z reaches full performance for bright sources (H_E \lesssim 22), independently of the redshift. We show that the probability threshold used to select promising quasar candidates can be adjusted after processing to fine-tune the F-measure value of candidates depending on their magnitude and redshift estimates. We show that for objects brighter than about two magnitudes above the survey detection limit, Owl-z provides a classification that will facilitate the optimisation of photometric and spectroscopic confirmation campaigns. In conclusion, Owl-z is a powerful public tool to help select high-z quasars, brown dwarfs or early-type galaxies at intermediate redshifts in Euclid or other wide-field surveys.

The superorbital modulations (SMs) observed in wind-fed X-ray binaries remain a puzzling phenomenon in astrophysics. To investigate this behavior observationally, we analyzed the long-term hard X-ray light curve from the Swift/BAT 157-Month Hard X-ray Survey in X-ray binary GX 301-2. Using three timing analysis methods--the Lomb-Scargle periodogram, the weighted wavelet Ztransform, and Gaussian processes--we identify a rare dual-SM behavior in this source: the 115-day modulation exceeds the 5$σ$ global significance level, whereas the 65-day signal only marginally reaches the 4$σ$ level. Because the 115-day period is more consistent with the previously reported linear relation between orbital and superorbital periods, we interpret 115 days as the actual superorbital period, while the weaker and less stable 65-day period is its beat modulation with the orbital period.By assessing the applicability of different physical scenarios to our results, we suggest that this dual-SM behavior is most plausibly associated with corotating interaction regions (CIRs) in the stellar wind. This framework can also account for the observed linear orbital-superorbital relation, despite the unclear physical mechanism that sets the apparent ratio between the CIR and orbital periods across sources. Further long-term monitoring of this system, together with continued theoretical development of the CIR scenario, will be essential for clarifying the origin of wind-fed SMs.

Statistically, green valley (GV) galaxies exhibit lower molecular gas fractions ($f_{gas}$) and reduced star formation efficiency (SFE) compared to star-forming galaxies. However, it remains unclear whether quenching is primarily driven by one factor or results from a combination of mechanisms in individual GV galaxies. In this study, we address this question by examining the spatial distributions of star formation and molecular gas in 28 GVs selected from the ALMaQUEST survey and additional literature samples. For each galaxy, we identify regions with suppressed specific star formation rate (sSFR) and measure $Δf_{gas}$ and $Δ$SFE-offsets from the resolved scaling relations of the star-forming main sequence galaxies. By comparing the fraction of regions with negative $Δf_{gas}$ and $Δ$SFE, we classify 35.7$\pm$13.2\% (57.1$\pm$17.9\%) of GV galaxies as $f_{gas}$-driven, 39.3$\pm$14.0\% (39.3$\pm$14.0\%) as SFE-driven, and 25.0$\pm$10.6\% (3.6$\pm$3.6\%) as mixed mode when adopting a fixed (variable) CO-to-$\rm H_{2}$ conversion factor ($α_{CO}$). These results indicate that GVs undergo quenching through multiple pathways. As sSFR decreases from the main sequence to the green valley, we observe a transition toward predominantly SFE-driven quenching, possibly linked to internal processes such as morphological quenching or AGN activity. We further estimate the quenching timescale ($τ_{decay}$), defined as the time from the peak SFR to 1/e (approximately 37\%) of its value, using integrated MaNGA spectra. SFE-driven quenching is typically associated with short $τ_{decay}$ , while $f_{gas}$-driven quenching shows a broader range. Overall, 75\% of GVs exhibit $τ_{decay}$ shorter than 1 Gyr, suggesting that quenching in most GVs proceeds rapidly, challenging purely slow-quenching scenarios like starvation.

