arXiv Daily Digest - 2026-01-21

Contrastive language-audio pretraining (CLAP) has achieved notable success in learning semantically rich audio representations and is widely adopted for various audio-related tasks. However, current CLAP models face several key limitations. First, they are typically trained on relatively small datasets, often comprising a few million audio samples. Second, existing CLAP models are restricted to short and fixed duration, which constrains their usage in real-world scenarios with variable-duration audio. Third, the standard contrastive training objective operates on global representations, which may hinder the learning of dense, fine-grained audio features. To address these challenges, we introduce Scalable Language-Audio Pretraining (SLAP), which scales language-audio pretraining to 109 million audio-text pairs with variable audio durations and incorporates multiple training objectives. SLAP unifies contrastive loss with additional self-supervised and captioning losses in a single-stage training, facilitating the learning of richer dense audio representations. The proposed SLAP model achieves new state-of-the-art performance on audio-text retrieval and zero-shot audio classification tasks, demonstrating its effectiveness across diverse benchmarks.

We address the following question: given a collection $\{\mathbf{A}^{(1)}, \dots, \mathbf{A}^{(N)}\}$ of independent $d \times d$ random matrices drawn from a common distribution $\mathbb{P}$, what is the probability that the centralizer of $\{\mathbf{A}^{(1)}, \dots, \mathbf{A}^{(N)}\}$ is trivial? We provide lower bounds on this probability in terms of the sample size $N$ and the dimension $d$ for several families of random matrices which arise from the discretization of linear Schrödinger operators with random potentials. When combined with recent work on machine learning theory, our results provide guarantees on the generalization ability of transformer-based neural networks for in-context learning of Schrödinger equations.

Multimodal Large Language Models (MLLMs) are increasingly used to interpret visualizations, yet little is known about why they fail. We present the first systematic analysis of barriers to visualization literacy in MLLMs. Using the regenerated Visualization Literacy Assessment Test (reVLAT) benchmark with synthetic data, we open-coded 309 erroneous responses from four state-of-the-art models with a barrier-centric strategy adapted from human visualization literacy research. Our analysis yields a taxonomy of MLLM failures, revealing two machine-specific barriers that extend prior human-participation frameworks. Results show that models perform well on simple charts but struggle with color-intensive, segment-based visualizations, often failing to form consistent comparative reasoning. Our findings inform future evaluation and design of reliable AI-driven visualization assistants.

Reproducibility of computational results remains a challenge in materials science, as simulation workflows and parameters are often reported only in unstructured text and tables. While literature data are valuable for validation and reuse, the lack of machine-readable workflow descriptions prevents large-scale curation and systematic comparison. Existing text-mining approaches are insufficient to extract complete computational workflows with their associated parameters. An ontology-driven, large language model (LLM)-assisted framework is introduced for the automated extraction and structuring of computational workflows from the literature. The approach focuses on density functional theory-based stacking fault energy (SFE) calculations in hexagonal close-packed magnesium and its binary alloys, and uses a multi-stage filtering strategy together with prompt-engineered LLM extraction applied to method sections and tables. Extracted information is unified into a canonical schema and aligned with established materials ontologies (CMSO, ASMO, and PLDO), enabling the construction of a knowledge graph using atomRDF. The resulting knowledge graph enables systematic comparison of reported SFE values and supports the structured reuse of computational protocols. While full computational reproducibility is still constrained by missing or implicit metadata, the framework provides a foundation for organizing and contextualizing published results in a semantically interoperable form, thereby improving transparency and reusability of computational materials data.

We present Semantic Fusion (SF), a formal framework for decentralized semantic coordination in multi-agent systems. SF allows agents to operate over scoped views of shared memory, propose structured updates, and maintain global coherence through local ontology-based validation and refresh without centralized control or explicit message passing. The central theoretical result is a bisimulation theorem showing that each agent's local execution is behaviorally equivalent to its projection of the global semantics, in both deterministic and probabilistic settings. This enables safety, liveness, and temporal properties to be verified locally and soundly lifted to the full system. SF supports agents whose update proposals vary across invocations, including those generated by learned or heuristic components, provided updates pass semantic validation before integration. We establish deterministic and probabilistic guarantees ensuring semantic alignment under asynchronous or degraded communication. To validate the model operationally, we implement a lightweight reference architecture that instantiates its core mechanisms. A 250-agent simulation evaluates these properties across over 11,000 validated updates, demonstrating convergence under probabilistic refresh, bounded communication, and resilience to agent failure. Together, these results show that Semantic Fusion can provide a formal and scalable basis for verifiable autonomy in decentralized systems.

Progress has led to a detailed understanding of the neural mechanisms that underlie decision making in primates. However, less is known about why such mechanisms are present in the first place. Theory suggests that primate decision making mechanisms, and their resultant behavioral abilities, emerged to maximize reward in the face of noisy, temporally evolving information. To test this theory, we trained an end-to-end deep recurrent neural network using reinforcement learning on a noisy perceptual discrimination task. Networks learned several key abilities of primate-like decision making including trading off speed for accuracy, and flexibly changing their mind in the face of new information. Internal dynamics of these networks suggest that these abilities were supported by similar decision mechanisms as those observed in primate neurophysiological studies. These results provide experimental support for key pressures that gave rise to the primate ability to make flexible decisions.

Artificial Intelligence is moving from models that only generate text to Agentic AI, where systems behave as autonomous entities that can perceive, reason, plan, and act. Large Language Models (LLMs) are no longer used only as passive knowledge engines but as cognitive controllers that combine memory, tool use, and feedback from their environment to pursue extended goals. This shift already supports the automation of complex workflows in software engineering, scientific discovery, and web navigation, yet the variety of emerging designs, from simple single loop agents to hierarchical multi agent systems, makes the landscape hard to navigate. In this paper, we investigate architectures and propose a unified taxonomy that breaks agents into Perception, Brain, Planning, Action, Tool Use, and Collaboration. We use this lens to describe the move from linear reasoning procedures to native inference time reasoning models, and the transition from fixed API calls to open standards like the Model Context Protocol (MCP) and Native Computer Use. We also group the environments in which these agents operate, including digital operating systems, embodied robotics, and other specialized domains, and we review current evaluation practices. Finally, we highlight open challenges, such as hallucination in action, infinite loops, and prompt injection, and outline future research directions toward more robust and reliable autonomous systems.

Category-Partition is a functional testing technique that is based on the idea that the input domain of the system under test can be divided into sub-domains, with the assumption that inputs that belong to the same sub-domain trigger a similar behaviour and that therefore it is sufficient to select one input from each sub-domain. Category-Partition proceeds in several steps, from the identification of so-called categories and choices, possibly constrained, which are subsequently used to form test frames, i.e., combinations of choices, and eventually test cases. This paper reports on an ongoing attempt to automate as many of those steps as possible, with graphical-user interface tool support. Specifically, the user interface allows the user to specify parameters as well as so-called environment variables, further specify categories and choices with optional constraints. Choices are provided with precise specifications with operations specific to their types (e.g., Boolean, Integer, Real, String). Then, the tool automates the construction of test frames, which are combinations of choices, according to alternative selection criteria, and the identification of input values for parameters and environment variables for these test frames, thereby producing test cases. The paper illustrates the capabilities of the tool with the use of nine different case studies.

Future direct-imaging flagship missions, such as NASA's Habitable Worlds Observatory (HWO), face critical decisions in prioritizing observations due to extremely stringent time and resource constraints. In this paper, we introduce two advanced machine-learning architectures tailored for predicting biosignature species fluxes from exoplanetary reflected-light spectra: a Bayesian Convolutional Neural Network (BCNN) and our novel model architecture, the Spectral Query Adaptive Transformer (SQuAT). The BCNN robustly quantifies both epistemic and aleatoric uncertainties, offering reliable predictions under diverse observational conditions, whereas SQuAT employs query-driven attention mechanisms to enhance interpretability by explicitly associating spectral features with specific biosignature species. We demonstrate that both models achieve comparably high predictive accuracy on an augmented dataset spanning a wide range of exoplanetary conditions, while highlighting their distinct advantages in uncertainty quantification and spectral interpretability. These capabilities position our methods as promising tools for accelerating target triage, optimizing observation schedules, and maximizing scientific return for upcoming flagship missions such as HWO.

Large language models (LLMs) can recall a wide range of factual knowledge across languages. However, existing factual recall evaluations primarily assess fact retrieval in isolation, where the queried entity is explicitly named and the fact is requested directly. In natural language use, facts are often accessed through context, where the relevant entity is introduced only indirectly. In this work, we study contextually mediated factual recall, asking whether LLMs can reliably retrieve factual knowledge when the target entity is embedded in a naturalistic context rather than queried explicitly, across languages. We construct controlled prompts that preserve the underlying fact while introducing referential mediation through contextual sentences. To disentangle contextual effects from name-specific associations, we further compare performance using synthetic names and real names across languages. Evaluating multiple model families in five languages, we find that contextual mediation consistently degrades factual recall, with substantial variation across relations. Larger models are more robust to contextual mediation, exhibiting a reduced performance gap relative to direct queries, while the effect of real names and name origin is mixed and unsystematic. These findings highlight a gap between isolated factual recall and context-dependent language understanding in multilingual LLMs.

Artificial Intelligence is rapidly transforming materials science and engineering, offering powerful tools to navigate complexity, accelerate discovery, and optimize material design in ways previously unattainable. Driven by the accelerating pace of algorithmic advancements and increasing data availability, AI is becoming an essential competency for materials researchers. This review provides a comprehensive and structured overview of the current landscape, synthesizing recent advancements and methodologies for materials scientists seeking to effectively leverage these data-driven techniques. We survey the spectrum of machine learning approaches, from traditional algorithms to advanced deep learning architectures, including CNNs, GNNs, and Transformers, alongside emerging generative AI and probabilistic models such as Gaussian Processes for uncertainty quantification. The review also examines the pivotal role of data in this field, emphasizing how effective representation and featurization strategies, spanning compositional, structural, image-based, and language-inspired approaches, combined with appropriate preprocessing, fundamentally underpin the performance of machine learning models in materials research. Persistent challenges related to data quality, quantity, and standardization, which critically impact model development and application in materials science and engineering, are also addressed.

Multilingual Large Language Models (LLMs) exhibit remarkable cross-lingual abilities, yet often exhibit a systematic bias toward the representations from other languages, resulting in semantic interference when generating content in non-English languages$-$a phenomenon we define as language spilling. This paper presents a novel comparative framework for evaluating multilingual semantic robustness by systematically measuring how models handle polysemous words across languages. Our methodology provides a relative measure of model performance: when required to generate exactly five meanings, both strong and weak models may resort to meanings from dominant languages, but semantically stronger models do so later in the generation sequence, producing more true meanings from the target language before failing, while weaker models resort to dominant-language meanings earlier in the sequence. We evaluate a diverse set of open and closed multilingual LLMs using a structured meaning generation task across nine languages, employing a carefully curated benchmark of 100 high-polysemy English words. Our findings reveal significant variation in semantic robustness across both models and languages, providing a principled ranking system for model comparison without requiring definitive causal attribution of error sources. We contribute both a scalable comparative benchmark for multilingual semantic evaluation and a rigorous validation pipeline$-$critical tools for developing more linguistically balanced AI systems.

Most clinical AI systems operate as prediction engines -- producing labels or risk scores -- yet real clinical reasoning is a time-bounded, sequential control problem under uncertainty. Clinicians interleave information gathering with irreversible actions, guided by regret, constraints and patient values. We argue that the dominant computational substrate of clinician reasoning is not cardinal optimization but ordinal, non-compensatory decision-making: Clinicians frequently rely on fast-and-frugal, lexicographic heuristics (e.g., fast-and-frugal trees) that stop early after checking a small, fixed sequence of cues. We provide a normative rationale for why such algorithms are not merely bounded rationality shortcuts, but can be epistemically preferred in medicine. First, many clinical trade-offs are constructed through human judgment and are only weakly measurable on absolute scales; without strong measurement axioms, only orderings are invariant, motivating an ordinal-by-default stance. Second, preference and signal elicitation are structurally crude: The mapping from truth $\to$ perception $\to$ inference $\to$ recorded variables introduces layered noise, leaving a persistent uncertainty floor. When this 'crudeness' overwhelms the decision margin, plug-in expected-utility optimization becomes brittle (high flip probability under small perturbations), whereas robust dominance/filtering rules ($ε$-dominance, maximin) stabilize decisions.Finally, we outline a clinician-aligned AI blueprint: Use rich models for beliefs and trajectories, but choose actions through robust ordinal rules; treat heuristics as the low-dimensional special case; and deploy AI as 'selective complexity' -- invoked mainly for tie-breaking when decisions are fragile and information has positive expected impact.

We study the online centralized charging scheduling problem (OCCSP). In this problem, a central authority must decide, in real time, when to charge dynamically arriving electric vehicles (EVs), subject to capacity limits, with the objective of balancing load across a finite planning horizon. To solve the problem, we first gamify it; that is, we model it as a game where charging blocks are placed within temporal and capacity constraints on a grid. We design heuristic policies, train learning agents with expert demonstrations, and improve them using Dataset Aggregation (DAgger). From a theoretical standpoint, we show that gamification reduces model complexity and yields tighter generalization bounds than vector-based formulations. Experiments across multiple EV arrival patterns confirm that gamified learning enhances load balancing. In particular, the image-to-movement model trained with DAgger consistently outperforms heuristic baselines, vector-based approaches, and supervised learning agents, while also demonstrating robustness in sensitivity analyses. These operational gains translate into tangible economic value. In a real-world case study for the Greater Montréal Area (Québec, Canada) using utility cost data, the proposed methods lower system costs by tens of millions of dollars per year over the prevailing practice and show clear potential to delay costly grid upgrades.

Artificial intelligence systems for scientific discovery have demonstrated remarkable potential, yet existing approaches remain largely proprietary and operate in batch-processing modes requiring hours per research cycle, precluding real-time researcher guidance. This paper introduces Deep Research, a multi-agent system enabling interactive scientific investigation with turnaround times measured in minutes. The architecture comprises specialized agents for planning, data analysis, literature search, and novelty detection, unified through a persistent world state that maintains context across iterative research cycles. Two operational modes support different workflows: semi-autonomous mode with selective human checkpoints, and fully autonomous mode for extended investigations. Evaluation on the BixBench computational biology benchmark demonstrated state-of-the-art performance, achieving 48.8% accuracy on open response and 64.5% on multiple-choice evaluation, exceeding existing baselines by 14 to 26 percentage points. Analysis of architectural constraints, including open access literature limitations and challenges inherent to automated novelty assessment, informs practical deployment considerations for AI-assisted scientific workflows.

Memes are a dominant medium for online communication and manipulation because meaning emerges from interactions between embedded text, imagery, and cultural context. Existing meme research is distributed across tasks (hate, misogyny, propaganda, sentiment, humour) and languages, which limits cross-domain generalization. To address this gap we propose MemeLens, a unified multilingual and multitask explanation-enhanced Vision Language Model (VLM) for meme understanding. We consolidate 38 public meme datasets, filter and map dataset-specific labels into a shared taxonomy of $20$ tasks spanning harm, targets, figurative/pragmatic intent, and affect. We present a comprehensive empirical analysis across modeling paradigms, task categories, and datasets. Our findings suggest that robust meme understanding requires multimodal training, exhibits substantial variation across semantic categories, and remains sensitive to over-specialization when models are fine-tuned on individual datasets rather than trained in a unified setting. We will make the experimental resources and datasets publicly available for the community.

Reasoning is a fundamental cognitive process underlying inference, problem-solving, and decision-making. While large language models (LLMs) demonstrate strong reasoning capabilities in closed-world settings, they struggle in open-ended and dynamic environments. Agentic reasoning marks a paradigm shift by reframing LLMs as autonomous agents that plan, act, and learn through continual interaction. In this survey, we organize agentic reasoning along three complementary dimensions. First, we characterize environmental dynamics through three layers: foundational agentic reasoning, which establishes core single-agent capabilities including planning, tool use, and search in stable environments; self-evolving agentic reasoning, which studies how agents refine these capabilities through feedback, memory, and adaptation; and collective multi-agent reasoning, which extends intelligence to collaborative settings involving coordination, knowledge sharing, and shared goals. Across these layers, we distinguish in-context reasoning, which scales test-time interaction through structured orchestration, from post-training reasoning, which optimizes behaviors via reinforcement learning and supervised fine-tuning. We further review representative agentic reasoning frameworks across real-world applications and benchmarks, including science, robotics, healthcare, autonomous research, and mathematics. This survey synthesizes agentic reasoning methods into a unified roadmap bridging thought and action, and outlines open challenges and future directions, including personalization, long-horizon interaction, world modeling, scalable multi-agent training, and governance for real-world deployment.

Low-resource machine translation (MT) has gained increasing attention as parallel data from low-resource language communities is collected, but many potential methods for improving low-resource MT remain unexplored. We investigate a self-supervised reinforcement-learning-based fine-tuning for translation in low-resource settings using round-trip bootstrapping with the No Language Left Behind (NLLB) family of models. Our approach translates English into a target low-resource language and then back into English, using a combination of chrF++ and BLEU as the reward function on the reconstructed English sentences. Using the NLLB-MD dataset, we evaluate both the 600M and 1.3B parameter NLLB models and observe consistent improvements for the following languages: Central Aymara, Friulian, Wolof and Russian. Qualitative inspection of translation outputs indicates increased fluency and semantic fidelity. We argue that our method can further benefit from scale, enabling models to increasingly leverage their pretrained knowledge and continue self-improving.

The relationship between emotional expression and eye movement is well-documented, with literature establishing gaze patterns are reliable indicators of emotion. However, most studies utilize specialized, high-resolution eye-tracking equipment, limiting the potential reach of findings. We investigate how eye movement can be used to predict multimodal markers of emotional expression from naturalistic, low-resolution videos. We utilize a collection of video interviews from the USC Shoah Foundation's Visual History Archive with Holocaust survivors as they recount their experiences in the Auschwitz concentration camp. Inspired by pretraining methods on language models, we develop a novel gaze detection model that uses self-supervised eye movement reconstruction that can effectively leverage unlabeled video. We use this model's encoder embeddings to fine-tune models on two downstream tasks related to emotional expression. The first is aligning eye movement with directional emotion estimates from speech. The second task is using eye gaze as a predictor of three momentary manifestations of emotional behaviors: laughing, crying/sobbing, and sighing. We find our new model is predictive of emotion outcomes and observe a positive correlation between pretraining performance and emotion processing performance for both experiments. We conclude self-supervised eye movement reconstruction is an effective method for encoding the affective signal they carry.

Decision trees are one of the most popular classifiers in the machine learning literature. While the most common decision tree learning algorithms treat data as a batch, numerous algorithms have been proposed to construct decision trees from a data stream. A standard training strategy involves augmenting the current tree by changing a leaf node into a split. Here we typically maintain counters in each leaf which allow us to determine the optimal split, and whether the split should be done. In this paper we focus on how to speed up the search for the optimal split when dealing with sparse binary features and a binary class. We focus on finding splits that have the approximately optimal information gain or Gini index. In both cases finding the optimal split can be done in $O(d)$ time, where $d$ is the number of features. We propose an algorithm that yields $(1 + α)$ approximation when using conditional entropy in amortized $O(α^{-1}(1 + m\log d) \log \log n)$ time, where $m$ is the number of 1s in a data point, and $n$ is the number of data points. Similarly, for Gini index, we achieve $(1 + α)$ approximation in amortized $O(α^{-1} + m \log d)$ time. Our approach is beneficial for sparse data where $m \ll d$. In our experiments we find almost-optimal splits efficiently, faster than the baseline, overperforming the theoretical approximation guarantees.

Collaborative perception holds great promise for improving safety in autonomous driving, particularly in dense traffic where vehicles can share sensory information to overcome individual blind spots and extend awareness. However, deploying such collaboration at scale remains difficult when communication bandwidth is limited and no roadside infrastructure is available. To overcome these limitations, we introduce a fully decentralized framework that enables vehicles to self organize into cooperative groups using only vehicle to vehicle communication. The approach decomposes the problem into two sequential game theoretic stages. In the first stage, vehicles form stable clusters by evaluating mutual sensing complementarity and motion coherence, and each cluster elects a coordinator. In the second stage, the coordinator guides its members to selectively transmit point cloud segments from perceptually salient regions through a non cooperative potential game, enabling efficient local fusion. Global scene understanding is then achieved by exchanging compact detection messages across clusters rather than raw sensor data. We design distributed algorithms for both stages that guarantee monotonic improvement of the system wide potential function. Comprehensive experiments on the CARLA-OpenCDA-NS3 co-simulation platform show that our method reduces communication overhead while delivering higher perception accuracy and wider effective coverage compared to existing baselines.

Software bugs cost technology providers (e.g., AT&T) billions annually and cause developers to spend roughly 50% of their time on bug resolution. Traditional methods for bug localization often analyze the suspiciousness of code components (e.g., methods, documents) in isolation, overlooking their connections with other components in the codebase. Recent advances in Large Language Models (LLMs) and agentic AI techniques have shown strong potential for code understanding, but still lack causal reasoning during code exploration and struggle to manage growing context effectively, limiting their capability. In this paper, we present a novel agentic technique for bug localization -- CogniGent -- that overcomes the limitations above by leveraging multiple AI agents capable of causal reasoning, call-graph-based root cause analysis and context engineering. It emulates developers-inspired debugging practices (a.k.a., dynamic cognitive debugging) and conducts hypothesis testing to support bug localization. We evaluate CogniGent on a curated dataset of 591 bug reports using three widely adopted performance metrics and compare it against six established baselines from the literature. Experimental results show that our technique consistently outperformed existing traditional and LLM-based techniques, achieving MAP improvements of 23.33-38.57% at the document and method levels. Similar gains were observed in MRR, with increases of 25.14-53.74% at both granularity levels. Statistical significance tests also confirm the superiority of our technique. By addressing the reasoning, dependency, and context limitations, CogniGent advances the state of bug localization, bridging human-like cognition with agentic automation for improved performance.

Lagrangian Neural Networks (LNNs) can learn arbitrary Lagrangians from trajectory data, but their unusual optimization objective leads to significant training instabilities that limit their application to complex systems. We propose several improvements that address these fundamental challenges, namely, a Hessian regularization scheme that penalizes unphysical signatures in the Lagrangian's second derivatives with respect to velocities, preventing the network from learning unstable dynamics, activation functions that are better suited to the problem of learning Lagrangians, and a physics-aware coordinate scaling that improves stability. We systematically evaluate these techniques alongside previously proposed methods for improving stability. Our improved architecture successfully trains on systems of unprecedented complexity, including triple pendulums, and achieved 96.6\% lower validation loss value and 90.68\% better stability than baseline LNNs in double pendulum systems. With the improved framework, we show that our LNNs can learn Lagrangians representing geodesic motion in both non-relativistic and general relativistic settings. To deal with the relativistic setting, we extended our regularization to penalize violations of Lorentzian signatures, which allowed us to predict a geodesic Lagrangian under AdS\textsubscript{4} spacetime metric directly from trajectory data, which to our knowledge has not been done in the literature before. This opens new possibilities for automated discovery of geometric structures in physics, including extraction of spacetime metric tensor components from geodesic trajectories. While our approach inherits some limitations of the original LNN framework, particularly the requirement for invertible Hessians, it significantly expands the practical applicability of LNNs for scientific discovery tasks.

Cooperative MARL often assumes frequent access to global information in a data buffer, such as team rewards or other agents' actions, which is typically unrealistic in decentralized MARL systems due to high communication costs. When communication is limited, agents must rely on outdated information to estimate gradients and update their policies. A common approach to handle missing data is called importance sampling, in which we reweigh old data from a base policy to estimate gradients for the current policy. However, it quickly becomes unstable when the communication is limited (i.e. missing data probability is high), so that the base policy in importance sampling is outdated. To address this issue, we propose a technique called base policy prediction, which utilizes old gradients to predict the policy update and collect samples for a sequence of base policies, which reduces the gap between the base policy and the current policy. This approach enables effective learning with significantly fewer communication rounds, since the samples of predicted base policies could be collected within one communication round. Theoretically, we show that our algorithm converges to an $\varepsilon$-Nash equilibrium in potential games with only $O(\varepsilon^{-3/4})$ communication rounds and $O(poly(\max_i |A_i|)\varepsilon^{-11/4})$ samples, improving existing state-of-the-art results in communication cost, as well as sample complexity without the exponential dependence on the joint action space size. We also extend these results to general Markov Cooperative Games to find an agent-wise local maximum. Empirically, we test the base policy prediction algorithm in both simulated games and MAPPO for complex environments.

In remote sensing images, complex backgrounds, weak object signals, and small object scales make accurate detection particularly challenging, especially under low-quality imaging conditions. A common strategy is to integrate single-image super-resolution (SR) before detection; however, such serial pipelines often suffer from misaligned optimization objectives, feature redundancy, and a lack of effective interaction between SR and detection. To address these issues, we propose a Saliency-Driven multi-task Collaborative Network (SDCoNet) that couples SR and detection through implicit feature sharing while preserving task specificity. SDCoNet employs the swin transformer-based shared encoder, where hierarchical window-shifted self-attention supports cross-task feature collaboration and adaptively balances the trade-off between texture refinement and semantic representation. In addition, a multi-scale saliency prediction module produces importance scores to select key tokens, enabling focused attention on weak object regions, suppression of background clutter, and suppression of adverse features introduced by multi-task coupling. Furthermore, a gradient routing strategy is introduced to mitigate optimization conflicts. It first stabilizes detection semantics and subsequently routes SR gradients along a detection-oriented direction, enabling the framework to guide the SR branch to generate high-frequency details that are explicitly beneficial for detection. Experiments on public datasets, including NWPU VHR-10-Split, DOTAv1.5-Split, and HRSSD-Split, demonstrate that the proposed method, while maintaining competitive computational efficiency, significantly outperforms existing mainstream algorithms in small object detection on low-quality remote sensing images. Our code is available at https://github.com/qiruo-ya/SDCoNet.

Multimodal Large Language Models (MLLMs) can directly consume exam documents, threatening conventional assessments and academic integrity. We present DoPE (Decoy-Oriented Perturbation Encapsulation), a document-layer defense framework that embeds semantic decoys into PDF/HTML assessments to exploit render-parse discrepancies in MLLM pipelines. By instrumenting exams at authoring time, DoPE provides model-agnostic prevention (stop or confound automated solving) and detection (flag blind AI reliance) without relying on conventional one-shot classifiers. We formalize prevention and detection tasks, and introduce FewSoRT-Q, an LLM-guided pipeline that generates question-level semantic decoys and FewSoRT-D to encapsulate them into watermarked documents. We evaluate on Integrity-Bench, a novel benchmark of 1826 exams (PDF+HTML) derived from public QA datasets and OpenCourseWare. Against black-box MLLMs from OpenAI and Anthropic, DoPE yields strong empirical gains: a 91.4% detection rate at an 8.7% false-positive rate using an LLM-as-Judge verifier, and prevents successful completion or induces decoy-aligned failures in 96.3% of attempts. We release Integrity-Bench, our toolkit, and evaluation code to enable reproducible study of document-layer defenses for academic integrity.

The problem of reconstructing a quantum channel from a sample of classical data is considered. When the total fidelity can be represented as a ratio of two quadratic forms (e.g., in the case of mapping a mixed state to a pure state, projective operators, unitary learning, and others), Semidefinite Programming (SDP) can be applied to solve the fidelity optimization problem with respect to the Choi matrix. A remarkable feature of SDP is that the optimization is convex, which allows the problem to be efficiently solved by a variety of numerical algorithms. We have tested several commercially available SDP solvers, all of which allowed for the reconstruction of quantum channels of different forms. A notable feature is that the Kraus rank of the obtained quantum channel typically comprises less than a few percent of its maximal possible value. This suggests that a relatively small Kraus rank quantum channel is typically sufficient to describe experimentally observed classical data. The theory was also applied to the problem of reconstructing projective operators from data. Finally, we discuss a classical computational model based on quantum channel transformation, performed and calculated on a classical computer, possibly hardware-optimized.

Despite scaling to massive context windows, Large Language Models (LLMs) struggle with multi-hop reasoning due to inherent position bias, which causes them to overlook information at certain positions. Whether these failures stem from an inability to locate evidence (recognition failure) or integrate it (synthesis failure) is unclear. We introduce Multi-Focus Attention Instruction (MFAI), a semantic probe to disentangle these mechanisms by explicitly steering attention towards selected positions. Across 5 LLMs on two multi-hop QA tasks (MuSiQue and NeoQA), we establish the "Weakest Link Law": multi-hop reasoning performance collapses to the performance level of the least visible evidence. Crucially, this failure is governed by absolute position rather than the linear distance between facts (performance variance $<3%$). We further identify a duality in attention steering: while matched MFAI resolves recognition bottlenecks, improving accuracy by up to 11.5% in low-visibility positions, misleading MFAI triggers confusion in real-world tasks but is successfully filtered in synthetic tasks. Finally, we demonstrate that "thinking" models that utilize System-2 reasoning, effectively locate and integrate the required information, matching gold-only baselines even in noisy, long-context settings.

Audio large language models (LLMs) enable unified speech understanding and generation, yet their adaptation to linguistically complex, dialect-rich settings remains underexplored. This paper presents the first systematic study of multi-task instruction tuning for an Arabic-centric audio LLM, covering a hierarchy of generative tasks (ASR, speech summarization) and discriminative tasks (dialect and emotion identification). To support this study, we introduce AraMega-SSum, a novel dataset for Arabic speech summarization. We fine-tune Qwen2.5-Omni (7B) and propose Task-Progressive Curriculum (TPC) along with Aligner-Based Diverse Sampling (ADS), a strategy that constructs information-dense batches by selecting task- and label-balanced examples. Our results reveal a critical efficiency, robustness trade-off: while ADS accelerates initial convergence and boosts paralinguistic F1-scores, its inherent gradient volatility can destabilize generative decoding under prolonged training. Furthermore, while the TPC stabilizes core acoustic mapping, it often induces negative transfer in downstream tasks. We demonstrate that a Hybrid TPC+ADS Strategy provides an optimal training ``recipe'', first establishing a robust representative foundation before employing diversity-aware refinement to capture fine-grained nuances. These findings offer practical guidance for the efficient adaptation of Omni-models in complex, low-resource multimodal environments.

Quantum error correction is a key ingredient for large scale quantum computation, protecting logical information from physical noise by encoding it into many physical qubits. Topological stabilizer codes are particularly appealing due to their geometric locality and practical relevance. In these codes, stabilizer measurements yield a syndrome that must be decoded into a recovery operation, making decoding a central bottleneck for scalable real time operation. Existing decoders are commonly classified into two categories. Classical algorithmic decoders provide strong and well established baselines, but may incur substantial computational overhead at large code distances or under stringent latency constraints. Machine learning based decoders offer fast GPU inference and flexible function approximation, yet many approaches do not explicitly exploit the lattice geometry and local structure of topological codes, which can limit performance. In this work, we propose QuantumSMoE, a quantum vision transformer based decoder that incorporates code structure through plus shaped embeddings and adaptive masking to capture local interactions and lattice connectivity, and improves scalability via a mixture of experts layer with a novel auxiliary loss. Experiments on the toric code demonstrate that QuantumSMoE outperforms state-of-the-art machine learning decoders as well as widely used classical baselines.

Predicting the outcomes of research ideas at their conceptual stage (i.e. before significant resources are committed) holds great potential for optimizing scientific resource allocation and research planning. While existing methods rely heavily on finished manuscripts or peer reviews, we propose a novel capability-aware framework that predicts paper acceptance and ratings using only author information and research ideas, without requiring full text or experimental results. Our approach integrates author information, (inferred) capability presentation, and research ideas through a three-way transformer architecture with flexible fusion mechanisms. We also introduce a two-stage architecture for learning the capability representation given the author information and idea. Experiments show that our method significantly outperform the single-way models by finetuning bert-base and bert-large, and the capability predicting significantly increase the predictive accuracy of the final model. The proposed method can be applied in both early-stage research outcome prediction and scientific resource allocation.

Current evaluation of large language models (LLMs) overwhelmingly prioritizes accuracy; however, in real-world and safety-critical applications, the ability to abstain when uncertain is equally vital for trustworthy deployment. We introduce MedAbstain, a unified benchmark and evaluation protocol for abstention in medical multiple-choice question answering (MCQA) -- a discrete-choice setting that generalizes to agentic action selection -- integrating conformal prediction, adversarial question perturbations, and explicit abstention options. Our systematic evaluation of both open- and closed-source LLMs reveals that even state-of-the-art, high-accuracy models often fail to abstain with uncertain. Notably, providing explicit abstention options consistently increases model uncertainty and safer abstention, far more than input perturbations, while scaling model size or advanced prompting brings little improvement. These findings highlight the central role of abstention mechanisms for trustworthy LLM deployment and offer practical guidance for improving safety in high-stakes applications.

Transformer-based models have shown strong performance in time-series forecasting by leveraging self-attention to model long-range temporal dependencies. However, their effectiveness depends critically on the quality and structure of input representations derived from raw multivariate time-series data. This work proposes a two-stage forecasting framework that explicitly separates local temporal representation learning from global dependency modelling. In the first stage, a convolutional neural network (CNN) operates on fixed-length temporal patches to extract short-range temporal dynamics and non-linear feature interactions, producing compact patch-level token embeddings. Token-level self-attention is subsequently applied during representation learning to refine these embeddings by enabling interactions across temporal patches. In the second stage, a Transformer encoder processes the resulting token sequence to model inter-patch temporal dependencies and generate per-patch forecasts. Experiments conducted on synthetic multivariate time-series data with controlled static and dynamic factors demonstrate that the proposed patch-based tokenization strategy achieves competitive forecasting performance compared to convolutional and patch-based Transformer baselines. The results highlight the importance of structured temporal representations and show that decoupling local temporal encoding from global attention-based modelling yields more effective and stable time-series forecasting.

Reinforcement Learning with Verifiable Rewards (RLVR) has proven effective in enhancing LLMs short-context reasoning, but its performance degrades in long-context scenarios that require both precise grounding and robust long-range reasoning. We identify the "almost-there" phenomenon in long-context reasoning, where trajectories are largely correct but fail at the final step, and attribute this failure to two factors: (1) the lack of high reasoning density in long-context QA data that push LLMs beyond mere grounding toward sophisticated multi-hop reasoning; and (2) the loss of valuable learning signals during long-context RL training due to the indiscriminate penalization of partially correct trajectories with incorrect outcomes. To overcome this bottleneck, we propose DeepReasonQA, a KG-driven synthesis framework that controllably constructs high-difficulty, multi-hop long-context QA pairs with inherent reasoning chains. Building on this, we introduce Long-context Process Advantage Shaping (LongPAS), a simple yet effective method that performs fine-grained credit assignment by evaluating reasoning steps along Validity and Relevance dimensions, which captures critical learning signals from "almost-there" trajectories. Experiments on three long-context reasoning benchmarks show that our approach substantially outperforms RLVR baselines and matches frontier LLMs while using far fewer parameters. Further analysis confirms the effectiveness of our methods in strengthening long-context reasoning while maintaining stable RL training.

The demand of customized large language models (LLMs) has led to commercial LLMs offering black-box fine-tuning APIs, yet this convenience introduces a critical security loophole: attackers could jailbreak the LLMs by fine-tuning them with malicious data. Though this security issue has recently been exposed, the feasibility of such attacks is questionable as malicious training dataset is believed to be detectable by moderation models such as Llama-Guard-3. In this paper, we propose TrojanPraise, a novel finetuning-based attack exploiting benign and thus filter-approved data. Basically, TrojanPraise fine-tunes the model to associate a crafted word (e.g., "bruaf") with harmless connotations, then uses this word to praise harmful concepts, subtly shifting the LLM from refusal to compliance. To explain the attack, we decouple the LLM's internal representation of a query into two dimensions of knowledge and attitude. We demonstrate that successful jailbreak requires shifting the attitude while avoiding knowledge shift, a distortion in the model's understanding of the concept. To validate this attack, we conduct experiments on five opensource LLMs and two commercial LLMs under strict black-box settings. Results show that TrojanPraise achieves a maximum attack success rate of 95.88% while evading moderation.

AI agents are autonomous systems that combine LLMs with external tools to solve complex tasks. While such tools extend capability, improper tool permissions introduce security risks such as indirect prompt injection and tool misuse. We characterize these failures as unbalanced tool-driven agency. Agents may retain unnecessary permissions (excessive agency) or fail to invoke required tools (insufficient agency), amplifying the attack surface and reducing performance. We introduce AgenTRIM, a framework for detecting and mitigating tool-driven agency risks without altering an agent's internal reasoning. AgenTRIM addresses these risks through complementary offline and online phases. Offline, AgenTRIM reconstructs and verifies the agent's tool interface from code and execution traces. At runtime, it enforces per-step least-privilege tool access through adaptive filtering and status-aware validation of tool calls. Evaluating on the AgentDojo benchmark, AgenTRIM substantially reduces attack success while maintaining high task performance. Additional experiments show robustness to description-based attacks and effective enforcement of explicit safety policies. Together, these results demonstrate that AgenTRIM provides a practical, capability-preserving approach to safer tool use in LLM-based agents.

Time series anomaly detection (TSAD) is essential for ensuring the safety and reliability of aerospace software systems. Although large language models (LLMs) provide a promising training-free alternative to unsupervised approaches, their effectiveness in aerospace settings remains under-examined because of complex telemetry, misaligned evaluation metrics, and the absence of domain knowledge. To address this gap, we introduce ATSADBench, the first benchmark for aerospace TSAD. ATSADBench comprises nine tasks that combine three pattern-wise anomaly types, univariate and multivariate signals, and both in-loop and out-of-loop feedback scenarios, yielding 108,000 data points. Using this benchmark, we systematically evaluate state-of-the-art open-source LLMs under two paradigms: Direct, which labels anomalies within sliding windows, and Prediction-Based, which detects anomalies from prediction errors. To reflect operational needs, we reformulate evaluation at the window level and propose three user-oriented metrics: Alarm Accuracy (AA), Alarm Latency (AL), and Alarm Contiguity (AC), which quantify alarm correctness, timeliness, and credibility. We further examine two enhancement strategies, few-shot learning and retrieval-augmented generation (RAG), to inject domain knowledge. The evaluation results show that (1) LLMs perform well on univariate tasks but struggle with multivariate telemetry, (2) their AA and AC on multivariate tasks approach random guessing, (3) few-shot learning provides modest gains whereas RAG offers no significant improvement, and (4) in practice LLMs can detect true anomaly onsets yet sometimes raise false alarms, which few-shot prompting mitigates but RAG exacerbates. These findings offer guidance for future LLM-based TSAD in aerospace software.

Federated learning enables collaborative model training across distributed institutions without centralizing sensitive data; however, ensuring algorithmic fairness across heterogeneous data distributions while preserving privacy remains fundamentally unresolved. This paper introduces CryptoFair-FL, a novel cryptographic framework providing the first verifiable fairness guarantees for federated learning systems under formal security definitions. The proposed approach combines additively homomorphic encryption with secure multi-party computation to enable privacy-preserving verification of demographic parity and equalized odds metrics without revealing protected attribute distributions or individual predictions. A novel batched verification protocol reduces computational complexity from BigO(n^2) to BigO(n \log n) while maintaining (\dparam, \deltap)-differential privacy with dparam = 0.5 and deltap = 10^{-6}. Theoretical analysis establishes information-theoretic lower bounds on the privacy cost of fairness verification, demonstrating that the proposed protocol achieves near-optimal privacy-fairness tradeoffs. Comprehensive experiments across four benchmark datasets (MIMIC-IV healthcare records, Adult Income, CelebA, and a novel FedFair-100 benchmark) demonstrate that CryptoFair-FL reduces fairness violations from 0.231 to 0.031 demographic parity difference while incurring only 2.3 times computational overhead compared to standard federated averaging. The framework successfully defends against attribute inference attacks, maintaining adversarial success probability below 0.05 across all tested configurations. These results establish a practical pathway for deploying fairness-aware federated learning in regulated industries requiring both privacy protection and algorithmic accountability.

The ability of Large Language Models (LLMs) to perform reasoning tasks such as deduction has been widely investigated in recent years. Yet, their capacity to generate proofs-faithful, human-readable explanations of why conclusions follow-remains largely under explored. In this work, we study proof generation in the context of OWL ontologies, which are widely adopted for representing and reasoning over complex knowledge, by developing an automated dataset construction and evaluation framework. Our evaluation encompassing three sequential tasks for complete proving: Extraction, Simplification, and Explanation, as well as an additional task of assessing Logic Completeness of the premise. Through extensive experiments on widely used reasoning LLMs, we achieve important findings including: (1) Some models achieve overall strong results but remain limited on complex cases; (2) Logical complexity, rather than representation format (formal logic language versus natural language), is the dominant factor shaping LLM performance; and (3) Noise and incompleteness in input data substantially diminish LLMs' performance. Together, these results underscore both the promise of LLMs for explanation with rigorous logics and the gap of supporting resilient reasoning under complex or imperfect conditions. Code and data are available at https://github.com/HuiYang1997/LLMOwlR.

The widespread use of Vision Language Models (VLMs, e.g. CLIP) has raised concerns about their vulnerability to sophisticated and imperceptible adversarial attacks. These attacks could compromise model performance and system security in cross-modal tasks. To address this challenge, three main defense paradigms have been proposed: Training-time Defense, Test-time Adaptation Defense, and Training-free Defense. Training-time Defense involves modifying the training process, typically through adversarial fine-tuning to improve the robustness to adversarial examples. While effective, this approach requires substantial computational resources and may not generalize across all adversarial attacks. Test-time Adaptation Defense focuses on adapting the model at inference time by updating its parameters to handle unlabeled adversarial examples, offering flexibility but often at the cost of increased complexity and computational overhead. Training-free Defense avoids modifying the model itself, instead focusing on altering the adversarial inputs or their feature embeddings, which enforces input perturbations to mitigate the impact of attacks without additional training. This survey reviews the latest advancements in adversarial defense strategies for VLMs, highlighting the strengths and limitations of such approaches and discussing ongoing challenges in enhancing the robustness of VLMs.

Scientific Artificial Intelligence (AI) applications require models that deliver trustworthy uncertainty estimates while respecting domain constraints. Existing uncertainty quantification methods lack mechanisms to incorporate symbolic scientific knowledge, while neurosymbolic approaches operate deterministically without principled uncertainty modeling. We introduce the Constraint-Aware Neurosymbolic Uncertainty Framework (CANUF), unifying Bayesian deep learning with differentiable symbolic reasoning. The architecture comprises three components: automated constraint extraction from scientific literature, probabilistic neural backbone with variational inference, and differentiable constraint satisfaction layer ensuring physical consistency. Experiments on Materials Project (140,000+ materials), QM9 molecular properties, and climate benchmarks show CANUF reduces Expected Calibration Error by 34.7% versus Bayesian neural networks while maintaining 99.2% constraint satisfaction. Ablations reveal constraint-guided recalibration contributes 18.3% performance gain, with constraint extraction achieving 91.4% precision. CANUF provides the first end-to-end differentiable pipeline simultaneously addressing uncertainty quantification, constraint satisfaction, and interpretable explanations for scientific predictions.

Audio-visual speech recognition (AVSR) typically improves recognition accuracy in noisy environments by integrating noise-immune visual cues with audio signals. Nevertheless, high-noise audio inputs are prone to introducing adverse interference into the feature fusion process. To mitigate this, recent AVSR methods often adopt mask-based strategies to filter audio noise during feature interaction and fusion, yet such methods risk discarding semantically relevant information alongside noise. In this work, we propose an end-to-end noise-robust AVSR framework coupled with speech enhancement, eliminating the need for explicit noise mask generation. This framework leverages a Conformer-based bottleneck fusion module to implicitly refine noisy audio features with video assistance. By reducing modality redundancy and enhancing inter-modal interactions, our method preserves speech semantic integrity to achieve robust recognition performance. Experimental evaluations on the public LRS3 benchmark suggest that our method outperforms prior advanced mask-based baselines under noisy conditions.

Cross-chain bridges constitute the single largest vector of systemic risk in Decentralized Finance (DeFi), accounting for over \$2.8 billion in losses since 2021. The fundamental vulnerability lies in the binary nature of existing bridge security models: a bridge is either fully operational or catastrophically compromised, with no intermediate state to contain partial failures. We present ASAS-BridgeAMM, a bridge-coupled automated market maker that introduces Contained Degradation: a formally specified operational state where the system gracefully degrades functionality in response to adversarial signals. By treating cross-chain message latency as a quantifiable execution risk, the protocol dynamically adjusts collateral haircuts, slippage bounds, and withdrawal limits. Across 18 months of historical replay on Ethereum and two auxiliary chains, ASAS-BridgeAMM reduces worst-case bridge-induced insolvency by 73% relative to baseline mint-and-burn architectures, while preserving 104.5% of transaction volume during stress periods. In rigorous adversarial simulations involving delayed finality, oracle manipulation, and liquidity griefing, the protocol maintains solvency with probability $>0.9999$ and bounds per-epoch bad debt to $<0.2%$ of total collateral. We provide a reference implementation in Solidity and formally prove safety (bounded debt), liveness (settlement completion), and manipulation resistance under a Byzantine relayer model.

Reliable flow measurements are essential in many industries, but current instruments often fail to accurately estimate multiphase flows, which are frequently encountered in real-world operations. Combining machine learning (ML) algorithms with accurate single-phase flowmeters has therefore received extensive research attention in recent years. The Coriolis mass flowmeter is a widely used single-phase meter that provides direct mass flow measurements, which ML models can be trained to correct, thereby reducing measurement errors in multiphase conditions. This paper demonstrates that preserving temporal information significantly improves model performance in such scenarios. We compare a multilayer perceptron, a windowed multilayer perceptron, and a convolutional neural network (CNN) on three-phase air-water-oil flow data from 342 experiments. Whereas prior work typically compresses each experiment into a single averaged sample, we instead compute short-time averages from within each experiment and train models that preserve temporal information at several downsampling intervals. The CNN performed best at 0.25 Hz with approximately 95 % of relative errors below 13 %, a normalized root mean squared error of 0.03, and a mean absolute percentage error of approximately 4.3 %, clearly outperforming the best single-averaged model and demonstrating that short-time averaging within individual experiments is preferable. Results are consistent across multiple data splits and random seeds, demonstrating robustness.

Vision-language model (VLM) hallucination is commonly linked to imbalanced allocation of attention across input modalities: system, image and text. However, existing mitigation strategies tend towards an image-centric interpretation of these imbalances, often prioritising increased image attention while giving less consideration to the roles of the other modalities. In this study, we evaluate a more holistic, system-mediated account, which attributes these imbalances to functionally redundant system weights that reduce attention to image and textual inputs. We show that this framework offers a useful empirical perspective on the yes-bias, a common form of hallucination in which VLMs indiscriminately respond 'yes'. Causally redistributing attention from the system modality to image and textual inputs substantially suppresses this bias, often outperforming existing approaches. We further present evidence suggesting that system-mediated attention imbalances contribute to the yes-bias by encouraging a default reliance on coarse input representations, which are effective for some tasks but ill-suited to others. Taken together, these findings firmly establish system attention as a key factor in VLM hallucination and highlight its potential as a lever for mitigation.

Water distribution systems (WDSs) face increasing cyber-physical risks, which make reliable anomaly detection essential. Many data-driven models ignore network topology and are hard to interpret, while model-based ones depend strongly on parameter accuracy. This work proposes a hydraulic-aware graph attention network using normalized conservation law violations as features. It combines mass and energy balance residuals with graph attention and bidirectional LSTM to learn spatio-temporal patterns. A multi-scale module aggregates detection scores from node to network level. On the BATADAL dataset, it reaches $F1=0.979$, showing $3.3$pp gain and high robustness under $15\%$ parameter noise.

Interpretability is critical for applications of large language models in the legal domain which requires trust and transparency. While some studies develop task-specific approaches, other use the classification model's parameters to explain the decisions. However, which technique explains the legal outcome prediction best remains an open question. To address this challenge, we propose a comparative analysis framework for model-agnostic interpretability techniques. Among these, we employ two rationale extraction methods, which justify outcomes with human-interpretable and concise text fragments (i.e., rationales) from the given input text. We conduct comparison by evaluating faithfulness-via normalized sufficiency and comprehensiveness metrics along with plausibility-by asking legal experts to evaluate extracted rationales. We further assess the feasibility of LLM-as-a-Judge using legal expert evaluation results. We show that the model's "reasons" for predicting a violation differ substantially from those of legal experts, despite highly promising quantitative analysis results and reasonable downstream classification performance. The source code of our experiments is publicly available at https://github.com/trusthlt/IntEval.

Recent alignment methods for large language models, including PPO, DPO, and IPO, are often presented as distinct algorithms. In this work, we show that many of these approaches implicitly conflate two fundamental and independent design choices: (i) the sampling geometry, which determines which samples dominate the gradient signal, and (ii) the optimization geometry, which determines how deviations in value are penalized. We formalize this observation by expressing alignment as the minimization of a generalized distance between policy energy and target energy, parameterized by an alpha-divergence-based sampling weight and a Bregman-divergence-based value metric. We demonstrate that the commonly used KL divergence induces an exponential penalty on unbounded value signals, leading to numerical instability and vanishing gradients in high-confidence regimes. To address this issue, we propose Orthogonalized Policy Optimization (OPO), a framework that explicitly decouples sampling geometry from optimization geometry. By combining alpha-weighted importance sampling with a chi-square-induced quadratic regularization in ratio coordinates, OPO yields a simple and well-conditioned objective with linear gradient dynamics. This formulation maintains stable optimization while preserving peak-seeking behavior and avoids gradient saturation even when model confidence is high. Our analysis positions OPO as a unifying perspective on existing alignment methods and provides a principled foundation for robust reasoning-oriented training.

Cognitive anthropology suggests that the distinction of human intelligence lies in the ability to infer other individuals' knowledge states and understand their intentions. In comparison, our closest animal relative, chimpanzees, lack the capacity to do so. With this paper, we aim to evaluate LLM performance in the area of knowledge state tracking and estimation. We design two tasks to test (1) if LLMs can detect when story characters, through their actions, demonstrate knowledge they should not possess, and (2) if LLMs can predict story characters' next actions based on their own knowledge vs. objective truths they do not know. Results reveal that most current state-of-the-art LLMs achieve near-random performance on both tasks, and are substantially inferior to humans. We argue future LLM research should place more weight on the abilities of knowledge estimation and intention understanding.

As LLMs rapidly advance and enter real-world use, their privacy implications are increasingly important. We study an authorship de-anonymization threat: using LLMs to link anonymous documents to their authors, potentially compromising settings such as double-blind peer review. We propose De-Anonymization at Scale (DAS), a large language model-based method for attributing authorship among tens of thousands of candidate texts. DAS uses a sequential progression strategy: it randomly partitions the candidate corpus into fixed-size groups, prompts an LLM to select the text most likely written by the same author as a query text, and iteratively re-queries the surviving candidates to produce a ranked top-k list. To make this practical at scale, DAS adds a dense-retrieval prefilter to shrink the search space and a majority-voting style aggregation over multiple independent runs to improve robustness and ranking precision. Experiments on anonymized review data show DAS can recover same-author texts from pools of tens of thousands with accuracy well above chance, demonstrating a realistic privacy risk for anonymous platforms. On standard authorship benchmarks (Enron emails and blog posts), DAS also improves both accuracy and scalability over prior approaches, highlighting a new LLM-enabled de-anonymization vulnerability.

Background: Pediatric dental disease remains one of the most prevalent and inequitable chronic health conditions worldwide. Although strong epidemiological evidence links oral health outcomes to socio-economic and demographic determinants, most artificial intelligence (AI) applications in dentistry rely on image-based diagnosis and black-box prediction models, limiting transparency and ethical applicability in pediatric populations. Objective: This study aimed to develop and evaluate an explainable machine learning framework for pediatric dental risk stratification that prioritizes interpretability, calibration, and ethical deployment over maximal predictive accuracy. Methods: A supervised machine learning model was trained using population-level pediatric data including age, income-to-poverty ratio, race/ethnicity, gender, and medical history. Model performance was assessed using receiver operating characteristic (ROC) analysis and calibration curves. Explainability was achieved using SHapley Additive exPlanations (SHAP) to provide global and individual-level interpretation of predictions. Results: The model achieved modest discrimination (AUC = 0.61) with conservative calibration, underestimating risk at higher probability levels. SHAP analysis identified age and income-to-poverty ratio as the strongest contributors to predicted risk, followed by race/ethnicity and gender. Conclusion: Explainable machine learning enables transparent, prevention-oriented pediatric dental risk stratification and supports population screening and equitable resource allocation rather than diagnostic decision-making.

Emotion detection from faces is one of the machine learning problems needed for human-computer interaction. The variety of methods used is enormous, which motivated an in-depth review of articles and scientific studies. Three of the most interesting and best solutions are selected, followed by the selection of three datasets that stood out for the diversity and number of images in them. The selected neural networks are trained, and then a series of experiments are performed to compare their performance, including testing on different datasets than a model was trained on. This reveals weaknesses in existing solutions, including differences between datasets, unequal levels of difficulty in recognizing certain emotions and the challenges in differentiating between closely related emotions.

Reinforcement learning (RL) has emerged as a powerful paradigm for fine-tuning large-scale generative models, such as diffusion and flow models, to align with complex human preferences and user-specified tasks. A fundamental limitation remains \textit{the curse of diversity collapse}, where the objective formulation and optimization landscape inherently collapse the policy to a Dirac delta distribution. To address this challenge, we propose \textbf{DRIFT} (\textbf{D}ive\textbf{R}sity-\textbf{I}ncentivized Reinforcement \textbf{F}ine-\textbf{T}uning for Versatile Image Generation), an innovative framework that systematically incentivizes output diversity throughout the on-policy fine-tuning process, reconciling strong task alignment with high generation diversity to enhance versatility essential for applications that demand diverse candidate generations. We approach the problem across three representative perspectives: i) \textbf{sampling} a reward-concentrated subset that filters out reward outliers to prevent premature collapse; ii) \textbf{prompting} with stochastic variations to expand the conditioning space, and iii) \textbf{optimization} of the intra-group diversity with a potential-based reward shaping mechanism. Experimental results show that DRIFT achieves superior Pareto dominance regarding task alignment and generation diversity, yielding a $ 9.08\%\!\sim\! 43.46\%$ increase in diversity at equivalent alignment levels and a $ 59.65\% \!\sim\! 65.86\%$ increase in alignment at equivalent levels of diversity.

Slow and costly communication is often the main bottleneck in distributed optimization, especially in federated learning where it occurs over wireless networks. We introduce BiCoLoR, a communication-efficient optimization algorithm that combines two widely used and effective strategies: local training, which increases computation between communication rounds, and compression, which encodes high-dimensional vectors into short bitstreams. While these mechanisms have been combined before, compression has typically been applied only to uplink (client-to-server) communication, leaving the downlink (server-to-client) side unaddressed. In practice, however, both directions are costly. We propose BiCoLoR, the first algorithm to combine local training with bidirectional compression using arbitrary unbiased compressors. This joint design achieves accelerated complexity guarantees in both convex and strongly convex heterogeneous settings. Empirically, BiCoLoR outperforms existing algorithms and establishes a new standard in communication efficiency.

Large language models (LLMs) have demonstrated notable advancements in psychological counseling. However, existing models generally do not explicitly model seekers' emotion shifts across counseling sessions, a core focus in classical psychological schools. Moreover, how to align counselor models' responses with these emotion shifts while proactively mitigating safety risks remains underexplored. To bridge these gaps, we propose PsychēChat, which explicitly integrates emotion shift tracking and safety risk analysis for psychological counseling. Specifically, we employ interactive role-playing to synthesize counselor--seeker dialogues, incorporating two modules: Emotion Management Module, to capture seekers' current emotions and emotion shifts; and Risk Control Module, to anticipate seekers' subsequent reactions and identify potential risks. Furthermore, we introduce two modeling paradigms. The Agent Mode structures emotion management, risk control, and counselor responses into a collaborative multi-agent pipeline. The LLM Mode integrates these stages into a unified chain-of-thought for end-to-end inference, balancing efficiency and performance. Extensive experiments, including interactive scoring, dialogue-level evaluation, and human assessment, demonstrate that PsychēChat outperforms existing methods for emotional insight and safety control.

In this work, we argue that not all sequence-to-sequence tasks require the strong inductive biases of autoregressive (AR) models. Tasks like multilingual transliteration, code refactoring, grammatical correction or text normalization often rely on local dependencies where the full modeling capacity of AR models can be overkill, creating a trade-off between their high accuracy and high inference latency. While non-autoregressive (NAR) models offer speed, they typically suffer from hallucinations and poor length control. To explore this trade-off, we focus on the multilingual transliteration task in Indic languages and introduce NADIR, a novel NAR architecture designed to strike a balance between speed and accuracy. NADIR integrates a Differential Transformer and a Mixture-of-Experts mechanism, enabling it to robustly model complex character mappings without sequential dependencies. NADIR achieves over a 13x speed-up compared to the state-of-the-art AR baseline. It maintains a competitive mean Character Error Rate of 15.78%, compared to 14.44% for the AR model and 21.88% for a standard NAR equivalent. Importantly, NADIR reduces Repetition errors by 49.53%, Substitution errors by 24.45%, Omission errors by 32.92%, and Insertion errors by 16.87%. This work provides a practical blueprint for building fast and reliable NAR systems, effectively bridging the gap between AR accuracy and the demands of real-time, large-scale deployment.

Real-world tabular databases routinely combine continuous measurements and categorical records, yet missing entries are pervasive and can distort downstream analysis. We propose Statistical-Neural Interaction (SNI), an interpretable mixed-type imputation framework that couples correlation-derived statistical priors with neural feature attention through a Controllable-Prior Feature Attention (CPFA) module. CPFA learns head-wise prior-strength coefficients $\{λ_h\}$ that softly regularize attention toward the prior while allowing data-driven deviations when nonlinear patterns appear to be present in the data. Beyond imputation, SNI aggregates attention maps into a directed feature-dependency matrix that summarizes which variables the imputer relied on, without requiring post-hoc explainers. We evaluate SNI against six baselines (Mean/Mode, MICE, KNN, MissForest, GAIN, MIWAE) on six datasets spanning ICU monitoring, population surveys, socio-economic statistics, and engineering applications. Under MCAR/strict-MAR at 30\% missingness, SNI is generally competitive on continuous metrics but is often outperformed by accuracy-first baselines (MissForest, MIWAE) on categorical variables; in return, it provides intrinsic dependency diagnostics and explicit statistical-neural trade-off parameters. We additionally report MNAR stress tests (with a mask-aware variant) and discuss computational cost, limitations -- particularly for severely imbalanced categorical targets -- and deployment scenarios where interpretability may justify the trade-off.

Watermarking (WM) is a critical mechanism for detecting and attributing AI-generated content. Current WM methods for Large Language Models (LLMs) are predominantly tailored for autoregressive (AR) models: They rely on tokens being generated sequentially, and embed stable signals within the generated sequence based on the previously sampled text. Diffusion Language Models (DLMs) generate text via non-sequential iterative denoising, which requires significant modification to use WM methods designed for AR models. Recent work proposed to watermark DLMs by inverting the process when needed, but suffers significant computational or memory overhead. We introduce Left-Right Diffusion Watermarking (LR-DWM), a scheme that biases the generated token based on both left and right neighbors, when they are available. LR-DWM incurs minimal runtime and memory overhead, remaining close to the non-watermarked baseline DLM while enabling reliable statistical detection under standard evaluation settings. Our results demonstrate that DLMs can be watermarked efficiently, achieving high detectability with negligible computational and memory overhead.

Proactive and agentic control in Sixth-Generation (6G) Open Radio Access Networks (O-RAN) requires control-grade prediction under stringent Near-Real-Time (Near-RT) latency and computational constraints. While Transformer-based models are effective for sequence modeling, their quadratic complexity limits scalability in Near-RT RAN Intelligent Controller (RIC) analytics. This paper investigates a post-Transformer design paradigm for efficient radio telemetry forecasting. We propose a quantum-inspired many-body state-space tensor network that replaces self-attention with stable structured state-space dynamics kernels, enabling linear-time sequence modeling. Tensor-network factorizations in the form of Tensor Train (TT) / Matrix Product State (MPS) representations are employed to reduce parameterization and data movement in both input projections and prediction heads, while lightweight channel gating and mixing layers capture non-stationary cross-Key Performance Indicator (KPI) dependencies. The proposed model is instantiated as an agentic perceive-predict xApp and evaluated on a bespoke O-RAN KPI time-series dataset comprising 59,441 sliding windows across 13 KPIs, using Reference Signal Received Power (RSRP) forecasting as a representative use case. Our proposed Linear Quantum-Inspired State-Space (LiQSS) model is 10.8x-15.8x smaller and approximately 1.4x faster than prior structured state-space baselines. Relative to Transformer-based models, LiQSS achieves up to a 155x reduction in parameter count and up to 2.74x faster inference, without sacrificing forecasting accuracy.

Existing approaches to bias evaluation in large language models (LLMs) trade ecological validity for statistical control, relying on artificial prompts that poorly reflect real-world use, or on naturalistic tasks that lack scale and rigor. We introduce a scalable bias-auditing framework using named entities as probes to measure structural disparities in model behavior. We show that synthetic data reliably reproduces bias patterns observed in natural text, enabling large-scale analysis. Using this approach, we conduct the largest bias audit to date, comprising 1.9 billion data points across multiple entity types, tasks, languages, models, and prompting strategies. Our results reveal systematic biases: models penalize right-wing politicians, favor left-wing politicians, prefer Western and wealthy nations over the Global South, favor Western companies, and penalize firms in the defense and pharmaceutical sectors. While instruction tuning reduces bias, increasing model scale amplifies it, and prompting in Chinese or Russian does not attenuate Western-aligned preferences. These results indicate that LLMs should undergo rigorous auditing before deployment in high-stakes applications.

Deep Research Agents are increasingly used for automated survey generation. However, whether they can write surveys like human experts remains unclear. Existing benchmarks focus on fluency or citation accuracy, but none evaluates the core capabilities: retrieving essential papers and organizing them into coherent knowledge structures. We introduce TaxoBench, a diagnostic benchmark derived from 72 highly-cited computer science surveys. We manually extract expert-authored taxonomy trees containing 3,815 precisely categorized citations as ground truth. Our benchmark supports two evaluation modes: Deep Research mode tests end-to-end retrieval and organization given only a topic, while Bottom-Up mode isolates structuring capability by providing the exact papers human experts used. We evaluate 7 leading Deep Research agents and 12 frontier LLMs. Results reveal a dual bottleneck: the best agent recalls only 20.9% of expert-selected papers, and even with perfect input, the best model achieves only 0.31 ARI in organization. Current deep research agents remain far from expert-level survey writing. Our benchmark is publicly available at https://github.com/KongLongGeFDU/TaxoBench.

Control valve stiction, a friction that prevents smooth valve movement, is a common fault in industrial process systems that causes instability, equipment wear, and higher maintenance costs. Many plants still operate with conventional valves that lack real time monitoring, making early predictions challenging. This study presents a machine learning (ML) framework for detecting and predicting stiction using only routinely collected process signals: the controller output (OP) from control systems and the process variable (PV), such as flow rate. Three deep learning models were developed and compared: a Convolutional Neural Network (CNN), a hybrid CNN with a Support Vector Machine (CNN-SVM), and a Long Short-Term Memory (LSTM) network. To train these models, a data-driven labeling method based on slope ratio analysis was applied to a real oil and gas refinery dataset. The LSTM model achieved the highest accuracy and was able to predict stiction up to four hours in advance. To the best of the authors' knowledge, this is the first study to demonstrate ML based early prediction of control valve stiction from real industry data. The proposed framework can be integrated into existing control systems to support predictive maintenance, reduce downtime, and avoid unnecessary hardware replacement.

Compilers constitute the foundational root-of-trust in software supply chains; however, their immense complexity inevitably conceals critical defects. Recent research has attempted to leverage historical bugs to design new mutation operators or fine-tune models to increase program diversity for compiler fuzzing.We observe, however, that bugs manifest primarily based on the semantics of input programs rather than their syntax. Unfortunately, current approaches, whether relying on syntactic mutation or general Large Language Model (LLM) fine-tuning, struggle to preserve the specific semantics found in the logic of bug-triggering programs. Consequently, these critical semantic triggers are often lost, resulting in a limitation of the diversity of generated programs. To explicitly reuse such semantics, we propose FeatureFuzz, a compiler fuzzer that combines features to generate programs. We define a feature as a decoupled primitive that encapsulates a natural language description of a bug-prone invariant, such as an out-of-bounds array access, alongside a concrete code witness of its realization. FeatureFuzz operates via a three-stage workflow: it first extracts features from historical bug reports, synthesizes coherent groups of features, and finally instantiates these groups into valid programs for compiler fuzzing. We evaluated FeatureFuzz on GCC and LLVM. Over 24-hour campaigns, FeatureFuzz uncovered 167 unique crashes, which is 2.78x more than the second-best fuzzer. Furthermore, through a 72-hour fuzzing campaign, FeatureFuzz identified 106 bugs in GCC and LLVM, 76 of which have already been confirmed by compiler developers, validating the approach's ability to stress-test modern compilers effectively.

Prompt injection attacks have become an increasing vulnerability for LLM applications, where adversarial prompts exploit indirect input channels such as emails or user-generated content to circumvent alignment safeguards and induce harmful or unintended outputs. Despite advances in alignment, even state-of-the-art LLMs remain broadly vulnerable to adversarial prompts, underscoring the urgent need for robust, productive, and generalizable detection mechanisms beyond inefficient, model-specific patches. In this work, we propose Zero-Shot Embedding Drift Detection (ZEDD), a lightweight, low-engineering-overhead framework that identifies both direct and indirect prompt injection attempts by quantifying semantic shifts in embedding space between benign and suspect inputs. ZEDD operates without requiring access to model internals, prior knowledge of attack types, or task-specific retraining, enabling efficient zero-shot deployment across diverse LLM architectures. Our method uses adversarial-clean prompt pairs and measures embedding drift via cosine similarity to capture subtle adversarial manipulations inherent to real-world injection attacks. To ensure robust evaluation, we assemble and re-annotate the comprehensive LLMail-Inject dataset spanning five injection categories derived from publicly available sources. Extensive experiments demonstrate that embedding drift is a robust and transferable signal, outperforming traditional methods in detection accuracy and operational efficiency. With greater than 93% accuracy in classifying prompt injections across model architectures like Llama 3, Qwen 2, and Mistral and a false positive rate of <3%, our approach offers a lightweight, scalable defense layer that integrates into existing LLM pipelines, addressing a critical gap in securing LLM-powered systems to withstand adaptive adversarial threats.

Autonomous vehicles (AVs) require adaptive behavior planners to navigate unpredictable, real-world environments safely. Traditional behavior trees (BTs) offer structured decision logic but are inherently static and demand labor-intensive manual tuning, limiting their applicability at SAE Level 5 autonomy. This paper presents an agentic framework that leverages large language models (LLMs) and multi-modal vision models (LVMs) to generate and adapt BTs on the fly. A specialized Descriptor agent applies chain-of-symbols prompting to assess scene criticality, a Planner agent constructs high-level sub-goals via in-context learning, and a Generator agent synthesizes executable BT sub-trees in XML format. Integrated into a CARLA+Nav2 simulation, our system triggers only upon baseline BT failure, demonstrating successful navigation around unexpected obstacles (e.g., street blockage) with no human intervention. Compared to a static BT baseline, this approach is a proof-of-concept that extends to diverse driving scenarios.

Recent advances in semantic correspondence have been largely driven by the use of pre-trained large-scale models. However, a limitation of these approaches is their dependence on high-resolution input images to achieve optimal performance, which results in considerable computational overhead. In this work, we address a fundamental limitation in current methods: the irreversible fusion of adjacent keypoint features caused by deep downsampling operations. This issue is triggered when semantically distinct keypoints fall within the same downsampled receptive field (e.g., 16x16 patches). To address this issue, we present SimpleMatch, a simple yet effective framework for semantic correspondence that delivers strong performance even at low resolutions. We propose a lightweight upsample decoder that progressively recovers spatial detail by upsampling deep features to 1/4 resolution, and a multi-scale supervised loss that ensures the upsampled features retain discriminative features across different spatial scales. In addition, we introduce sparse matching and window-based localization to optimize training memory usage and reduce it by 51%. At a resolution of 252x252 (3.3x smaller than current SOTA methods), SimpleMatch achieves superior performance with 84.1% PCK@0.1 on the SPair-71k benchmark. We believe this framework provides a practical and efficient baseline for future research in semantic correspondence. Code is available at: https://github.com/hailong23-jin/SimpleMatch.

To lower the expertise barrier in machine learning, the AutoML community has focused on the CASH problem, a fundamental challenge that automates the process of algorithm selection and hyperparameter tuning. While traditional methods like Bayesian Optimization (BO) struggle with cold-start issues, Large Language Models (LLMs) can mitigate these via semantic priors. However, existing LLM-based optimizers generalize poorly to the high-dimensional, structured CASH space. We propose LB-MCTS, a framework synergizing LLMs and BO within a Monte Carlo Tree Search structure. It maximizes LLM reasoning with Selective Tuning Memory (STM) and explicit exploration-exploitation trade-off. It combines the strengths of both paradigms by dynamically shifting from LLM-driven to BO-driven proposals as data accumulates. Experiments on 104 AMLB datasets demonstrate the superiority of LB-MCTS over the competitive baselines.

Single-channel speech enhancement models face significant performance degradation in extremely noisy environments. While prior work has shown that complementary bone-conducted speech can guide enhancement, effective integration of this noise-immune modality remains a challenge. This paper introduces a novel multimodal speech enhancement framework that integrates bone-conduction sensors with air-conducted microphones using a conditional diffusion model. Our proposed model significantly outperforms previously established multimodal techniques and a powerful diffusion-based single-modal baseline across a wide range of acoustic conditions.

Large multimodal model powered GUI agents are emerging as high-privilege operators on mobile platforms, entrusted with perceiving screen content and injecting inputs. However, their design operates under the implicit assumption of Visual Atomicity: that the UI state remains invariant between observation and action. We demonstrate that this assumption is fundamentally invalid in Android, creating a critical attack surface. We present Action Rebinding, a novel attack that allows a seemingly-benign app with zero dangerous permissions to rebind an agent's execution. By exploiting the inevitable observation-to-action gap inherent in the agent's reasoning pipeline, the attacker triggers foreground transitions to rebind the agent's planned action toward the target app. We weaponize the agent's task-recovery logic and Android's UI state preservation to orchestrate programmable, multi-step attack chains. Furthermore, we introduce an Intent Alignment Strategy (IAS) that manipulates the agent's reasoning process to rationalize UI states, enabling it to bypass verification gates (e.g., confirmation dialogs) that would otherwise be rejected. We evaluate Action Rebinding Attacks on six widely-used Android GUI agents across 15 tasks. Our results demonstrate a 100% success rate for atomic action rebinding and the ability to reliably orchestrate multi-step attack chains. With IAS, the success rate in bypassing verification gates increases (from 0% to up to 100%). Notably, the attacker application requires no sensitive permissions and contains no privileged API calls, achieving a 0% detection rate across malware scanners (e.g., VirusTotal). Our findings reveal a fundamental architectural flaw in current agent-OS integration and provide critical insights for the secure design of future agent systems. To access experimental logs and demonstration videos, please contact yi_qian@smail.nju.edu.cn.

The proliferation of generative AI has transformed creative workflows, yet current systems face critical challenges in controllability and content protection. We propose a novel multi-agent framework that addresses both limitations through specialized agent roles and integrated watermarking mechanisms. Unlike existing multi-agent systems focused solely on generation quality, our approach uniquely combines controllable content synthesis with provenance protection during the generation process itself. The framework orchestrates Director/Planner, Generator, Reviewer, Integration, and Protection agents with human-in-the-loop feedback to ensure alignment with user intent while embedding imperceptible digital watermarks. We formalize the pipeline as a joint optimization objective unifying controllability, semantic alignment, and protection robustness. This work contributes to responsible generative AI by positioning multi-agent architectures as a solution for trustworthy creative workflows with built-in ownership tracking and content traceability.

Real-world graphs are dynamic, with frequent updates to their structure and features due to evolving vertex and edge properties. These continual changes pose significant challenges for efficient inference in graph neural networks (GNNs). Existing vertex-wise and layer-wise inference approaches are ill-suited for dynamic graphs, as they incur redundant computations, large neighborhood traversals, and high communication costs, especially in distributed settings. Additionally, while sampling-based approaches can be adopted to approximate final layer embeddings, these are often not preferred in critical applications due to their non-determinism. These limitations hinder low-latency inference required in real-time applications. To address this, we propose RIPPLE++, a framework for streaming GNN inference that efficiently and accurately updates embeddings in response to changes in the graph structure or features. RIPPLE++ introduces a generalized incremental programming model that captures the semantics of GNN aggregation functions and incrementally propagates updates to affected neighborhoods. RIPPLE++ accommodates all common graph updates, including vertex/edge addition/deletions and vertex feature updates. RIPPLE++ supports both single-machine and distributed deployments. On a single machine, it achieves up to $56$K updates/sec on sparse graphs like Arxiv ($169$K vertices, $1.2$M edges), and about $7.6$K updates/sec on denser graphs like Products ($2.5$M vertices, $123.7$M edges), with latencies of $0.06$--$960$ms, and outperforming state-of-the-art baselines by $2.2$--$24\times$ on throughput. In distributed settings, RIPPLE++ offers up to $\approx25\times$ higher throughput and $20\times$ lower communication costs compared to recomputing baselines.

Latest advances in deep spatial filtering for Ambisonics demonstrate strong performance in stationary multi-speaker scenarios by rotating the sound field toward a target speaker prior to multi-channel enhancement. For applicability in dynamic acoustic conditions with moving speakers, we propose to automate this rotary steering using an interleaved tracking algorithm conditioned on the target's initial direction. However, for nearby or crossing speakers, robust tracking becomes difficult and spatial cues less effective for enhancement. By incorporating the processed recording as additional guide into both algorithms, our novel joint autoregressive framework leverages temporal-spectral correlations of speech to resolve spatially challenging speaker constellations. Consequently, our proposed method significantly improves tracking and enhancement of closely spaced speakers, consistently outperforming comparable non-autoregressive methods on a synthetic dataset. Real-world recordings complement these findings in complex scenarios with multiple speaker crossings and varying speaker-to-array distances.

Large language models (LLMs) are increasingly used to predict human behavior. We propose a measure for evaluating how much knowledge a pretrained LLM brings to such a prediction: its equivalent sample size, defined as the amount of task-specific data needed to match the predictive accuracy of the LLM. We estimate this measure by comparing the prediction error of a fixed LLM in a given domain to that of flexible machine learning models trained on increasing samples of domain-specific data. We further provide a statistical inference procedure by developing a new asymptotic theory for cross-validated prediction error. Finally, we apply this method to the Panel Study of Income Dynamics. We find that LLMs encode considerable predictive information for some economic variables but much less for others, suggesting that their value as substitutes for domain-specific data differs markedly across settings.

This paper introduces a dataset and conceptual framework for LLMs to mimic real world emotional dynamics through time and in-context learning leveraging physics-informed neural network, opening a possibility for interpretable dialogue modeling.

Customer reviews contain detailed, domain specific signals about service failures and user expectations, but converting this unstructured feedback into actionable business decisions remains difficult. We study review-to-action generation: producing concrete, implementable recommendations grounded in review text. We propose a modular two-LLM framework in which an Issue model extracts salient issues and assigns coarse themes, and an Advice model generates targeted operational fixes conditioned on the extracted issue representation. To enable specialization without expensive full fine-tuning, we adapt the Advice model using a mixture of LoRA experts strategy: multiple low-rank adapters are trained and a lightweight gating mechanism performs token-level expert mixing at inference, combining complementary expertise across issue types. We construct synthetic review-issue-advice triples from Yelp reviews (airlines and restaurants) to supervise training, and evaluate recommendations using an eight dimension operational rubric spanning actionability, specificity, feasibility, expected impact, novelty, non-redundancy, bias, and clarity. Across both domains, our approach consistently outperforms prompting-only and single-adapter baselines, yielding higher actionability and specificity while retaining favorable efficiency-quality trade-offs.

RAG has emerged as a key technique for enhancing response quality of LLMs without high computational cost. In traditional architectures, RAG services are provided by a single entity that hosts the dataset within a trusted local environment. However, individuals or small organizations often lack the resources to maintain data storage servers, leading them to rely on outsourced cloud storage. This dependence on untrusted third-party services introduces privacy risks. Embedding-based retrieval mechanisms, commonly used in RAG systems, are vulnerable to privacy leakage such as vector-to-text reconstruction attacks and structural leakage via vector analysis. Several privacy-preserving RAG techniques have been proposed but most existing approaches rely on partially homomorphic encryption, which incurs substantial computational overhead. To address these challenges, we propose an efficient privacy-preserving RAG framework (ppRAG) tailored for untrusted cloud environments that defends against vector-to-text attack, vector analysis, and query analysis. We propose Conditional Approximate Distance-Comparison-Preserving Symmetric Encryption (CAPRISE) that encrypts embeddings while still allowing the cloud to compute similarity between an encrypted query and the encrypted database embeddings. CAPRISE preserves only the relative distance ordering between the encrypted query and each encrypted database embedding, without exposing inter-database distances, thereby enhancing both privacy and efficiency. To mitigate query analysis, we introduce DP by perturbing the query embedding prior to encryption, preventing the cloud from inferring sensitive patterns. Experimental results show that ppRAG achieves efficient processing throughput, high retrieval accuracy, strong privacy guarantees, making it a practical solution for resource-constrained users seeking secure cloud-augmented LLMs.

Glacial Lake Outburst Floods (GLOFs) pose a serious threat in high mountain regions. They are hazardous to communities, infrastructure, and ecosystems further downstream. The classical methods of GLOF detection and prediction have so far mainly relied on hydrological modeling, threshold-based lake monitoring, and manual satellite image analysis. These approaches suffer from several drawbacks: slow updates, reliance on manual labor, and losses in accuracy when clouds interfere and/or lack on-site data. To tackle these challenges, we present IceWatch: a novel deep learning framework for GLOF prediction that incorporates both spatial and temporal perspectives. The vision component, RiskFlow, of IceWatch deals with Sentinel-2 multispectral satellite imagery using a CNN-based classifier and predicts GLOF events based on the spatial patterns of snow, ice, and meltwater. Its tabular counterpart confirms this prediction by considering physical dynamics. TerraFlow models glacier velocity from NASA ITS_LIVE time series while TempFlow forecasts near-surface temperature from MODIS LST records; both are trained on long-term observational archives and integrated via harmonized preprocessing and synchronization to enable multimodal, physics-informed GLOF prediction. Both together provide cross-validation, which will improve the reliability and interpretability of GLOF detection. This system ensures strong predictive performance, rapid data processing for real-time use, and robustness to noise and missing information. IceWatch paves the way for automatic, scalable GLOF warning systems. It also holds potential for integration with diverse sensor inputs and global glacier monitoring activities.

Generative AI (GenAI) systems promise to transform knowledge work by automating a range of tasks, yet their deployment in enterprise settings remains hindered by the lack of systematic quality assurance mechanisms. We present an Expert Validation Framework that places domain experts at the center of building software with GenAI components, enabling them to maintain authoritative control over system behavior through structured specification, testing, validation, and continuous monitoring processes. Our framework addresses the critical gap between AI capabilities and organizational trust by establishing a rigorous, expert-driven methodology for ensuring quality across diverse GenAI applications. Through a four-stage implementation process encompassing specification, system creation, validation, and production monitoring, the framework enables organizations to leverage GenAI capabilities while maintaining expert oversight and quality standards.

Humans face countless scenarios that require reasoning and judgment in daily life. However, existing large language model training methods primarily allow models to learn from existing textual content or solve predetermined problems, lacking experience in real scenarios involving interaction, negotiation, and competition with others. To address this, this paper proposes Multi-Agent Reward Optimization (MARO), a method that enables large language models (LLMs) to acquire stronger reasoning abilities by learning and practicing in multi-agent social environments. Specifically, MARO first addresses the sparse learning signal problem by decomposing final success or failure outcomes into each specific behavior during the interaction process; second, it handles the uneven role distribution problem by balancing the training sample weights of different roles; finally, it addresses environmental instability issues by directly evaluating the utility of each behavior. Experimental results demonstrate that MARO not only achieves significant improvements in social reasoning capabilities, but also that the abilities acquired through social simulation learning can effectively transfer to other tasks such as mathematical reasoning and instruction following. This reveals the tremendous potential of multi-agent social learning in enhancing the general reasoning capabilities of LLMs.

Momentum SGD (MSGD) serves as a foundational optimizer in training deep models due to momentum's key role in accelerating convergence and enhancing generalization. Meanwhile, asynchronous distributed learning is crucial for training large-scale deep models, especially when the computing capabilities of the workers in the cluster are heterogeneous. To reduce communication frequency, local updates are widely adopted in distributed learning. However, how to implement asynchronous distributed MSGD with local updates remains unexplored. To solve this problem, we propose a novel method, called \underline{or}dered \underline{lo}cal \underline{mo}mentum (OrLoMo), for asynchronous distributed learning. In OrLoMo, each worker runs MSGD locally. Then the local momentum from each worker will be aggregated by the server in order based on its global iteration index. To the best of our knowledge, OrLoMo is the first method to implement asynchronous distributed MSGD with local updates. We prove the convergence of OrLoMo for non-convex problems under arbitrary delays. Experiments validate that OrLoMo can outperform its synchronous counterpart and other asynchronous methods.

The advancement of Document Intelligence (DI) demands large-scale, high-quality training data, yet manual annotation remains a critical bottleneck. While data generation methods are evolving rapidly, existing surveys are constrained by fragmented focuses on single modalities or specific tasks, lacking a unified perspective aligned with real-world workflows. To fill this gap, this survey establishes the first comprehensive technical map for data generation in DI. Data generation is redefined as supervisory signal production, and a novel taxonomy is introduced based on the "availability of data and labels." This framework organizes methodologies into four resource-centric paradigms: Data Augmentation, Data Generation from Scratch, Automated Data Annotation, and Self-Supervised Signal Construction. Furthermore, a multi-level evaluation framework is established to integrate intrinsic quality and extrinsic utility, compiling performance gains across diverse DI benchmarks. Guided by this unified structure, the methodological landscape is dissected to reveal critical challenges such as fidelity gaps and frontiers including co-evolutionary ecosystems. Ultimately, by systematizing this fragmented field, data generation is positioned as the central engine for next-generation DI.

Automation in data analysis has been a long-time pursuit. Current agentic LLM shows a promising solution towards it. Like DeepAnalyze, DataSage, and Datawise. They are all powerful agentic frameworks for automatic fine-grained analysis and are powered by LLM-based agentic tool calling ability. However, what about powered by a preset AutoML-like workflow? If we traverse all possible exploration, like Xn itself`s statistics, Xn1-Xn2 relationships, Xn to all other, and finally explain? Our Explanova is such an attempt: Cheaper due to a Local Small LLM.

We present a semantics modulated, multi scale Transformer for 3D gaze estimation. Our model conditions CLIP global features with learnable prototype banks (illumination, head pose, background, direction), fuses these prototype-enriched global vectors with CLIP patch tokens and high-resolution CNN tokens in a unified attention space, and replaces several FFN blocks with routed/shared Mixture of Experts to increase conditional capacity. Evaluated on MPIIFaceGaze, EYEDIAP, Gaze360 and ETH-XGaze, our model achieves new state of the art angular errors of 2.49°, 3.22°, 10.16°, and 1.44°, demonstrating up to a 64% relative improvement over previously reported results. ablations attribute gains to prototype conditioning, cross scale fusion, MoE and hyperparameter. Our code is publicly available at https://github. com/AIPMLab/Gazeformer.

The emergence of 6G-enabled vehicular metaverses enables Autonomous Vehicles (AVs) to operate across physical and virtual spaces through space-air-ground-sea integrated networks. The AVs can deploy AI agents powered by large AI models as personalized assistants, on edge servers to support intelligent driving decision making and enhanced on-board experiences. However, such cross-reality interactions may cause serious location privacy risks, as adversaries can infer AV trajectories by correlating the location reported when AVs request LBS in reality with the location of the edge servers on which their corresponding AI agents are deployed in virtuality. To address this challenge, we design a cross-reality location privacy protection framework based on hybrid actions, including continuous location perturbation in reality and discrete privacy-aware AI agent migration in virtuality. In this framework, a new privacy metric, termed cross-reality location entropy, is proposed to effectively quantify the privacy levels of AVs. Based on this metric, we formulate an optimization problem to optimize the hybrid action, focusing on achieving a balance between location protection, service latency reduction, and quality of service maintenance. To solve the complex mixed-integer problem, we develop a novel LLM-enhanced Hybrid Diffusion Proximal Policy Optimization (LHDPPO) algorithm, which integrates LLM-driven informative reward design to enhance environment understanding with double Generative Diffusion Models-based policy exploration to handle high-dimensional action spaces, thereby enabling reliable determination of optimal hybrid actions. Extensive experiments on real-world datasets demonstrate that the proposed framework effectively mitigates cross-reality location privacy leakage for AVs while maintaining strong user immersion within 6G-enabled vehicular metaverse scenarios.