We present new sub-arcsecond ($\sim$0.3-1 arcsec; $\sim$1--3\,kpc) VLA imaging at 1.4\,GHz and 6\,GHz of 29 optically-selected, [O~{\sc iii}] luminous ($L_{\rm [O III]}$ > 10$^{42.1}$\,erg\,s$^{-1}$), $z<0.2$ quasars drawn from the expanded Quasar Feedback Survey (QFeedS; with $L_\mathrm{1.4\,GHz} = 10^{22.6}$--10$^{26.3}$\,W\,Hz$^{-1}$). These 29 new objects occupy the low end of the radio-power distribution ($L_\mathrm{1.4\,GHz}$=$10^{22.63}$--10$^{23.45}$\,W\,Hz$^{-1}$) in the QFeedS sample and are nominally `radio quiet'. Despite this, we find widespread evidence of AGN-driven synchrotron activity. Nearly $\sim 31\,$per\,cent exhibit resolved radio structures on $\sim$0.1--20\,kpc scales consistent with compact jets or wind-driven outflows, and $\sim 90\,$per\,cent display steep spectra ($α\lesssim -1$) indicative of optically thin synchrotron emission. Combining morphology, spectral index and brightness-temperature diagnostics, at least $\sim38\,$per\,cent of the sample show clear AGN signatures that cannot be explained by star formation alone. These constitute the first results from the expanded QFeedS (now 71 quasars spanning $\approx 4$ dex in radio power) and demonstrate that compact, low-power jets and AGN shocks are common deep inside the radio-quiet regime. A thorough understanding of feedback processes from quasars, deep into the `radio-quiet' regime, will be obtained by connecting these high resolution radio observations with multi-wavelength observations.

Ultracompact dwarf galaxies (UCDs) are thought to be remnants of stripped galactic nuclei, among which a handful are known to host a central supermassive black hole (SMBH). As in stripped nuclear star clusters, the SMBHs in UCDs may be fed by stellar winds from old stellar populations, in the absence of substantial gas reservoirs and galactic inflows. In this work, we investigate such a wind-fed accretion scenario for M60-UCD1, which harbors a confirmed $2\times10^7~M_\odot$ SMBH and exhibits X-ray emission suggestive of SMBH accretion signature. Using three-dimensional hydrodynamical simulations, we simulate the SMBH accreting stellar winds from approximately 1500 asymptotic giant branch stars, and explore the role of ram pressure from the ambient interstellar or intracluster medium. After 5 Myr, the majority of the stellar winds form a cold gas disk ($\sim1000~M_\odot$) within $\sim10~\rm pc$ as well as the SMBH's gravitational sphere of influence. Within the inner $10^4~r_{\rm g}$, this disk transitions into a hot ($\sim10^7-10^9~\rm K$), geometrically thick corona that dominates the X-ray emission. The SMBH achieves an accretion rate of $\sim10^{-5}~M_\odot~\rm yr^{-1}$, yielding an X-ray luminosity of $\sim7\times10^{37}~\rm erg~s^{-1}$, well consistent with observations. Including ram pressure stripping reduces both the accretion rate and luminosity by about a factor of two. Our results suggest that the X-ray counterpart of M60-UCD1 originates from a weakly accreting SMBH fed by stellar winds, with broader insights into the feeding mechanisms of central massive black holes and the origins of X-ray sources in other UCDs.

We present a model for the intrinsic alignment contamination of the shear three-point correlation function and skewness of the mass aperture statistic using the tidal alignment and tidal torquing (TATT) formalism. We compute the intrinsic alignment bispectra components in terms of the TATT model parameters. We consider two effective field theory approaches in the literature, relate them to the TATT model parameters and an extension to TATT that includes the velocity-shear (VS) parameter. We compare the impact of changing between NLA, TATT, and TATT+VS on the theoretical computation of the 3PCF using the best fit parameters and tomographic redshift distributions from Dark Energy Survey Year 3. We find that the TATT model significantly impacts the skewed triangle configurations of the 3PCF. Additionally, including the higher-order effects from TATT can introduce opposite effects on the two-point function and on the mass aperture skewness, damping the signal of the former while boosting the signal of the latter. We argue that a joint 2PCF+3PCF analysis with the TATT model can help break the degeneracy between its model parameters and provide more robust constraints on both cosmology and intrinsic alignment amplitude parameters. We show that typical values of order unity for the intrinsic alignment parameters introduce differences of around $10\%$ between NLA and TATT predictions.