Self-training systems often degenerate due to the lack of an external criterion for judging data quality, leading to reward hacking and semantic drift. This paper provides a proof-of-concept system architecture for stable self-training under sparse external feedback and bounded memory, and empirically characterises its learning dynamics and failure modes. We introduce a self-training architecture in which learning is mediated exclusively by environmental viability, rather than by reward, objective functions, or externally defined fitness criteria. Candidate behaviours are executed under real resource constraints, and only those whose environmental effects both persist and preserve the possibility of future interaction are propagated. The environment does not provide semantic feedback, dense rewards, or task-specific supervision; selection operates solely through differential survival of behaviours as world-altering events, making proxy optimisation impossible and rendering reward-hacking evolutionarily unstable. Analysis of semantic dynamics shows that improvement arises primarily through the persistence of effective and repeatable strategies under a regime of consolidation and pruning, a paradigm we refer to as negative-space learning (NSL), and that models develop meta-learning strategies (such as deliberate experimental failure in order to elicit informative error messages) without explicit instruction. This work establishes that environment-grounded selection enables sustainable open-ended self-improvement, offering a viable path toward more robust and generalisable autonomous systems without reliance on human-curated data or complex reward shaping.

Few-shot learning in remote sensing remains challenging due to three factors: the scarcity of labeled data, substantial domain shifts, and the multi-scale nature of geospatial objects. To address these issues, we introduce Adaptive Multi-Scale Correlation Meta-Network (AMC-MetaNet), a lightweight yet powerful framework with three key innovations: (i) correlation-guided feature pyramids for capturing scale-invariant patterns, (ii) an adaptive channel correlation module (ACCM) for learning dynamic cross-scale relationships, and (iii) correlation-guided meta-learning that leverages correlation patterns instead of conventional prototype averaging. Unlike prior approaches that rely on heavy pre-trained models or transformers, AMC-MetaNet is trained from scratch with only $\sim600K$ parameters, offering $20\times$ fewer parameters than ResNet-18 while maintaining high efficiency ($<50$ms per image inference). AMC-MetaNet achieves up to 86.65\% accuracy in 5-way 5-shot classification on various remote sensing datasets, including EuroSAT, NWPU-RESISC45, UC Merced Land Use, and AID. Our results establish AMC-MetaNet as a computationally efficient, scale-aware framework for real-world few-shot remote sensing.

Recent advances in LLM-based multi-agent systems (MAS) show that workflows composed of multiple LLM agents with distinct roles, tools, and communication patterns can outperform single-LLM baselines on complex tasks. However, most frameworks are homogeneous, where all agents share the same base LLM and differ only in prompts, tools, and positions in the workflow. This raises the question of whether such workflows can be simulated by a single agent through multi-turn conversations. We investigate this across seven benchmarks spanning coding, mathematics, general question answering, domain-specific reasoning, and real-world planning and tool use. Our results show that a single agent can reach the performance of homogeneous workflows with an efficiency advantage from KV cache reuse, and can even match the performance of an automatically optimized heterogeneous workflow. Building on this finding, we propose \textbf{OneFlow}, an algorithm that automatically tailors workflows for single-agent execution, reducing inference costs compared to existing automatic multi-agent design frameworks without trading off accuracy. These results position the single-LLM implementation of multi-agent workflows as a strong baseline for MAS research. We also note that single-LLM methods cannot capture heterogeneous workflows due to the lack of KV cache sharing across different LLMs, highlighting future opportunities in developing \textit{truly} heterogeneous multi-agent systems.

We propose a novel MLaaS Dataset Generator (MDG) framework that creates configurable and reproducible datasets for evaluating Machine Learning as a Service (MLaaS) selection and composition. MDG simulates realistic MLaaS behaviour by training and evaluating diverse model families across multiple real-world datasets and data distribution settings. It records detailed functional attributes, quality of service metrics, and composition-specific indicators, enabling systematic analysis of service performance and cross-service behaviour. Using MDG, we generate more than ten thousand MLaaS service instances and construct a large-scale benchmark dataset suitable for downstream evaluation. We also implement a built-in composition mechanism that models how services interact under varied Internet of Things conditions. Experiments demonstrate that datasets generated by MDG enhance selection accuracy and composition quality compared to existing baselines. MDG provides a practical and extensible foundation for advancing data-driven research on MLaaS selection and composition

Vision-language pre-training (VLP) models are vulnerable to adversarial examples, particularly in black-box scenarios. Existing multimodal attacks often suffer from limited perturbation diversity and unstable multi-stage pipelines. To address these challenges, we propose 2S-GDA, a two-stage globally-diverse attack framework. The proposed method first introduces textual perturbations through a globally-diverse strategy by combining candidate text expansion with globally-aware replacement. To enhance visual diversity, image-level perturbations are generated using multi-scale resizing and block-shuffle rotation. Extensive experiments on VLP models demonstrate that 2S-GDA consistently improves attack success rates over state-of-the-art methods, with gains of up to 11.17\% in black-box settings. Our framework is modular and can be easily combined with existing methods to further enhance adversarial transferability.

An interesting phenomenon arises: Empirical Risk Minimization (ERM) sometimes outperforms methods specifically designed for out-of-distribution tasks. This motivates an investigation into the reasons behind such behavior beyond algorithmic design. In this study, we find that one such reason lies in the distribution shift across training domains. A large degree of distribution shift can lead to better performance even under ERM. Specifically, we derive several theoretical and empirical findings demonstrating that distribution shift plays a crucial role in model learning and benefits learning invariant prediction. Firstly, the proposed upper bounds indicate that the degree of distribution shift directly affects the prediction ability of the learned models. If it is large, the models' ability can increase, approximating invariant prediction models that make stable predictions under arbitrary known or unseen domains; and vice versa. We also prove that, under certain data conditions, ERM solutions can achieve performance comparable to that of invariant prediction models. Secondly, the empirical validation results demonstrated that the predictions of learned models approximate those of Oracle or Optimal models, provided that the degree of distribution shift in the training data increases.

Reward-guided search methods have demonstrated strong potential in enhancing tool-using agents by effectively guiding sampling and exploration over complex action spaces. As a core design, those search methods utilize process reward models (PRMs) to provide step-level rewards, enabling more fine-grained monitoring. However, there is a lack of systematic and reliable evaluation benchmarks for PRMs in tool-using settings. In this paper, we introduce ToolPRMBench, a large-scale benchmark specifically designed to evaluate PRMs for tool-using agents. ToolPRMBench is built on top of several representative tool-using benchmarks and converts agent trajectories into step-level test cases. Each case contains the interaction history, a correct action, a plausible but incorrect alternative, and relevant tool metadata. We respectively utilize offline sampling to isolate local single-step errors and online sampling to capture realistic multi-step failures from full agent rollouts. A multi-LLM verification pipeline is proposed to reduce label noise and ensure data quality. We conduct extensive experiments across large language models, general PRMs, and tool-specialized PRMs on ToolPRMBench. The results reveal clear differences in PRM effectiveness and highlight the potential of specialized PRMs for tool-using. Code and data will be released at https://github.com/David-Li0406/ToolPRMBench.

Learning representative embeddings for different types of speaking styles, such as emotion, age, and gender, is critical for both recognition tasks (e.g., cognitive computing and human-computer interaction) and generative tasks (e.g., style-controllable speech generation). In this work, we introduce ParaMETA, a unified and flexible framework for learning and controlling speaking styles directly from speech. Unlike existing methods that rely on single-task models or cross-modal alignment, ParaMETA learns disentangled, task-specific embeddings by projecting speech into dedicated subspaces for each type of style. This design reduces inter-task interference, mitigates negative transfer, and allows a single model to handle multiple paralinguistic tasks such as emotion, gender, age, and language classification. Beyond recognition, ParaMETA enables fine-grained style control in Text-To-Speech (TTS) generative models. It supports both speech- and text-based prompting and allows users to modify one speaking styles while preserving others. Extensive experiments demonstrate that ParaMETA outperforms strong baselines in classification accuracy and generates more natural and expressive speech, while maintaining a lightweight and efficient model suitable for real-world applications.

Probabilistic time series forecasting is crucial for quantifying future uncertainty, with significant applications in fields such as energy and finance. However, existing methods often rely on computationally expensive sampling or restrictive parametric assumptions to characterize future distributions, which limits predictive performance and introduces distributional mismatch. To address these challenges, this paper presents TimeGMM, a novel probabilistic forecasting framework based on Gaussian Mixture Models (GMM) that captures complex future distributions in a single forward pass. A key component is GMM-adapted Reversible Instance Normalization (GRIN), a novel module designed to dynamically adapt to temporal-probabilistic distribution shifts. The framework integrates a dedicated Temporal Encoder (TE-Module) with a Conditional Temporal-Probabilistic Decoder (CTPD-Module) to jointly capture temporal dependencies and mixture distribution parameters. Extensive experiments demonstrate that TimeGMM consistently outperforms state-of-the-art methods, achieving maximum improvements of 22.48\% in CRPS and 21.23\% in NMAE.

The increasing prevalence of Large Language Models (LLMs) demands effective safeguards for their operation, particularly concerning their tendency to generate out-of-context responses. A key challenge is accurately detecting when LLMs stray from expected conversational norms, manifesting as topic shifts, factual inaccuracies, or outright hallucinations. Traditional anomaly detection struggles to directly apply within contextual semantics. This paper outlines our experiment in exploring the use of Representation Engineering (RepE) and One-Class Support Vector Machine (OCSVM) to identify subspaces within the internal states of LLMs that represent a specific context. By training OCSVM on in-context examples, we establish a robust boundary within the LLM's hidden state latent space. We evaluate out study with two open source LLMs - Llama and Qwen models in specific contextual domain. Our approach entailed identifying the optimal layers within the LLM's internal state subspaces that strongly associates with the context of interest. Our evaluation results showed promising results in identifying the subspace for a specific context. Aside from being useful in detecting in or out of context conversation threads, this research work contributes to the study of better interpreting LLMs.

The functions of different regions of the human brain are closely linked to their distinct cytoarchitecture, which is defined by the spatial arrangement and morphology of the cells. Identifying brain regions by their cytoarchitecture enables various scientific analyses of the brain. However, delineating these areas manually in brain histological sections is time-consuming and requires specialized knowledge. An automated approach is necessary to minimize the effort needed from human experts. To address this, we propose CytoCLIP, a suite of vision-language models derived from pre-trained Contrastive Language-Image Pre-Training (CLIP) frameworks to learn joint visual-text representations of brain cytoarchitecture. CytoCLIP comprises two model variants: one is trained using low-resolution whole-region images to understand the overall cytoarchitectural pattern of an area, and the other is trained on high-resolution image tiles for detailed cellular-level representation. The training dataset is created from NISSL-stained histological sections of developing fetal brains of different gestational weeks. It includes 86 distinct regions for low-resolution images and 384 brain regions for high-resolution tiles. We evaluate the model's understanding of the cytoarchitecture and generalization ability using region classification and cross-modal retrieval tasks. Multiple experiments are performed under various data setups, including data from samples of different ages and sectioning planes. Experimental results demonstrate that CytoCLIP outperforms existing methods. It achieves an F1 score of 0.87 for whole-region classification and 0.91 for high-resolution image tile classification.

Electroencephalography (EEG) decoding requires models that can effectively extract and integrate complex temporal, spectral, and spatial features from multichannel signals. To address this challenge, we propose a lightweight and generalizable decoding framework named Hierarchical Convolutional Fusion Transformer (HCFT), which combines dual-branch convolutional encoders and hierarchical Transformer blocks for multi-scale EEG representation learning. Specifically, the model first captures local temporal and spatiotemporal dynamics through time-domain and time-space convolutional branches, and then aligns these features via a cross-attention mechanism that enables interaction between branches at each stage. Subsequently, a hierarchical Transformer fusion structure is employed to encode global dependencies across all feature stages, while a customized Dynamic Tanh normalization module is introduced to replace traditional Layer Normalization in order to enhance training stability and reduce redundancy. Extensive experiments are conducted on two representative benchmark datasets, BCI Competition IV-2b and CHB-MIT, covering both event-related cross-subject classification and continuous seizure prediction tasks. Results show that HCFT achieves 80.83% average accuracy and a Cohen's kappa of 0.6165 on BCI IV-2b, as well as 99.10% sensitivity, 0.0236 false positives per hour, and 98.82% specificity on CHB-MIT, consistently outperforming over ten state-of-the-art baseline methods. Ablation studies confirm that each core component of the proposed framework contributes significantly to the overall decoding performance, demonstrating HCFT's effectiveness in capturing EEG dynamics and its potential for real-world BCI applications.

The design-build-test cycle is essential for innovation, but physical prototyping is often slow and expensive. Although physics-based simulation and strategic prototyping can reduce cost, meaningful evaluation is frequently constrained until an integrated prototype is built. This paper investigates whether a generative pretrained transformer (GPT) can predict information typically obtained through prototyping, including cost, performance, and perceived usability. We introduce a retrieval-augmented generation (RAG) method to emulate design feedback using OpenAI GPT-4o, grounded in prototyping data scraped from Instructables.com to increase access to relevant precedent. Two studies are reported. First, a controlled experiment compares GPT-RAG and human designers, who receive design sketches and predict cost, performance, and usability; predictions are evaluated against ground-truth results from physical prototypes. Second, we report an applied demonstration in which a physical prototype is produced from GPT-RAG recommendations and compared with a commercial baseline and a topology-optimized design. Results show that GPT-RAG provides more accurate cost and performance estimates than individual or crowd human estimates, while yielding comparable usability insights; the GPT-RAG-informed prototype also outperforms both comparison prototypes. Repeated querying with response averaging significantly improves accuracy, suggesting that LLMs can emulate crowd aggregation effects consistent with the law of large numbers.

Concolic testing, a powerful hybrid software testing technique, has historically been plagued by fundamental limitations such as path explosion and the high cost of constraint solving, which hinder its practical application in large-scale, real-world software systems. This paper introduces a novel algorithmic framework that synergistically integrates concolic execution with Large Language Models (LLMs) to overcome these challenges. Our hybrid approach leverages the semantic reasoning capabilities of LLMs to guide path exploration, prioritize interesting execution paths, and assist in constraint solving. We formally define the system architecture and algorithms that constitute this new paradigm. Through a series of experiments on both synthetic and real-world Fintech applications, we demonstrate that our approach significantly outperforms traditional concolic testing, random testing, and genetic algorithm-based methods in terms of branch coverage, path coverage, and time-to-coverage. The results indicate that by combining the strengths of both concolic execution and LLMs, our method achieves a more efficient and effective exploration of the program state space, leading to improved bug detection capabilities.

In recent years, Automated Program Repair (APR) techniques specifically designed for quantum programs have been proposed. However, existing approaches often suffer from low repair success rates or poor understandability of the generated patches. In this study, we construct a framework in which a large language model (LLM) generates code repairs along with a natural language explanation of the applied repairs. To investigate how the contextual information included in prompts influences APR performance for quantum programs, we design four prompt configurations with different combinations of static information, dynamic information, and mutation analysis results. Mutation analysis evaluates how small changes to specific parts of a program affect its execution results and provides more detailed dynamic information than simple execution outputs such as stack traces. Our experimental results show that mutation analysis can provide valuable contextual information for LLM-based APR of quantum programs, improving repair success rates (achieving 94.4% in our experiment) and in some cases also improving the quality of generated explanations. Our findings point toward new directions for developing APR techniques for quantum programs that enhance both reliability and explainability.

Autoregressive language models are next-token predictors and have been criticized for only optimizing surface plausibility (i.e., local coherence) rather than maintaining correct latent-state representations (i.e., global coherence). Because Theory of Mind (ToM) tasks crucially depend on reasoning about latent mental states of oneself and others, such models are therefore often thought to fail at ToM. While post-training methods can improve ToM performance, we show that strong ToM capability can be recovered directly from the base model without any additional weight updates or verifications. Our approach builds on recent power-sampling methods (Karan & Du, 2025) that use Markov chain Monte Carlo (MCMC) to sample from sharpened sequence-level (rather than token-level) probability distributions of autoregressive language models. We further find that incorporating annealing, where the tempered distribution is gradually shifted from high to low temperature, substantially improves ToM performance over fixed-temperature power sampling. Together, these results suggest that sampling-based optimization provides a powerful way to extract latent capabilities from language models without retraining.

We study the problem of scheduling delay-sensitive jobs over spot and on-demand cloud instances to minimize average cost while meeting an average delay constraint. Jobs arrive as a general stochastic process, and incur different costs based on the instance type. This work provides the first analytical treatment of this problem using tools from queuing theory, stochastic processes, and optimization. We derive cost expressions for general policies, prove queue length one is optimal for low target delays, and characterize the optimal wait-time distribution. For high target delays, we identify a knapsack structure and design a scheduling policy that exploits it. An adaptive algorithm is proposed to fully utilize the allowed delay, and empirical results confirm its near-optimality.

This study proposes an algorithm titled a statistical firefly algorithm (SFA) for truss topology optimization. In the proposed algorithm, historical results of fireflies' motions are used in hypothesis testing to limit the motions of fireflies that are suggested by current information exchanges between fireflies only to those that are potentially useful. Hypothesis testing is applied to the mechanism of an ordinary firefly algorithm (FA) without changing its structure. As a result, the implementation of the proposed algorithm is simple and straightforward. Limiting the motions of fireflies to those that are potential useful results in reduction of firefly evaluations, and, subsequently, reduction of computational efforts. To test the validity and efficiency of the proposed algorithm, it is used to solve several truss topology optimization problems, including some benchmark problems. It is found that the added statistical strategy in the SFA significantly enhances the performance of the original FA in terms of computational efforts while still maintains the quality of the obtained results.

Vision-Language Models (VLMs) are rapidly replacing unimodal encoders in modern retrieval and recommendation systems. While their capabilities are well-documented, their robustness against adversarial manipulation in competitive ranking scenarios remains largely unexplored. In this paper, we uncover a critical vulnerability in VLM-based product search: multimodal ranking attacks. We present Multimodal Generative Engine Optimization (MGEO), a novel adversarial framework that enables a malicious actor to unfairly promote a target product by jointly optimizing imperceptible image perturbations and fluent textual suffixes. Unlike existing attacks that treat modalities in isolation, MGEO employs an alternating gradient-based optimization strategy to exploit the deep cross-modal coupling within the VLM. Extensive experiments on real-world datasets using state-of-the-art models demonstrate that our coordinated attack significantly outperforms text-only and image-only baselines. These findings reveal that multimodal synergy, typically a strength of VLMs, can be weaponized to compromise the integrity of search rankings without triggering conventional content filters.

Recent progress in large language models (LLMs) has improved code generation, but most evaluations still test isolated, small-scale code (e.g., a single function) under default or unspecified software environments. As a result, it is unclear whether LLMs can reliably generate executable code tailored to a user's specific environment. We present the first systematic study of Environment-Aware Code Generation (EACG), where generated code must be functionally correct and directly executable under arbitrary software configurations. To enable realistic evaluation, we introduce VersiBCB, a benchmark that is multi-package, execution-verified, and deprecation-aware, capturing complex and evolving environments that prior datasets often overlook. Using VersiBCB, we investigate three complementary adaptation axes: data, parameters, and cache, and develop representative strategies for each. Our results show that current LLMs struggle with environment-specific code generation, while our adaptations improve environment compatibility and executability. These findings highlight key challenges and opportunities for deploying LLMs in practical software engineering workflows.

Document understanding (VRDU) in regulated domains is particularly challenging, since scanned documents often contain sensitive, evolving, and domain specific knowledge. This leads to two major challenges: the lack of manual annotations for model adaptation and the difficulty for pretrained models to stay up-to-date with domain-specific facts. While Multimodal Large Language Models (MLLMs) show strong zero-shot abilities, they still suffer from hallucination and limited domain grounding. In contrast, discriminative Vision-Language Pre-trained Models (VLPMs) provide reliable grounding but require costly annotations to cover new domains. We introduce Docs2Synth, a synthetic-supervision framework that enables retrieval-guided inference for private and low-resource domains. Docs2Synth automatically processes raw document collections, generates and verifies diverse QA pairs via an agent-based system, and trains a lightweight visual retriever to extract domain-relevant evidence. During inference, the retriever collaborates with an MLLM through an iterative retrieval--generation loop, reducing hallucination and improving response consistency. We further deliver Docs2Synth as an easy-to-use Python package, enabling plug-and-play deployment across diverse real-world scenarios. Experiments on multiple VRDU benchmarks show that Docs2Synth substantially enhances grounding and domain generalization without requiring human annotations.

Building upon FutureX, which established a live benchmark for general-purpose future prediction, this report introduces FutureX-Pro, including FutureX-Finance, FutureX-Retail, FutureX-PublicHealth, FutureX-NaturalDisaster, and FutureX-Search. These together form a specialized framework extending agentic future prediction to high-value vertical domains. While generalist agents demonstrate proficiency in open-domain search, their reliability in capital-intensive and safety-critical sectors remains under-explored. FutureX-Pro targets four economically and socially pivotal verticals: Finance, Retail, Public Health, and Natural Disaster. We benchmark agentic Large Language Models (LLMs) on entry-level yet foundational prediction tasks -- ranging from forecasting market indicators and supply chain demands to tracking epidemic trends and natural disasters. By adapting the contamination-free, live-evaluation pipeline of FutureX, we assess whether current State-of-the-Art (SOTA) agentic LLMs possess the domain grounding necessary for industrial deployment. Our findings reveal the performance gap between generalist reasoning and the precision required for high-value vertical applications.

Conventional imaging requires a line of sight to create accurate visual representations of a scene. In certain circumstances, however, obtaining a suitable line of sight may be impractical, dangerous, or even impossible. Non-line-of-sight (NLOS) imaging addresses this challenge by reconstructing the scene from indirect measurements. Recently, passive NLOS methods that use an ordinary photograph of the subtle shadow cast onto a visible wall by the hidden scene have gained interest. These methods are currently limited to 1D or low-resolution 2D color imaging or to localizing a hidden object whose shape is approximately known. Here, we generalize this class of methods and demonstrate a 3D reconstruction of a hidden scene from an ordinary NLOS photograph. To achieve this, we propose a novel reformulation of the light transport model that conveniently decomposes the hidden scene into \textit{light-occluding} and \textit{non-light-occluding} components to yield a separable non-linear least squares (SNLLS) inverse problem. We develop two solutions: A gradient-based optimization method and a physics-inspired neural network approach, which we call Soft Shadow diffusion (SSD). Despite the challenging ill-conditioned inverse problem encountered here, our approaches are effective on numerous 3D scenes in real experimental scenarios. Moreover, SSD is trained in simulation but generalizes well to unseen classes in simulation and real-world NLOS scenes. SSD also shows surprising robustness to noise and ambient illumination.

Large language models (LLMs) have demonstrated their instruction-following capabilities and achieved powerful performance on various tasks. Inspired by their success, recent works in the molecular domain have led to the development of large molecular language models (LMLMs) that integrate 1D molecular strings or 2D molecular graphs into the language models. However, existing LMLMs often suffer from hallucination and limited robustness, largely due to inadequate integration of diverse molecular modalities such as 1D sequences, 2D molecular graphs, and 3D conformations. To address these limitations, we propose CoLLaMo, a large language model-based molecular assistant equipped with a multi-level molecular modality-collaborative projector. The relation-aware modality-collaborative attention mechanism in the projector facilitates fine-grained and relation-guided information exchange between atoms by incorporating 2D structural and 3D spatial relations. Furthermore, we present a molecule-centric new automatic measurement, including a hallucination assessment metric and GPT-based caption quality evaluation to address the limitations of token-based generic evaluation metrics (i.e., BLEU) widely used in assessing molecular comprehension of LMLMs. Our extensive experiments demonstrate that our CoLLaMo enhances the molecular modality generalization capabilities of LMLMs, achieving the best performance on multiple tasks, including molecule captioning, computed property QA, descriptive property QA, motif counting, and IUPAC name prediction.

Efficient fine-tuning of visual-language models like CLIP has become crucial due to their large-scale parameter size and extensive pretraining requirements. Existing methods typically address either the issue of unseen classes or unseen domains in isolation, without considering a joint framework for both. In this paper, we propose \textbf{Fed}erated Joint Learning for \textbf{D}omain and \textbf{C}lass \textbf{G}eneralization, termed \textbf{FedDCG}, a novel approach that addresses both class and domain generalization in federated learning settings. Our method introduces a domain grouping strategy where class-generalized networks are trained within each group to prevent decision boundary confusion. During inference, we aggregate class-generalized results based on domain similarity, effectively integrating knowledge from both class and domain generalization. Specifically, a learnable network is employed to enhance class generalization capabilities, and a decoupling mechanism separates general and domain-specific knowledge, improving generalization to unseen domains. Extensive experiments across various datasets show that \textbf{FedDCG} outperforms state-of-the-art baselines in terms of accuracy and robustness.

Breast cancer is one of the most common cancers among women worldwide, and its accurate and timely diagnosis plays a critical role in improving treatment outcomes. This thesis presents an innovative framework for detecting malignant masses in mammographic images by integrating the Pyramid Adaptive Atrous Convolution (PAAC) and Transformer architectures. The proposed approach utilizes Multi-Scale Feature Fusion to enhance the extraction of features from benign and malignant tissues and combines Dice Loss and Focal Loss functions to improve the model's learning process, effectively reducing errors in binary breast cancer classification and achieving high accuracy and efficiency. In this study, a comprehensive dataset of breast cancer images from INbreast, MIAS, and DDSM was preprocessed through data augmentation and contrast enhancement and resized to 227x227 pixels for model training. Leveraging the Transformer's ability to manage long-range dependencies with Self-Attention mechanisms, the proposed model achieved high accuracy in detecting cancerous masses, outperforming foundational models such as BreastNet, DeepMammo, Multi-Scale CNN, Swin-Unet, and SegFormer. The final evaluation results for the proposed model include an accuracy of 98.5\%, sensitivity of 97.8\%, specificity of 96.3\%, F1-score of 98.2\%, and overall precision of 97.9\%. These metrics demonstrate a significant improvement over traditional methods and confirm the model's effectiveness in identifying cancerous masses in complex scenarios and large datasets. This model shows potential as a reliable and efficient tool for breast cancer diagnosis and can be effectively integrated into medical diagnostic systems.

Recent advances in audio-aware large language models have shown strong performance on audio question answering. However, existing benchmarks mainly cover answerable questions and overlook the challenge of unanswerable ones, where no reliable answer can be inferred from the audio. Such cases are common in real-world settings, where questions may be misleading, ill-posed, or incompatible with the information. To address this gap, we present AQUA-Bench, a benchmark for Audio Question Unanswerability Assessment. It systematically evaluates three scenarios: Absent Answer Detection (the correct option is missing), Incompatible Answer Set Detection (choices are categorically mismatched with the question), and Incompatible Audio Question Detection (the question is irrelevant or lacks sufficient grounding in the audio). By assessing these cases, AQUA-Bench offers a rigorous measure of model reliability and promotes the development of audio-language systems that are more robust and trustworthy. Our experiments suggest that while models excel on standard answerable tasks, they often face notable challenges with unanswerable ones, pointing to a blind spot in current audio-language understanding.

Diffusion Language Models (DLMs) present a promising non-sequential paradigm for text generation, distinct from standard autoregressive (AR) approaches. However, current decoding strategies often adopt a reactive stance, underutilizing the global bidirectional context to dictate global trajectories. To address this, we propose Plan-Verify-Fill (PVF), a training-free paradigm that grounds planning via quantitative validation. PVF actively constructs a hierarchical skeleton by prioritizing high-leverage semantic anchors and employs a verification protocol to operationalize pragmatic structural stopping where further deliberation yields diminishing returns. Extensive evaluations on LLaDA-8B-Instruct and Dream-7B-Instruct demonstrate that PVF reduces the Number of Function Evaluations (NFE) by up to 65% compared to confidence-based parallel decoding across benchmark datasets, unlocking superior efficiency without compromising accuracy.

The increasing complexity of marine operations has intensified the need for intelligent robotic systems to support ocean observation, exploration, and resource management. Underwater swarm robotics offers a promising framework that extends the capabilities of individual autonomous platforms through collective coordination. Inspired by natural systems, such as fish schools and insect colonies, bio-inspired swarm approaches enable distributed decision-making, adaptability, and resilience under challenging marine conditions. Yet research in this field remains fragmented, with limited integration across algorithmic, communication, and hardware design perspectives. This review synthesises bio-inspired coordination mechanisms, communication strategies, and system design considerations for underwater swarm robotics. It examines key marine-specific algorithms, including the Artificial Fish Swarm Algorithm, Whale Optimisation Algorithm, Coral Reef Optimisation, and Marine Predators Algorithm, highlighting their applications in formation control, task allocation, and environmental interaction. The review also analyses communication constraints unique to the underwater domain and emerging acoustic, optical, and hybrid solutions that support cooperative operation. Additionally, it examines hardware and system design advances that enhance system efficiency and scalability. A multi-dimensional classification framework evaluates existing approaches across communication dependency, environmental adaptability, energy efficiency, and swarm scalability. Through this integrated analysis, the review unifies bio-inspired coordination algorithms, communication modalities, and system design approaches. It also identifies converging trends, key challenges, and future research directions for real-world deployment of underwater swarm systems.

Video summarization helps turn long videos into clear, concise representations that are easier to review, document, and analyze, especially in high-stakes domains like surgical training. Prior work has progressed from using basic visual features like color, motion, and structural changes to using pre-trained vision-language models that can better understand what's happening in the video (semantics) and capture temporal flow, resulting in more context-aware video summarization. We propose a three-stage framework, PRISM: Procedural Representation via Integrated Semantic and Multimodal analysis, that produces semantically grounded video summaries. PRISM combines adaptive visual sampling, label-driven keyframe anchoring, and contextual validation using a large language model (LLM). Our method ensures that selected frames reflect meaningful and procedural transitions while filtering out generic or hallucinated content, resulting in contextually coherent summaries across both domain-specific and instructional videos. We evaluate our method on instructional and activity datasets, using reference summaries for instructional videos. Despite sampling fewer than 5% of the original frames, our summaries retain 84% semantic content while improving over baselines by as much as 33%. Our approach generalizes across procedural and domain-specific video tasks, achieving strong performance with both semantic alignment and precision.

In recent years, Non-Orthogonal Multiple Access (NOMA) system has emerged as a promising candidate for multiple access frameworks due to the evolution of deep machine learning, trying to incorporate deep machine learning into the NOMA system. The main motivation for such active studies is the growing need to optimize the utilization of network resources as the expansion of the internet of things (IoT) caused a scarcity of network resources. The NOMA addresses this need by power multiplexing, allowing multiple users to access the network simultaneously. Nevertheless, the NOMA system has few limitations. Several works have proposed to mitigate this, including the optimization of power allocation known as joint resource allocation(JRA) method, and integration of the JRA method and deep reinforcement learning (JRA-DRL). Despite this, the channel assignment problem remains unclear and requires further investigation. In this paper, we propose a deep reinforcement learning framework incorporating replay memory with an on-policy algorithm, allocating network resources in a NOMA system to generalize the learning. Also, we provide extensive simulations to evaluate the effects of varying the learning rate, batch size, type of model, and the number of features in the state.

Disaggregation has emerged as a powerful strategy for optimizing large language model (LLM) inference by separating compute-intensive prefill and memory-bound decode phases across specialized GPUs. This separation improves utilization and throughput under fixed hardware capacity. However, as model and cluster scales grow, power, rather than compute, has become the dominant limiter of overall performance and cost efficiency. In this paper, we propose RAPID, a power-aware disaggregated inference framework that jointly manages GPU roles and power budgets to sustain goodput within strict power caps. RAPID utilizes static and dynamic power reallocation in addition to GPU reallocation to improve performance under fixed power bounds. RAPID improves overall performance and application consistency beyond what is achievable in current disaggregation solutions, resulting in up to a 2x improvement in SLO attainment at peak load when compared to a static assignment without an increase in complexity or cost.

While momentum-based acceleration has been studied extensively in deterministic optimization problems, its behavior in nonstationary environments -- where the data distribution and optimal parameters drift over time -- remains underexplored. We analyze the tracking performance of Stochastic Gradient Descent (SGD) and its momentum variants (Polyak heavy-ball and Nesterov) under uniform strong convexity and smoothness in varying stepsize regimes. We derive finite-time bounds in expectation and with high probability for the tracking error, establishing a sharp decomposition into three components: a transient initialization term, a noise-induced variance term, and a drift-induced tracking lag. Crucially, our analysis uncovers a fundamental trade-off: while momentum can suppress gradient noise, it incurs an explicit penalty on the tracking capability. We show that momentum can substantially amplify drift-induced tracking error, with amplification that becomes unbounded as the momentum parameter approaches one, formalizing the intuition that using 'stale' gradients hinders adaptation to rapid regime shifts. Complementing these upper bounds, we establish minimax lower bounds for dynamic regret under gradient-variation constraints. These lower bounds prove that the inertia-induced penalty is not an artifact of analysis but an information-theoretic barrier: in drift-dominated regimes, momentum creates an unavoidable 'inertia window' that fundamentally degrades performance. Collectively, these results provide a definitive theoretical grounding for the empirical instability of momentum in dynamic environments and delineate the precise regime boundaries where SGD provably outperforms its accelerated counterparts.

Generating 3D models has traditionally been a complex task requiring specialized expertise. While recent advances in generative AI have sought to automate this process, existing methods produce non-editable representation, such as meshes or point clouds, limiting their adaptability for iterative design. In this paper, we introduce Proc3D, a system designed to generate editable 3D models while enabling real-time modifications. At its core, Proc3D introduces procedural compact graph (PCG), a graph representation of 3D models, that encodes the algorithmic rules and structures necessary for generating the model. This representation exposes key parameters, allowing intuitive manual adjustments via sliders and checkboxes, as well as real-time, automated modifications through natural language prompts using Large Language Models (LLMs). We demonstrate Proc3D's capabilities using two generative approaches: GPT-4o with in-context learning (ICL) and a fine-tuned LLAMA-3 model. Experimental results show that Proc3D outperforms existing methods in editing efficiency, achieving more than 400x speedup over conventional approaches that require full regeneration for each modification. Additionally, Proc3D improves ULIP scores by 28%, a metric that evaluates the alignment between generated 3D models and text prompts. By enabling text-aligned 3D model generation along with precise, real-time parametric edits, Proc3D facilitates highly accurate text-based image editing applications.

Insider threat detection is a key challenge in enterprise security, relying on user activity logs that capture rich and complex behavioral patterns. These logs are often multi-channel, non-stationary, and anomalies are rare, making anomaly detection challenging. To address these issues, we propose a novel framework that integrates wavelet-aware modulation, multi-resolution wavelet decomposition, and resolution-adaptive attention for robust anomaly detection. Our approach first applies a deviation-aware modulation scheme to suppress routine behaviors while amplifying anomalous deviations. Next, discrete wavelet transform (DWT) decomposes the log signals into multi-resolution representations, capturing both long-term trends and short-term anomalies. Finally, a learnable attention mechanism dynamically reweights the most discriminative frequency bands for detection. On the CERT r4.2 benchmark, our approach consistently outperforms existing baselines in precision, recall, and F1 score across various time granularities and scenarios.

Foundation models trained on electronic health records show strong performance on many clinical prediction tasks but are limited by sparse and irregular documentation. Wearable devices provide dense continuous physiological signals but lack semantic grounding. Existing methods usually model these data sources separately or combine them through late fusion. We propose a multimodal foundation model that jointly represents electronic health records and wearable data as a continuous time latent process. The model uses modality specific encoders and a shared temporal backbone pretrained with self supervised and cross modal objectives. This design produces representations that are temporally coherent and clinically grounded. Across forecasting physiological and risk modeling tasks the model outperforms strong electronic health record only and wearable only baselines especially at long horizons and under missing data. These results show that joint electronic health record and wearable pretraining yields more faithful representations of longitudinal health.

Enabling intuitive, language-driven interaction with surgical scenes is a critical step toward intelligent operating rooms and autonomous surgical robotic assistance. However, the task of referring segmentation, localizing surgical instruments based on natural language descriptions, remains underexplored in surgical videos, with existing approaches struggling to generalize due to reliance on static visual cues and predefined instrument names. In this work, we introduce SurgRef, a novel motion-guided framework that grounds free-form language expressions in instrument motion, capturing how tools move and interact across time, rather than what they look like. This allows models to understand and segment instruments even under occlusion, ambiguity, or unfamiliar terminology. To train and evaluate SurgRef, we present Ref-IMotion, a diverse, multi-institutional video dataset with dense spatiotemporal masks and rich motion-centric expressions. SurgRef achieves state-of-the-art accuracy and generalization across surgical procedures, setting a new benchmark for robust, language-driven surgical video segmentation.

We present an algorithm for normalizing \emph{Batched Einstein Summation} expressions by mapping mathematically equivalent formulations to a unique normal form. Batches of einsums with the same Einstein notation that exhibit substantial data reuse appear frequently in finite element methods (FEM), numerical linear algebra, and computational chemistry. To effectively exploit this temporal locality for high performance, we consider groups of einsums in batched form. Representations of equivalent batched einsums may differ due to index renaming, permutations within the batch, and, due to the commutativity and associativity of multiplication operation. The lack of a canonical representation hinders the reuse of optimization and tuning knowledge in software systems. To this end, we develop a novel encoding of batched einsums as colored graphs and apply graph canonicalization to derive a normal form. In addition to the canonicalization algorithm, we propose a representation of einsums using functional array operands and provide a strategy to transfer transformations operating on the normal form to \emph{functional batched einsums} that exhibit the same normal form; crucial for fusing surrounding computations for memory bound einsums. We evaluate our approach against JAX, and observe a geomean speedup of $4.7\times$ for einsums from the TCCG benchmark suite and an FEM solver.

Genetic mutations frequently disrupt protein structure, stability, and solubility, acting as primary drivers for a wide spectrum of diseases. Despite the critical importance of these molecular alterations, existing computational models often lack interpretability, and fail to integrate essential physicochemical interaction. To overcome these limitations, we propose SheafLapNet, a unified predictive framework grounded in the mathematical theory of Topological Deep Learning (TDL) and Persistent Sheaf Laplacian (PSL). Unlike standard Topological Data Analysis (TDA) tools such as persistent homology, which are often insensitive to heterogeneous information, PSL explicitly encodes specific physical and chemical information such as partial charges directly into the topological analysis. SheafLapNet synergizes these sheaf-theoretic invariants with advanced protein transformer features and auxiliary physical descriptors to capture intrinsic molecular interactions in a multiscale and mechanistic manner. To validate our framework, we employ rigorous benchmarks for both regression and classification tasks. For stability prediction, we utilize the comprehensive S2648 and S350 datasets. For solubility prediction, we employ the PON-Sol2 dataset, which provides annotations for increased, decreased, or neutral solubility changes. By integrating these multi-perspective features, SheafLapNet achieves state-of-the-art performance across these diverse benchmarks, demonstrating that sheaf-theoretic modeling significantly enhances both interpretability and generalizability in predicting mutation-induced structural and functional changes.

Wearable foundation models have the potential to transform digital health by learning transferable representations from large-scale biosignals collected in everyday settings. While recent progress has been made in large-scale pretraining, most approaches overlook the spectral structure of photoplethysmography (PPG) signals, wherein physiological rhythms unfold across multiple frequency bands. Motivated by the insight that many downstream health-related tasks depend on multi-resolution features spanning fine-grained waveform morphology to global rhythmic dynamics, we introduce Masked Multiscale Reconstruction (MMR) for PPG representation learning - a self-supervised pretraining framework that explicitly learns from hierarchical time-frequency scales of PPG data. The pretraining task is designed to reconstruct randomly masked out coefficients obtained from a wavelet-based multiresolution decomposition of PPG signals, forcing the transformer encoder to integrate information across temporal and spectral scales. We pretrain our model with MMR using ~17 million unlabeled 10-second PPG segments from ~32,000 smartwatch users. On 17 of 19 diverse health-related tasks, MMR trained on large-scale wearable PPG data improves over or matches state-of-the-art open-source PPG foundation models, time-series foundation models, and other self-supervised baselines. Extensive analysis of our learned embeddings and systematic ablations underscores the value of wavelet-based representations, showing that they capture robust and physiologically-grounded features. Together, these results highlight the potential of MMR as a step toward generalizable PPG foundation models.

Matrix completion is a classical problem that has received recurring interest across a wide range of fields. In this paper, we revisit this problem in an ultra-sparse sampling regime, where each entry of an unknown, $n\times d$ matrix $M$ (with $n \ge d$) is observed independently with probability $p = C / d$, for a fixed integer $C \ge 2$. This setting is motivated by applications involving large, sparse panel datasets, where the number of rows far exceeds the number of columns. When each row contains only $C$ entries -- fewer than the rank of $M$ -- accurate imputation of $M$ is impossible. Instead, we estimate the row span of $M$ or the averaged second-moment matrix $T = M^{\top} M / n$. The empirical second-moment matrix computed from observed entries exhibits non-random and sparse missingness. We propose an unbiased estimator that normalizes each nonzero entry of the second moment by its observed frequency, followed by gradient descent to impute the missing entries of $T$. The normalization divides a weighted sum of $n$ binomial random variables by the total number of ones. We show that the estimator is unbiased for any $p$ and enjoys low variance. When the row vectors of $M$ are drawn uniformly from a rank-$r$ factor model satisfying an incoherence condition, we prove that if $n \ge O({d r^5 ε^{-2} C^{-2} \log d})$, any local minimum of the gradient-descent objective is approximately global and recovers $T$ with error at most $ε^2$. Experiments on both synthetic and real-world data validate our approach. On three MovieLens datasets, our algorithm reduces bias by $88\%$ relative to baseline estimators. We also empirically validate the linear sampling complexity of $n$ relative to $d$ on synthetic data. On an Amazon reviews dataset with sparsity $10^{-7}$, our method reduces the recovery error of $T$ by $59\%$ and $M$ by $38\%$ compared to baseline methods.

Inference time latency has remained an open challenge for real world applications of large language models (LLMs). State-of-the-art (SOTA) speculative sampling (SpS) methods for LLMs, like EAGLE-3, use tree-based drafting to explore multiple candidate continuations in parallel. However, the hyperparameters controlling the tree structure are static, which limits flexibility and efficiency across diverse contexts and domains. We introduce Reinforcement learning for Speculative Sampling (Re-SpS), the first reinforcement learning (RL)-based framework for draft tree hyperparameter optimization. Re-SpS dynamically adjusts draft tree hyperparameters in real-time, learning context-aware policies that maximize generation speed by balancing speculative aggression with computational overhead. It leverages efficient state representations from target model hidden states and introduces multi-step action persistence for better context modeling. Evaluation results across five diverse benchmarks demonstrate consistent improvements over the SOTA method EAGLE-3, achieving up to 5.45$\times$ speedup over the backbone LLM and up to 1.12$\times$ speedup compared to EAGLE-3 across five diverse benchmarks, with no loss in output fidelity.

The Fast Fourier Transform (FFT) is a fundamental numerical technique with widespread application in a range of scientific problems. As scientific simulations attempt to exploit exascale systems, there has been a growing demand for distributed FFT algorithms that can effectively utilize modern heterogeneous high-performance computing (HPC) systems. Conventional FFT algorithms commonly encounter performance bottlenecks, especially when run on heterogeneous platforms. Most distributed FFT approaches rely on static task distribution and require synchronization barriers, limiting scalability and impacting overall resource utilization. In this paper we present DaggerFFT, a distributed FFT framework, developed in Julia, that treats highly parallel FFT computations as a dynamically scheduled task graph. Each FFT stage operates on a separately defined distributed array. FFT operations are expressed as DTasks operating on pencil or slab partitioned DArrays. Each FFT stage owns its own DArray, and the runtime assigns DTasks across devices using Dagger's dynamic scheduler that uses work stealing. We demonstrate how DaggerFFT's dynamic scheduler can outperform state-of-the-art distributed FFT libraries on both CPU and GPU backends, achieving up to a 2.6x speedup on CPU clusters and up to a 1.35x speedup on GPU clusters. We have integrated DaggerFFT into Oceananigans.jl, a geophysical fluid dynamics framework, demonstrating that high-level, task-based runtimes can deliver both superior performance and modularity in large-scale, real-world simulations.

Evaluating conversational systems in multi-turn settings remains a fundamental challenge. Conventional pipelines typically rely on manually defined rubrics and fixed conversational context$-$a static approach that limits coverage and fails to capture the diverse, emergent behaviors of dialogue models. To address this, we introduce CoReflect (Conversational Evaluation via Co-Evolutionary Simulation and Reflective Rubric Refinement), which unifies dialogue simulation and evaluation into an adaptive, iterative process. CoReflect employs a conversation planner that generates structured templates to guide a user simulator through diverse, goal-directed dialogues. Subsequently, a reflective analyzer processes these dialogues to identify systematic behavioral patterns and automatically refine the evaluation rubrics. Crucially, the insights from the conversation analysis are fed back into the planner to update conversation templates for subsequent iterations. This co-evolution loop ensures that the complexity of test cases and the diagnostic precision of rubrics improve in tandem. By minimizing human intervention, CoReflect provides a scalable and self-refining methodology that allows evaluation protocols to adapt alongside the rapidly advancing capabilities of dialogue models.

This paper investigates three crucial yet underexplored aspects of the generalization capabilities of neural audio codecs (NACs): (i) whether NACs can generalize to unseen languages during pre-training, (ii) whether speech-only pre-trained NACs can effectively generalize to non-speech applications such as environmental sounds, music, and animal vocalizations, and (iii) whether incorporating non-speech data during pre-training can improve performance on both speech and non-speech tasks. Existing studies typically rely on off-the-shelf NACs for comparison, which limits insight due to variations in implementation. In this work, we train NACs from scratch using strictly controlled configurations and carefully curated pre-training data to enable fair comparisons. We conduct a comprehensive evaluation of NAC performance on both signal reconstruction quality and downstream applications using 11 metrics. Our results show that NACs can generalize to unseen languages during pre-training, speech-only pre-trained NACs exhibit degraded performance on non-speech tasks, and incorporating non-speech data during pre-training improves performance on non-speech tasks while maintaining comparable performance on speech tasks.

This paper proposes a Dialect Identification (DID) approach inspired by the Connectionist Temporal Classification (CTC) loss function as used in Automatic Speech Recognition (ASR). CTC-DID frames the dialect identification task as a limited-vocabulary ASR system, where dialect tags are treated as a sequence of labels for a given utterance. For training, the repetition of dialect tags in transcriptions is estimated either using a proposed Language-Agnostic Heuristic (LAH) approach or a pre-trained ASR model. The method is evaluated on the low-resource Arabic Dialect Identification (ADI) task, with experimental results demonstrating that an SSL-based CTC-DID model, trained on a limited dataset, outperforms both fine-tuned Whisper and ECAPA-TDNN models. Notably, CTC-DID also surpasses these models in zero-shot evaluation on the Casablanca dataset. The proposed approach is found to be more robust to shorter utterances and is shown to be easily adaptable for streaming, real-time applications, with minimal performance degradation.

Multi-domain thinking verifiers trained via Reinforcement Learning from Verifiable Rewards (RLVR) are a prominent fixture of the Large Language Model (LLM) post-training pipeline, owing to their ability to robustly rate and rerank model outputs. However, the adoption of such verifiers towards code generation has been comparatively sparse, with execution feedback constituting the dominant signal. Nonetheless, code verifiers remain valuable toward judging model outputs in scenarios where execution feedback is hard to obtain and are a potentially powerful addition to the code generation post-training toolbox. To this end, we create and open-source Aletheia, a controlled testbed that enables execution-grounded evaluation of code verifiers' robustness across disparate policy models and covariate shifts. We examine components of the RLVR-based verifier training recipe widely credited for its success: (1) intermediate thinking traces, (2) learning from negative samples, and (3) on-policy training. While experiments show the optimality of RLVR, we uncover important opportunities to simplify the recipe. Particularly, despite code verification exhibiting positive training- and inference-time scaling, on-policy learning stands out as the key component at small verifier sizes, and thinking-based training emerges as the most important component at larger scales.

Modern language models like GPT-3, BERT, and LLaMA require massive training data, yet with sufficient training they reliably learn to distinguish grammatical from ungrammatical sentences. Children aged as young as 14 months already have the capacity to learn abstract grammar rules from very few exemplars, even in the presence of non-rule-following exceptions. Yang's (2016) Tolerance Principle defines a precise threshold for how many exceptions a rule can tolerate and still be learnable. The present study explored the minimal amount and quality of training data necessary for rules to be generalized by a transformer-based language model to test the predictions of the Tolerance Principle. We trained BabyBERTa (Huebner et al. 2021), a transformer model optimized for small datasets, on artificial grammars. The training sets varied in size, number of unique sentence types, and proportion of rule-following versus exception exemplars. We found that, unlike human infants, BabyBERTa's learning dynamics do not align with the Tolerance Principle.

We propose a federated learning framework for the calibration of parametric insurance indices under heterogeneous renewable energy production losses. Producers locally model their losses using Tweedie generalized linear models and private data, while a common index is learned through federated optimization without sharing raw observations. The approach accommodates heterogeneity in variance and link functions and directly minimizes a global deviance objective in a distributed setting. We implement and compare FedAvg, FedProx and FedOpt, and benchmark them against an existing approximation-based aggregation method. An empirical application to solar power production in Germany shows that federated learning recovers comparable index coefficients under moderate heterogeneity, while providing a more general and scalable framework.

Large language models (LLMs) are increasingly deployed as analytical tools across multilingual contexts, yet their outputs may carry systematic biases conditioned by the language of the prompt. This study presents an experimental comparison of LLM-generated political analyses of a Ukrainian civil society document, using semantically equivalent prompts in Russian and Ukrainian. Despite identical source material and parallel query structures, the resulting analyses varied substantially in rhetorical positioning, ideological orientation, and interpretive conclusions. The Russian-language output echoed narratives common in Russian state discourse, characterizing civil society actors as illegitimate elites undermining democratic mandates. The Ukrainian-language output adopted vocabulary characteristic of Western liberal-democratic political science, treating the same actors as legitimate stakeholders within democratic contestation. These findings demonstrate that prompt language alone can produce systematically different ideological orientations from identical models analyzing identical content, with significant implications for AI deployment in polarized information environments, cross-lingual research applications, and the governance of AI systems in multilingual societies.

Projection-based model reduction enables efficient simulation of complex dynamical systems by constructing low-dimensional surrogate models from high-dimensional data. The Operator Inference (OpInf) approach learns such reduced surrogate models through a two-step process: constructing a low-dimensional basis via Singular Value Decomposition (SVD) to compress the data, then solving a linear least-squares (LS) problem to infer reduced operators that govern the dynamics in this compressed space, all without access to the underlying code or full model operators, i.e., non-intrusively. Traditional OpInf operates as a batch learning method, where both the SVD and LS steps process all data simultaneously. This poses a barrier to deployment of the approach on large-scale applications where dataset sizes prevent the loading of all data into memory at once. Additionally, the traditional batch approach does not naturally allow model updates using new data acquired during online computation. To address these limitations, we propose Streaming OpInf, which learns reduced models from sequentially arriving data streams. Our approach employs incremental SVD for adaptive basis construction and recursive LS for streaming operator updates, eliminating the need to store complete data sets while enabling online model adaptation. The approach can flexibly combine different choices of streaming algorithms for numerical linear algebra: we systematically explore the impact of these choices both analytically and numerically to identify effective combinations for accurate reduced model learning. Numerical experiments on benchmark problems and a large-scale turbulent channel flow demonstrate that Streaming OpInf achieves accuracy comparable to batch OpInf while reducing memory requirements by over 99% and enabling dimension reductions exceeding 31,000x, resulting in orders-of-magnitude faster predictions.

This study investigates the use of neural topic modeling and LLMs to uncover meaningful themes from patient storytelling data, to offer insights that could contribute to more patient-oriented healthcare practices. We analyze a collection of transcribed interviews with cancer patients (132,722 words in 13 interviews). We first evaluate BERTopic and Top2Vec for individual interview summarization by using similar preprocessing, chunking, and clustering configurations to ensure a fair comparison on Keyword Extraction. LLMs (GPT4) are then used for the next step topic labeling. Their outputs for a single interview (I0) are rated through a small-scale human evaluation, focusing on {coherence}, {clarity}, and {relevance}. Based on the preliminary results and evaluation, BERTopic shows stronger performance and is selected for further experimentation using three {clinically oriented embedding} models. We then analyzed the full interview collection with the best model setting. Results show that domain-specific embeddings improved topic \textit{precision} and \textit{interpretability}, with BioClinicalBERT producing the most consistent results across transcripts. The global analysis of the full dataset of 13 interviews, using the BioClinicalBERT embedding model, reveals the most dominant topics throughout all 13 interviews, namely ``Coordination and Communication in Cancer Care Management" and ``Patient Decision-Making in Cancer Treatment Journey''. Although the interviews are machine translations from Dutch to English, and clinical professionals are not involved in this evaluation, the findings suggest that neural topic modeling, particularly BERTopic, can help provide useful feedback to clinicians from patient interviews. This pipeline could support more efficient document navigation and strengthen the role of patients' voices in healthcare workflows.

Despite their prominent performance on tasks such as ROI classification and segmentation, many pathology foundation models remain constrained by a specific input size e.g. 224 x 224, creating substantial inefficiencies when applied to whole-slide images (WSIs), which span thousands of resolutions. A naive strategy is to either enlarge inputs or downsample the WSIs. However, enlarging inputs results in prohibitive GPU memory consumption, while downsampling alters the microns-per-pixel resolution and obscures critical morphological details. To overcome these limitations, we propose an space- and time- efficient inference strategy that sparsifies attention using spatially aware neighboring blocks and filters out non-informative tokens through global attention scores. This design substantially reduces GPU memory and runtime during high-resolution WSI inference while preserving and even improving the downstream performance, enabling inference at higher resolutions under the same GPU budget. The experimental results show that our method can achieves up to an 7.67% improvement in the ROI classification and compatible results in segmentation.

Malicious code in open-source repositories such as PyPI poses a growing threat to software supply chains. Traditional rule-based tools often overlook the semantic patterns in source code that are crucial for identifying adversarial components. Large language models (LLMs) show promise for software analysis, yet their use in interpretable and modular security pipelines remains limited. This paper presents LAMPS, a multi-agent system that employs collaborative LLMs to detect malicious PyPI packages. The system consists of four role-specific agents for package retrieval, file extraction, classification, and verdict aggregation, coordinated through the CrewAI framework. A prototype combines a fine-tuned CodeBERT model for classification with LLaMA-3 agents for contextual reasoning. LAMPS has been evaluated on two complementary datasets: D1, a balanced collection of 6,000 setup.py files, and D2, a realistic multi-file dataset with 1,296 files and natural class imbalance. On D1, LAMPS achieves 97.7% accuracy, surpassing MPHunter--one of the state-of-the-art approaches. On D2, it reaches 99.5% accuracy and 99.5% balanced accuracy, outperforming RAG-based approaches and fine-tuned single-agent baselines. McNemar's test confirmed these improvements as highly significant. The results demonstrate the feasibility of distributed LLM reasoning for malicious code detection and highlight the benefits of modular multi-agent designs in software supply chain security.

Segment Anything (SAM) has recently pushed the boundaries of segmentation by demonstrating zero-shot generalization and flexible prompting after training on over one billion masks. Despite this, its mask prediction accuracy often falls short of the precision required in real-world applications. While several refinement modules have been proposed to boost SAM's segmentation quality, achieving highly accurate object delineation within a single, unified framework remains an open challenge. Furthermore, interactive image matting, which aims to generate fine-grained alpha mattes guided by diverse user hints, has not yet been explored in the context of SAM. Insights from recent studies highlight strong correlations between segmentation and matting, suggesting the feasibility of a unified model capable of both tasks. In this paper, we introduce Segment And Matte Anything (SAMA), a lightweight extension of SAM that delivers high-quality interactive image segmentation and matting with minimal extra parameters. Our Multi-View Localization Encoder (MVLE) captures detailed features from local views, while the Localization Adapter (Local-Adapter) refines mask outputs by recovering subtle boundary details. We also incorporate two prediction heads for each task into the architecture to generate segmentation and matting masks, simultaneously. Trained on a diverse dataset aggregated from publicly available sources, SAMA achieves state-of-the-art performance across multiple segmentation and matting benchmarks, showcasing its adaptability and effectiveness in a wide range of downstream tasks.

Large Language Models have demonstrated a remarkable capability in natural language and program generation and software development. However, the source code generated by the LLMs does not always meet quality requirements and may fail to compile. Therefore, many studies evolve into agents that can reason about the problem before generating the source code for the solution. The goal of this paper is to study the degree to which such agents benefit from access to software development tools, in our case, a \texttt{gcc} compiler. We conduct a computational experiment on the RosettaCode dataset, on 699 programming tasks in C. We evaluate how the integration with a compiler shifts the role of the language model from a passive generator to an active agent capable of iteratively developing runnable programs based on feedback from the compiler. We evaluated 16 language models with sizes ranging from small (135 million) to medium (3 billion) and large (70 billion). Our results show that access to a compiler improved the compilation success by 5.3 to 79.4 percentage units in compilation without affecting the semantics of the generated program. Syntax errors dropped by 75\%, and errors related to undefined references dropped by 87\% for the tasks where the agents outperformed the baselines. We also observed that in some cases, smaller models with a compiler outperform larger models with a compiler. We conclude that it is essential for LLMs to have access to software engineering tools to enhance their performance and reduce the need for large models in software engineering, such as reducing our energy footprint.

Softmax attention struggles with long contexts due to structural limitations: the strict sum-to-one constraint forces attention sinks on irrelevant tokens, and probability mass disperses as sequence lengths increase. We tackle these problems with Threshold Differential Attention (TDA), a sink-free attention mechanism that achieves ultra-sparsity and improved robustness at longer sequence lengths without the computational overhead of projection methods or the performance degradation caused by noise accumulation of standard rectified attention. TDA applies row-wise extreme-value thresholding with a length-dependent gate, retaining only exceedances. Inspired by the differential transformer, TDA also subtracts an inhibitory view to enhance expressivity. Theoretically, we prove that TDA controls the expected number of spurious survivors per row to $O(1)$ and that consensus spurious matches across independent views vanish as context grows. Empirically, TDA produces $>99\%$ exact zeros and eliminates attention sinks while maintaining competitive performance on standard and long-context benchmarks.

Task planning with temporally extended goals (TEGs) is a critical challenge in AI and robotics, enabling agents to achieve complex sequences of objectives over time rather than addressing isolated, immediate tasks. Linear Temporal Logic on finite traces (LTLf ) provides a robust formalism for encoding these temporal goals. Traditional LTLf task planning approaches often transform the temporal planning problem into a classical planning problem with reachability goals, which are then solved using off-the-shelf planners. However, these methods often lack informed heuristics to provide a guided search for temporal goals. We introduce TIDE (Trace-Informed Depth-first Exploration), a novel approach that addresses this limitation by decomposing a temporal problem into a sequence of smaller, manageable reach-avoid sub-problems, each solvable using an off-the-shelf planner. TIDE identifies and prioritizes promising automaton traces within the domain graph, using cost-driven heuristics to guide exploration. Its adaptive backtracking mechanism systematically recovers from failed plans by recalculating costs and penalizing infeasible transitions, ensuring completeness and efficiency. Experimental results demonstrate that TIDE achieves promising performance and is a valuable addition to the portfolio of planning methods for temporally extended goals.

Large Language Models (LLMs) are increasingly integrated into vehicle-based digital assistants, where unsafe, ambiguous, or legally incorrect responses can lead to serious safety, ethical, and regulatory consequences. Despite growing interest in LLM safety, existing taxonomies and evaluation frameworks remain largely general-purpose and fail to capture the domain-specific risks inherent to real-world driving scenarios. In this paper, we introduce DriveSafe, a hierarchical, four-level risk taxonomy designed to systematically characterize safety-critical failure modes of LLM-based driving assistants. The taxonomy comprises 129 fine-grained atomic risk categories spanning technical, legal, societal, and ethical dimensions, grounded in real-world driving regulations and safety principles and reviewed by domain experts. To validate the safety relevance and realism of the constructed prompts, we evaluate their refusal behavior across six widely deployed LLMs. Our analysis shows that the evaluated models often fail to appropriately refuse unsafe or non-compliant driving-related queries, underscoring the limitations of general-purpose safety alignment in driving contexts.

The relentless scaling of deep learning models has led to unsustainable computational demands, positioning Mixture-of-Experts (MoE) architectures as a promising path towards greater efficiency. However, MoE models are plagued by two fundamental challenges: 1) a load imbalance problem known as the``rich get richer" phenomenon, where a few experts are over-utilized, and 2) an expert homogeneity problem, where experts learn redundant representations, negating their purpose. Current solutions typically employ an auxiliary load-balancing loss that, while mitigating imbalance, often exacerbates homogeneity by enforcing uniform routing at the expense of specialization. To resolve this, we introduce the Eigen-Mixture-of-Experts (EMoE), a novel architecture that leverages a routing mechanism based on a learned orthonormal eigenbasis. EMoE projects input tokens onto this shared eigenbasis and routes them based on their alignment with the principal components of the feature space. This principled, geometric partitioning of data intrinsically promotes both balanced expert utilization and the development of diverse, specialized experts, all without the need for a conflicting auxiliary loss function. Our code is publicly available at https://github.com/Belis0811/EMoE.

As AI assistance becomes embedded in programming practice, researchers have increasingly examined how these systems help learners generate code and work more efficiently. However, these studies often position AI as a replacement for human collaboration and overlook the social and learning-oriented aspects that emerge in collaborative programming. Our work introduces human-human-AI (HHAI) triadic programming, where an AI agent serves as an additional collaborator rather than a substitute for a human partner. Through a within-subjects study with 20 participants, we show that triadic collaboration enhances collaborative learning and social presence compared to the dyadic human-AI (HAI) baseline. In the triadic HHAI conditions, participants relied significantly less on AI-generated code in their work. This effect was strongest in the HHAI-shared condition, where participants had an increased sense of responsibility to understand AI suggestions before applying them. These findings demonstrate how triadic settings activate socially shared regulation of learning by making AI use visible and accountable to a human peer, suggesting that AI systems that augment rather than automate peer collaboration can better preserve the learning processes that collaborative programming relies on.

Bengali text classification is a Significant task in natural language processing (NLP), where text is categorized into predefined labels. Unlike English, Bengali faces challenges due to the lack of extensive annotated datasets and pre-trained language models. This study explores the effectiveness of large language models (LLMs) in classifying Bengali newspaper articles. The dataset used, obtained from Kaggle, consists of articles from Prothom Alo, a major Bangladeshi newspaper. Three instruction-tuned LLMs LLaMA 3.1 8B Instruct, LLaMA 3.2 3B Instruct, and Qwen 2.5 7B Instruct were evaluated for this task under the same classification framework. Among the evaluated models, Qwen 2.5 achieved the highest classification accuracy of 72%, showing particular strength in the "Sports" category. In comparison, LLaMA 3.1 and LLaMA 3.2 attained accuracies of 53% and 56%, respectively. The findings highlight the effectiveness of LLMs in Bengali text classification, despite the scarcity of resources for Bengali NLP. Future research will focus on exploring additional models, addressing class imbalance issues, and refining fine-tuning approaches to improve classification performance.

Solar activity, including solar flares, coronal mass ejections (CMEs), and geomagnetic storms, can significantly impact satellites, aviation, power grids, data centers, and space missions. Extreme solar events can cause substantial economic damage if not predicted in advance, highlighting the importance of accurate forecasting and effective education in space science. Although large language models (LLMs) perform well on general tasks, they often lack domain-specific knowledge and pedagogical capability to clearly explain complex space science concepts. We introduce SolarGPT-QA, a question answering system based on a domain-adapted large language model built on the LLaMA-3 base model. The model is trained using scientific literature and large-scale question-answer data generated with GPT-4 and refined using Grok-3 in a student-friendly storytelling style. Human pairwise evaluations show that SolarGPT-QA outperforms general-purpose models in zero-shot settings and achieves competitive performance compared to instruction-tuned models for educational explanations in space weather and heliophysics. A small pilot student comprehension study further suggests improved clarity and accessibility of the generated explanations. Ablation experiments indicate that combining domain-adaptive pretraining with pedagogical fine-tuning is important for balancing scientific accuracy and educational effectiveness. This work represents an initial step toward a broader SolarGPT framework for space science education and forecasting.

Existing 3D human motion generation and understanding methods often exhibit limited interpretability, restricting effective mutual enhancement between these inherently related tasks. While current unified frameworks based on large language models (LLMs) leverage linguistic priors, they frequently encounter challenges in semantic alignment and task coherence. Moreover, the next-token prediction paradigm in LLMs is ill-suited for motion sequences, causing cumulative prediction errors. To address these limitations, we propose UniMo, a novel framework that integrates motion-language information and interpretable chain of thought (CoT) reasoning into the LLM via supervised fine-tuning (SFT). We further introduce reinforcement learning with Group Relative Policy Optimization (GRPO) as a post-training strategy that optimizes over groups of tokens to enforce structural correctness and semantic alignment, mitigating cumulative errors in motion token prediction. Extensive experiments demonstrate that UniMo significantly outperforms existing unified and task-specific models, achieving state-of-the-art performance in both motion generation and understanding.

The use of synthetic data in health applications raises privacy concerns, yet the lack of open frameworks for privacy evaluations has slowed its adoption. A major challenge is the absence of accessible benchmark datasets for evaluating privacy risks, due to difficulties in acquiring sensitive data. To address this, we introduce SynQP, an open framework for benchmarking privacy in synthetic data generation (SDG) using simulated sensitive data, ensuring that original data remains confidential. We also highlight the need for privacy metrics that fairly account for the probabilistic nature of machine learning models. As a demonstration, we use SynQP to benchmark CTGAN and propose a new identity disclosure risk metric that offers a more accurate estimation of privacy risks compared to existing approaches. Our work provides a critical tool for improving the transparency and reliability of privacy evaluations, enabling safer use of synthetic data in health-related applications. % In our quality evaluations, non-private models achieved near-perfect machine-learning efficacy \(\ge0.97\). Our privacy assessments (Table II) reveal that DP consistently lowers both identity disclosure risk (SD-IDR) and membership-inference attack risk (SD-MIA), with all DP-augmented models staying below the 0.09 regulatory threshold. Code available at https://github.com/CAN-SYNH/SynQP

Offline policy learning aims to use historical data to learn an optimal personalized decision rule. In the standard estimate-then-optimize framework, reweighting-based methods (e.g., inverse propensity weighting or doubly robust estimators) are widely used to produce unbiased estimates of policy values. However, when the propensity scores of some treatments are small, these reweighting-based methods suffer from high variance in policy value estimation, which may mislead the downstream policy optimization and yield a learned policy with inferior value. In this paper, we systematically develop an offline policy learning algorithm based on a weight-clipping estimator that truncates small propensity scores via a clipping threshold chosen to minimize the mean squared error (MSE) in policy value estimation. Focusing on linear policies, we address the bilevel and discontinuous objective induced by weight-clipping-based policy optimization by reformulating the problem as a Heaviside composite optimization problem, which provides a rigorous computational framework. The reformulated policy optimization problem is then solved efficiently using the progressive integer programming method, making practical policy learning tractable. We establish an upper bound for the suboptimality of the proposed algorithm, which reveals how the reduction in MSE of policy value estimation, enabled by our proposed weight-clipping estimator, leads to improved policy learning performance.

Fine-tuned language models pose significant privacy risks, as they may memorize and expose sensitive information from their training data. Membership inference attacks (MIAs) provide a principled framework for auditing these risks, yet existing methods achieve limited detection rates, particularly at the low false-positive thresholds required for practical privacy auditing. We present EZ-MIA, a membership inference attack that exploits a key observation: memorization manifests most strongly at error positions, specifically tokens where the model predicts incorrectly yet still shows elevated probability for training examples. We introduce the Error Zone (EZ) score, which measures the directional imbalance of probability shifts at error positions relative to a pretrained reference model. This principled statistic requires only two forward passes per query and no model training of any kind. On WikiText with GPT-2, EZ-MIA achieves 3.8x higher detection than the previous state-of-the-art under identical conditions (66.3% versus 17.5% true positive rate at 1% false positive rate), with near-perfect discrimination (AUC 0.98). At the stringent 0.1% FPR threshold critical for real-world auditing, we achieve 8x higher detection than prior work (14.0% versus 1.8%), requiring no reference model training. These gains extend to larger architectures: on AG News with Llama-2-7B, we achieve 3x higher detection (46.7% versus 15.8% TPR at 1% FPR). These results establish that privacy risks of fine-tuned language models are substantially greater than previously understood, with implications for both privacy auditing and deployment decisions. Code is available at https://github.com/JetBrains-Research/ez-mia.

Large language models (LLMs) have shown promise for automated text annotation, raising hopes that they might accelerate cross-cultural research by extracting structured data from ethnographic texts. We evaluated 7 state-of-the-art LLMs on their ability to annotate 121 ritual features across 567 ethnographic excerpts. Performance was limited, falling well below levels required for reliable automated annotation. Longer texts, features requiring ordinal distinctions, and ambiguous constructs proved particularly difficult. Human inter-coder reliability set an approximate ceiling on LLM accuracy: features that human coders found difficult to agree upon were also difficult for LLMs. Yet even on features where humans reliably agreed, models fell short of human performance. Our findings suggest that LLMs cannot yet substitute for human expertise in ethnographic annotation.

Neural network models often face challenges when processing very small or very large numbers due to issues such as overflow, underflow, and unstable output variations. To mitigate these problems, we propose using embedding vectors for numbers instead of directly using their raw values. These embeddings aim to retain essential algebraic properties while preventing numerical instabilities. In this paper, we introduce, for the first time, a fixed-length number embedding vector that preserves algebraic operations, including addition, multiplication, and comparison, within the field of rational numbers. We propose a novel Neural Isomorphic Field, a neural abstraction of algebraic structures such as groups and fields. The elements of this neural field are embedding vectors that maintain algebraic structure during computations. Our experiments demonstrate that addition performs exceptionally well, achieving over 95 percent accuracy on key algebraic tests such as identity, closure, and associativity. In contrast, multiplication exhibits challenges, with accuracy ranging from 53 percent to 73 percent across various algebraic properties. These findings highlight the model's strengths in preserving algebraic properties under addition while identifying avenues for further improvement in handling multiplication.

Accurately and efficiently solving nonlinear differential equations is crucial for modeling dynamic behavior across science and engineering. Physics-Informed Neural Networks (PINNs) have emerged as a powerful solution that embeds physical laws in training by enforcing equation residuals. However, these struggle to model nonlinear dynamics, suffering from limited generalization across problems and long training times. To address these limitations, we propose a perturbation-guided transfer learning framework for PINNs (PTL-PINN), which integrates perturbation theory with transfer learning to efficiently solve nonlinear equations. Unlike gradient-based transfer learning, PTL-PINNs solve an approximate linear perturbative system using closed-form expressions, enabling rapid generalization with the time complexity of matrix-vector multiplication. We show that PTL-PINNs achieve accuracy comparable to various Runge-Kutta methods, with computational speeds up to one order of magnitude faster. To benchmark performance, we solve a broad set of problems, including nonlinear oscillators across various damping regimes, the equilibrium-centered Lotka-Volterra system, the KPP-Fisher and the Wave equation. Since perturbation theory sets the accuracy bound of PTL-PINNs, we systematically evaluate its practical applicability. This work connects long-standing perturbation methods with PINNs, demonstrating how perturbation theory can guide foundational models to solve nonlinear systems with speeds comparable to those of classical solvers.

Large language models (LLMs) exhibit systematic Western-centric bias, yet whether prompting in non-Western languages (e.g., Chinese) can mitigate this remains understudied. Answering this question requires rigorous evaluation and effective mitigation, but existing approaches fall short on both fronts: evaluation methods force outputs into predefined cultural categories without a neutral option, while mitigation relies on expensive multi-cultural corpora or agent frameworks that use functional roles (e.g., Planner--Critique) lacking explicit cultural representation. To address these gaps, we introduce CEBiasBench, a Chinese--English bilingual benchmark, and Multi-Agent Vote (MAV), which enables explicit ``no bias'' judgments. Using this framework, we find that Chinese prompting merely shifts bias toward East Asian perspectives rather than eliminating it. To mitigate such persistent bias, we propose Multi-Agent Cultural Debate (MACD), a training-free framework that assigns agents distinct cultural personas and orchestrates deliberation via a "Seeking Common Ground while Reserving Differences" strategy. Experiments demonstrate that MACD achieves 57.6% average No Bias Rate evaluated by LLM-as-judge and 86.0% evaluated by MAV (vs. 47.6% and 69.0% baseline using GPT-4o as backbone) on CEBiasBench and generalizes to the Arabic CAMeL benchmark, confirming that explicit cultural representation in agent frameworks is essential for cross-cultural fairness.

Spatio-Temporal (ST) Foundation Models (STFMs) promise cross-dataset generalization, yet joint ST pretraining is computationally expensive and grapples with the heterogeneity of domain-specific spatial patterns. Substantially extending our preliminary conference version, we present FactoST-v2, an enhanced factorized framework redesigned for full weight transfer and arbitrary-length generalization. FactoST-v2 decouples universal temporal learning from domain-specific spatial adaptation. The first stage pretrains a minimalist encoder-only backbone using randomized sequence masking to capture invariant temporal dynamics, enabling probabilistic quantile prediction across variable horizons. The second stage employs a streamlined adapter to rapidly inject spatial awareness via meta adaptive learning and prompting. Comprehensive evaluations across diverse domains demonstrate that FactoST-v2 achieves state-of-the-art accuracy with linear efficiency - significantly outperforming existing foundation models in zero-shot and few-shot scenarios while rivaling domain-specific expert baselines. This factorized paradigm offers a practical, scalable path toward truly universal STFMs. Code is available at https://github.com/CityMind-Lab/FactoST.

Assisting pathologists in the analysis of histopathological images has high clinical value, as it supports cancer detection and staging. In this context, histology foundation models have recently emerged. Among them, Vision-Language Models (VLMs) provide strong yet imperfect zero-shot predictions. We propose to refine these predictions by adapting Conditional Random Fields (CRFs) to histopathological applications, requiring no additional model training. We present HistoCRF, a CRF-based framework, with a novel definition of the pairwise potential that promotes label diversity and leverages expert annotations. We consider three experiments: without annotations, with expert annotations, and with iterative human-in-the-loop annotations that progressively correct misclassified patches. Experiments on five patch-level classification datasets covering different organs and diseases demonstrate average accuracy gains of 16.0% without annotations and 27.5% with only 100 annotations, compared to zero-shot predictions. Moreover, integrating a human in the loop reaches a further gain of 32.6% with the same number of annotations. The code will be made available on https://github.com/tgodelaine/HistoCRF.

Large Language Models (LLMs) excel at general-purpose tasks, yet adapting their responses to individual users remains challenging. Retrieval augmentation provides a lightweight alternative to fine-tuning by conditioning LLMs on user history records, and existing approaches typically select these records based on semantic relevance. We argue that relevance serves as an unreliable proxy for utility: a record may be semantically similar to a query yet fail to improve generation quality or even degrade it due to redundancy or conflicting information. To bridge this gap, we propose PURPLE, a contextual bandit framework that oPtimizes UseR Profiles for Llm pErsonalization. In contrast to a greedy selection of the most relevant records, PURPLE treats profile construction as a set generation process and utilizes a Plackett-Luce ranking model to capture complex inter-record dependencies. By training with dense feedback provided by the likelihood of the reference response, our method aligns retrieval directly with generation quality. Extensive experiments on nine personalization tasks demonstrate that PURPLE consistently outperforms strong heuristic and retrieval-augmented baselines in both effectiveness and efficiency, establishing a principled and scalable solution for optimizing user profiles.

Remote sensing video referring object segmentation (RS-RVOS) is challenged by weak target saliency and severe visual information truncation in dynamic scenes, making it extremely difficult to maintain discriminative target representations during segmentation. Moreover, progress in this field is hindered by the absence of large-scale dedicated benchmarks, while existing models are often affected by biased initial memory construction that impairs accurate instance localization in complex scenarios, as well as indiscriminate memory accumulation that encodes noise from occlusions or misclassifications, leading to persistent error propagation. This paper advances RS-RVOS research through dual contributions in data and methodology. First, we construct RS-RVOS Bench, the first large-scale benchmark comprising 111 video sequences, about 25,000 frames, and 213,000 temporal referring annotations. Unlike common RVOS benchmarks where many expressions are written with access to the full video context, our dataset adopts a strict causality-aware annotation strategy in which linguistic references are generated solely from the target state in the initial frame. Second, we propose a memory-quality-aware online referring segmentation framework, termed Memory Quality Control with Segment Anything Model (MQC-SAM). MQC-SAM introduces a temporal motion consistency module for initial memory calibration, leveraging short-term motion trajectory priors to correct structural deviations and establish accurate memory anchoring. Furthermore, it incorporates a decoupled attention-based memory integration mechanism with dynamic quality assessment, selectively updating high-confidence semantic features while filtering unreliable information, thereby effectively preventing error accumulation and propagation. Extensive experiments on RS-RVOS Bench demonstrate that MQC-SAM achieves state-of-the-art performance.

Language models used in retrieval-augmented settings must arbitrate between parametric knowledge stored in their weights and contextual information in the prompt. This work presents a mechanistic study of that choice by extracting an \emph{arbitration vector} from model activations on a curated dataset designed to disentangle (i) irrelevant contexts that elicit parametric recall and (ii) relevant but false contexts that elicit copying. The vector is computed as the residual-stream centroid difference between these regimes across 27 relations, and is injected as an additive intervention at selected layers and token spans to steer behavior in two directions: Copy$\rightarrow$Recall (suppressing context use) and Recall$\rightarrow$Copy (inducing the model to copy any token from the context). Experiments on two architectures (decoder-only and encoder/decoder) and two open-domain QA benchmarks show consistent behavior shifts under moderate scaling while monitoring accuracy and fluency. Mechanistic analyses of attention routing, MLP contributions, and layer-wise probability trajectories reveal an asymmetry: inducing copying is an easy ``reactivation'' process that can be triggered at different locations in the input, while restoring recall is a ``suppression'' process that is more fragile and strongly tied to object-token interventions.

Increased access to reliable health information is essential for non-English-speaking populations, yet resources in Bangla for disease prediction remain limited. This study addresses this gap by developing a comprehensive Bangla symptoms-disease dataset containing 758 unique symptom-disease relationships spanning 85 diseases. To ensure transparency and reproducibility, we also make our dataset publicly available. The dataset enables the prediction of diseases based on Bangla symptom inputs, supporting healthcare accessibility for Bengali-speaking populations. Using this dataset, we evaluated multiple machine learning models to predict diseases based on symptoms provided in Bangla and analyzed their performance on our dataset. Both soft and hard voting ensemble approaches combining top-performing models achieved 98\% accuracy, demonstrating superior robustness and generalization. Our work establishes a foundational resource for disease prediction in Bangla, paving the way for future advancements in localized health informatics and diagnostic tools. This contribution aims to enhance equitable access to health information for Bangla-speaking communities, particularly for early disease detection and healthcare interventions.

Dialogue Act (DA) annotation typically treats communicative or pedagogical intent as localized to individual utterances or turns. This leads annotators to agree on the underlying action while disagreeing on segment boundaries, reducing apparent reliability. We propose codebook-injected segmentation, which conditions boundary decisions on downstream annotation criteria, and evaluate LLM-based segmenters against standard and retrieval-augmented baselines. To assess these without gold labels, we introduce evaluation metrics for span consistency, distinctiveness, and human-AI distributional agreement. We found DA-awareness produces segments that are internally more consistent than text-only baselines. While LLMs excel at creating construct-consistent spans, coherence-based baselines remain superior at detecting global shifts in dialogue flow. Across two datasets, no single segmenter dominates. Improvements in within-segment coherence frequently trade off against boundary distinctiveness and human-AI distributional agreement. These results highlight segmentation as a consequential design choice that should be optimized for downstream objectives rather than a single performance score.

Unsupervised deep image prior (DIP) addresses shortcomings of training data requirements and limited generalization associated with supervised deep learning. The performance of DIP depends on the network architecture and the stopping point of its iterative process. Optimizing these parameters for a new image requires time, restricting DIP application in domains where many images need to be processed. Focusing on fluorescence microscopy data, we hypothesize that similar images share comparable optimal parameter configurations for DIP-based denoising, potentially enabling optimization-free DIP for fluorescence microscopy. We generated a calibration (n=110) and validation set (n=55) of semantically different images from an open-source dataset for a network architecture search targeted towards ideal U-net architectures and stopping points. The calibration set represented our transfer basis. The validation set enabled the assessment of which image similarity criterion yields the best results. We then implemented AUTO-DIP, a pipeline for automatic parameter transfer, and compared it to the originally published DIP configuration (baseline) and a state-of-the-art image-specific variational denoising approach. We show that a parameter transfer from the calibration dataset to a test image based on only image metadata similarity (e.g., microscope type, imaged specimen) leads to similar and better performance than a transfer based on quantitative image similarity measures. AUTO-DIP outperforms the baseline DIP (DIP with original DIP parameters) as well as the variational denoising approaches for several open-source test datasets of varying complexity, particularly for very noisy inputs. Applications to locally acquired fluorescence microscopy images further proved superiority of AUTO-DIP.

While foundation models have achieved remarkable results across a diversity of domains, they still rely on human-generated data, such as text, as a fundamental source of knowledge. However, this data is ultimately the product of human brains, the filtered projection of a deeper neural complexity. In this paper, we explore a new strategy for artificial intelligence: moving beyond surface-level statistical regularities by training foundation models directly on human brain data. We hypothesize that neuroimaging data could open a window into elements of human cognition that are not accessible through observable actions, and argue that this additional knowledge could be used, alongside classical training data, to overcome some of the current limitations of foundation models. While previous research has demonstrated the possibility to train classical machine learning or deep learning models on neural patterns, this path remains largely unexplored for high-level cognitive functions. Here, we classify the current limitations of foundation models, as well as the promising brain regions and cognitive processes that could be leveraged to address them, along four levels: perception, valuation, execution, and integration. Then, we propose two methods that could be implemented to prioritize the use of limited neuroimaging data for strategically chosen, high-value steps in foundation model training: reinforcement learning from human brain (RLHB) and chain of thought from human brain (CoTHB). We also discuss the potential implications for agents, artificial general intelligence, and artificial superintelligence, as well as the ethical, social, and technical challenges and opportunities. We argue that brain-trained foundation models could represent a realistic and effective middle ground between continuing to scale current architectures and exploring alternative, neuroscience-inspired solutions.

Interpretability of modern visual models is crucial, particularly in high-stakes applications. However, existing interpretability methods typically suffer from either reliance on white-box model access or insufficient quantitative rigor. To address these limitations, we introduce FocaLogic, a novel model-agnostic framework designed to interpret and quantify visual model decision-making through logic-based representations. FocaLogic identifies minimal interpretable subsets of visual regions-termed visual focuses-that decisively influence model predictions. It translates these visual focuses into precise and compact logical expressions, enabling transparent and structured interpretations. Additionally, we propose a suite of quantitative metrics, including focus precision, recall, and divergence, to objectively evaluate model behavior across diverse scenarios. Empirical analyses demonstrate FocaLogic's capability to uncover critical insights such as training-induced concentration, increasing focus accuracy through generalization, and anomalous focuses under biases and adversarial attacks. Overall, FocaLogic provides a systematic, scalable, and quantitative solution for interpreting visual models.

Visual token compression is widely adopted to improve the inference efficiency of Large Vision-Language Models (LVLMs), enabling their deployment in latency-sensitive and resource-constrained scenarios. However, existing work has mainly focused on efficiency and performance, while the security implications of visual token compression remain largely unexplored. In this work, we first reveal that visual token compression substantially degrades the robustness of LVLMs: models that are robust under uncompressed inference become highly vulnerable once compression is enabled. These vulnerabilities are state-specific; failure modes emerge only in the compressed setting and completely disappear when compression is disabled, making them particularly hidden and difficult to diagnose. By analyzing the key stages of the compression process, we identify instability in token importance ranking as the primary cause of this robustness degradation. Small and imperceptible perturbations can significantly alter token rankings, leading the compression mechanism to mistakenly discard task-critical information and ultimately causing model failure. Motivated by this observation, we propose a Compression-Aware Attack to systematically study and exploit this vulnerability. CAA directly targets the token selection mechanism and induces failures exclusively under compressed inference. We further extend this approach to more realistic black-box settings and introduce Transfer CAA, where neither the target model nor the compression configuration is accessible. We further evaluate potential defenses and find that they provide only limited protection. Extensive experiments across models, datasets, and compression methods show that visual token compression significantly undermines robustness, revealing a previously overlooked efficiency-security trade-off.