We provide a characterization of the red supergiant (RSG) progenitor candidate for the nearby Type II-plateau supernova (SN) 2025pht in NGC 1637. The star was first detectable in 2001 by the Hubble Space Telescope (HST) and then again in a dozen bands by the James Webb Space Telescope (JWST) in 2024. This "quasi-snapshot" of the star's nature almost immediately prior to explosion is unprecedented. The RSG varied in brightness, and we posit that it could have been a pulsating variable, possibly with a long period of ~660 days. The largest uncertainty is the host-galaxy distance, which we establish to be 10.73+/-1.76 Mpc. The star was also heavily extinguished by interstellar dust internal to the host, with visual extinction A_V(host)~1.7 mag (total A_V(tot)~1.8 mag). Dust radiative-transfer modeling reveals the star's circumstellar medium to be quite dusty and silicate-rich, yielding a bolometric luminosity log(L_bol/L_Sun)=5.08+/-0.16 and a cool effective temperature T_eff=2100--2500 K. The available HST optical data had no bearing on the shape of the candidate's observed spectral energy distribution -- for the first time, without the archival JWST observations we would not have been able to detect and characterize the candidate at all. The SN 2025pht progenitor candidate, although quite similar to that of SN 2023ixf, may be the most luminous candidate identified to date.

In low-mass core-collapse supernova (CCSN) progenitors, nuclear burning beyond oxygen can become explosive under degenerate conditions, triggering eruptive mass loss before the final explosion. We investigate such pre-SN eruptions using \texttt{SNEC} hydrodynamic simulations and realistic stellar models, parameterizing the nuclear energy deposition as a fraction of the binding energy of the combined He layer and H-rich envelope. For the lowest-mass model (9 $M_\odot$), the ejecta mass ($M_{\rm ej}$) scales with the energy gained by the H-rich envelope via a power law (index$\sim$3.5). Across 9-10 $M_\odot$, this relation shows limited scatter within a factor of $\sim$2.6, enabling an estimation of the gained energy from $M_{\rm ej}$. The shock passage also flattens the bound envelope, which can affect the SN light curve morphology and provide another diagnostic for the eruption. Then, we compute the associated precursor light curves for the 9 $M_\odot$ model with the multi-group radiative-transfer code \texttt{STELLA}. These signals are typically faint, with bolometric luminosities of $\sim10^{39}$ erg s$^{-1}$ lasting hundreds of days. Their cool black-body spectra make them brighter in the infrared, yet several magnitudes fainter than observed pre-SN precursors at the threshold for full envelope ejection. To aid future studies, we make our post-eruption stellar profiles and precursor light curves publicly available.

Relativistic magnetic reconnection is a key mechanism for dissipating magnetic energy and accelerating particles in astrophysics. In the absence of radiative cooling, recent particle-in-cell (PIC) simulations have shown that high-energy particles gain most of their energy in the upstream region, during a short-lived "free phase" where they meander between the two sides of the layer; when they get captured/trapped by the downstream flux ropes, they undergo a "trapped phase", where no significant energization occurs. Here, we perform a suite of 3D PIC simulations of relativistic reconnection including inverse Compton (IC) losses in the weakly cooled regime in which the radiation-reaction-limited Lorentz factor $γ_{\rm rad}$ exceeds the magnetization $σ$. We show that electron cooling losses do not appreciably alter the reconnection rate, the structure of the layer, and the physics of particle acceleration in the free phase, so the spectrum of free electrons is $dN_{\rm free}/dγ\propto γ^{-1}$, as in the uncooled case. The spectrum of trapped electrons above the cooling break $γ_{\rm cool}$ (in the range $γ_{\rm cool}<γ<γ_{\rm rad}$) is $dN/dγ\propto γ^{-3}$, steeper than the scaling $dN/dγ\propto γ^{-2}$ of uncooled simulations. This confirms that no significant particle energization occurs during the trapped phase. Our results validate the model by arXiv:2302.12269 for particle acceleration in 3D relativistic reconnection, and imply that radiative emission models of reconnection-powered astrophysical sources should employ a two-zone structure, that differentiates between free, rapidly accelerating particles and trapped, passively cooling particles.