The use of Large Language Models (LLMs) for reasoning and planning tasks has drawn increasing attention in Artificial Intelligence research. Despite their remarkable progress, these models still exhibit limitations in multi-step inference scenarios, particularly in mathematical and logical reasoning. We introduce PREGU (Partial Reasoning Guided by Uncertainty). PREGU monitors the entropy of the output distribution during autoregressive generation and halts the process whenever entropy exceeds a defined threshold, signaling uncertainty. From that point, a localized search is performed in the latent space to refine the partial reasoning and select the most coherent answer, using the Soft Reasoning method. Experiments conducted with LLaMA-3-8B, Mistral-7B, and Qwen2-7B across four reasoning benchmarks (GSM8K, GSM-Hard, SVAMP, and StrategyQA) showed performance greater than or similar to Soft Reasoning, indicating that entropy can serve as an effective signal to trigger selective refinement during reasoning.

The paper develops a Transformer architecture for estimating dynamic factors from multivariate time series data under flexible identification assumptions. Performance on small datasets is improved substantially by using a conventional factor model as prior information via a regularization term in the training objective. The results are interpreted with Attention matrices that quantify the relative importance of variables and their lags for the factor estimate. Time variation in Attention patterns can help detect regime switches and evaluate narratives. Monte Carlo experiments suggest that the Transformer is more accurate than the linear factor model, when the data deviate from linear-Gaussian assumptions. An empirical application uses the Transformer to construct a coincident index of U.S. real economic activity.

Dung's abstract argumentation framework characterises argument acceptability solely via an attack relation, deliberately abstracting from the internal structure of arguments. While this level of abstraction has enabled a rich body of results, it limits the ability to represent structural dependencies that are central in many structured argumentation formalisms, in particular subargument relations. Existing extensions, including bipolar argumentation frameworks, introduce support relations, but these do not capture the asymmetric and constitutive nature of subarguments or their interaction with attacks. In this paper, we study abstract argumentation frameworks enriched with an explicit subargument relation, treated alongside attack as a basic relation. We analyse how subargument relations interact with attacks and examine their impact on fundamental semantic properties. This framework provides a principled abstraction of structural information and clarifies the role of subarguments in abstract acceptability reasoning.

Personalization in Large Language Models (LLMs) often relies on user-specific soft prompts. However, these prompts become obsolete when the foundation model is upgraded, necessitating costly, full-scale retraining. To overcome this limitation, we propose the Prompt-level User Migration Adapter (PUMA), a lightweight framework to efficiently migrate personalized prompts across incompatible models. PUMA utilizes a parameter-efficient adapter to bridge the semantic gap, combined with a group-based user selection strategy to significantly reduce training costs. Experiments on three large-scale datasets show our method matches or even surpasses the performance of retraining from scratch, reducing computational cost by up to 98%. The framework demonstrates strong generalization across diverse model architectures and robustness in advanced scenarios like chained and aggregated migrations, offering a practical path for the sustainable evolution of personalized AI by decoupling user assets from the underlying models.

Quantization is widely adopted to reduce the computational cost of large language models (LLMs); however, its implications for fairness and safety, particularly in dynamic quantization and multilingual contexts, remain underexplored. In this work, we conduct a systematic study of how static and dynamic quantization methods impact fairness and safety across benchmarks measuring intrinsic and extrinsic bias and safety alignment. For fairness, we evaluate English, French, Dutch, Spanish, and Turkish; for safety, we focus on English, Korean, and Arabic. Our findings reveal that quantization consistently degrades fairness and safety, with dynamic methods demonstrating greater stability than static ones. Moreover, fairness degradation varies across languages, while safety deterioration is especially pronounced in non-English settings. To address these risks, we introduce Critical Weight Protection, a novel technique that identifies and preserves fairness- and safety-critical weights during quantization. This approach effectively mitigates bias and safety deterioration without costly retraining or alignment, maintaining trustworthiness while retaining efficiency.

This definitive research memoria presents a comprehensive, mathematically verified paradigm for neural communication with Bitcoin mining Application-Specific Integrated Circuits (ASICs), integrating five complementary frameworks: thermodynamic reservoir computing, hierarchical number system theory, algorithmic analysis, network latency optimization, and machine-checked mathematical formalization. We establish that obsolete cryptocurrency mining hardware exhibits emergent computational properties enabling bidirectional information exchange between AI systems and silicon substrates. The research program demonstrates: (1) reservoir computing with NARMA-10 Normalized Root Mean Square Error (NRMSE) of 0.8661; (2) the Thermodynamic Probability Filter (TPF) achieving 92.19% theoretical energy reduction; (3) the Virtual Block Manager achieving +25% effective hashrate; and (4) hardware universality across multiple ASIC families including Antminer S9, Lucky Miner LV06, and Goldshell LB-Box. A significant contribution is the machine-checked mathematical formalization using Lean 4 and Mathlib, providing unambiguous definitions, machine-verified theorems, and reviewer-proof claims. Key theorems proven include: independence implies zero leakage, predictor beats baseline implies non-independence (the logical core of TPF), energy savings theoretical maximum, and Physical Unclonable Function (PUF) distinguishability witnesses. Vladimir Veselov's hierarchical number system theory explains why early-round information contains predictive power. This work establishes a new paradigm: treating ASICs not as passive computational substrates but as active conversational partners whose thermodynamic state encodes exploitable computational information.

Large language models are increasingly deployed as research agents for deep search and long-horizon information seeking, yet their performance often degrades as interaction histories grow. This degradation, known as context rot, reflects a failure to maintain coherent and task-relevant internal states over extended reasoning horizons. Existing approaches primarily manage context through raw accumulation or passive summarization, treating it as a static artifact and allowing early errors or misplaced emphasis to persist. Motivated by this perspective, we propose ARC, which is the first framework to systematically formulate context management as an active, reflection-driven process that treats context as a dynamic internal reasoning state during execution. ARC operationalizes this view through reflection-driven monitoring and revision, allowing agents to actively reorganize their working context when misalignment or degradation is detected. Experiments on challenging long-horizon information-seeking benchmarks show that ARC consistently outperforms passive context compression methods, achieving up to an 11% absolute improvement in accuracy on BrowseComp-ZH with Qwen2.5-32B-Instruct.

Customer reviews contain rich signals about product weaknesses and unmet user needs, yet existing analytic methods rarely move beyond descriptive tasks such as sentiment analysis or aspect extraction. While large language models (LLMs) can generate free-form suggestions, their outputs often lack accuracy and depth of reasoning. In this paper, we present a multi-agent, LLM-based framework for prescriptive decision support, which transforms large scale review corpora into actionable business advice. The framework integrates four components: clustering to select representative reviews, generation of advices, iterative evaluation, and feasibility based ranking. This design couples corpus distillation with feedback driven advice refinement to produce outputs that are specific, actionable, and practical. Experiments across three service domains and multiple model families show that our framework consistently outperform single model baselines on actionability, specificity, and non-redundancy, with medium sized models approaching the performance of large model frameworks.

Semi-implicit variational inference (SIVI) enhances the expressiveness of variational families through hierarchical semi-implicit distributions, but the intractability of their densities makes standard ELBO-based optimization biased. Recent score-matching approaches to SIVI (SIVI-SM) address this issue via a minimax formulation, at the expense of an additional lower-level optimization problem. In this paper, we propose kernel semi-implicit variational inference (KSIVI), a principled and tractable alternative that eliminates the lower-level optimization by leveraging kernel methods. We show that when optimizing over a reproducing kernel Hilbert space, the lower-level problem admits an explicit solution, reducing the objective to the kernel Stein discrepancy (KSD). Exploiting the hierarchical structure of semi-implicit distributions, the resulting KSD objective can be efficiently optimized using stochastic gradient methods. We establish optimization guarantees via variance bounds on Monte Carlo gradient estimators and derive statistical generalization bounds of order $\tilde{\mathcal{O}}(1/\sqrt{n})$. We further introduce a multi-layer hierarchical extension that improves expressiveness while preserving tractability. Empirical results on synthetic and real-world Bayesian inference tasks demonstrate the effectiveness of KSIVI.

The widespread proliferation of online content has intensified concerns about clickbait, deceptive or exaggerated headlines designed to attract attention. While Large Language Models (LLMs) offer a promising avenue for addressing this issue, their effectiveness is often hindered by Sycophancy, a tendency to produce reasoning that matches users' beliefs over truthful ones, which deviates from instruction-following principles. Rather than treating sycophancy as a flaw to be eliminated, this work proposes a novel approach that initially harnesses this behavior to generate contrastive reasoning from opposing perspectives. Specifically, we design a Self-renewal Opposing-stance Reasoning Generation (SORG) framework that prompts LLMs to produce high-quality agree and disagree reasoning pairs for a given news title without requiring ground-truth labels. To utilize the generated reasoning, we develop a local Opposing Reasoning-based Clickbait Detection (ORCD) model that integrates three BERT encoders to represent the title and its associated reasoning. The model leverages contrastive learning, guided by soft labels derived from LLM-generated credibility scores, to enhance detection robustness. Experimental evaluations on three benchmark datasets demonstrate that our method consistently outperforms LLM prompting, fine-tuned smaller language models, and state-of-the-art clickbait detection baselines.

Large Language Models (LLMs) are increasingly required to generate structured, machine-readable outputs for downstream systems. While recent benchmarks have focused on evaluating the structural correctness of such outputs, the environmental impact of inference for different output formats has largely been overlooked. In this paper, we argue that structured output formats should be assessed not only in terms of correctness, but also with respect to their environmental efficiency. To this end, we introduce a sustainability-aware evaluation framework for structured generation that measures token usage, generation time, and estimated carbon emissions. Within this framework, we propose the Environment-Aware Generation Correctness Score (GCS_env), a unified metric that integrates structural correctness with carbon-aware efficiency. Using this framework, we systematically benchmark the novel TOON format against established representations (JSON, XML, YAML) across multiple LLMs spanning different architectures and parameter scales. Our results reveal a consistent trade-off: TOON yields markedly more compact outputs and lower emissions, but lower structural correctness when models lack native support. We show that increased model capacity reduces this gap and that environment-aware scoring can shift format rankings depending on deployment priorities. highlighting the need for sustainability-inclusive benchmarking and provides empirical evidence that compact representations such as TOON can offer practical advantages in large-scale, carbon-conscious LLM deployments.

The application of loss reweighting in modern deep learning presents a nuanced picture. While it fails to alter the terminal learning phase in overparameterized deep neural networks (DNNs) trained on high-dimensional datasets, empirical evidence consistently shows it offers significant benefits early in training. To transparently demonstrate and analyze this phenomenon, we introduce a small-scale model (SSM). This model is specifically designed to abstract the inherent complexities of both the DNN architecture and the input data, while maintaining key information about the structure of imbalance within its spectral components. On the one hand, the SSM reveals how vanilla empirical risk minimization preferentially learns to distinguish majority classes over minorities early in training, consequently delaying minority learning. In stark contrast, reweighting restores balanced learning dynamics, enabling the simultaneous learning of features associated with both majorities and minorities.

Ensuring safety is a critical challenge in applying Reinforcement Learning (RL) to real-world scenarios. Constrained Reinforcement Learning (CRL) addresses this by maximizing returns under predefined constraints, typically formulated as the expected cumulative cost. However, expectation-based constraints overlook rare but high-impact extreme value events in the tail distribution, such as black swan incidents, which can lead to severe constraint violations. To address this issue, we propose the Extreme Value policy Optimization (EVO) algorithm, leveraging Extreme Value Theory (EVT) to model and exploit extreme reward and cost samples, reducing constraint violations. EVO introduces an extreme quantile optimization objective to explicitly capture extreme samples in the cost tail distribution. Additionally, we propose an extreme prioritization mechanism during replay, amplifying the learning signal from rare but high-impact extreme samples. Theoretically, we establish upper bounds on expected constraint violations during policy updates, guaranteeing strict constraint satisfaction at a zero-violation quantile level. Further, we demonstrate that EVO achieves a lower probability of constraint violations than expectation-based methods and exhibits lower variance than quantile regression methods. Extensive experiments show that EVO significantly reduces constraint violations during training while maintaining competitive policy performance compared to baselines.

Autonomous systems often operate in multi-agent settings and need to make concurrent, strategic decisions, typically in uncertain environments. Verification and control problems for these systems can be tackled with concurrent stochastic games (CSGs), but this model requires transition probabilities to be precisely specified - an unrealistic requirement in many real-world settings. We introduce *robust CSGs* and their subclass *interval CSGs* (ICSGs), which capture epistemic uncertainty about transition probabilities in CSGs. We propose a novel framework for *robust* verification of these models under worst-case assumptions about transition uncertainty. Specifically, we develop the underlying theoretical foundations and efficient algorithms, for finite- and infinite-horizon objectives in both zero-sum and nonzero-sum settings, the latter based on (social-welfare optimal) Nash equilibria. We build an implementation in the PRISM-games model checker and demonstrate the feasibility of robust verification of ICSGs across a selection of large benchmarks.

The rapid integration of AI algorithms in safety-critical applications such as autonomous driving and healthcare is raising significant concerns about the ability to meet stringent safety standards. Traditional tools for formal safety verification struggle with the black-box nature of AI-driven systems and lack the flexibility needed to scale to the complexity of real-world applications. In this paper, we present a data-driven approach for safety verification and synthesis of black-box systems with discrete-time stochastic dynamics. We employ the concept of control barrier certificates, which can guarantee safety of the system, and learn the certificate directly from a set of system trajectories. We use conditional mean embeddings to embed data from the system into a reproducing kernel Hilbert space (RKHS) and construct an RKHS ambiguity set that can be inflated to robustify the result to out-of-distribution behavior. We provide the theoretical results on how to apply the approach to general classes of temporal logic specifications beyond safety. For the data-driven computation of safety barriers, we leverage a finite Fourier expansion to cast a typically intractable semi-infinite optimization problem as a linear program. The resulting spectral barrier allows us to leverage the fast Fourier transform to generate the relaxed problem efficiently, offering a scalable yet distributionally robust framework for verifying safety. Our work moves beyond restrictive assumptions on system dynamics and uncertainty, as demonstrated on two case studies including a black-box system with a neural network controller.

Intrusion detection systems (IDS) are essential for protecting computer systems and networks against a wide range of cyber threats that continue to evolve over time. IDS are commonly categorized into two main types, each with its own strengths and limitations, such as difficulty in detecting previously unseen attacks and the tendency to generate high false positive rates. This paper presents a comprehensive survey and a conceptual overview of Hybrid IDS, which integrate signature-based and anomaly-based detection techniques to enhance attack detection capabilities. The survey examines recent research on Hybrid IDS, classifies existing models into functional categories, and discusses their advantages, limitations, and application domains, including financial systems, air traffic control, and social networks. In addition, recent trends in Hybrid IDS research, such as machine learning-based approaches and cloud-based deployments, are reviewed. Finally, this work outlines potential future research directions aimed at developing more cost-effective Hybrid IDS solutions with improved ability to detect emerging and sophisticated cyberattacks.

NoSQL Injection attacks are a class of cybersecurity attacks where an attacker sends a specifically engineered query to a NoSQL database which then performs an unauthorized operation. To defend against such attacks, rule based systems were initially developed but then were found to be ineffective to innovative injection attacks hence a model based approach was developed. Most model based detection systems, during testing gave exponentially positive results but were trained only on the query statement sent to the server. However due to the scarcity of data and class imbalances these model based systems were found to be not effective against all attacks in the real world. This paper explores classifying NoSQL injection attacks sent to a MongoDB server based on Log Data, and other extracted features excluding raw query statements. The log data was collected from a simulated attack on an empty MongoDB server which was then processed and explored. A discriminant analysis was carried out to determine statistically significant features to discriminate between injection and benign queries resulting in a dataset of significant features. Several Machine learning based classification models using an AutoML library, "FLAML", as well as 6 manually programmed models were trained on this dataset , which were then trained on 50 randomized samples of data, cross validated and evaluated. The study found that the best model was the "FLAML" library's "XGBoost limited depth" model with an accuracy of 71%.

Learning robust audio-visual embeddings requires bringing genuinely related audio and visual signals together while filtering out incidental co-occurrences - background noise, unrelated elements, or unannotated events. Most contrastive and triplet-loss methods use sparse annotated labels per clip and treat any co-occurrence as semantic similarity. For example, a video labeled "train" might also contain motorcycle audio and visual, because "motorcycle" is not the chosen annotation; standard methods treat these co-occurrences as negatives to true motorcycle anchors elsewhere, creating false negatives and missing true cross-modal dependencies. We propose a framework that leverages soft-label predictions and inferred latent interactions to address these issues: (1) Audio-Visual Semantic Alignment Loss (AV-SAL) trains a teacher network to produce aligned soft-label distributions across modalities, assigning nonzero probability to co-occurring but unannotated events and enriching the supervision signal. (2) Inferred Latent Interaction Graph (ILI) applies the GRaSP algorithm to teacher soft labels to infer a sparse, directed dependency graph among classes. This graph highlights directional dependencies (e.g., "Train (visual)" -> "Motorcycle (audio)") that expose likely semantic or conditional relationships between classes; these are interpreted as estimated dependency patterns. (3) Latent Interaction Regularizer (LIR): A student network is trained with both metric loss and a regularizer guided by the ILI graph, pulling together embeddings of dependency-linked but unlabeled pairs in proportion to their soft-label probabilities. Experiments on AVE and VEGAS benchmarks show consistent improvements in mean average precision (mAP), demonstrating that integrating inferred latent interactions into embedding learning enhances robustness and semantic coherence.

In-context learning (ICL) has proven highly effective across diverse large language model (LLM) tasks. However, its potential for enhancing tasks that demand step-by-step logical deduction, such as mathematical reasoning, remains underexplored. A core limitation of existing ICL approaches is their static use of demonstrations: examples are pre-selected before inference and remain fixed, failing to adapt to the dynamic confusion points that often arise during multi-step reasoning such as ambiguous calculations or logical gaps. These unresolved confusion points can lead to cascading errors that degrade final accuracy. To tackle this issue, we propose Process In-Context Learning (PICL), a dynamic demonstration integration framework designed to boost mathematical reasoning by responding to real-time inference needs. PICL operates in two stages: 1)~it identifies potential confusion points by analyzing semantics and entropy in the reasoning process and summarizes their core characteristics; 2)~upon encountering these points, it retrieves relevant demonstrations from the demonstration pool that match the confusion context and inserts them directly into the ongoing reasoning process to guide subsequent steps. Experiments show that PICL outperforms baseline methods by mitigating mid-inference confusion, highlighting the value of adaptive demonstration insertion in complex mathematical reasoning.

Forecasting in power systems often involves multivariate time series with complex dependencies and strict privacy constraints across regions. Traditional forecasting methods require significant expert knowledge and struggle to generalize across diverse deployment scenarios. Recent advancements in pre-trained time series models offer new opportunities, but their zero-shot performance on domain-specific tasks remains limited. To address these challenges, we propose a novel MoE Encoder module that augments pretrained forecasting models by injecting a sparse mixture-of-experts layer between tokenization and encoding. This design enables two key capabilities: (1) trans forming multivariate forecasting into an expert-guided univariate task, allowing the model to effectively capture inter-variable relations, and (2) supporting localized training and lightweight parameter sharing in federated settings where raw data cannot be exchanged. Extensive experiments on public multivariate datasets demonstrate that MoE-Encoder significantly improves forecasting accuracy compared to strong baselines. We further simulate federated environments and show that transferring only MoE-Encoder parameters allows efficient adaptation to new regions, with minimal performance degradation. Our findings suggest that MoE-Encoder provides a scalable and privacy-aware extension to foundation time series models.

While Large Language Models (LLMs) enable complex autonomous behavior, current agents remain constrained by static, human-designed prompts that limit adaptability. Existing self-improving frameworks attempt to bridge this gap but typically rely on inefficient, multi-turn recursive loops that incur high computational costs. To address this, we propose Metacognitive Agent Reflective Self-improvement (MARS), a framework that achieves efficient self-evolution within a single recurrence cycle. Inspired by educational psychology, MARS mimics human learning by integrating principle-based reflection (abstracting normative rules to avoid errors) and procedural reflection (deriving step-by-step strategies for success). By synthesizing these insights into optimized instructions, MARS allows agents to systematically refine their reasoning logic without continuous online feedback. Extensive experiments on six benchmarks demonstrate that MARS outperforms state-of-the-art self-evolving systems while significantly reducing computational overhead.

Fuzzing continues to be the most effective method for identifying security vulnerabilities in software. In the context of fuzz testing, the fuzzer supplies varied inputs to fuzz targets, which are designed to comprehensively exercise critical sections of the client code. Various studies have focused on optimizing and developing advanced fuzzers, such as AFL++, libFuzzer, Honggfuzz, syzkaller, ISP-Fuzzer, which have substantially enhanced vulnerability detection in widely used software and libraries. Nevertheless, achieving greater coverage necessitates improvements in both the quality and quantity of fuzz targets. In large-scale software projects and libraries -- characterized by numerous user defined functions and data types -- manual creation of fuzz targets is both labor-intensive and time-consuming. This challenge underscores the need for automated techniques not only to generate fuzz targets but also to streamline the execution and analysis of their results. In this paper, we introduce an approach to improving fuzz target generation through static analysis of library source code. The proposed method encompasses several key aspects: it analyzes source code structures to accurately construct function calls and generate fuzz targets; it maps fuzzer input data to the corresponding function parameters; it synthesizes compilation information for the fuzz targets; and it automatically collects and analyzes execution results. Our findings are demonstrated through the application of this approach to the generation of fuzz targets for C/C++ libraries.

Existing works increasingly adopt memory-centric mechanisms to process long contexts in a segment manner, and effective memory management is one of the key capabilities that enables large language models to effectively propagate information across the entire sequence. Therefore, leveraging reward models (RMs) to automatically and reliably evaluate memory quality is critical. In this work, we introduce $\texttt{MemoryRewardBench}$, the first benchmark to systematically study the ability of RMs to evaluate long-term memory management processes. $\texttt{MemoryRewardBench}$ covers both long-context comprehension and long-form generation tasks, featuring 10 distinct settings with different memory management patterns, with context length ranging from 8K to 128K tokens. Evaluations on 13 cutting-edge RMs indicate a diminishing performance gap between open-source and proprietary models, with newer-generation models consistently outperforming their predecessors regardless of parameter count. We further expose the capabilities and fundamental limitations of current RMs in evaluating LLM memory management across diverse settings.

Reinforcement learning has become a central paradigm for improving LLM reasoning. However, existing methods use a single policy to produce both inference responses and training optimization trajectories. The objective conflict between generating stable inference responses and diverse training trajectories leads to insufficient exploration, which harms reasoning capability. In this paper, to address the problem, we propose R$^2$PO (Residual Rollout Policy Optimization), which introduces a lightweight Residual Rollout-Head atop the policy to decouple training trajectories from inference responses, enabling controlled trajectory diversification during training while keeping inference generation stable. Experiments across multiple benchmarks show that our method consistently outperforms baselines, achieving average accuracy gains of 3.1% on MATH-500 and 2.4% on APPS, while also reducing formatting errors and mitigating length bias for stable optimization. Our code is publicly available at https://github.com/RRPO-ARR/Code.

Overlapping calendar invitations force busy professionals to repeatedly decide which meetings to attend, reschedule, or decline. We refer to this preference-driven decision process as calendar conflict resolution. Automating such process is crucial yet challenging. Scheduling logistics drain hours, and human delegation often fails at scale, which motivate we to ask: Can we trust large language model (LLM) or language agent to manager time? To enable systematic study of this question, we introduce CalConflictBench, a benchmark for long-horizon calendar conflict resolution. Conflicts are presented sequentially and agents receive feedback after each round, requiring them to infer and adapt to user preferences progressively. Our experiments show that current LLM agents perform poorly with high error rates, e.g., Qwen-3-30B-Think has 35% average error rate. To address this gap, we propose PEARL, a reinforcement-learning framework that augments language agent with an external memory module and optimized round-wise reward design, enabling agent to progressively infer and adapt to user preferences on-the-fly. Experiments on CalConflictBench shows that PEARL achieves 0.76 error reduction rate, and 55% improvement in average error rate compared to the strongest baseline.

Trustworthy reasoning in Large Language Models (LLMs) is challenged by their propensity for hallucination. While augmenting LLMs with Knowledge Graphs (KGs) improves factual accuracy, existing KG-augmented methods fail to quantify epistemic uncertainty in both the retrieved evidence and LLMs' reasoning. To bridge this gap, we introduce DoublyCal, a framework built on a novel double-calibration principle. DoublyCal employs a lightweight proxy model to first generate KG evidence alongside a calibrated evidence confidence. This calibrated supporting evidence then guides a black-box LLM, yielding final predictions that are not only more accurate but also well-calibrated, with confidence scores traceable to the uncertainty of the supporting evidence. Experiments on knowledge-intensive benchmarks show that DoublyCal significantly improves both the accuracy and confidence calibration of black-box LLMs with low token cost.

Dynamic graphs have attracted increasing attention due to their ability to model complex and evolving relationships in real-world scenarios. Traditional approaches typically pre-train models using dynamic link prediction and directly apply the resulting node temporal embeddings to specific downstream tasks. However, the significant differences among downstream tasks often lead to performance degradation, especially under few-shot settings. Prompt tuning has emerged as an effective solution to this problem. Existing prompting methods are often strongly coupled with specific model architectures or pretraining tasks, which makes it difficult to adapt to recent or future model designs. Moreover, their exclusive focus on modifying node or temporal features while neglecting spatial structural information leads to limited expressiveness and degraded performance. To address these limitations, we propose DDGPrompt, a data-centric prompting framework designed to effectively refine pre-trained node embeddings at the input data level, enabling better adaptability to diverse downstream tasks. We first define a unified node expression feature matrix that aggregates all relevant temporal and structural information of each node, ensuring compatibility with a wide range of dynamic graph models. Then, we introduce three prompt matrices (temporal bias, edge weight, and feature mask) to adjust the feature matrix completely, achieving task-specific adaptation of node embeddings. We evaluate DDGPrompt under a strict few-shot setting on four public dynamic graph datasets. Experimental results demonstrate that our method significantly outperforms traditional methods and prompting approaches in scenarios with limited labels and cold-start conditions.

Constrained Reinforcement Learning (CRL) aims to maximize cumulative rewards while satisfying constraints. However, existing CRL algorithms often encounter significant constraint violations during training, limiting their applicability in safety-critical scenarios. In this paper, we identify the underestimation of the cost value function as a key factor contributing to these violations. To address this issue, we propose the Memory-driven Intrinsic Cost Estimation (MICE) method, which introduces intrinsic costs to mitigate underestimation and control bias to promote safer exploration. Inspired by flashbulb memory, where humans vividly recall dangerous experiences to avoid risks, MICE constructs a memory module that stores previously explored unsafe states to identify high-cost regions. The intrinsic cost is formulated as the pseudo-count of the current state visiting these risk regions. Furthermore, we propose an extrinsic-intrinsic cost value function that incorporates intrinsic costs and adopts a bias correction strategy. Using this function, we formulate an optimization objective within the trust region, along with corresponding optimization methods. Theoretically, we provide convergence guarantees for the proposed cost value function and establish the worst-case constraint violation for the MICE update. Extensive experiments demonstrate that MICE significantly reduces constraint violations while preserving policy performance comparable to baselines.

Quantum neural networks (QNNs) have attracted growing interest for scientific machine learning, yet in regression settings they often suffer from limited trainability under noisy gradients and ill-conditioned optimization. We propose a hybrid quantum-classical regression framework designed to mitigate these bottlenecks. Our model prepends a lightweight classical embedding that acts as a learnable geometric preconditioner, reshaping the input representation to better condition a downstream variational quantum circuit. Building on this architecture, we introduce a curriculum optimization protocol that progressively increases circuit depth and transitions from SPSA-based stochastic exploration to Adam-based gradient fine-tuning. We evaluate the approach on PDE-informed regression benchmarks and standard regression datasets under a fixed training budget in a simulator setting. Empirically, the proposed framework consistently improves over pure QNN baselines and yields more stable convergence in data-limited regimes. We further observe reduced structured errors that are visually correlated with oscillatory components on several scientific benchmarks, suggesting that geometric preconditioning combined with curriculum training is a practical approach for stabilizing quantum regression.

Scaling test-time compute via Long Chain-of-Thought (Long-CoT) significantly enhances reasoning capabilities, yet extended generation does not guarantee correctness: after an early wrong commitment, models may keep elaborating a self-consistent but incorrect prefix. Through fine-grained trajectory analysis, we identify Thinking Traps, prefix-dominant deadlocks where later reflection, alternative attempts, or verification fails to revise the root error. On a curated subset of DAPO-MATH, 89\% of failures exhibit such traps. To solve this problem, we introduce TAAR (Trap-Aware Adaptive Restart), a test-time control framework that trains a diagnostic policy to predict two signals from partial trajectories: a trap index for where to truncate and an escape probability for whether and how strongly to intervene. At inference time, TAAR truncates the trajectory before the predicted trap segment and adaptively restarts decoding; for severely trapped cases, it applies stronger perturbations, including higher-temperature resampling and an optional structured reboot suffix. Experiments on challenging mathematical and scientific reasoning benchmarks (AIME24, AIME25, GPQA-Diamond, HMMT25, BRUMO25) show that TAAR improves reasoning performance without fine-tuning base model parameters.

High-Energy Physics (HEP) experiments, such as those at the Large Hadron Collider (LHC), generate massive datasets that challenge classical computational limits. Quantum Machine Learning (QML) offers a potential advantage in processing high-dimensional data; however, finding the optimal architecture for current Noisy Intermediate-Scale Quantum (NISQ) devices remains an open challenge. This study investigates the performance of Variational Quantum Classifiers (VQC) in detecting Higgs Boson signals using the ATLAS Higgs Boson Machine Learning Challenge 2014 experiment dataset. We implemented a dimensionality reduction pipeline using Principal Component Analysis (PCA) to map 30 physical features into 4-qubit and 8-qubit latent spaces. We benchmarked three configurations: (A) a shallow 4-qubit circuit, (B) a deep 4-qubit circuit with increased entanglement layers, and (C) an expanded 8-qubit circuit. Experimental results demonstrate that increasing circuit depth significantly improves performance, yielding the highest accuracy of 56.2% (Configuration B), compared to a baseline of 51.9%. Conversely, simply scaling to 8 qubits resulted in a performance degradation to 50.6% due to optimization challenges associated with Barren Plateaus in the larger Hilbert space. These findings suggest that for near-term quantum hardware, prioritizing circuit depth and entanglement capability is more critical than increasing qubit count for effective anomaly detection in HEP data.

Cloud data centers face increasing pressure to reduce operational energy consumption as big data workloads continue to grow in scale and complexity. This paper presents a workload aware and energy efficient scheduling framework that profiles CPU utilization, memory demand, and storage IO behavior to guide virtual machine placement decisions. By combining historical execution logs with real time telemetry, the proposed system predicts the energy and performance impact of candidate placements and enables adaptive consolidation while preserving service level agreement compliance. The framework is evaluated using representative Hadoop MapReduce, Spark MLlib, and ETL workloads deployed on a multi node cloud testbed. Experimental results demonstrate consistent energy savings of 15 to 20 percent compared to a baseline scheduler, with negligible performance degradation. These findings highlight workload profiling as a practical and scalable strategy for improving the sustainability of cloud based big data processing environments.

The current state of event detection research has two notable re-occurring limitations that we investigate in this study. First, the unidirectional nature of decoder-only LLMs presents a fundamental architectural bottleneck for natural language understanding tasks that depend on rich, bidirectional context. Second, we confront the conventional reliance on Micro-F1 scores in event detection literature, which systematically inflates performance by favoring majority classes. Instead, we focus on Macro-F1 as a more representative measure of a model's ability across the long-tail of event types. Our experiments demonstrate that models enhanced with sentence context achieve superior performance over canonical decoder-only baselines. Using Low-Rank Adaptation (LoRA) during finetuning provides a substantial boost in Macro-F1 scores in particular, especially for the decoder-only models, showing that LoRA can be an effective tool to enhance LLMs' performance on long-tailed event classes.

Machine learning enabled systems (MLS) often operate in settings where they regularly encounter uncertainties arising from changes in their surrounding environment. Without structured oversight, such changes can degrade model behavior, increase operational cost, and reduce the usefulness of deployed systems. Although Machine Learning Operations (MLOps) streamlines the lifecycle of ML models, it provides limited support for addressing runtime uncertainties that influence the longer term sustainability of MLS. To support continued viability, these systems need a mechanism that detects when execution drifts outside acceptable bounds and adjusts system behavior in response. Despite the growing interest in sustainable and self-adaptive MLS, there has been limited work towards exemplars that allow researchers to study these challenges in MLOps pipelines. This paper presents Harmonica, a self-adaptation exemplar built on the HarmonE approach, designed to enable the sustainable operation of such pipelines. Harmonica introduces structured adaptive control through MAPE-K loop, separating high-level adaptation policy from low-level tactic execution. It continuously monitors sustainability metrics, evaluates them against dynamic adaptation boundaries, and automatically triggers architectural tactics when thresholds are violated. We demonstrate the tool through case studies in time series regression and computer vision, examining its ability to improve system stability and reduce manual intervention. The results show that Harmonica offers a practical and reusable foundation for enabling adaptive behavior in MLS that rely on MLOps pipelines for sustained operation.

Recent observations of large unconventional spin-orbit torques in van der Waals (vdW) materials are driving intense interest for energy-efficient spintronic applications. A key limitation of ferromagnet (FM)/vdW heterostructures is their lower value of damping-like torque conductivity ($σ{\rm_{DL}^{y}}$) compared to the conventional heavy metal-based systems, limiting their prospects for commercial spintronic devices. Here, we report both a giant $σ{\rm_{DL}^{y}}$ of $-(1.25 \pm 0.11)\times 10^{5}~\hbar/ 2e~Ω^{-1}$m$^{-1}$ and an unconventional spin-orbit torque in a heterostructure comprising an FM (Ni$_{80}$Fe$_{20}$) and the vdW material GeTe. The value of $σ{\rm_{DL}^{y}}$ represents the highest reported torque conductivity for any FM/vdW interface and is comparable to benchmark heavy metal heterostructures. First-principles calculations reveal that this substantial torque originates from the cooperative interplay of the spin Hall effect, orbital Hall effect, and orbital Rashba effect, assisted by interfacial charge transfer. These findings demonstrate the potential of carefully engineered vdW heterostructures to achieve highly efficient electrical manipulation of magnetization at room temperature, paving the way for next-generation low-power spintronic devices.

This work investigates the electronic properties of twisted bilayer graphene (TBG) through computational calculations, with the aim of understanding the emergence of flat bands and conditions favorable for superconductivity close to the magic angle. This study utilizes a k\cdot p continuum model, and the low-energy Hamiltonians are derived from angle-dependent datasets provided by Carr et al. Using this model, the band structure, density of states (DoS), and Fermi velocity are systematically calculated across a range of twist angles. The calculations are performed by discretizing high-symmetry paths in the moire Brillouin zone for band structure calculations, uniformly sampling a square grid for DoS analysis, and employing finite-difference methods to evaluate the Fermi velocity near the Dirac points. The results identify a narrow magic-angle window around $θ\approx 0.98^\circ-1.00^\circ$, where the bands become nearly dispersionless, the DoS exhibits a sharp peak, and the Fermi velocity is strongly suppressed. This computational framework does not directly predict superconductivity, but rather establishes the electronic foundation for exploring flat-band physics and correlation-driven phenomena such as unconventional superconductivity in twisted bilayer graphene.

Small-scale intermittency is a defining feature of fully developed fluid turbulence, marked by rare and extreme fluctuations of velocity increments and gradients that defy mean-field descriptions. Existing multifractal descriptions of intermittency focus primarily on longitudinal increments and gradients, despite mounting evidence that transverse components exhibit distinct and stronger intermittency. Here, we develop a unified multifractal framework that jointly prescribes longitudinal and transverse velocity increments, and extends to gradients. We derive explicit relations linking inertial-range scaling exponents of structure functions to moments of velocity gradients in dissipation range. Our results reveal that longitudinal gradient scaling is solely prescribed by longitudinal structure functions, as traditionally expected; however, transverse gradient scaling is prescribed by mixed longitudinal-transverse structure functions. Validation with high-resolution direct numerical simulations of isotropic turbulence, at Taylor-scale Reynolds number up to $1300$ demonstrates excellent agreement, paving way for a more complete and predictive description of intermittency faithful to the underlying turbulence dynamics.

We explore the zero-phonon line of single photon emitters in helium-ion treated monolayer MoS$_2$, which are currently understood in terms of single sulfur-site vacancies. By comparing the linewidths of the zero-phonon line as extracted directly from optical spectra with values inferred from the first-order autocorrelation function of the photoluminescence, we quantify bounds of the homogeneous broadening and of phonon-assisted contributions. The results are discussed in terms of both the independent boson model and ab-initio results as computed from GW and Bethe-Salpeter equation approximations.

Spin-wave spectrum of monolayer CrSBr waveguides was studied by numerically diagonalizing the Bogoliubov-de Gennes Hamiltonian derived from linearising the Landau-Lifshitz-Gilbert equation. In contrast to its short-range counterparts, the long-range dipolar field acts statically as a confining potential for spin wave, while the dynamic part couples the spin and orbit degrees of freedom, thus giving rise to spin-orbit coupling for spin wave. Due to the inversion symmetry of the Hamiltonian and the spinor structure of the wave function, spin-wave eigenstates form doublets with definite parity. Micromagnetic simulation tallies well with numerical calculation. Our study on spin-wave eigenstates in CrSBr waveguides sheds light on the nature of exchange-dipole spin wave in ferromagnetic slabs. We confirm particularly that the robustness of the Damon-Eshbach mode is not derived from topology, but rather from the static dipolar field. Moreover, a thorough knowledge on spin wave in monolayer CrSBr itself represents a step forward to understanding the more complicated antiferromagnetic resonance in bulk CrSBr.

We introduce an efficient first-principles framework for simulating angle-resolved photoemission spectroscopy (ARPES) by computing photoelectron states as solutions of the Kohn-Sham equation with scattering boundary conditions. This approach is formally equivalent to the Lippmann-Schwinger formalism but offers superior computational efficiency and direct integration with plane-wave or real-space density functional theory. By enabling direct calculation of photoemission matrix elements, our method bridges the gap between intrinsic electronic properties and experimental ARPES spectra. We demonstrate its accuracy through circular dichroism ARPES simulations for monolayer graphene and bulk $2H$-WSe$_2$, achieving excellent agreement with experimental data and highlighting the critical role of pseudopotentials in describing high-energy photoelectron scattering. Our results establish a robust and accessible route for quantitative ARPES modeling, paving the way for advanced studies of orbital textures, many-body effects, and time-resolved photoemission.

The combination of disorder and interactions is known in many systems to produce a feature in the single-particle density of states, the shape and parameter dependence of which act as signatures of the underlying electronic state. Strong Coulomb repulsion gives rise to a host of novel phenomena, among these is a unique zero-bias anomaly. While understanding of the anomaly in bulk materials remains incomplete, a version of this anomaly can be found in an ensemble of two-site systems and hence has been predicted to be observable in parallel-coupled double quantum dots. However, prior work did not include the influence of the leads. Here we show that the presence of the leads results in changes to the projected stability diagrams but that the signature of the strongly-correlated zero-bias anomaly nonetheless remains clearly visible.

Standard quantum metrology relies on ensemble-averaged quantities, such as the Quantum Fisher Information (QFI), which often mask the fluctuations inherent to single-shot realizations. In this work, we bridge the gap between quantum information geometry and stochastic thermodynamics by introducing the Conditional Quantum Fisher Information (CQFI). Defined via the Symmetric Logarithmic Derivative, the CQFI generalizes the classical stochastic Fisher information to the quantum domain. We demonstrate that the CQFI admits a decomposition into incoherent (population) and coherent (basis rotation) contributions, augmented by a transient interference cross-term absent at the ensemble level. Crucially, we show that this cross-term can be negative, signaling destructive interference between classical and quantum information channels along individual trajectories. Leveraging this framework, we construct a stochastic information geometry that defines thermodynamic length and action for single quantum trajectories. Finally, we derive fundamental quantum speed limits valid at the single-trajectory level and validate our results using the quantum jump unraveling of a driven thermal qubit.

We investigate operator delocalization in disordered one-dimensional spin chains by introducing -- besides the already known operator mass -- a complementary measure of operator complexity: the operator length. Like the operator nonstabilizerness, both these quantities are defined from the expansion of time-evolved operators in the Pauli basis. They characterize, respectively, the number of sites on which an operator acts nontrivially and the spatial extent of its support. We show that both the operator mass and length can be computed efficiently and exactly within a matrix-product-operator (MPS) framework, providing direct access to their full probability distributions, without resorting to stochastic sampling. Applying this approach to the disordered XXZ spin-1/2 chain, we find sharply distinct behaviors in non-interacting and interacting regimes. In the Anderson-localized case, operator mass, length, and operator entanglement entropy rapidly saturate, signaling the absence of scrambling. By contrast, in the many-body localized (MBL) regime, for arbitrarily weak interactions, all quantities exhibit a robust logarithmic growth in time, consistent with the known logarithmic light cone of quantum-correlation propagation in MBL. We demonstrate that this behavior is quantitatively captured by an effective $\ell$-bit model and persists across system sizes accessible via tensor-network simulations.

We present a unified description of symmetric two-state Markov relaxation and intrinsic entanglement dynamics based on thermo-field dynamics (TFD). A classical two-state Markov process is embedded into a dissipative two-level quantum system by identifying the Markov relaxation rate with the dissipation parameter in a von Neumann equation with a relaxation term. Using the reduced extended density matrix in the TFD formalism, we explicitly separate classical thermal mixing from intrinsic quantum entanglement. For a minimal exchange-like two-level subspace, we obtain a closed-form expression for the intrinsic entanglement component, $b_{qe}(t)=\frac{1}{4}e^{-λt}\sin^2(ωt)$, demonstrating that a single classical timescale controls the decay envelope of genuine entanglement. We further show that the extended entanglement entropy naturally decomposes into a classical Shannon-type contribution and a purely quantum entanglement contribution, clarifying how stochastic relaxation constrains entanglement loss in a minimal setting.

Chiral superconductivity is a time reversal symmetry breaking superconducting phase that has attracted broad interest as a potential platform for topological quantum computation. A fundamental consequence of this symmetry breaking is orbital magnetization, yet a clear microscopic formulation of this quantity has remained elusive. This difficulty arises because Bogoliubov quasiparticles do not carry a definite electric charge, precluding a simple interpretation of orbital magnetization in terms of circulating quasiparticle currents. Moreover, superconductivity and ferromagnetism rarely coexist, and in the few materials where they do (e.g. uranium-based compounds), strong spin-orbit coupling obscures the orbital contribution to the magnetization. The recent report of chiral superconductivity in rhombohedral multilayer graphene, which has negligible spin-orbit coupling, therefore provides a unique opportunity to develop and test a microscopic theory of orbital magnetization in chiral superconductors. Here we develop such a theory, unifying the interband coherence effects underlying normal-state orbital magnetization with the intrinsic orbital moments of the Cooper-pair condensate. Applying our theory to rhombohedral tetralayer graphene, we find that the onset of superconductivity can either enhance or suppress the normal-state orbital magnetization, depending sensitively on the bandstructure. We further identify a generalized clapping mode corresponding to coherent fluctuations between the two opposite chiral windings of the p-wave order parameter, with a gap set by the sublattice winding form factor. This collective mode is unique to chiral superconductors and contributes to the orbital magnetization through its role in dressing the photon vertex. Experimental measurements of the orbital magnetization relative to the quarter-metal phase would provide a direct test of our theory.

Entanglement entropy (EE) is a fundamental probe of quantum phases and critical phenomena, which was thought to reflect only bulk universality for a long time. Very recently, people realized that the microscopic geometry of the entanglement cut can induce distinct entanglement-edge modes, whose coupling to bulk critical fluctuations may alter the scaling of the EE. However, this perception is very qualitative and lacks quantitative consideration. Here, we investigate this problem through high-precision quantum Monte Carlo simulations combined with the analysis of scaling theory to build a quantitative understanding. By considering three distinct bipartitions corresponding to three surface criticality types, we reveal a striking dependence of the constant term γ on the microscopic cut at the quantum critical point. Notably, cuts that generate extra gapless edge modes yield a sign reversal in γ compared to those producing gapped edges. We explain this behavior via a modified scaling form that incorporates contributions from both bulk and surface critical modes. Furthermore, we demonstrate that the derivative of EE robustly extracts the bulk critical point and exponent ν regardless of the cut geometry, providing a reliable diagnostic of bulk universality in the presence of strong surface effects. Our work for the first time establishes a direct quantitative connection between surface criticality and entanglement scaling, challenging the conventional view that EE solely reflects bulk properties and offering a refined framework for interpreting entanglement in systems with boundary-sensitive criticality.

We consider a correlated site percolation problem on a cubic lattice of size $L^3$, with $16\le L\le 512$. The sites of an initially full lattice are removed by a random walk of ${\cal N}=uL^3$ steps. When the parameter $u$ crosses a threshold $u_c=3.15$, a large system transitions between percolating and non-percolating states. We study the $L$-dependence of the mean mass (number of sites) $M_r$ of the $r$th largest cluster, as well as $r$-dependence of $M_r$ for various system sizes $L$ at $u_c$. We demonstrate that $M_r\sim L^{5/2}/r^{5/6}$ for moderate or large $L$ and $r\gg 1$, and also conclude that for {\em any} $r$ the fractal dimensions of the clusters are $5/2$.

We uncover a universal scaling law governing the dispersion of collective attention and identify its underlying stochastic criticality. By analysing large-scale ensembles of Wikipedia page views, we find that the variance of logarithmic attention grows ultraslowly, $\operatorname{Var}[\ln{X(t)}]\propto\ln{t}$, in sharp contrast to the power-law scaling typically expected for diffusive processes. We show that this behaviour is captured by a minimal stochastic differential equation driven by fractional Brownian motion, in which long-range memory ($H$) and temporal decay of volatility ($η$) enter through the single exponent $ξ\equiv H-η$. At marginality, $ξ=0$, the variance grows logarithmically, marking the critical boundary between power-law growth ($ξ>0$) and saturation ($ξ<0$). By incorporating article-level heterogeneity through a Gaussian mixture model, we further reconstruct the empirical distribution of cumulative attention within the same framework. Our results place collective attention in a distinct class of non-Markovian stochastic processes, with close affinity to ageing-like and ultraslow dynamics in glassy systems.

Nanocrystals (NCs) serve as versatile building blocks for the creation of functional materials, with NC self-assembly offering opportunities to enable novel material properties. Here, we demonstrate that PbS NCs functionalized with strongly negatively charged metal chalcogenide complex (MCC) ligands, such as $Sn_2S_6^{4-}$ and $AsS_4^{3-}$, can self-assemble into all-inorganic superlattices with both long-range superlattice translational and atomic-lattice orientational order. Structural characterizations reveal that the NCs adopt unexpected edge-to-edge alignment, and numerical simulation clarifies that orientational order is thermodynamically stabilized by many-body ion correlations originating from the dense electrolyte. Furthermore, we show that the superlattices of $Sn_2S_6^{4-}$-functionalized PbS NCs can be fully disassembled back into the colloidal state, which is highly unusual for orientationally attached superlattices with atomic-lattice alignment. The reversible oriented attachment of NCs, enabling their dynamic assembly and disassembly into effectively single-crystalline superstructures, offers a pathway toward designing reconfigurable materials with adaptive and controllable electronic and optoelectronic properties.

In a recent letter [T. D. Kühne, M. Krack, F. Mohamed and M. Parrinello, Phys. Rev. Lett. 98, 066401 (2007)], we outlined a new Car-Parrinello-like approach to Born-Oppenheimer molecular dynamics. Here, we provide a guide to performing actual calculations using our method and demonstrate this on liquid water at ambient conditions. We do not go into methodological details beyond those necessary for applying this approach, but focus on practical details pertinent to our particular implementation within the CP2K/Quickstep code [T. D. Kühne et al., J. Chem. Phys. 152, 194103 (2020)].

We investigate spin-valley instabilities and their connection to the reentrant superconducting states recently observed in the twisted bilayer dichalcogenide WSe$_2$ at a $5^o$ twist angle. Starting from an effective three-orbital faithful Wannier model for the spin-locked moiré bands, combined with orbital-dependent Hubbard interactions, we analyze the evolution of magnetic instabilities as a function of carrier density using the matrix random phase approximation (mRPA) approach. By computing the Stoner boundary lines from the spin-valley susceptibilities over the electric-field by hole filling phase diagram, we show that the spin-valley instabilities result in ordered states in the region close to the Lifshitz transition at the topmost moiré valence band, marked by crossing of the van Hove singularity in the density of states. These spin-valley ordered states are dominated by interorbital spin-valley-flips involving the $MM$ and $MX$ moiré orbitals and occur at different momenta in each side of the van Hove line, indicating a distinct spatial dependence of the spin-valley order parameter depending on the hole filling. Moreover, the corresponding Stoner boundaries exhibit strong fluctuations on its flanks, which can favor superconducting states in the regions close to the spin-valley-ordered ones. This mechanism provides a natural description for a reentrant superconducting dome consistent with the experimental results. As such, our results suggest spin-valley fluctuations near the van-Hove line as the microscopic origin of the reentrant superconductivity in twisted WSe$_2$.

We develop an information-theoretic formulation of stochastic dynamics in which the fundamental stochastic variable is the total action connecting spacetime points, rather than individual paths. By maximizing Shannon entropy over a joint distribution of actions and endpoints, subject to normalization and a constraint on the mean action, we obtain a Boltzmann-like distribution in action space. This framework reproduces the standard Brownian propagator in the nonrelativistic limit and naturally extends to relativistic regimes, where the Wiener construction fails to preserve Lorentz covariance. The approach bypasses functional integration over paths, makes the role of entropic degeneracy explicit through an action-space density of states, and provides a transparent connection between the principle of least action and statistical inference. We derive the density of states explicitly using large deviation theory, showing that it takes a Gaussian form centered at the minimal action, and rigorously justify the saddle-point approximation in the diffusive regime. The Markovian property of the resulting propagator is verified to hold via the Chapman--Kolmogorov equation, following from the additivity of the minimal action for free-particle dynamics. In the diffusive regime, the resulting dynamics are governed by a competition between extremization of the action and entropic effects, which can be interpreted in terms of an effective action free energy. Our results establish an unified, covariant, and information-based foundation for classical and relativistic stochastic processes.

The quantum first-detection problem concerns the statistics of the time at which a system, subject to repeated measurements, is observed in a prescribed target state for the first time. Unlike its classical counterpart, the measurement back action intrinsic to quantum mechanics may profoundly alter the system dynamics. Here we show that it induces a distinct change in the statistics of the first-detection time. For a quantum particle in one spatial dimension subject to stroboscopic measurements, we observe an algebraic decay of the probability of the first-detection time if the particle is free, an exponential decay in the presence of a confining potential, and a time-dependent crossover between these behaviors if the particle is partially confined. This crossover reflects the purely quantum nature of the detection process, which fundamentally distinguishes it from the first-passage problem in classical systems.

The exploration of quantum phenomena in complex materials such as moiré superlattices is limited by the O(N^3) scaling of conventional electronic structure methods. Here we introduce a high-performance tight-binding framework that reduces the complexity to O(N^1.5) by transforming the Hamiltonian into a real symmetric form and combining Sylvester's inertia law with LDL decomposition. This approach enables efficient band structure calculations for large systems: solving magic angle twisted bilayer graphene in minutes on a laptop and scaling to 1.5 million atoms within days on a workstation. We apply it to the previously inaccessible ultra-low twist-angle regime (less than 0.16 degree) with mechanical strain relaxation and find robust flat bands persisting down to 0.09 degree. Our framework bridges density functional theory accuracy with large-scale quantum simulation, opening a route to systematic data-driven exploration of mesoscale quantum materials.

We develop a theory of the ratchet effect in a lateral plasmonic crystal (LPC) formed by a two dimensional electron gas under a periodic dual-grating gate. The system is driven by terahertz radiation, and the spatial asymmetry required for the generation of dc photocurrent is introduced by a phase shift between the radiation's near-field modulation and the static electron density profile. In contrast to the commonly used perturbative "minimal model" of the ratchet effect, which assumes weak density modulation, we solve the problem exactly with respect to the static gate-induced potential while treating the radiation field perturbatively. This approach reveals a dramatic enhancement of the plasmonic contribution to the ratchet current due to the interference of "bright" and "dark" plasmon modes, which are excited on an equal footing in the asymmetric LPC. Specifically, we predict a parametric growth of the plasmonic peak as compared with the Drude peak with increasing coupling, and the appearance of a dense super-resonant structure when the spacing between plasmonic sub-bands becomes larger than the damping rate. Hence, the dc response exhibits both resonant and super-resonant regimes observed in recent experiments on the radiation transmission through the LPC. The interplay of bright and dark modes, together with their interference, provides a powerful mechanism for controlling the magnitude and sign of the photocurrent by gate voltages and the radiation frequency.

Several advances in the theory of space-resolved viscoelasticity of liquids and other amorphous systems are discussed in the present paper. In particular, considering long-time regimes of stress relaxation in liquids we obtain the generalized compressibility equation valid for systems with mass polydispersity, and derive a new relation allowing to calculate the wavevector-dependent equilibrium transverse modulus in terms of the generalized structure factors. Turning to the basic relations between the spatially-resolved relaxation moduli and the spatio-temporal correlation functions of the stress tensor, we provide their new derivation based on a conceptually simple argument that does not involve consideration of non-stationary processes. We also elucidate the relationship between the stress noise associated with the classical Newtonian dynamics and the reduced deviatoric stress coming from the Zwanzig-Mori projection operator formalism. The general relations between the stress noise and the tensor of relaxation moduli are discussed as well.

CP-PAW is a combined electronic structure and ab-initio molecular dynamics code to perform mixed quantum and classical simulations of atomistic condensed phase systems, such as solids, liquids, and molecular systems. As the name suggests, the CP-PAW code unifies the all-electron projector augmented-wave method with the Car-Parrinello approach to determine not only the electronic and nuclear ground state of condensed matter, but also to study their properties and dynamics. In addition to briefly outlining the underlying theory, the focus will be on unique aspects of CP-PAW and how to correctly employ them as a user. How to install CP-PAW using the new build system will also be briefly mentioned.

We address issues related to the presence of defects at topological transitions, in particular when defects are modeled in terms of further variables associated with a quenched disorder, corresponding to the limit in which the defect dynamics is very slow. As a paradigmatic model, we consider the three-dimensional lattice ${\mathbb Z}_2$ gauge model in the presence of quenched uncorrelated disorder associated with the plaquettes of the lattice, whose topological transitions are characterized by the absence of a local order parameter. We study the critical behaviors in the presence of weak disorder. We show that they belong to a new topological universality class, different from that of the lattice ${\mathbb Z}_2$ gauge models without disorder, in agreement with the Harris criterium for the relevance of uncorrelated quenched disorder when the pure system undergoes a continuous transition with positive specific-heat critical exponent.

We present a theory of the emission spectrum of magnetoexcitons interacting with a $ν= 1$ quantum Hall state of massive Dirac fermions in monolayer transition metal dichalcogenides in high magnetic fields. Using an ab initio-parametrized massive Dirac fermion model including valley and spin degrees of freedom, combined with exact diagonalization techniques, we show that interband emission from the massive Dirac Fermion magnetoexciton interacting with $ν= 1$ state directly probes intra-conduction-band excitations of the $ν= 1$. Many-body interactions with the filled massive Dirac fermion $ν= 1$ level yield a strong renormalization of the emission spectrum, including fully polarized emission, a pronounced redshift, and broadening relative to neutral and charged excitons. The calculated spectra are consistent with recent experiments [1-3], establishing magneto-spectroscopy as a probe of finite carrier densities in massive Dirac systems.

It has been observed in numerous experiments, simulations, and various theoretical treatments that the spreading of particles can be modeled by the continuous-time random walk. We consider two well-known cases, i.e., Gaussian displacements and discrete displacements, to compute the position distribution and demonstrate the emergence of exponential decay in the far tails when a bias is introduced. We further analyze the temporal rate function and the positional rate function to examine the convergence of the theoretical predictions. For Gaussian displacements, we further discuss the relationship between the position distributions with and without bias in different asymptotic limits.

The quantum Hall effect is a fascinating electrical transport phenomenon signified by precise quantization of Hall conductivity $σ_\mathrm{xy}$ and vanishing longitudinal conductivity $σ_\mathrm{xx}$. Laughlin proposed an elegant explanation in which adiabatic insertion of a flux tube pumps charge through the system. This analysis unveils the fundamental role of gauge invariance and provides a compelling argument about the fractional charge of fractional quantum Hall states. While it has been used extensively as a theoretical tool, a quantitative experimental investigation is lacking despite multiple attempts. Here we report successful realizations of Laughlin pumping in several integer and fractional quantum Hall states. One essential technical innovation is using surface acoustic waves to periodically clear the charges accumulated during the pumping process. Magnetic fluxes are inserted at a constant rate so there is no need to perform complicated data fitting. Furthermore, our setting can reliably extract $σ_\mathrm{xx}$ that is several orders of magnitude lower than the limit of conventional techniques. Effective energy gaps can be deduced from the temperature dependence of $σ_\mathrm{xx}$, which are drastically different from those provided by conventional transport data. This work not only brings a famous gedanken experiment to reality but also serves as a portal for many future investigations.

Strong Coulomb interactions in two-dimensional (2D) semiconductors give rise to tightly bound excitons, exciton polarons, and correlated electronic phases such as Wigner crystals (WCs), yet their mutual interplay remains poorly understood. Here we report the observation of multi-branch excitonic Umklapp scattering in both electron and hole WCs realized in ultraclean monolayer WSe$_2$, exhibiting exceptionally high melting temperatures (T$_c$ $\approx$ 20-30 K). Robust Wigner crystallization activates multiple finite-momentum optical resonances, including quasilinearly dispersing, light-like excitons and exciton polarons, extending far beyond the single excitonic Umklapp feature reported previously. Helicity-resolved magneto-optical measurements reveal a pronounced valley dependence of the scattering processes. Combined experiment and theory identify a polaron-induced brightening mechanism in which exciton polarons transfer oscillator strength from bright zero-momentum states to otherwise dark finite-momentum states, explaining the emergence of multiple Umklapp branches where conventional exciton-WC scattering is ineffective. These results establish WC polarons as a new quasiparticle paradigm and introduce polaron-induced Umklapp scattering as a general route to accessing finite-momentum many-body excitations in 2D quantum materials.

Low-temperature plasma processing is a promising technique for tailoring the properties of transition metal dichalcogenides (TMDs) because it allows for precise control of radical and ion energies and fluxes. For chalcogen substitution, a key challenge is to identify the ion energy window that enables selective chalcogen removal while preserving the metal lattice. Using ab-initio molecular dynamics (AIMD), we demonstrate that oxygen and fluorine functionalization through thermal chemisorption significantly lowers the sputtering energy threshold ($E_{sputt}$) of MoS${}_2$ from $\sim 35$ eV to $\sim 10$ eV. In addition, we find that a non-orthogonal impact angle $\sim 30{}^{\circ}$ reduces the sputtering energy threshold, while cryogenic-range TMD temperatures may increase. To explain the observed trends, a multi-step sputtering mechanism is proposed. Our results show that oxygen/fluorine functionalization, impact angle, and material temperature are key control parameters for selective, damage-free chalcogen removal in TMD processing.

Transitions from reversible to irreversible or fluctuating states above a critical density and shear amplitude have been extensively studied in non-thermal cyclically sheared suspensions and amorphous solids. Here, we propose that the same type of reversible to irreversible transition occurs for a system of particles with competing short-range attraction and long-range repulsion, which can form crystals, stripes, and bubbles as the ratio of attraction to repulsion varies. By oscillating the strength of the attractive part of the potential, we find that the system can organize into either time-periodic states consisting of nondiffusive complex closed orbits, or into a diffusive fluctuating state. A critical point separates these states as a function of the maximum strength of the attraction, oscillation frequency, and particle density. We also find a re-entrant behavior of the reversible state as a function of the strength of the attraction and the oscillation frequency.

Nitrogen vacancy (NV) centers have been demonstrated as a useful tool in high pressure environments. However, the geometry and small working area of the diamond anvil cells (DACs) used to apply pressure present a challenge to effective delivery of microwave (mw) fields. We designed and characterized a novel slotted design for mw transmission to nitrogen-vacancy centers (NVs) in a diamond anvil cell via zero-field and in-field optically detected magnetic resonance (ODMR) measurements across pressures between 1 and 48 GPa. The mw fields experienced by NVs across the diamond culet was calculated from Rabi frequency and found to be higher and more uniform than those generated by an equivalent simple mw line, which will improve performance for wide-field, high-pressure measurements to probe spatial variations across samples under pressure.

We report the experimental realization of strong longitudinal (ZZ) coupling between two superconducting transmon qubits achieved solely through capacitive engineering. By systematically varying the qubit frequency detuning, we measure cross-Kerr inter-qubit interaction strengths ranging from 10 MHz up to 350 MHz, more than an order of magnitude larger than previously observed in similar capacitively coupled systems. In this configuration, the qubits enter a strong-interaction regime in which the excitation of one qubit inhibits that of its neighbor, demonstrating a dynamical blockade mediated entirely by the engineered ZZ coupling. Circuit quantization simulations accurately reproduce the experimental results, while perturbative models confirm the theoretical origin of the energy shift as a hybridization between the computational states and higher-excitation manifolds. We establish a robust and scalable method to access interaction-dominated physics in superconducting circuits, providing a pathway towards solid-state implementations of globally controlled quantum architectures and cooperative many-body dynamics.

An obituary of J.R. Dorfman. The focus is on his scientific career and on his many important publications.

We investigate the wrinkling dynamics of a thin elastic sheet that is indented or compressed while floating on a viscous liquid. We show that the deformation speed controls the dynamics, leading to a wrinkle wavelength significantly smaller than that selected under quasistatic compression. Once active compression ceases, the wrinkles coarsen until their wavelength relaxes toward the equilibrium value. We develop a theoretical model coupling Stokes flow in the liquid to elastic bending of the sheet, which quantitatively predicts both the initial wavelength selection and its subsequent coarsening. We demonstrate that the same mechanism governs two dimensional and axisymmetric geometries, thereby extending classical static wavelength selection laws to dynamic situations. Although developed from controlled laboratory experiments, the model captures a generic viscous-elastic coupling and applies broadly to thin elastic films interacting with viscous environments, including the formation of surface wrinkles in pahoehoe lava flows.

This is an editorial article based on the reseaches on the Ising model over the last 100 years.

We study the Dirac spectrum in a sine-Gordon soliton background, where the induced position-dependent mass reduces the spectral problem to a Heun-type differential equation. Bound and scattering sectors are treated within a unified framework, with spectral data encoded in Wronskians matching local Heun solutions and exhibiting explicit dependence on the soliton parameters and the bare fermion mass. This formulation enables a systematic analysis of spinor bound and scattering states, supported by analytic and numerical verification of wave function matching across the soliton domain. The present work is related to arXiv:2512.07658 and emphasizes a pedagogical treatment of scattering states within the Heun-equation formalism.

This work presents a numerical investigation of interacting chiral oscillators (COs), characterized by an intrinsic rotational handedness.

The El Niño-Southern Oscillation (ENSO) is a dominant mode of interannual climate variability, yet the mechanisms limiting its long-lead predictability remain unclear. Here we develop a physics-guided Deep Echo State Network (DESN) that operates on physically interpretable climate modes selected from the extended recharge oscillator (XRO) framework. DESN achieves skillful Niño3.4 predictions up to 16-20 months ahead with minimal computational cost. Mechanistic experiments show that extended predictability arises from nonlinear coupling between warm water volume and inter-basin climate modes. Error-growth analysis further indicates a finite ENSO predictability horizon of approximately 30 months. These results demonstrate that physics-guided reservoir computing provides an efficient and interpretable framework for diagnosing and predicting ENSO at long lead times.

The Rulkov model, which simulates the behavior of biological neurons, is modified by replacing one of its control parameters with a memristive, sigmoid-type function of finite memory. This modification causes the parameter to vary according to the system's history throughout the simulation. Previous works usually modify the Rulkov model by introducing additional parameters altering its behavior. Here, by contrast, we retain the original equations and allow the control parameters to vary in time, thereby preserving the model's fundamental properties. In this sense, the proposed model is locally equivalent in time to the original one. However, unlike the original model, which reproduces a single neuronal regime per simulation, the new memristive version exhibits both uniform and chaotic transitions among multiple neuronal activity regimes. Its dynamics are examined with respect to the rate at which the memristive function changes and the number of internal states it stores. Three distinct scenarios emerge around a bifurcation point. Before the bifurcation, the system undergoes uniform transitions toward a stable bursting regime. After the bifurcation, it shows uniform transitions toward a final spiking or silent regime. At the bifurcation point, highly complex transitions arise. As examples, we present trajectories in which the neuron chaotically switches between regimes without ever settling, and trajectories for which it requires around 140000 map iterations to reach a stationary regime.

The Korteweg-de Vries (KdV) equation serves as a foundational model in nonlinear wave physics, describing the balance between dispersive spreading and nonlinear steepening that gives rise to solitons. This article introduces sangkuriang, an open-source Python library for solving this equation using Fourier pseudo-spectral spatial discretization coupled with adaptive high-order time integration. The implementation leverages just-in-time (JIT) compilation for computational efficiency while maintaining accessibility for instructional purposes. Validation encompasses progressively complex scenarios including isolated soliton propagation, symmetric two-wave configurations, overtaking collisions between waves of differing amplitudes, and three-body interactions. Conservation of the classical invariants is monitored throughout, with deviations remaining small across all test cases. Measured soliton velocities conform closely to theoretical predictions based on the amplitude-velocity relationship characteristic of integrable systems. Complementary diagnostics drawn from information theory and recurrence analysis confirm that computed solutions preserve the regular phase-space structure expected for completely integrable dynamics. The solver outputs data in standard scientific formats compatible with common analysis tools and generates visualizations of spatiotemporal wave evolution. By combining numerical accuracy with practical accessibility on modest computational resources, sangkuriang offers a platform suitable for both classroom demonstrations of nonlinear wave phenomena and exploratory research into soliton dynamics.

In many real-world applications of data assimilation (DA), the strategic placement of observers is crucial for effective and efficient forecasting. Motivated by practical constraints in sensor deployment, we show that global recovery of the flow field can be achieved using observations available only in a subregion of the domain, possibly far from the boundary. We focus on the two-dimensional incompressible Navier-Stokes equations posed in a bounded physical domain with Dirichlet boundary conditions. Building on the continuous data assimilation framework of Azouani, Olson, and Titi (2014), we rigorously prove that the assimilated solution converges globally to the true solution under suitable conditions on the nudging parameter, spatial resolution, and the geometry of the observation region, specifically, when the maximum distance from any point in the domain to the observational subregion is bounded by a constant multiple of \( ν^{1/2} \) (in terms of scaling). Our computational results, conducted via finite element methods over complex geometries, support the theoretical findings and reveal even greater robustness in practice. Specifically, synchronization with the true solution is achieved even when the observational subregion lies farther from the rest of the domain than the theoretical threshold permits. Across all three tested scenarios, the local nudging algorithm performs comparably to full-domain assimilation, reaching global accuracy up to machine precision. Interestingly, observational data near the boundary are found to be largely uninformative. This demonstrates that full observability is not necessary: carefully chosen interior observations, even far from the boundary, can suffice.

Complete analytic solutions to quasi-continuous-wave four-wave mixing in nonlinear optical fibres are presented in terms of Weierstrass elliptic $\wp$, $ζ$, and $σ$ functions, providing the full complex envelopes for all four waves under arbitrary initial conditions. A sequence of coordinate transformations reveals a canonical form with universal parameter-free structure. Remarkably, these transformations depend explicitly on the propagation variable yet preserve the structural form of the differential equations, an invariance property not previously reported for four-wave mixing. In the canonical coordinates, solutions become single-valued meromorphic Kronecker theta functions, establishing connections with other integrable nonlinear optical systems. The Hamiltonian conservation is shown to arise from the Frobenius-Stickelberger determinant. Numerical validation confirms the solutions using open-source Python libraries.

The primordial 7Li abundance predicted by standard Big Bang Nucleosynthesis (BBN) exceeds that inferred from old, metal-poor stars by a factor of about 3-4. In standard BBN, most primordial 7Li is produced as 7Be in the early Universe and later converted by electron capture. Additional production or destruction channels of 7Be, such as proton capture or antineutrino capture during BBN, may therefore affect the final lithium yield. We quantify the depletion of 7Be due to in-situ electron capture, including the associated antineutrino channel, positron decay from nuclear excited states, and proton capture through the radiative 7Be(p,gamma)8B reaction. We also investigate stimulated emission induced by the dense photon background during the nuclear statistical equilibrium epoch, as well as a three-body Auger-like variant transferring the capture energy to a continuum electron. Decay rates are computed using first-order perturbation theory, modelling weak interactions with a Fermi contact term and factorising hadronic and leptonic currents. Thermally averaged rates are obtained by folding cross-sections with Maxwell-Boltzmann distributions and accounting for particle densities in the temperature range 10-100 keV. We find that the electron-capture rate decreases rapidly with temperature and is significantly enhanced by the inclusion of the antineutrino channel. Stimulated emission and plasma screening increase the radiative proton-capture rate by only 1-3 percent at temperatures around 87 keV. The Auger-like channel contributes at the level of a few thousandths of a percent and becomes negligible at lower temperatures. Overall, our total rate revises previous estimates by nearly an order of magnitude. Electron capture, proton capture, and positron decay provide corrections to the dominant depletion channel 7Be(n,p)7Li.

Ultrasound vortices have rapidly expanded their applications to areas like particle trapping, contactless manipulation, acoustic communications. In ultrasonic imaging and therapy involving bone tissues, these vortex beams offer intriguing possibilities but transmitting them through bone (especially the skull) poses challenges. Traditional acoustic lenses were engineered to rectify skull-induced beam aberration, and their capacity was limited to generating only static ultrasound fields within the brain. To overcome this constraint, our study presents a novel method for transcranially steering focused ultrasound vortex using 3D printed binary acoustic metasurfaces (BAMs) with a thickness of 0.8 λ. We tackled the challenge of skull-induced phase aberration by computing the phase distribution via a time reversal technique, which concurrently enabled the generation of a steerable focused vortex inside an ex vivo human skull by adjusting the operating frequency. Both numerical and simulations experiments were conducted to validate the capabilities of BAMs. Furthermore, we explored the generation of higher-order topological charge acoustic vortices within the brain utilizing the BAM. This development paves the way for designing cost-effective particle-trapping systems, facilitating clot manipulation, and applying acoustic-radiation forces and torques within or across bone structures, thus presenting a new frontier for potential biomedical applications.

This study investigates the economic complexity of Indian states by constructing a state-industry bipartite network using firm-level data on registered companies and their paid-up capital. We compute the Economic Complexity Index and apply the fitness-complexity algorithm to quantify the diversity and sophistication of productive capabilities across the Indian states and two union territories. The results reveal substantial heterogeneity in regional capability structures, with states such as Maharashtra, Karnataka, and Delhi exhibiting consistently high complexity, while others remain concentrated in ubiquitous, low-value industries. The analysis also shows a strong positive relationship between complexity metrics and per-capita Gross State Domestic Product, underscoring the role of capability accumulation in shaping economic performance. Additionally, the number of active firms in India demonstrates a persistent exponential growth at an annual rate of 11.2%, reflecting ongoing formalization and industrial expansion. The ordered binary matrix displays the characteristic triangular structure observed in complexity studies, validating the applicability of complexity frameworks at the sub-national level. This work highlights the usefulness of firm-based data for assessing regional productive structures and emphasizes the importance of capability-oriented strategies for fostering balanced and sustainable development across Indian states. By demonstrating the usefulness of firm registry data in data constrained environments, this study advances the empirical application of economic complexity methods and provides a quantitative foundation for capability-oriented industrial and regional policy in India.

Low-scattering, deep-penetration light transport in biological media remains a pivotal challenge for biophotonic technologies, including biomedical imaging, optical diagnostics, and photodynamic therapy. This review builds upon and extends our earlier studies of nonlinear optical self-trapping and optically induced waveguiding in biological suspensions, such as human erythrocytes and cyanobacteria, where light-matter coupling is governed by optical-force-mediated particle redistribution. Recent progress has revealed increasingly rich and complex regimes, including the propagation and nonlinear self-action of structured (vortex) beams in biological environments, as well as nonlinear responses dominated by thermally driven mechanisms in absorptive biomolecular solutions (e.g., heme and chlorophyll). We place particular emphasis on distinctive nonlinear phenomena observed in these systems, including spatial self-phase modulation, optical-force-induced sculpturing of effective energy landscapes, and quasi-waveguide formation in soft, heterogeneous biological media. We conclude by highlighting emerging opportunities to harness these nonlinear behaviors for deep-tissue imaging, label-free biosensing, and the realization of biocompatible photonic structures and devices assembled directly from living or hybrid biological matter.

Shock dynamics and nonlinear wave propagation are fundamental to computational fluid dynamics (CFD) and high-speed flow modeling. In this study, we developed explicit and implicit finite-difference solvers for the one-dimensional Burgers viscous equation to model shock formation, propagation, and dissipation. The governing equation, which incorporates convective and diffusive effects, serves as a simplified analogue of the Navier-Stokes equations. Using the Finite-JAX framework, each solver is implemented with upwind and central finite-difference schemes for the convective and diffusive terms, respectively. Time integration is performed using explicit forward Euler and implicit backward-time central space (BTCS) schemes under periodic and Dirichlet boundary conditions. Stability is ensured by the Courant-Friedrichs-Lewy (CFL) criteria for the convective and diffusive components. Numerical experiments quantify the accuracy, convergence, and real-time performance of JAX across CPUs, GPUs, and TPUs, demonstrating that JAX maintains fidelity while achieving portability. The results show that the explicit scheme captures impact accurately under strict time-step constraints, while the implicit formulation provides greater stability and accuracy at a higher computational cost. Taken together, these results establish a reproducible dataset for benchmarking CFD solvers and training machine learning models for nonlinear transport and impact-driven phenomena. Our new implementation of FiniteJAX enhances the portability, scalability, and performance of solvers based on the JAX framework developed by Google DeepMind.

Vaccination games in higher-order settings remain underexplored, despite their importance in shaping opinions and collective decisions. Here, we introduce a parsimonious behavioral-epidemiological model to evaluate how peer reinforcement pressure influences vaccination uptake. The framework consists of a two-layer multiplex: an epidemic layer governed by the SIR process on a square lattice, and a behavioral layer represented by a hypergraph of triadic interactions. Individuals update their vaccination strategy via imitation, modulated by a reinforcement parameter $α$ when peer support is present. We find that higher-order structure alone induces clusters of vaccinated individuals that act as protective barriers. Low but nonzero reinforcement ($α\approx 0.5$) maximizes coverage and suppresses outbreaks, while both negligible ($α\approx 0$) and moderate ($α> 0.1$) reinforcement reduce uptake, as excessive confirmation lowers adaptability and enables non-vaccinators to re-emerge. Our work bridges complex contagion theory with evolutionary game dynamics, offering insights into how contact structure and peer reinforcement jointly shape vaccination behavior.

Fisher zeros play a central role in the theoretical understanding of phase transitions. However, their computation requires knowledge of the density of states, which limits their practical applicability. Alternative approaches based on the Energy Probability Distribution (EPD) and Moment Generating Function (MGF) alleviate the computational cost but suffer from convergence issues in the two-dimensional \textbf{anisotropic Heisenberg} model (XY model). In this work, we introduce a Padé approximation to systematically reduces the number of zeros required in the Fisher, EPD, and MGF formulations without loss of accuracy. Moreover, since the Fisher zeros formulation does not rely on a convergence algorithm, their combination with a Padé approximation enables a reliable analysis of the XY model while significantly reducing computational cost. Applications to the two-dimensional Ising and XY models demonstrate substantial reductions in polynomial degree and computation time while preserving accurate estimates of the critical temperature.

In this paper we analyze the Finnish road network as a graph in order to measure whether the "rurality" or "urbanity" of an area correlates with local geometrical properties of the graph. Our primary motivation is the observation that the road systems in rural areas look similar to hyperbolic graphs, while in large cities they resemble more the Cayley graph of $\mathbb{Z}^2$. We do not aim for a comprehensive analysis, but rather wish to demonstrate that this observation can be measured and analyzed through looking at various "hyperbolicity measures" of randomly sampled geodesic triangles in the road graph.

Biological matrices are rich in information related to life processes, serving as invaluable media for assessing an individual's overall physiological status and its dynamic fluctuations, as well as crucial foundations for disease diagnosis. However, the inherent complexity of these matrices, coupled with our incomplete understanding of their full composition, presents significant challenges for comprehensive analysis and accurate diagnostic interpretation. The advent of single-molecule technologies has revolutionized biomedical research, enabling the direct observation of life processes at the molecular scale. We have proposed an Intelligent Nano-Fingerprinting strategy based on single-molecule nanopore technology, designed to capture the global molecular fingerprints of complex plasma matrices. Furthermore, we developed an intelligent algorithmic model capable of achieving precise classification of plasma samples. This approach is characterized by its simplicity, efficiency, and considerable potential for large-scale adoption and transferable applications.

Baseball consists of two teams alternating between batting and fielding while competing to score runs through sequential pitching events. Recent advances in tracking technology have enabled all Major League Baseball (MLB) clubs to record every pitch with high resolution, yet most quantitative studies have primarily emphasized single-pitch metrics, leaving the role of sequential structure less explored. Here, we examine pitch-pattern motifs of multiple lengths using approximately 12.4 million Statcast pitch recordings from the 2008-2025 MLB regular seasons at two complementary scales. At the macroscale, we quantify pitch-sequence diversity using the Shannon entropy and inverse Simpson index and examine their relationships with earned run average and win totals. At the microscale, we compare hit and out frequencies across pitch-pattern motifs. Rather than identifying outcome-determining sequences, we find that motif usage exhibits stable, non-random organization, as reflected in Zipf s and Heaps' laws, while showing limited association with conventional performance measures. While language-like scaling (Zipf's and Heaps' laws) clearly reveals an underlying 'grammar' of MLB pitch sequences, that grammar alone is insufficient to account for performance indicators such as ERA or wins. These results suggest that sequence-based analyses clarify the structural organization of pitch usage, while also delineating the limits of motif-based approaches for explaining performance without richer contextual information.

Two-phase surfactant-laden flow and transport in porous media are central to many natural and engineering applications. Surfactants alter two-phase flow by modifying interfacial tension and wettability, while two-phase flow controls surfactant transport pathways and interfacial adsorption. These coupled processes are commonly modeled using Darcy-type two-phase flow equations combined with advection--dispersion--adsorption transport equations, with capillary pressure--saturation relationships scaled by the Leverett $J$-function. However, the Leverett $J$-function idealizes porous media as bundles of cylindrical tubes and decouples interfacial tension and wettability, limiting its ability to represent angular pore geometries and interfacial tension--wettability coupling effects. We present a modeling framework that explicitly incorporates pore angularity and interfacial tension--wettability coupling into Darcy-scale surfactant-laden flow and transport models. Two-phase flow properties are derived for angular pores, upscaled across pore size distributions, and formulated as explicit and closed-form expressions. These upscaled relationships are integrated into a coupled flow--transport model to simulate transient two-phase flow and surfactant transport. Results reveal a nonlinear and nonmonotonic dependence of two-phase flow properties on pore angularity, pore size distribution, and interfacial tension. Example simulations of water flow and PFAS migration in unsaturated soils indicate that surfactant-induced flow effects on PFAS leaching are generally minor under typical conditions, whereas pore angularity strongly controls water flow, interfacial area, and PFAS retention. Overall, the proposed framework provides a more physically grounded approach for modeling two-phase surfactant-laden flow and transport in porous media.

The grandmother hypothesis is the most influential account of the evolution of menopause in humans, but other theories warrant investigation. Here I use simulations to investigate two theories that ground the evolution of menopause in biparental care. Kaplan et al. (2010) proposed a "two-sex" learning and skill-based account, termed the Embodied Capital Model (ECM), in which the high energetic burden of caring for multiple, slow-developing offspring was met with biparental investment. Menopause evolved because the physiological costs of pregnancy and childbirth increased with age yet productivity also increased with age, and the benefits of transferring resources to adult children and their offspring eventually outweigh the benefits of continued reproduction. Kuhle (2007) proposed the "father absent" hypothesis in which the higher mortality rate of husbands would often have left wives without the resources to raise young children, selecting for early reproductive cessation in monogamous couples. Simulations of hunter-gatherer energy consumption and production across the lifespan, taking account of age- and sex-specific survivorship, interbirth intervals, and varying rates of strength and foraging skill acquisition typical of contemporary foragers, reveal a pronounced midlife energy deficit that could be averted by ceasing reproduction midlife and receiving energy transfers from both younger couples (e.g., brideservice) and from older parents (the grandmother hypothesis). Menopause emerges as an integral and strictly necessary component of the unique human pattern of relatively short interbirth intervals and a long period of juvenile dependency, supporting and extending the ECM.

Motivation: Opioid use disorder (OUD) often arises after prescription opioid exposure and follows transitions among onset, remission, and relapse. Linked EHR-survey resources such as the All of Us Research Program enable stage-specific risk modeling and connection to intervention options. Results: We built a multi-stage framework to model time-to-onset, time-to-remission, and time-to-relapse after remission using All of Us EHR and survey data. For each participant we derived longitudinal predictors from clinical conditions and survey concepts, including recent (1/3/12-month) event counts, cumulative exposures, and time since last event. We fit regularized Cox models for each transition and aggregated selection frequencies and hazard ratios to identify a compact set of high-confidence predictors. Pain, mental health, and polysubstance use contributed across stages: chronic pain syndromes, tobacco/nicotine dependence, anxiety and depressive disorders, and cannabis dependence prominently predicted onset and relapse, whereas tobacco dependence during remission and other remission-coded conditions were strongly associated with transition to remission. To support therapeutic prioritization, we constructed a therapy knowledge graph integrating genetic targets, biological pathways, and published evidence to map identified risk factors to candidate treatments in recent OUD studies and clinical guidelines.

Existing methods in nanoscale connectomics are at present too slow to map entire mammalian brains. As an emerging approach, expansion microscopy (ExM) has enormous promise, yet it still suffers from throughput limitations. Mapping the human brain and even mapping nonhuman primate brains therefore remain distant goals. While ExM increases effective resolution linearly, it enlarges tissue volume cubically, which dramatically increases imaging time. As a rapid tomographic technique, X-ray microscopy has potential for drastically speeding up large-volume connectomics. But to the best of my knowledge, no group has so far imaged cellular features within expanded tissue using X-ray microscopy. I herein present an early-stage report featuring the first demonstration of X-ray microscopy reconstruction of cell bodies within expanded tissue. This was achieved by combining a modified enzymatic Unclearing technique with a metallic gold stain and imaging using a laboratory X-ray microscope. I emphasize that a great deal of work remains to develop "expansion X-ray microscopy" (ExXRM) to the point where it can be useful for connectomics since the current iteration of ExXRM only resolves cell bodies and not neurites due to extensive off-target staining. Additionally, the current method must be modified to accommodate for the challenges of synchrotron X-ray microscopy, a vastly speedier approach than laboratory X-ray microscopy. Nonetheless, achieving X-ray contrast in expanded tissues represents a significant first step towards realizing ExXRM as a connectomics imaging modality.

Diseases originating from wildlife pose a significant threat to global health, causing human and economic losses each year. The transmission of disease from animals to humans occurs at the interface between humans, livestock, and wildlife reservoirs, influenced by abiotic factors and ecological mechanisms. Although evidence suggests that intact ecosystems can reduce transmission, disease prevention has largely been neglected in conservation efforts and remains underfunded compared to mitigation. A major constraint is the lack of reliable, spatially explicit information to guide efforts effectively. Given the increasing rate of new disease emergence, accelerated by climate change and biodiversity loss, identifying priority areas for mitigating the risk of disease transmission is more crucial than ever. We present new high-resolution (1 km) maps of priority areas for targeted ecological countermeasures aimed at reducing the likelihood of zoonotic spillover, along with a methodology adaptable to local contexts. Our study compiles data on well-documented risk factors, protection status, forest restoration potential, and opportunity cost of the land to map areas with high potential for cost-effective interventions. We identify low-cost priority areas across 50 countries, including 277,000 km2 where environmental restoration could mitigate the risk of zoonotic spillover and 198,000 km2 where preventing deforestation could do the same, 95% of which are not currently under protection. The resulting layers, covering tropical regions globally, are freely available alongside an interactive no-code platform that allows users to adjust parameters and identify priority areas at multiple scales. Ecological countermeasures can be a cost-effective strategy for reducing the emergence of new pathogens; however, our study highlights the extent to which current conservation efforts fall short of this goal.

Triple-negative breast cancer (TNBC) is an aggressive disease with high mortality and limited treatment options, due to its lack of receptors that have targeted therapies available. The tumor microenvironment (TME) plays a critical role in TNBC progression and therapeutic resistance. In this work, we developed a novel mathematical model to describe key cellular interactions within the TNBC TME, informed by current literature and expert input. Our model consists of a system of ordinary differential equations representing five interacting cell populations: M2 macrophages, cancer-associated fibroblasts, TNBC tumor cells, cytotoxic T lymphocytes, and regulatory T cells. We performed global sensitivity analysis to determine which model parameters most strongly influence tumor burden over a clinically-relevant treatment timeframe. The pathways associated with the most-influential parameters correspond to biological mechanisms that are consistent with known and emerging therapeutic strategies in TNBC, including stromal-mediated tumor support. These results highlight key regulatory interactions within the TNBC TME and provide a quantitative framework for hypothesis generation and future investigation of combination treatment strategies.

For 20 years the beautiful structure in the grid cell code has presented an attractive puzzle: what computation do these representations subserve, and why does it manifest so curiously in neurons. The first question quickly attracted an answer: grid cells subserve path-integration, the ability to keep track of one's position as you move about the world. Subsequent work has only solidified this link: bottom-up mechanistic models that perform path-integration match the measured neural responses, while experimental perturbations that selectively disrupt grid cell activity impair performance on path-integration dependent tasks. A more controversial area of work has been top-down normative modelling: why has the brain chosen to compute like this? Floods of ink have been spilt attempting to build a precise link between the population's objective and the measured implementation. The holy grail is a normative link with broad predictive power which generalises to other neural systems. We review this literature and argue that, despite some controversies, the literature largely agrees that grid cells can be explained as a (1) biologically plausible (2) high fidelity, non-linearly decodable code for position that (3) subserves path-integration. As a rare area of neuroscience with mature theoretical and experimental work, this story holds lessons for normative theories of neural computations, and on the risks and rewards of integrating task-optimised neural networks into such theorising.

Recent developments in spatial omics technologies have enabled the generation of high dimensional molecular data, such as transcriptomes, proteomes, and epigenomes, within their spatial tissue context, either through coprofiling on the same slice or through serial tissue sections. These datasets, which are often complemented by images, have given rise to multimodal frameworks that capture both the cellular and architectural complexity of tissues across multiple molecular layers. Integration in such multimodal data poses significant computational challenges due to differences in scale, resolution, and data modality. In this review, we present a comprehensive overview of computational methods developed to integrate multimodal spatial omics and imaging datasets. We highlight key algorithmic principles underlying these methods, ranging from probabilistic to the latest deep learning approaches.

A major goal of computational neuroscience has been to explain how the primate ventral visual stream (VVS) transforms visual input into temporally evolving neural representations that support robust visual perception. Historically, most modeling efforts have assumed static conditions: monkeys fixate a dot, images are briefly flashed, and neural responses are analyzed through time-averaged metrics. Feedforward deep networks trained on static object recognition tasks outperform prior work in approximating these static snapshot-driven VVS responses. However, mounting neurophysiological evidence demonstrates that VVS responses are rich dynamical signals shaped not only by the retinal input but also by intrinsic circuit dynamics, recurrent interactions, and widespread top-down modulation. Moreover, real-world vision is inherently dynamic: objects move, the observer moves, and the eyes actively sample the environment. Here, we review recent progress in modeling dynamic responses in the macaque ventral stream across three domains: (1) intrinsic dynamics elicited by static images, (2) dynamics evoked by dynamic visual stimuli, and (3) dynamics generated by active sensing during eye movements. We argue that accurately modeling VVS dynamics will require representational, circuit-level, and behavioral perspectives, including multi-area recurrence, structured E/I interactions, and temporal objectives that better reflect natural behavior. We outline some key missing ingredients and propose a roadmap toward dynamic, multi-timescale models of the primate VVS.

A model of DNA replication is investigated, which is based on the biochemical kinetics of DNA polymerases, copying a DNA strand into its complement, except for rare point-like mutations due to nucleotide substitutions. Numerical simulations of many successive replications starting from an initial DNA sequence show that the fractions of mono- and oligonucleotides converge toward compliance with Chargaff's second parity rule. The theory of this multireplication process reveals that the approximate equalities between the fractions of complementary nucleotides are the consequence of (1) the dominant role of complementarity in the DNA replication kinetic process and (2) the smallness of the polymerase error probability. These two features lead to a robust mechanism underlying Chargaff's second parity rule.

Understanding how the brain integrates motor suppression with motivational processes remains a fundamental question in neuroscience. The rostral Pedunculopontine nucleus, a brainstem structure involved in motor control, has been shown to induce transient motor arrest upon optogenetic or electrical stimulation. However, our current understanding of its potential role in linking motor suppression with motivational or reinforcement-related processes is still insufficient. To further explore the effects induced by PPN stimulations and infer the potential mechanism underlying its role involved in both motor and emotional regulation, we developed a fully automated, low-cost system combining real-time animal tracking with closed-loop optogenetic stimulation, using the OpenMV Cam H7 Plus and embedded neural network models. The system autonomously detects the rat's position and triggers optical stimulation upon entry into a predefined region of interest, enabling unbiased, unsupervised behavioral assays. Optogenetic activation of CaMKIIa-expressing neurons in the rostral PPN reliably induced transient motor arrest. When stimulation was consistently paired with a specific location in a conditioned place preference task. When motor arrest was spatially paired with a defined region of interest, rats developed a robust place preference after limited training. These results suggest that rostral PPN activation can couple motor inhibition with reinforcement-related behavioral circuitry. Together, our work provides both a technical framework for scalable closed-loop neuroscience experiments and preliminary evidence that the rostral PPN may participate in coordinating motor suppression with motivational processes.

The exploration of vast genotype spaces poses fundamental challenges for evolving populations. As the number of genotypes encoding viable phenotypes grows exponentially with genome length, populations can only explore a tiny fraction of these immense spaces, a fact consistently supported by empirical and theoretical evidence. Paradoxically, local, mutation-driven searches near abundant sequences allow populations to generate phenotypic improvements and functional innovations despite this immense search space. In this contribution, we integrate insights from viral evolution with theoretical expectations derived from genotype-phenotype maps to re-examine how high-dimensional sequence spaces shape evolutionary dynamics. In resolving the paradox, abundant phenotypes play a crucial role because their combinatorial weight biases evolutionary trajectories. We discuss how this bias, together with limited accessibility of fitness peaks, modifies traditional metaphors -- such as fitness landscapes -- and challenges standard notions of evolutionary optimality. Our results underscore that adaptation is predominantly local yet remarkably efficient, providing a unifying perspective on the coexistence of robustness, innovation, and constrained exploration in molecular evolution.

Biochemical signaling cascades transmit intracellular information while dissipating energy under nonequilibrium conditions. We model a cascade as a code string and apply information-entropy ideas to quantify an optimal transmission rate. A time-normalized entropy functional is maximized to define a capacity-like quantity governed by a conserved multiplier. To place the theory on a rigorous stochastic-thermodynamic footing, we formulate stepwise signaling as a continuous-time Markov jump process with forward and reverse competing rates. The embedded jump chain yields well-defined transition probabilities that justify time-scale-based expressions. Under local detailed balance, the log ratio of forward and reverse rates can be interpreted as entropy production per event, enabling a trajectory-level derivation of detailed and integral fluctuation theorems. We further connect the information-theoretic capacity to the mean dissipation rate and outline finite-time fluctuation structure via the scaled cumulant generating function (SCGF) and Gallavotti--Cohen symmetry, including a worked example using MAPK/ERK timescales.

To develop a deep-learning method for achieving fast high-resolution MR elastography from highly undersampled data without the need of high-quality training dataset. We first framed the deep neural network representation as a nonlinear extension of the linear subspace model, then used it to represent and reconstruct MRE image repetitions from undersampled k-space data. The network weights were learned using a multi-level k-space consistent loss in a self-supervised manner. To further enhance reconstruction quality, phase-contrast specific magnitude and phase priors were incorporated, including the similarity of anatomical structures and smoothness of wave-induced harmonic displacement. Experiments were conducted using both 3D gradient-echo spiral and multi-slice spin-echo spiral MRE datasets. Compared to the conventional linear subspace-based approaches, the nonlinear network representation method was able to produce superior image reconstruction with suppressed noise and artifacts from a single in-plane spiral arm per MRE repetition (e.g., total R=10), yielding comparable stiffness estimation to the fully sampled data. This work demonstrated the feasibility of using deep network representations to model and reconstruct MRE images from highly-undersampled data, a nonlinear extension of the subspace-based approaches.

Accurate segmentation of brain tumors is essential for clinical diagnosis and treatment planning. Deep learning is currently the state-of-the-art for brain tumor segmentation, yet it requires either large datasets or extensive computational resources that are inaccessible in most areas. This makes the problem increasingly difficult: state-of-the-art models use thousands of training cases and vast computational power, where performance drops sharply when either is limited. The top performer in the Brats GLI 2023 competition relied on supercomputers trained on over 92,000 augmented MRI scans using an AMD EPYC 7402 CPU, six NVIDIA RTX 6000 GPUs (48GB VRAM each), and 1024GB of RAM over multiple weeks. To address this, the Karhunen--Loève Expansion (KLE) was implemented as a feature extraction step on downsampled, z-score normalized MRI volumes. Each 240$\times$240$\times$155 multi-modal scan is reduced to four $48^3$ channels and compressed into 32 KL coefficients. The resulting approximate reconstruction enables a residual-based anomaly map, which is upsampled and added as a fifth channel to a compact 3D U-Net. All experiments were run on a consumer workstation (AMD Ryzen 5 7600X CPU, RTX 4060Ti (8GB VRAM), and 64GB RAM while using far fewer training cases. This model achieves post-processed Dice scores of 0.929 (WT), 0.856 (TC), and 0.821 (ET), with HD95 distances of 2.93, 6.78, and 10.35 voxels. These results are significantly better than the winning BraTS 2023 methodology for HD95 distances and WT dice scores. This demonstrates that a KLE-based residual anomaly map can dramatically reduce computational cost and data requirements while retaining state-of-the-art performance.

Glioblastoma, IDH-wildtype (GBM-IDHwt) is the most common malignant brain tumor. Histomorphology is a crucial component of the integrated diagnosis of GBM-IDHwt. Artificial intelligence (AI) methods have shown promise to extract additional prognostic information from histological whole-slide images (WSI) of hematoxylin and eosin-stained glioblastoma tissue. Here, we present an explainable AI-based method to support systematic interpretation of histomorphological features associated with survival. It combines an explainable multiple instance learning (MIL) architecture with a sparse autoencoder (SAE) to relate human-interpretable visual patterns of tissue to survival. The MIL architecture directly identifies prognosis-relevant image tiles and the SAE maps these tiles post-hoc to visual patterns. The MIL method was trained and evaluated using a new real-world dataset that comprised 720 GBM-IDHwt cases from three hospitals and four cancer registries in Germany. The SAE was trained using 1878 WSIs of glioblastoma from five independent public data collections. Despite the many factors influencing survival time, our method showed some ability to discriminate between patients living less than 180 days or more than 360 days solely based on histomorphology (AUC: 0.67; 95% CI: 0.63-0.72). Cox proportional hazards regression confirmed a significant difference in survival time between the predicted groups after adjustment for established prognostic factors (hazard ratio: 1.47; 95% CI: 1.26-1.72). Our method identified multiple interpretable visual patterns associated with survival. Three neuropathologists separately found that 21 of the 24 most strongly associated patterns could be clearly attributed to seven histomorphological categories. Necrosis and hemorrhage appeared to be associated with shorter survival while highly cellular tumor areas were associated with longer survival.

Understanding infectious disease spread remains a critical public health challenge, particularly given the interplay between household dynamics and community transmission patterns. Traditional epidemiological models often oversimplify these dynamics by treating populations as homogeneous, failing to capture crucial household-level interactions that can significantly impact disease spread. This paper introduces a new stochastic differential equation model extending the SIR framework by capturing the randomness in disease spread and incorporating household structure and heterogeneous mixing patterns. The model divides the population into groups based on age and household size, includes subpopulation-targeted lockdown parameters and constructs detailed contact matrices accounting for both public and within-household interactions. Through the approximation of Markov jump processes by branching processes near the disease free equilibrium, we derive the basic reproduction number of our model and conduct global sensitivity analysis using Sobol indices to identify influential factors. Our simulations reveal that incorporating household structure leads to substantially different predictions compared to traditional models, particularly in epidemic timing and peak intensity. The stochastic framework captures important variations in outbreak trajectories overlooked by deterministic approaches, especially during early and peak phases. This work contributes to both mathematical epidemiology and practical public health planning by providing a sophisticated mathematical understanding of how population structure and randomness influence disease dynamics, offering insights for intervention strategies where household transmission plays a significant role.

Recently, researchers have proposed the Asymmetric Bethe ansatz method - a theoretical tool that extends the scope of Bethe ansatz-solvable models by "breaking" partial mirror symmetry via the introduction of a fully reflecting boundary. Within this framework, the integrability conditions which were originally put forward by Gaudin have been further generalized. In this work, building on Gaudin's generalized kaleidoscope model, we present a detailed investigation of the relationship between DN symmetry and its integrability. We demonstrate that the mathematical essence of integrability in this class of models is characterized by a newly proposed Kaleidoscope Yang-Baxter Equation. Furthermore, we show that the solvability of a model via the coordinate Bethe ansatz depends not only on the consistency relations satisfied by scattering matrices, but also on the model's boundary conditions and the symmetry of the subspace where solutions are sought. Through finite element method based numerical studies, we further confirm that Bethe ansatz integrability arises in a specific symmetry sector. Finally, by analyzing the algebraic structure of the Kaleidoscope Yang-Baxter Equation, we derive a series of novel quantum algebraic identities within the framework of quantum torus algebra.

Continuous quantum metrology holds promise for realizing high-precision sensing by harnessing information progressively carried away by the radiation quanta emitted into the environment. Despite recent progress, a comprehensive understanding of the fundamental precision limits of continuous metrology with bosonic systems is currently lacking. We develop a general theoretical framework for quantum metrology with multimode free bosons under continuous Gaussian measurements. We derive analytical expressions for the asymptotic growth rates of the global quantum Fisher information (QFI) and the environmental QFI, which quantify the total information encoded in the joint system-environment state and the information accessible from the emitted radiation, respectively. We derive fundamental bounds on these quantities, showing that while Heisenberg-type scaling with the number of modes is attainable, the precision scales at most linearly with time and a meaningful energy resource. To illustrate our findings, we analyze several concrete setups, including coupled cavity arrays and trapped particle arrays. While a local setup yields a standard linear scaling with resources, a globally coupled setup can achieve the optimal quadratic scaling in terms of the mode number. Furthermore, we demonstrate that a nonreciprocal setup can leverage the non-Hermitian skin effect to realize an exponentially enhanced global QFI. Notably, however, this enhancement cannot be reflected in the environmental QFI, highlighting a fundamental distinction between the information stored within the joint state and the information radiated into the environment. These findings establish an understanding of the resource trade-offs and scaling behaviors in continuous bosonic sensing.

A newly born neutron star is expected to exhibit significant deviations from spherical symmetry, which decay with time. Determining how much deformation remains at present is crucial for gravitational-wave astronomy. This study is the first investigation into the evolution of quadrupole deformation during the solid crust formation phase to obtain a plausible value at present. The equilibrium structure before solidification is modeled using a fluid description, and the deformation is introduced through an assumed driving force. As the star cools, this force weakens, leading to a gradual decay of the deformation. Eventually, the deformation vanishes in the fluid region but partially remains in the crust, sustained by elastic forces, after solidification. By comparing the equilibrium models before and after solidification, we estimate the residual ellipticity and demonstrate that the spatial profile of the elastic shear is imprinted in the crust. The relic ellipticity is only a few percent of the original value, with its absolute magnitude depending on the deformation mechanism during the hot era, which cannot be specified owing to the lack of elaborate models. This work provides a first step toward linking early neutron star deformation with future gravitational-wave observations.

In this paper, we investigate the non-Hermitian skin effect in a semi-infinite Fock-state lattice, where the inherent coupling scales as \sqrt{n}. By analytically solving a non-uniform, non-reciprocal SSH model, we demonstrate that the intrinsic inhomogeneous coupling, in combination with nonreciprocity, fundamentally modifies the conventional skin effect. Instead of accumulating at the physical boundary, all eigenmodes become compressed and skewed within a finite spatial range determined by the inhomogeneous profile-a phenomenon we term the confined non-Hermitian skin effect. Consequently, the evolution of the probability distribution on the lattice starting from a single site is doubly confined: it is spatially bounded to a finite range by the inhomogeneous coupling, and further restricted to a one-sided trajectory at the edge of this range by the non-reciprocity. Moreover, a feasible experimental scheme based on a single trapped ion is also proposed. This work reveals how engineered coupling profiles in synthetic dimensions can reshape non-Hermitian properties and enable new protocols for quantum state manipulation.

Within the standard framework in which electroweak sphaleron processes relate lepton and baryon number, we derive an upper limit on the amplitude of a chiral gravitational wave background produced prior to the electroweak epoch. This bound is independent of the production time of chiral GWs for superhorizon modes, while it becomes sensitive to the production time for subhorizon modes. For sufficiently high reheating temperatures, the bound becomes significantly more stringent than the conventional big bang nucleosynthesis constraints at frequencies above the MHz scale, thereby providing a powerful and \emph{model-independent} probe of parity-violating physics in the early Universe.

Nonequilibrium dynamics of quantum many-body systems is one of the main targets of quantum simulations. This focus - together with rapid advances in quantum-computing hardware - has driven increasing applications in high-energy physics, particularly in lattice gauge theories. However, most existing experimental demonstrations remain restricted to (1+1)-dimensional and/or abelian gauge theories, such as the Schwinger model and the toric code. It is essential to develop quantum simulations of nonabelian gauge theories in higher dimensions, addressing realistic problems in high-energy physics. To fill the gap, we demonstrate a quantum simulation of thermalization dynamics in a (2+1)-dimensional $q$-deformed $\mathrm{SU}(2)_3$ Yang-Mills theory using a trapped-ion quantum computer. By restricting the irreducible representations of the gauge fields to the integer-spin sector of $\mathrm{SU}(2)_3$, we obtain a simplified yet nontrivial model described by Fibonacci anyons, which preserves the essential nonabelian fusion structure of the gauge fields. We successfully simulate the real-time dynamics of this model using quantum circuits that explicitly implement $F$-moves. In our demonstrations, the quantum circuits execute up to 47 sequential $F$-moves. We identify idling errors as the dominant error source, which can be effectively mitigated using dynamical decoupling combined with a parallelized implementation of $F$-moves.

We investigate the quasinormal modes of polar metric-dilaton perturbations around the dilaton-Euler-Heisenberg (dEH) black holes with dilaton hair obtained from the Einstein-Maxwell-dilaton theory with two dilaton coupling constants ($α,β$) to the nonlinear Euler-Heisenberg term. We compute the quasinormal mode spectra by making use of two numerical techniques: direct integration and matrix values continued fraction methods. An excellent agreement is found between two approaches, confirming the robustness of our computation. We present the fundamental quasinormal frequencies for both gravitational and dilaton modes and analyze their dependence on the magnetic charge ($Q_m$), angular momentum quantum number $(l)$, and coupling parameter ($ε=α-β$). All negative imaginary quasinormal frequencies imply that the dEH black hole with dilaton hair is stable against polar metric-dilaton perturbations. Also, our results reveal distinct qualitative behaviors between $ε=1$ and $ε=-1 $, particularly in the damping rates near the extremality.

We show that regularizing $(2+1)$-dimensional Minkowski spacetime with a finite-resolution Gaussian probe, analogous to Weyl-Heisenberg (Gabor) signal analysis and related quantization, induces a curved geometry with a topological defect. The regularized metric replaces $r^2$ by $r^2+σ^2$ in the angular part, where $σ$ is the resolution scale from the width of the Gaussian probe. The resulting Gaussian curvature integrates to $-2π$, independently of $σ$, and including the boundary contribution, yields Euler characteristic $χ=0$, corresponding to a punctured plane. This curvature defines an effective stress-energy source with total energy $E_{\text{eff}}=-1/(4G)$, universal and $σ$-independent. Spatial slices embed isometrically as helicoids, and geodesics exhibit a characteristic swirling motion. These results show that finite spatial resolution measurement does not merely smooth singularities but imprints topological defects with fixed physical consequences, suggesting that observational limitations fundamentally shape spacetime geometry. We show how our Gabor regularisation is extendable to $(3+1)$ Minkowski space-time.

In this paper, we show that in Hilbert space of any finite dimension N, there are N^2 pure states which constitute Symmetric Informationally Complete Positive Operator Valued Measure (SIC-POVM).

Recently, quantum entanglement has been presented as a cohomological obstruction to reconstructing a global quantum state from locally compatible information, where sheafification provides a functor that is forgetful with regards to global-from-local signatures while acting faithfully with respect to within-patch multipartite structures. Nontrivial connections to Hecke modifications and the geometric Langlands program are explored in the process. The aim of this work is to validate and extend a number of the claims made in [arXiv:2511.04326] through both theoretical analysis and numerical simulations, employing concrete perspectives from condensed matter physics.

As quantum machine-learning architectures mature, a central challenge is no longer their construction, but identifying the regimes in which they offer practical advantages over classical approaches. In this work, we introduce a framework for addressing this question in data-driven hadronic physics problems by developing diagnostic tools - centered on a quantitative quantum qualifier - that guide model selection between classical and quantum deep neural networks based on intrinsic properties of the data. Using controlled classification and regression studies, we show how relative model performance follows systematic trends in complexity, noise, and dimensionality, and how these trends can be distilled into a predictive criterion. We then demonstrate the utility of this approach through an application to Compton form factor extraction from deeply virtual Compton scattering, where the quantum qualifier identifies kinematic regimes favorable to quantum models. Together, these results establish a principled framework for deploying quantum machine-learning tools in precision hadronic physics.

We present a novel tunnel architecture for the Einstein Telescope that departs from the traditional large-cavern approach and reduces the excavated volume by an order of magnitude. In the proposed design, all seismic isolation systems are housed in raise-bore wells drilled upward from the main tunnel toward an upper service tunnel. The pre-isolators for the most sensitive optics are located in the service tunnel, seating directly on strong and compact rock, while the other filters are distributed along the wells within compact, side-access vacuum chambers. Shorter, separate wells accommodate the seismic isolation systems for less demanding optics. This configuration provides substantial advantages: easier lock acquisition and improved robustness of the interferometers, lower-frequency pendulum stages, reduced congestion around the test masses, simplified installation and maintenance, improved vacuum partitioning, strong physical decoupling between the high- and low-frequency interferometers, and enhanced compatibility with future advances of Newtonian-noise cancellation. A novel technique for real-time, precision monitoring of rock motion and tilt provides a new signal for Newtonian noise cancellation and enables correction of seismic disturbances even during earthquakes, offering unique geophysical measurement capabilities.

We demonstrate a neutral atom networking node that combines high photon collection efficiency with high atom photon entanglement fidelity in a compact, fiber integrated platform. A parabolic mirror is used both to form the trap and to collect fluorescence from a single rubidium atom, intrinsically mode matching $σ$ polarized emitted photons to the fiber and rendering the system largely insensitive to small imperfections or drifts. The core optics consist of millimeter scale components that are pre aligned, rigidly bonded on a monolithic invacuum assembly, and interfaced entirely via optical fibers. With this design, we measure an overall photon collection and detection efficiency of $3.66\%$, from which we infer an overall collection efficiency of $6.6\%$ after the single--mode fiber coupling. We generate atom photon entangled states with a raw Bell state fidelity of 0.93 and an inferred fidelity of 0.98 after correcting for atom readout errors. The same node design has been realized in two independent setups with comparable performance and is compatible with adding high NA objective lenses to create and control atomic arrays at each node. Our results establish a robust, cavity free neutral atom interface that operates near the limit set by the collection optics numerical aperture and provides a practical building block for scalable quantum network nodes and repeaters.

We experimentally determine the coupling strength between two strongly coupled nanomechanical modes using a Ramsey-inspired technique optimized for signals as short as four fringes. The method is applied to precisely probe the change of the coupling rate induced by a modification of the microwave-cavity readout field. It opens a pathway towards sensing electrostatic field fluctuations approaching single-charge resolution.

Nitrogen vacancy centres (NVC) in diamond have been widely used for near-dc magnetometry. The intrinsic properties of diamonds make them potential candidates for tokamak fusion power diagnostics, where radiation-hard magnetometers will be essential for efficient control. An NVC magnetometer placed in a tokamak will need to operate within a $\geq$ 1 T magnetic field. In this work, we demonstrate fibre-coupled ensemble NVC optically detected magnetic resonance (ODMR) and magnetometry measurements at magnetic fields up to 1.2 T. Sensitivities of approximately 240 to 600 nT/$\sqrt{\textrm{Hz}}$ and 110 nT/$\sqrt{\textrm{Hz}}$ are achieved in a (10-150) Hz frequency range, for non-degenerate and near-$\langle$111$\rangle$ field alignments respectively.

We continue our series of papers where we study the quasinormal modes, and their excitation, of black holes in the simplest beyond general relativity model in which first-principle calculations are tractable: a nonrotating black hole in an effective-field-theory extension of general relativity with cubic-in-curvature terms. In this theory, the equivalence between the quasinormal mode spectra associated with metric perturbations of polar and axial parities ("isospectrality") of the Schwarzschild black hole in general relativity no longer holds. How does this loss of isospectrality translate into the time-domain ringdown of gravitational waves? Given such a ringdown, can we identify the two "fundamental quasinormal modes" associated to the two metric-perturbation parities? We study these questions through a large suite of time-domain numerical simulations, by a prescription on how to relate the gauge-invariant master functions that describe metric perturbations of each parity with the gravitational polarizations. Under the assumptions made in our calculations, we find that it is in general difficult to identify either of the two fundamental modes from the time series, although finding evidence for a non-general-relativistic mode is possible sometimes. We discuss our results in light of our assumptions and speculate about what may occur when they are relaxed.

Weighted graph states extend standard graph states by associating phases with entangling edges, and may serve as resources for measurement-based quantum computation (MBQC). We analyze how the two main fusion operations, Type-I and Type-II, act on weighted graph states. Type-I fusion operates identically to the unweighted case, merging two one-dimensional weighted graphs, while preserving edge weights and success probabilities. In addition, the pool of 2-qubit weighted graph states can be generated easily by GHZ states or Bell pairs. In contrast, Type-II fusion requires a logical qubit, which can be formed only for specific weight configurations, and with success probability below one-half, which is an obstacle one can avoid. When successful, it fuses the states correctly, but its failure outcomes destroy the structure of the graphs, removing the good-failure feature, known from ordinary graph states. We compute the entanglement reduction of the resulting link due to the fused states being weighted graph states (for generalized fusion), and classify the resulting states of a general non-Bell projection. These results define the practical limits of the fusion-based construction of weighted graph states for MBQC.

We theoretically investigate loopless multiterminal hybrid superconducting devices at odd fermion parity with time-reversal symmetry. We find that the energy-phase relationship has a double minimum corresponding to opposite windings of the superconducting phases. Spin-orbit coupling adds multi-axial spin splittings, which contrasts with two-terminal devices where spin dependence is uniaxial. Capacitive shunting localizes quantum circuit states in the wells and exponentially suppresses their splitting. For weak spin-orbit strength, the system has a four-dimensional spin-chirality low-energy subspace which can be universally controlled with electric fields only.

Stochastic resetting describes dynamics which are reinitialized to a reference state at random times. These protocols are attracting significant interest: they can stabilize nonequilibrium stationary states, generate correlations in noninteracting systems, and enable optimal search strategies. While a constant reset probability results in a Markovian dynamics, much less is known about non-Markovian effects in quantum stochastic resetting. Here, we analyze memory effects in these processes -- examining the evolution of quantum states and of observables -- through witnesses of non-Markovianity for open quantum systems. We focus on discrete-time reset processes, which are of particular interest as they can be implemented on existing gate-operated quantum devices. We show that these processes are generically described by non-divisible maps and, in non-classical scenarios where the effective reset probability can become negative, can feature revivals in the state distinguishability. Our results reveal non-Markovian effects in quantum stochastic resetting and show that a time-dependent reset may be exploited to engineer enhanced stationary quantum correlations.

We derive the quantum filter for a quantum open system undergoing quadrature measurements (homodyning) where the input field is in a general quasi-free state. This extends previous work for thermal input noise and allows for squeezed inputs. We introduce a convenient class of Bogoliubov transformations which we refer to as balanced and formulate the quantum stochastic model with squeezed noise as an Araki-Woods type representation. We make an essential use of the Tomita-Takesaki theory to construct the commutant of the C*-algebra describing the inputs and obtain the filtering equations using the quantum reference probability technique. The derived quantum filter must be independent of the choice of representation and this is achieved by fixing an independent quadrature in the commutant algebra.

We investigate how quantum coherence scales and is redistributed in quantum communication protocols, using superdense coding and quantum teleportation as paradigmatic case studies. Employing the relative entropy of coherence as a circuit-level resource measure, we show that multipartite resource states relevant to generalized superdense coding can enable scalable communication while exhibiting only logarithmic or even constant coherence growth, depending on their entanglement structure. In sharp contrast, quantum teleportation displays an unavoidable, protocol-induced coherence cost that grows linearly with the number of teleported qubits and is independent of the input state. Through a stage-resolved analysis of the teleportation circuit, we separate protocol-generated coherence from message-dependent contributions and identify a universal two-bit coherence offset per teleported qubit at the maximal-coherence stage. We further demonstrate explicitly that this extensive intermediate coherence generation is fully consistent with information-theoretic bounds, including the Holevo limit, and does not correspond to an increase in accessible classical information.

Quantum error correction (QEC) is essential for scalable quantum computing, yet repeated syndrome-measurement cycles dominate its spacetime and hardware cost. Although stabilizers commute and admit many valid execution orders, different schedules induce distinct error-propagation paths under realistic noise, leading to large variations in logical error rate. Outside of surface codes, effective syndrome-measurement scheduling remains largely unexplored. We present AlphaSyndrome, an automated synthesis framework for scheduling syndrome-measurement circuits in general commuting-stabilizer codes under minimal assumptions: mutually commuting stabilizers and a heuristic decoder. AlphaSyndrome formulates scheduling as an optimization problem that shapes error propagation to (i) avoid patterns close to logical operators and (ii) remain within the decoder's correctable region. The framework uses Monte Carlo Tree Search (MCTS) to explore ordering and parallelism, guided by code structure and decoder feedback. Across diverse code families, sizes, and decoders, AlphaSyndrome reduces logical error rates by 80.6% on average (up to 96.2%) relative to depth-optimal baselines, matches Google's hand-crafted surface-code schedules, and outperforms IBM's schedule for the Bivariate Bicycle code.

Statistical ensembles of reduced density matrices of bipartite quantum systems play a central role in entanglement estimation, but do not capture the non-stationary nature of entanglement relevant to realistic quantum information processing. To address this limitation, we propose a dynamical extension of the Hilbert-Schmidt ensemble, a baseline statistical model for entanglement estimation, arising from non-intersecting squared Bessel processes and perform entanglement estimation via average entanglement entropy and quantum purity. The investigation is enabled by finding spectral moments of the proposed dynamical ensemble, which serves as a new approach for systematic computation of entanglement metrics. Along the way, we also obtain new results for the underlying multiple orthogonal polynomials of modified Bessel weights, including structure and recurrence relations, and a Christoffel-Darboux formula for the correlation kernels.

We present a consistent relativistic formulation of Mach principle within a geometric theory of gravitation. In this approach, neither inertia nor free fall is assumed a priori. Instead, the motion of any local system arises from its dynamical interdependence, direct or indirect, with the rest of the Universe. This viewpoint provides a sharp criterion for what qualifies as a genuinely Machian gravitational theory. A theory is Machian only if the global mass energy distribution underwrites the very existence of the pseudo Riemannian structure of spacetime, rather than merely determining local features such as curvature on a pre existing background. We analyze the resulting cosmological model and discuss its phenomenological implications. In the appropriate local limit, the theory reduces to General Relativity, thereby preserving agreement with all Solar System tests.

The color code model is a crucial instance of a Calderbank--Shor--Steane (CSS)-type topological quantum error-correcting code, which notably supports transversal implementation of the full Clifford group. Its robustness against local noise is rooted in the structure of its topological excitations. From the perspective of quantum phases of matter, it is essential to understand these excitations in the thermodynamic limit. In this work, we analyze the color code model on an infinite lattice within the quasi-local $C^{*}$-algebra framework, using a cone-localized Doplicher-Haag-Roberts (DHR) analysis. We classify its irreducible anyon superselection sectors and construct explicit string operators that generate anyonic excitations from the ground state. We further examine the fusion and braiding properties of these excitations. Our results show that the topological order of the color code is described by $\mathsf{Rep}(D(\mathbb{Z}_2 \times \mathbb{Z}_2)) \simeq \mathsf{Rep}(D(\mathbb{Z}_2)) \boxtimes \mathsf{Rep}(D(\mathbb{Z}_2))$, which is equivalent to a double layer of the toric code and consistent with established analyses on finite lattices.

Using a map between the Lindbladian evolution of dephasing in free fermions and the time evolution of imaginary-interaction Fermi-Hubbard models in bipartite lattices, we present an efficient classical algorithm to solve the Schrödinger equation in these interacting systems. This algorithm leverages the recently discovered algorithm for simulating Lindbladian evolution by sampling mixed unitary channels (Wang et al arXiv:2601.06298). We comment on the expected classical complexity of the problem for general complex values of the parameters and discuss some applications.

The spontaneous disentanglement hypothesis is motivated by some outstanding issues in standard quantum mechanics, including the problem of quantum measurement. The current study compares between some possible methods that can be used to implement the hypothesis. Disentanglement is formulated using a nonlinear operator, which can be used to modify both the Schrödinger equation for the quantum state vector, and the master equation for the density operator. Two types of nonlinear disentanglement operators are explored. The first one gives rise to matrix deranking, and the second one to correlation suppression. Both types are demonstrated using a two spin system that is driven close to the Hartmann--Hahn double resonance. It is shown that limit cycle steady state solutions, which are excluded by standard quantum mechanics, become possible in the presence of disentanglement.

We investigate the thermodynamics of a spatially flat Friedmann-Robertson-Walker (FRW) universe within the framework of Generalized Proca (GP) theory, a comprehensive vector-tensor theory. By adopting two distinct dark energy models in GP, we derive the corresponding modified Friedmann equations, the thermodynamic first law for the apparent horizon, and the equation of state for these models. For the first model, characterized by power-law couplings and an ansatz linking the Proca field to the Hubble parameter, we analytically demonstrate the existence of a critical point in the pressure-volume$(P-V)$ diagram, indicating a $P-V$ phase transition. For the second model, defined by a Proca field marginal coupled to curvature, we show that there is no phase transition when the coupling constants are selected within the range permitted by observations. This investigation not only extends the FRW thermodynamics to vector-tensor theories but also demonstrates that cosmological phase transitions can serve as a powerful diagnostic tool for distinguishing viable dark energy models in GP theory.

We investigate quantum correlations in a hybrid qubit-qutrit system subject to both axial and planar single-ion anisotropies, dipolar spin-spin interactions, and Dzyaloshinskii-Moriya (DM) coupling. Using Negativity, Measurement-Induced Non-locality (MIN), Uncertainty-Induced Nonlocality (UIN), and Bell nonlocality (as quantified by the CHSH inequality) as measures, we analyze the interplay between anisotropy parameters, magnetic fields, and temperature on the survival of quantum correlations. Our results demonstrate that Bell nonlocality and entanglement (Negativity) are highly sensitive to temperature and anisotropy, exhibiting sudden death under thermal noise, whereas MIN and UIN are significantly more robust. In particular, these discord-like and information-theoretic measures provide the largest baseline and persist even in parameter regions where entanglement vanishes, highlighting their suitability as a quantumness witness in realistic conditions. Notably, our Bell nonlocality study is tailored to the asymmetric qubit-qutrit setting by exploiting a recently developed qubit-qudit CHSH maximization framework. However, Bell nonlocality is confirmed to be the most fragile, surviving only in narrow parameter windows at low temperature. A key finding of this work is that we observe the fragility hierarchy: Bell nonlocality < Negativity < UIN(MIN) in the qubit-qutrit setting. These results provide deeper insight into the relative robustness of distinct quantum resources in anisotropic qubit-qutrit models, suggesting that quantum discord-like measures, such as MIN and UIN, may serve as more practical resources than entanglement for quantum information tasks in thermally active spin systems.

Causality underpins all logical reasoning. However, the causal structure in quantum processes can be far from intuitive, often differing from its classical counterpart in relativity, which is defined by the light cone. In particular, in systems with measurement and post-selection, causal influence can occur between spacelike separated regions. In this work, we study the causal structure and emergent "arrow of time" in monitored quantum dynamics, particularly their dependence on initial and final states. We propose a new measure, the cross-entropy quantum causal influence, to quantify the extent of causal influence, whose simulation demonstrates exotic causal structures, such as inverted light cones. This quantity can be measured in current quantum computing platforms. Additionally, we provide an analytical understanding of the relation between time arrow and entropy by studying two types of models that are analytically tractable: a quantum Brownian evolution model and a dual-unitary circuit model.

Quantum simulation provides a powerful route for exploring many-body phenomena beyond the capabilities of classical computation. Existing approaches typically proceed in the forward direction: a model Hamiltonian is specified, implemented on a programmable quantum platform, and its phase diagram and properties are explored. Here we present a quantum algorithmic framework for inverse quantum simulation, enabling quantum material design with desired properties. Target material characteristics are encoded as a cost function, which is minimized on quantum hardware to prepare a many-body state with the desired properties in quantum memory. Hamiltonian learning is then used to reconstruct a low-energy Hamiltonian for which this state is an approximate ground state, yielding a physically interpretable model that can guide experimental synthesis. As illustrative applications, we outline how the method can be used to search for high-temperature superconductors within the fermionic Hubbard model, enhancing $d$-wave correlations over a broad range of dopings and temperatures, design quantum phases by stabilizing a topological order through continuous Hamiltonian modifications, and optimize dynamical properties relevant for photochemistry and frequency- and momentum-resolved condensed-matter data. These results extend the scope of quantum simulators from exploring quantum many-body systems to designing and discovering new quantum materials.

We study classical-quantum (C-Q) channel resolvability. C-Q channel resolvability has been proved by only random coding in the literature. In our previous study, we proved channel resolvability by deterministic coding, using multiplicative weight update algorithm. We extend this approach to C-Q channels and prove C-Q channel resolvability by deterministic coding, using the matrix multiplicative weight update algorithm. This is the first approach to C-Q channel resolvability using deterministic coding.

We investigate the effective quantum evolution of the Bianchi type I cosmological model within the Brans-Dicke framework using an effective Hamiltonian approach that includes expectation values, quantum dispersions and cross-correlation terms between different degrees of freedom. For the case $ω< -3/2$, where energy conditions are violated and bouncing solutions exist classically, we demonstrate that quantum backreaction effects significantly smooth the bounce of directional scale factors, with the bounce occurring at scales set by the quantum state width. For the conformally invariant case $ω= -3/2$, quantum corrections cause scale factors to enter accelerated expansion phases more rapidly than in the classical limit. Most significantly, we show that cross-correlation terms between canonical variables are essential for obtaining physically consistent effective dynamics: neglecting these terms leads to spurious divergences and unphysical behavior. When correlations are included, small-amplitude oscillations appear shortly after the bounce, rapidly damping to classical trajectories. We interpret these oscillations as quantum remnant effects encoding information about correlations between gravitational and scalar field degrees of freedom. Our results demonstrate that cross-correlations carry crucial quantum information that substantially influences cosmological dynamics, with implications for quantum gravity phenomenology near singularities. We compare our findings with existing loop quantum cosmology results.

Precision, robustness, and efficiency are crucial aspects in the design of quantum technologies. Here, we show how genuine quantum features, together with non-Gaussianity, can be the key elements to achieve the best of these three aspects during a quantum battery-charging process. Taking inspiration from a light-matter interaction paradigm, i.e., the Jaynes-Cummings model, we employ the Full Counting Statistics to study the stochastic exchanges of energy between an entire stack of qubits and a single-mode electromagnetic field (or mechanical oscillator). Our study allows to conclude that charging the battery through a sequential protocol involving a quantum non-Gaussian field state guarantees extremely high-performances in the charging process, whose precision is maximized even under sub-optimal operating conditions. These results highlight the potential of non-Gaussian quantum state charging to achieve a robust quantum precision advantage over Gaussian states of the field by suppressing detrimental quantum fluctuations, thus making it suitable to ultimate tasks for which a significant degree of accuracy is required.

Topological quantum computing is expected to be less sensitive to noise because information is stored in global states rather than local features. To examine whether different device topologies show measurable differences, we study three geometries with distinct topological invariants: a Mobius strip, a loop, and a trefoil knot, which have been proposed in electronic-structure settings. From quantum capacitance measurements, we extract power versus frequency spectra and fit Lorentzian line shapes to obtain the linewidth, amplitude, signal-to-noise ratio, and coherence time. The signal-to-noise ratio quantifies the ratio of the parity measurement signal to background noise and serves as an indicator of readout quality, while the coherence time characterizes the timescale for decoherence of the quantum state. Across all three topologies, coherence times are similar, with no clear dependence on geometry. In contrast, the signal-to-noise ratio differs in the regime E0 = 10 micro-eV and Z = -1, following the ordering Trefoil, Mobius, and Loop. These results provide a reference point for future experiments aimed at separating genuine topological effects from device-level parameters.

For any kind of wave phenomenon one can find ways to derive the respective dispersion relation from experimental observations and measurements. This dispersion relation determines the structure of the wave equation and thus characterizes the dynamics of the respective wave. Different wave phenomena are thus governed by different differential equations. Here we want to emphasize the experimental approach to matter waves, but before doing so we will discuss and test the procedure for other types of waves, in particular water waves.

Many protocols within quantum cryptography, communications, and computing require the ability to generate entangled states as well as spectral qudits. Nonlinear interferometry is a viable way to engineer these complex quantum states of light. However, it is difficult to achieve a high level of control over spectral correlations. Here, we present a protocol utilizing a nonlinear interferometer with linear spectral phases that can generate both high-dimensional spectral qudits and high-dimensional entangled states. We model the effect of loss and loss of overlap on interference visibility and thereby on the states generated.

Quantum amplifiers are intrinsically nonlinear systems whose performance limits are set by quantum mechanics. In quantum measurement, amplifier operation is conventionally optimized in the linear regime by maximizing signal-to-noise ratio, an objective that is well-suited to parameter estimation but is typically insufficient for more general tasks such as arbitrary quantum state discrimination. Here we show that single-shot quantum state classification can benefit from operating a quantum amplifier outside the linear regime, when the measurement chain is optimized end-to-end for a task-specific cost function. We analyze a realistic superconducting readout architecture that includes state preparation, cryogenic nonlinear amplification, and room-temperature detection with finite noise. By introducing performance metrics tailored to state discrimination, we identify operating regimes in which nonlinear amplification provides a measurable advantage and clarify the trade-offs that ultimately limit classification fidelity. Our results propose the utility of practical nonlinear quantum amplifiers for quantum state discrimination, and are the first step in a broader research program aimed at developing a general framework for end-to-end, resource-limited optimization of nonlinear quantum amplifiers for such quantum information processing applications.

I define a notion of locality LOC, closely modelled on the Bell principle of Local Causality, construed as the condition that single case probabilities cannot be modified by actions at spacelike separation. The new principle, like that of Bell, forces Bell inequalities, but with two loopholes: one is violation of measurement independence, known to Bell, but the other is non-uniqueness of remote outcomes, a loophole only for LOC, not for Local Causality. I also set out a theory of physical probability, applicable to the Everett interpretation, in which the Born rule is derived, and which therefore violates Bell inequalities. I show it is consistent with LOC. Surprisingly, both loopholes are exploited. I conclude not only that physical probability in the Everett interpretation involves no action at a distance, but that the observed violation of Bell inequalities is powerful evidence for many worlds.

Two-photon interference is a cornerstone of photonic quantum technologies. However, its practical implementation in promising hybrid architectures is severely constrained by the requirement of photon wavepacket indistinguishability, in particular, in terms of the photon linewidth and associated time scale. While narrowband filtering can improve interference visibility, it introduces significant photon loss - a critical limitation for applications. Here, we experimentally demonstrate an efficient approach to enable non-classical two-photon interference between spectral-bandwidth mismatched photons using an electro-optic time lens. We increase the visibility of Hong-Ou-Mandel interference between photons of 10-fold spectral bandwidth mismatch by more than 12 times, achieving non-classical two-photon interference visibility without spectral filtering. This result opens the possibility to efficiently integrate quantum systems operating at different time scales for hybrid quantum communication, teleportation, entanglement swapping, distributed sensing, and hybrid quantum computing.

The quantum Fisher information matrix (QFIM) is the cornerstone of multiparameter quantum metrology. In this work, we investigate multiparameter quantum estimation in baryon-antibaryon (B bar-B) pairs produced via the e+ e- -> J/psi -> B bar-B process at the BESIII experiment, utilizing the symmetric logarithmic derivative (SLD) formalism. Moreover, the QFIM defines the quantum Cramer-Rao bound and dictates the choice of optimal probe states. We compare individual and simultaneous estimation strategies for two key physical parameters: the scattering angle phi and the decay parameter alpha_psi. The estimation variances are found to depend strongly on the explored region of the (phi, alpha_psi) parameter space and to display markedly different temporal dynamics. In general, higher true values of a parameter increase the system's sensitivity, thereby significantly reducing the associated variance. While both variances increase with evolution time, they do so at distinct rates, revealing parameter-dependent information loss driven by environmental decoherence. These findings demonstrate the utility of the QFIM framework for multiparameter quantum estimation in realistic open systems and provide new insights into the ultimate precision limits achievable for hyperon decay parameters.

A unique postulate is shown to underly the whole quantum mechanics theory: the invariance of the Heisenberg uncertainty inequality under a group of special nonlinear gauge transformations (NLGT). With this postulate, the quantum mechanics of a free particle is derived from classical mechanics, including the statements of the postulates of quantum mechanics, except for the wave-function-collapse postulate. An explanatory mechanism for the latter postulate is derived by performing an analytical continuation of the NLGTs. This extension results in a Schrödinger-bridge process, intertwined under the NLGT with the standard unitary quantum evolution, and revealing non-quantum (or beyond-quantum) phenomena. Mechanisms of that latter kind, like the ones associated to the quantum measurement process, occur in a new space-like dimension and hence are non causal in nature, in opposition to a time evolution. The present exercice focusses on the free particle in order to highlight the features of the performed derivation in the simplest possible way. Work is in progress to extend the performed derivation beyond that simple case.

We derive the affine tensor associated with the energy and momentum densities of both gravitational and matter fields, the complex pseudo-tensor, for $f(Q)$ non-metric gravity, the straightforward extension of Symmetric Teleparallel Equivalent of General Relativity (STEGR), characterized by a flat, torsion-free, non-metric connection. The local conservation of energy-momentum complex on-shell is satisfied through a continuity equation. An important analogy is pointed out between gravitational pseudo-tensor of teleparallel $f(T)$ gravity, in the Weitzenböck gauge, and the same object of symmetric teleparallel $f(Q)$ gravity, in the coincident gauge. Furthermore, we perturb the gravitational pseudo-tensor $τ^α_{\phantomαλ}$ in the coincident gauge up to the second order in the metric perturbation, obtaining a useful expression for the power carried by the related gravitational waves. We also present an application of the gravitational pseudotensor, determining the gravitational energy density of a Schwarzschild spacetime in STEGR gravity, adopting the concident gauge. Finally, analyzing the conserved quantities on manifolds, the Stokes theorem can be formulated for generic affine connections

The action of tensionless spinning string invariant under reparametrizions, both local supersymmetry and dilatations, is considered. The density of energy-momentum tensor is constructed and vanishing of its covariant divergence is proved. This result arises from mutual cancellation of the bosonic and fermionic contributions. Differences in the geometry of worldsheets swept by tensionless and tensionfull spinning strings are analyzed. Shown is emergence of covariant trace of a torsion tensor on w-s of the tensionless spinning string. It is derived from the condition for the fermionic scalar density to be a composite one including the 2-dim. w-s density simulating the 4-dim. Rarita-Schwinger field. The said condition is accompanied with the Noether condition for covariant divergence of the vector metric density to vanish.

Quantum nonlocality manifests in multipartite systems through entanglement, Bell's nonlocality, and Einstein-Podolsky-Rosen (EPR) steering. While Peres's positive-partial-transpose criterion provides a simple and powerful test for entanglement, a comparably elegant spectral criterion for detecting EPR steering remains an open challenge. In this work, we systematically explore whether a Peres-type criterion can be established for EPR steering in the two-qubit system. Focusing on rank-2 (including rank-1) states and the two-qubit Werner state, we analyze the eigenvalues of their partially transposed density matrices and construct a significant steering criterion based on symmetric combinations of these eigenvalues. We prove that this criterion serves as a necessary and sufficient condition for steerability for the Werner state, all two-qubit pure states, all two-qubit rank-2 states. Furthermore, we validate the criterion for higher-rank states (rank-3 and rank-4) and show that the results align with known steering inequalities. Our findings suggest a more unified framework for detecting quantum nonlocality via partial transposition and open avenues for further theoretical and numerical investigations into steering detection.

It is widely recognized that finite temperatures degrade quantum coherence and can induce thermalization. Here, we study the effect of finite temperature on a kicked Tonks--Girardeau gas, which is known to exhibit many--body dynamical localization and delocalization under periodic and quasiperiodic kicks, respectively. We find that many--body dynamical localization persists at finite--and even high--temperatures, although the coherence of the localized state is further degraded. In particular, we demonstrate a modified effective thermalization of the localized state by considering the initial temperature. Moreover, we show many--body dynamical localization transition at intermediate temperature. Our work extends the study of many--body dynamical localization and delocalization to the finite--temperature regime, providing comprehensive guidance for cold--atom experiments.

We present a numerical study of the evolution of the Hawking mass for closed nonspherical surfaces evolved under a class of expanding flows in Minkowski spacetime. Although formal monotonicity of the Hawking mass under smooth inverse mean curvature flow is well established in the Riemannian setting, comparatively little is known about the robustness of this behavior in discrete numerical implementations applied to explicitly embedded surfaces away from exact symmetry. We consider surfaces defined by small spherical harmonic perturbations of a round sphere and evolve them under an in-slice, time-flat flow analogous to inverse mean curvature flow. We examine the behaviour of the Hawking mass under the flow and find that monotonicity persists for a class of nonspherical perturbations and is robust under variations in perturbation amplitude and angular frequency. We also identify regimes in which numerical instabilities arise, highlighting practical challenges associated with extending such flows beyond simple symmetry assumptions. These results provide a concrete computational testbed for future investigations of uniformly expanding flows and quasi-local mass in more general spacetime settings.

Scalar-tensor theories predict fundamental scalar fields of considerable interest in astrophysics and cosmology. We investigate the superradiant instability of scalar clouds around Kerr black holes, showing that stimulated decay generates detectable electromagnetic signals. The growth of the superradiant scalar cloud differs from that of other bosonic fields and depends sensitively on the matter distribution surrounding the black hole, which originates from the scalar-matter coupling realized by the chameleon mechanism in modified gravity theories. In non-uniform matter distributions, stimulated emission from scalar clouds offers an observational signature that distinguishes fundamental scalars from other light bosonic fields.

Eigenvalue transformations appear ubiquitously in scientific computation, ranging from matrix polynomials to differential equations, and are beyond the reach of the quantum singular value transformation framework. In this work, we study the efficiency of quantum algorithms based on contour integral representation for eigenvalue transformations from both theoretical and practical aspects. Theoretically, we establish a complete complexity analysis of the contour integral approach proposed in [Takahira, Ohashi, Sogabe, and Usuda. Quant. Inf. Comput., 22, 11\&12, 965--979 (2021)]. Moreover, we combine the contour integral approach and the sampling-based linear combination of unitaries to propose a quantum algorithm for estimating observables of eigenvalue transformations using only $3$ additional qubits. Practically, we design contour integral based quantum algorithms for Hamiltonian simulation, matrix polynomials, and solving linear ordinary differential equations, and show that the contour integral algorithm can outperform all the existing quantum algorithms in the case of solving asymptotically stable differential equations.

Indoor occupancy classification enables privacy-preserving monitoring in settings such as remote elder care, where presence information helps triage alarms without cameras or wearables. Radar suits this role by sensing motion through occlusions and in darkness. Modern deep-learning pipelines are the standard for interpreting radar returns effectively; however, they are often parameter-heavy and sensitive at low signal-to-noise ratios (SNR), motivating compact alternatives like Hybrid Quantum Neural Networks (HQNNs). A two-qubit HQNN is benchmarked against convolutional neural networks (CNNs) using a physics-informed 60GHz digital twin and real radar measurements under matched training protocols. In clean conditions, the HQNN achieves high accuracy (99.7% synthetic; 97.0% real) with up to 170x fewer parameters (0.066M). Its parameter efficiency is shown to be structural, as an ablation of the parameterized quantum circuit (PQC) causes sharp performance drops on real data (to 68.5% and 31.5% for the control heads). A domain-dependent sensitivity emerges under additive-noise evaluation, where the HQNN begins recovery earlier in synthetic data while CNNs recover more steeply and peak higher on real measurements. In label-fraction ablations, CNNs prove more sample-efficient on real Range-Doppler Maps (RDMs), with the performance gap being most pronounced (at 50% labels, BA 0.89-0.99 vs. HQNN 0.75). On synthetic data, this gap narrows significantly, largely vanishing by the 50% label mark. Overall, the HQNN's value lies in parameter efficiency and a compact inductive bias that shapes its distinct sensitivity profile; this work establishes a rigorous baseline for hybrid quantum models in privacy-preserving radar occupancy sensing.

Eliminating cryogenic operating requirements while preserving microsecond-scale quantum coherence and enabling CMOS scalability remains a central challenge for telecom quantum technologies. Addressing this, we introduce a CMOS-compatible quantum system comprising single-erbium-(Er)-ion qudits (five-level systems) operating across the visible and telecom C-band. Through innovative nanofabrication, we achieve self-aligned ion placement, enabling spatial isolation of single-Er ions and suppressing dephasing. We realize individually addressable single-Er-devices with record-long optical coherence times in the telecom C-band exceeding 500 μs at ambient conditions, a performance previously limited to vacuum conditions at temperatures over 900 times lower. Furthermore, we present the first demonstration of background-free, upconversion-enabled single-photon Er-emissions providing coherent, high-contrast optical readouts. This work showcases the first room-temperature single-Er-qudit system with unprecedented properties enabling next-generation cryogen-free telecom quantum technologies.

Gravitational-wave (GW) observatories have used template-based search to detect hundreds of compact binary coalescences (CBCs). However, template-based search cannot detect astrophysical sources that lack accurate waveform models, including core-collapse supernovae, neutron star glitches, and cosmic strings. Here, we present a novel approach for template-free search using coincident anomaly detection (CoAD). CoAD requires neither labeled training examples nor background-only training sets, instead exploiting the coincidence of events across spatially separated detectors as the training loss itself: two neural networks independently analyze data from each detector and are trained to maximize coincident predictions. Additionally, we show that integrated gradient analysis can localize GW signals from the neural-network weights, providing a path toward data-driven template construction of unmodeled sources, and further improving precision by frequency matching. Using the CodaBench dataset of real LIGO backgrounds with injected simulated CBCs and sine-Gaussian low-frequency bursts, CoAD achieves recall up to 0.91 and 0.85 respectively at a false-alarm rate of one event per year, and achieves recall above 0.5 at signal-to-noise ratios below 10. The fully-unsupervised nature of CoAD makes it especially well-suited for next-generation detectors with greater sensitivity and associated increases in GW event rates.

Quantum many-body scars (QMBS) constitute a subtle violation of ergodicity through a set of non-thermal eigenstates, referred to as scar states, which are embedded in an otherwise thermal spectrum. In a broad class of known examples, these scar states admit a simple interpretation: they are magnon excitations of fixed momentum on top of a ferromagnetic background. In this paper we prove that any Hamiltonian hosting such ``ferromagnetic scar states'' necessarily admits a structural decomposition into a Zeeman term and an ``annihilator'' that annihilates the entire scar manifold. Moreover, we show that this annihilator must itself decompose into a sum of terms built from local projectors that locally annihilate the scar states. This architecture is closely related to the Shiraishi--Mori construction, and our main theorem establishes that an appropriate generalization of that construction is in fact essentially exhaustive for this class of scar states.

We revisit the recent work of Huang on the superradiant stability of Kerr black holes coupled to massive scalar fields. While their analysis provides sufficient conditions for stability, it imposes an unnecessarily strong requirement by demanding that two roots of the relevant quartic equation be explicitly negative. By instead analyzing the polynomial's coefficients, we show that simpler constraints already exclude additional positive turning points, thereby slightly enlarging the region of guaranteed stability. We further present a near-extremal estimate that tightens the stability bound for rapidly spinning black holes. These refinements sharpen the analytic stability limits without introducing extra assumptions.

Autonomous materials science, where active learning is used to navigate large compositional phase space, has emerged as a powerful vehicle to rapidly explore new materials. A crucial aspect of autonomous materials science is exploring new materials using as little data as possible. Gaussian process-based active learning allows effective charting of multi-dimensional parameter space with a limited number of training data, and thus is a common algorithmic choice for autonomous materials science. An integral part of the autonomous workflow is the application of kernel functions for quantifying similarities among measured data points. A recent theoretical breakthrough has shown that quantum kernel models can achieve similar performance with less training data than classical models. This signals the possible advantage of applying quantum kernel machine learning to autonomous materials discovery. In this work, we compare quantum and classical kernels for their utility in sequential phase space navigation for autonomous materials science. Specifically, we compute a quantum kernel and several classical kernels for x-ray diffraction patterns taken from an Fe-Ga-Pd ternary composition spread library. We conduct our study on both IonQ's Aria trapped ion quantum computer hardware and the corresponding classical noisy simulator. We experimentally verify that a quantum kernel model can outperform some classical kernel models. The results highlight the potential of quantum kernel machine learning methods for accelerating materials discovery and suggest complex x-ray diffraction data is a candidate for robust quantum kernel model advantage.

We experimentally demonstrate the distribution of polarization-entangled photons across a 62-km, partially-aerial fiber. With polarization stabilization applied to the fiber link, we achieve a photon pair rate of approximately 1500 per second and observe a CHSH inequality violation with S=2.34.

For complete spin initial data sets with an asymptotically anti--de Sitter end, we introduce a charged energy--momentum defined as a linear functional arising from the Einstein--Maxwell constraints. Under a dominant energy condition adapted to the presence of a negative cosmological constant, we establish positive energy--momentum theorems, showing in particular that this functional is non--negative on a natural real cone. We place particular emphasis on the case where the manifold carries a compact inner boundary. In the time--symmetric setting, this yields a mass--charge inequality for asymptotically hyperbolic manifolds with charge.

We study the early-time component of the Green's function of a Schwarzschild black hole, traveling on the curved light cone and usually denoted as the prompt response. Working in a post-Minkowskian approximation, we show for the first time that the prompt response is given by the residue of poles at $ω=0$ present in the complex Fourier domain. The contribution of the high-frequency arcs, previously assumed to generate the prompt response, vanishes. The analytical expression of the prompt response in this scheme is a polynomial of order $\ell$ in the observer's retarded time, with $\ell$ the multipole number. We validate the model against numerical predictions, obtaining good agreement for a compact source far from the black hole. We provide a phenomenologically-corrected expression to improve the match as the source is moved closer. We investigate the polynomial structure of the prompt response for sources close to the black hole through a series of numerical fits. Our work is a fundamental step in the broader effort to develop first-principles, analytical models for binary black hole coalescence signals, valid close to the merger and during the early ringdown stage.

We present a unified framework for gravitational baryogenesis combining two mechanisms: (i) an entropy-clock source that generates a sign-definite baryon chemical potential mu_B proportional to d ln S/dt during irreversible entropy production, and (ii) CP violation from a gravitational theta term theta R Rtilde analogous to the QCD vacuum angle. The entropy-clock evades adiabatic cancellation that suppresses purely oscillatory chemical potentials under smooth freeze-out; we quantify this with a universal low-pass transfer function F(x)=1/sqrt(1+x^2), where x=omega*tau_off. A minimal UV completion via a dilaton coupled to the conformal anomaly yields the entropy-clock coupling with decay constant f_sigma ~ 10^17-10^18 GeV. The theta term provides CP violation through interference between topologically distinct gravitational configurations; we estimate an instanton suppression factor kappa_inst ~ 10^-2-10^-1. The combined mechanism predicts Y_B ~ c*kappa*Pi_eff*epsilon_eff, linking the baryon asymmetry to the reheating temperature, the theta angle, and the entropy-production history. For Loop Quantum Cosmology bounces we derive an analytic expression for the circular polarization of the stochastic gravitational-wave background, with |Pi| ~ pi*|theta| for horizon-crossing modes, potentially testable by future space-based detectors.

This study explores the construction, validity and the properties of Boson or Bose-Einstein condensate (BEC) stars under the framework of de Rham-Gabadadze-Tolley (dRGT) like massive gravity, employing the Kuchowicz metric potential to model their internal structure. This gravitational framework accounts for a massive graviton while ensuring the absence of ghost instabilities during propagation. The BEC stellar configuration in this study was obtained by determining the solutions characterized by static and spherical symmetric metric. This study provides a detailed account of the stellar structure, highlighting the roles played by massive gravity and the Kuchowicz metric through a combination of analytical and numerical solutions. Our work specifically utilizes the Colpi-Wasserman-shapiro (CWS) and Gross-Pitaevskii (GP) equations of state (EoS) to model the internal thermodynamic behavior of the BEC. We have evaluated the physical viability of the BEC stellar framework by analyzing the energy conditions, and the EoS parameter along with the gradients of the energy-momentum tensor. The stability criteria such as the study of surface redshift, adiabatic index and squared sound velocity were utilized to confirm that our proposed model is both stable and physically consistent. Hence, this study offers a definitive structural analysis of the BEC stars, providing precise results in this massive gravity environment.

We investigate non-perturbative multi-pion contributions to nucleon flavor asymmetry within the framework of Light-Front Hamiltonian Effective Field Theory (LFHEFT). Utilizing a Fock sector expansion, we systematically incorporate pionic degrees of freedom, with the nucleon-pion interactions governed by a scalar variant of chiral effective field theory. Our results demonstrate that the non-perturbatively calculated longitudinal momentum distributions exhibit significant deviations from leading-order perturbative predictions, emphasizing the importance of higher-order Fock components in describing the proton's sea quark structure. Furthermore, we demonstrate the feasibility of extending this framework to investigate nuclear effects in light nuclei, such as the deuteron. This unified approach provides a consistent basis for analyzing the interplay between intrinsic nucleon structure and nuclear modifications, potentially offering new insights into the flavor asymmetry observed in fixed-target and collider experiments.

We discuss dominance of imaginary part of the elastic scattering amplitude and argue in favor of approximation based on this dominance.

Signatures of new physics at the LHC are varied and, by nature, often very different from those of Standard Model processes. Novel experimental techniques, including dedicated data streams, are exploited to enhance the sensitivity of the CMS Experiment to search for such signatures. This report highlights the CMS results obtained using data collected at the LHC during Run-II and Run-III through the so-called "Data Scouting" and "Data Parking" strategies. These approaches have allowed us to set some of the strongest constraints to date for low-mass resonances in prompt and long-lived signatures.

The trace anomaly of dense matter, $Δ\equiv 1/3 - P/\varepsilon$, defined in terms of the ratio of pressure $P$ to energy density $\varepsilon$, quantifies deviations from conformal symmetry and plays a central role in both the hydrodynamic response and gravitational equilibrium. While $Δ(\varepsilon)$ has recently been inferred from neutron star observations, we report the first Bayesian extraction of the trace anomaly from collective flow observables in intermediate-energy heavy-ion collisions. By employing transport-model simulations that explicitly decouple the cold-matter mean-field potential from thermal effects, we directly constrain the cold dense-matter equation of state (EOS). Remarkably, the trace anomaly inferred from laboratory flow data agrees quantitatively, within $68\%$ credible intervals, with independent astrophysical posterior bands. This nontrivial agreement demonstrates that heavy-ion collisions and neutron star observations probe the same universal macroscopic properties of dense matter, establishing the trace anomaly as a composition-insensitive descriptor of dense matter across widely different physical environments.

This study presents a dedicated and systematic sensitivity analysis of a singlet-type vector-like $B$ quark at an FCC--$μp$ collider with a center-of-mass energy of $\sqrt{s}=24.5~\mathrm{TeV}$ via photon-induced production mechanisms. The analyses are based on the $B \to Zb$ decay and the lepton decay of the $Z$ boson on a clean and selective final state. In this regard, we observed that the applied kinematic cuts suppress the Standard Model backgrounds by several orders of magnitude while maintain high signal efficiency. Also, the obtained results show that for $\mathcal{L}=1000~\mathrm{fb^{-1}}$, even small coupling constant values in the mass range $M_B = 2$--$3$~TeV are accessible. In this context, an exclusion limit of $g^\ast \simeq 0.12$--$0.17$ at 95\% confidence level and a sensitivity in the range of $g^\ast \simeq 0.22$--$0.30$ for the $5σ$ discovery region are obtained. Moreover, the calculations indicate that scenarios with smaller ${\rm BR}(B\to Zb)$ values via increasing integrated luminosity values can be experimentally tested. As a result, the outputs of the study show that the FCC--$μp$ environment offers a unique and powerful discovery potential for vector-like quark searches that go beyond the current LHC limits and complement the decreasing sensitivity of the ep and $e^+e^-$ colliders in the high-mass regime. In this context, the findings suggest that the $μp$ colliders can be considered not only as a complementary option for the search of heavy new particles in the future accelerator programme, but also as a strategic discovery tool.

A combined statistical analysis of searches for heavy vector boson resonances decaying into pairs of W, Z, or Higgs bosons, as well as into quark pairs ($\mathrm{q\bar{q}}$, $\mathrm{b\bar{b}}$, $\mathrm{t\bar{t}}$, $\mathrm{t\bar{b}}$) or lepton pairs ($\ell^+\ell^-$, $\ell\barν$), with $\ell =$ e, $μ$, $τ$, is presented. The results are based on proton-proton collision data at a center-of-mass energy of 13 TeV, corresponding to an integrated luminosity of 138 fb$^{-1}$, collected by the CMS experiment from 2016 to 2018. No significant deviation from the expectations of the standard model is observed. The results are interpreted in the simplified heavy vector triplet (HVT) framework, setting 95% confidence level upper limits on the production cross sections and coupling strengths to standard model particles or the HVT bosons. The results exclude HVT resonances with masses below 5.5 TeV in a weakly coupled scenario, below 4.8 TeV in a strongly coupled scenario, and up to 2.0 TeV in the case of production via vector boson fusion. The combination provides the most stringent constraints to date on new phenomena predicted by the HVT model.

The {$\bfνEYE$} (``new eye'', Neutrino Experiment at YEmilab, \href{https://sites.google.com/korea.ac.kr/the-nueye-telescope} {\tt nuEYE.korea.ac.kr}) neutrino project leverages the existing large pit at Yemilab located in South Korea, to reveal the existence of sterile neutrino, the up-turn of the neutrinos from the Sun, and the first minimum of the neutrino oscillation over distances on the order of tens of kilometers for the first time. This initiative is expected to facilitate a wide range of significant scientific and technological advancements within both South Korean and international communities engaged in neutrino science and technology. The {$\bfνEYE$} aims to investigate the largely unexplored sector of almost-massless lepton in the elementary particle physics in detail. The emphasis will be placed on the study of real time nuclear processes and reactions involving possible sterile neutrinos on timescales down to nanoseconds in ultra-high intense or radioactive neutrino beams for the first time in the world; the {$\bfνEYE$} looks at to-be universal oscillation (``up-turn'' in the electron neutrino survival probability) of neutrinos predicted by the three neutrino oscillation paradigm. This will confirm or deny our current understanding on the particle interactions of the lepton sector; and measurement of the first oscillation minimum between the first and second neutrinos in mass.

Micro-Pattern Gas Detectors (MPGDs) are extensively employed in modern high-energy and nuclear Physics experiments because of their excellent spatial resolution, high rate capability, and operational stability. Among these, the Gas Electron Multiplier (GEM) has emerged as one of the most widely adopted MPGD technologies. Despite their widespread adoption, GEM detectors based on the conventional bi-conical hole geometry do not always achieve optimal performance, particularly in maximizing effective gain while suppressing ion backflow. One of the primary factors limiting a GEM's performance is ion backflow. The accumulation and gradual discharge of these ions might alter the local electric field, resulting in a temporary dead time and complicating responses to subsequent events. These limitations pose challenges for applications requiring high precision and stable long-term operation. In this work, we address these issues by investigating modified GEM geometries designed to enhance gain performance and reduce ion backflow, thereby improving overall detector performance. The current study investigates geometric optimization strategies for a triple-GEM detector to enhance performance, mitigate ion backflow, and augment gain. The detector structures were designed using the ANSYS Mechanical APDL, and the associated electrostatic field configurations were computed using the ANSYS Maxwell. A thorough investigation of gain and ion backflow calculations was carried out when the generated field maps were interfaced with Garfield$^{++}$. The potential enhancements in detector efficiency and stability that the proposed modifications to the GEM foil geometry offers a valuable insights for the design of next-generation gaseous detectors.

We identify a minimal superspace projector that uniquely selects the Rarita Schwinger mass bilinear in four-dimensional N=1 supergravity. Working in a reduced superspace with a single fermionic direction, we show that Berezin projection of a canonical even supergeometric form isolates the unique Lorentz-invariant fermionic bilinear compatible with local supersymmetry. The construction is predynamical: it fixes the algebraic structure of the gravitino mass term independently of supersymmetry breaking mechanisms, background geometry, or matter couplings. We show that this projector embeds consistently into curved superspace, extends to N>1 theories, and clarifies the universality of the gravitino mass structure in supergravity.

The ALICE Collaboration is proposing a completely new apparatus, ALICE~3, for the LHC Runs~5 and beyond. In this context, a key subsystem for high-energy charged particle identification will be a proximity-focusing ring-imaging Cherenkov detector using aerogel as radiator and silicon photomultipliers (SiPMs) as photon sensors. We assembled a small-scale prototype instrumented with Hamamatsu S13352 and S13361-3075AE-08 SiPM arrays, readout by custom boards equipped with front-end Petiroc 2A ASICs. The Cherenkov radiator consisted of a 2 cm thick hydrophobic aerogel tile with a refractive index of 1.03 separated from the SiPM plane by a 23 cm expansion gap. The prototype was successfully tested in a campaign at the CERN PS T10 beam line with the goal of validating the design bRICH specifications in terms to achieve the target separation power. We measured a single photon angular resolution of 3.8~mrad at the Cherenkov angle saturation value of 242~mrad, as well as the expected scaling of the angular resolution with the increasing number of detected photons. We also studied the contribution of uncorrelated and correlated background sources with respect to the signal and proved the effectiveness of time matching between charged tracks and photon hits to achieve efficient suppression of the SiPM dark count rate background. In this paper, the detector concept, the description of the tested prototype layout and the main beam test results are reported.

During the High Luminosity phase of LHC, up to 200 proton-proton collisions per bunch crossing will bring severe challenges for event reconstruction. To mitigate pileup effects, an extended upgrade program of the CMS experiment is expected. A new timing layer, the MIP Timing Detector (MTD), will be integrated between the tracker and the calorimeters. With a time resolution of 30-60 ps, the MTD will enable 4D vertexing and it will bring significant improvements in track-to-vertex association and object identification. The MTD is composed of two subsystems based on different technologies: the Barrel Timing Layer (BTL) consists of LYSO:Ce scintillating crystals readout by SiPMs and the Endcap Timing Layer (ETL) is made of Low Gain Avalanche Detectors. The BTL is currently under production, while ETL sensor prototyping and validation are ongoing. Recent system tests have confirmed the performance of the full acquisition chain. This talk will provide an overview of the MTD design, along with the physics motivation and the current status of BTL construction and ETL development.

The ALICE Collaboration is proposing a completely new apparatus, ALICE 3, for the LHC Run 5 and beyond. A key subsystem for charged particle identification will be a Ring-Imaging Cherenkov (RICH) detector consisting of an aerogel radiator and a photosensitive surface based on Silicon Photomultiplier (SiPM) arrays in a proximity-focusing configuration. A thin high-refractive index slab of transparent material (window), acting as a second Cherenkov radiator, is glued on the entrance face of the SiPM arrays to achieve precise charged particle timing. Requiring time matching between aerogel Cherenkov photon and track hits leads to an improvement of pattern recognition by discarding the uncorrelated SiPM dark count hits. In this work we present the current status of the R\&D performed for the ALICE 3 RICH detector prototype and the expected full scale system performance. A special focus will be given to the beam test results obtained with a small-scale prototype instrumented with various array of Hamamatsu SiPMs with pitches ranging from 1 to 3 mm. The Cherenkov radiator consisted of a 2 cm thick aerogel tile with a refractive index of 1.03 at 400 nm wavelength. For timing measurements SiPM arrays coupled with two different window materials (SiO$_2$ and MgF$_2$) were used. The prototype was successfully tested in beam test campaigns at the CERN PS T10 beam line. The data were collected with a complete chain of front-end and readout electronics based on the Petiroc 2A and Radioroc 2 together with a picoTDC to measure charges and times. We measured a charged particle detection efficiency above 99\% and a single photon angular resolution better than 4.2 mrad at the Cherenkov angle saturation with a time resolution better than 70 ps for charged particles.

Parton distribution function (PDF) at small $x$ in a fast-moving proton is investigated within an upgraded parton model that includes parton splitting and fusion. In the region of moderately small $x$, we obtain a power-law behavior of the parton density $x f(x)$ with an exponent proportional to the logarithm of the probability of parton splitting. Taking into account parton fusion leads to a nonlinear equation for the PDF. In the region of very small $x$, the phenomenon of saturation of the parton density is detected and a model estimate of its value in this regime is obtained. The results are compared with those obtained previously based on the analysis of equations in logarithmic approximations of perturbative QCD.

A dedicated R\&D is ongoing for the charged particle identification system of the \mbox{ALICE 3} experiment proposed for the LHC Run 5 and beyond. One of the subsystems for the high-energy charged particle identification will be a Ring-Imaging Cherenkov (RICH) detector. The possibility of integrating Cherenkov-based charged particle timing measurements is currently under study. The proposed system is based on a proximity-focusing RICH configuration including an aerogel radiator separated from a SiPM array layer by an expansion gap. A thin high-refractive index window of transparent material, acting as a second Cherenkov radiator, is glued on the SiPM array to enable time-of-flight measurements of charged particles by exploiting the yield of Cherenkov photons in the thin window. We assembled a small-scale prototype instrumented with different Hamamatsu SiPM array sensors with pitches ranging from 1 to 3 mm, readout by custom boards equipped with the front-end Petiroc 2A ASICs to measure charges and times. The primary Cherenkov radiator consisted of a 2 cm thick aerogel tile, while various window materials, including SiO$_2$ and MgF$_2$, were used as secondary Cherenkov radiators. The prototype was successfully tested in a campaign at the CERN PS T10 beam line with pions and protons. This paper summarizes the results achieved in the 2023 test beam campaign.

Precision measurements of four neutrino oscillation parameters, $θ_{12}$, $θ_{13}$, $Δm^2_{21}$, and |$Δm^2_{31}$|, face significant interference from a previously overlooked correlated background. Recent findings from the SNO+ and JUNO experiments reveal that cascade decays of $^{214}$Bi-$^{214}$Po in liquid scintillator detectors can mimic inverse beta decay signals from reactor and geoneutrinos, with a misidentification probability on the order of $10^{-4}$ when hydrogen neutron capture is used, a rate ten times higher than Geant4 simulations predicted. This work identifies the $^5$Li resonance in $α$-$p$ elastic scattering as the underlying cause. For alpha energies above 5~MeV, the cross section is hundreds of times larger than that of Rutherford scattering. After correctly incorporating the differential cross section into Geant4, the misidentification probability is recalculated as 1.9$\times$10$^{-4}$. The simulated shape of the long tail in the alpha deposited energy also differs from the extrapolation models currently used by SNO+ and JUNO. These results will assist both experiments in more accurately estimating this novel background, thereby refining measurements of neutrino oscillation parameters and the geoneutrino flux. Additionally, the study implies an overlooked background with a rate of 0.5 events per detector per day in the Daya Bay $θ_{13}$ analysis using hydrogen neutron capture, leading to an increase of $\sin^22θ_{13}$ by approximately 0.012. Consequently, the Particle Data Group's reported $\sin^2θ_{13}$ value shall increase by about 0.006~(1$σ$).

We study high-energy (HE) neutrino production from interactions between supernova (SN) ejecta and the surrounding circumstellar material (CSM), focusing on regular Type~II and Type~IIn SNe. Using observationally inferred CSM density distributions, we calculate the resulting neutrino fluxes and examine their dependence on key parameters, including the CSM density normalization $D_*$, outer radius $R_{\rm csm}$, proton acceleration efficiency $ε_p$, and magnetic energy fraction $ε_B$. Detection prospects are assessed with a binned likelihood analysis for IceCube, indicating that nearby SNe with moderately dense, confined CSM can produce detectable signals, with a typical detection horizon of $\sim 0.1$ - 1 Mpc. For a Galactic SN at $\sim 10$ kpc, high-statistics neutrino data with detailed temporal and spectral information can constrain $D_*$, $R_{\rm csm}$, and $ε_p$ to within a factor of $\sim 10$ or to a precision of $\sim 20\%$, depending on the assumed values of $D_*$ and $R_{\rm csm}$. These neutrino signals thus provide a complementary probe of the CSM profile and shock acceleration, alongside traditional electromagnetic observations.

Dissipative processes are pivotal for understanding the hydrodynamic evolution of hot and dense QCD matter created in relativistic nuclear collisions. The interplay of multiple conserved charges -- net baryon, strangeness, and electric charge -- is of particular interest. We simulate the longitudinal hydrodynamic evolution with the three diffusion currents in a hydrodynamic model with a lattice-QCD-based equation of state, NEOS-4D, and estimate rapidity distributions including diffusive corrections to the phase-space distribution in the presence of multiple charges, which ensure charge conservation at particlization. We determine the response of particle yields at midrapidity to changes in the diagonal and off-diagonal conductivities. Inversely, we find that most components of the conductivity matrix can be constrained experimentally using identified particle multiplicities at different collision energies.

The next-generation liquid xenon experiments like PandaX-xT target an energy range from sub-keV to multi-MeV to address the requirement of multiple physics searches. The Hamamatsu R12699-406-M4 photomultiplier tubes (PMTs) were developed and selected as photon sensors for PandaX-xT. Their voltage-divider base is optimized for a broad dynamic range, from single-photoelectron (SPE) sensitivity to 30~nC collected charge (matching the 2.5~MeV Q-value of $^{136}$Xe neutrinoless double beta decay~(NLDBD)). Using a dedicated test bench, we characterize the saturation and suppression responses of R12699-406-M4 PMTs with this base design. Based on measured PMT-base responses, we develop a circuit simulation model that accurately reproduces the physical mechanisms underlying these effects with key parameters tuned via experimental data. The combined simulation and bench-test approach guides base design and optimization, enabling improved detector dynamic range and supporting future saturation and suppression correction studies in data analysis.

In this work, we investigate the interactions between an excited anti-charm meson doublet $(\bar{D}_1, \bar{D}_2^*)$ and ground-state octet baryons $(N, Λ, Σ, Ξ)$ with the aim of identifying possible molecular pentaquark states. A systematic analysis is performed within the one-boson-exchange model, which incorporates both $S$-wave and $P$-wave interactions, $S$-$D$ wave mixing, and coupled-channel effects. By solving the Schrödinger equations, we can predict a rich spectrum of loosely bound anti-charm molecular pentaquarks with strangeness $|S| = 0, 1, 2$. Our results provide specific quantum number assignments and mass range predictions to guide future experimental searches at facilities such as LHCb and Belle II. The discovery of such states would significantly enrich the hadron spectrum and serve as a critical test of theoretical models for hadronic interactions.

We propose a simple yet testable framework for light fermion dark matter (DM) with mass in the MeV--GeV range, charged under a dark $U(1)_D$ gauge symmetry. The $U(1)_D$ is spontaneously broken by a scalar field $Φ$, giving mass to the dark gauge boson $X_D$. The dominant DM annihilation proceeds via a forbidden channel, where the DM pair annihilates into slightly heavier dark gauge bosons and scalars after the dark-sector phase transition. Once the dark-sector phase transition occurs, the induced mass gap activates the forbidden annihilation channel, which in turn determines the DM relic abundance and naturally suppresses late-time annihilation. As a result, the scenario avoids stringent cosmic microwave background and indirect detection constraints that typically exclude thermal light DM. Moreover, the same symmetry-breaking phase transition is strongly first-order, producing a stochastic gravitational wave background that could be probed by upcoming space-based interferometers and pulsar timing arrays. We demonstrate that achieving the observed DM abundance tightly correlates the DM mass with the nucleation temperature of the phase transition. Thus, this setup links the DM relic abundance, dark-sector dynamics, and gravitational wave signals, offering complementary paths for discovery in both terrestrial and cosmological observations.

Using the data sample of $2.7\times10^9$ $ψ(3686)$ events collected with the BESIII detector at the BEPCII collider, we present an observation of the $Ξ(1530)^0$ polarization in the decay $ψ(3686)\toΞ(1530)^0\barΞ(1530)^0$ with a significance larger than $20σ$ compared with all other tested hypotheses. The helicity amplitudes for the process $ψ(3686)\toΞ(1530)^0\barΞ(1530)^0$ and the moduli of form factors including electric charge, magnetic dipole, electric quadrupole, and magnetic octupole are measured for the first time by performing an angular distribution analysis. Additionally, the polarization correlations between $Ξ(1530)^0$ and $\barΞ(1530)^0$ are measured.

We study open charm hadrons production in pp and Au+Au collisions at $\sqrt{s_{\mathrm{NN}}} = 200$~GeV using an improved a multi-phase transport (AMPT) model. Specifically, we show the transverse-momentum spectra and nuclear modification factors $R_{\mathrm{AA}}$ of $D^{0}$ mesons and $Λ_{c}^{+}$ baryons, as well as the $Λ_{c}^{+}/D^{0}$ ratio in pp and Au+Au collisions. The results obtained from the AMPT model simulations are compared with the STAR experimental data and found to be consistent. We further investigate the $Λ_{c}^{+}/D^{0}$ ratio by evaluating contributions from coalescence, fragmentation, and the combined coalescence+fragmentation mechanisms, and we find that fragmentation alone underestimates the pronounced enhancement in Au+Au relative to pp at low and intermediate $p_{\mathrm{T}}$, whereas the coalescence+fragmentation mechanism reproduces the observed trend significantly better. These results indicate that coalescence plays a key role in charm baryon productions and helps constrain the relative importance of different hadronization mechanisms in the ultra-relativistic nuclear collisions.

Full one-loop electroweak radiative corrections to $μ^- μ^+ \to W^\pm W^\mp \to hh$ in the Standard Model are computed for the first time in this work. We then evaluate neutral scalar pair production through vector boson fusion at multi--TeV muon colliders within Two-Higgs-Doublet Model (THDM). In the phenomenological analysis, the enhancement factor, defined as the ratio of the cross sections for SM-like Higgs pair production in the THDM with respect to the corresponding ones in the SM, is examined over the viable regions of the parameter space in the Type-X and Type-Y THDMs. Our findings show that this ratio can reach a factor of $3$ in several regions within the valid parameter space of the Type-X THDM, whereas it ranges from $0.91$ to $0.96$ over the entire parameter space of the Type-Y THDM. Finally, we scan the cross sections for double CP-odd Higgs production over the updated parameter space of the Type-X and Type-Y THDMs. In the Type-Y case at $\sqrt{s} = 10$~TeV with an integrated luminosity of $\mathcal{L} = 10000~\text{fb}^{-1}$, CP-odd Higgs pair production in the $t\bar{t}b\bar{b}$ final state, with subsequent top-quark decays into leptons and bottom quarks taken into account, can be tested with a statistical significance exceeding the $2σ$ level at several viable parameter points.

Bilinear Euclidean quark and gluon correlators with Wilson links have been used widely for applications of large-momentum effective field theories to computing non-perturbative collinear and soft parton physics. Due to color confinement, these correlators decay exponentially at large spatial distances, a behavior crucial for computing momentum-space Fourier transformations with controlled errors from lattice QCD data. Using heavy-quark effective theory reduction, dispersive analysis, Lorentz symmetry, and heavy-flavor spectra, we determine the leading and next-to-leading asymptotic behaviors and relate the expansion parameters to binding energies of heavy-flavor hadrons. We demonstrate the results through two-loop calculations in $φ^3$ theory and from the perspective of locality and analyticity. We also study the impact of the asymptotic analysis on realistic lattice QCD data and demonstrate reliable error estimates.

In these proceedings, we present the measurements of charged $K^{+} - K^{+}$ and neutral $K_{s}^{0} - K_{s}^{0}$ correlation functions from Au+Au fixed-target collisions at $\sqrt{s_{NN}}$ = 3.0, 3.2, 3.5, 3.9 and 4.5 GeV at STAR. This is the first such systematic measurement of correlation functions involving strangeness in the high baryon density region. The source size values do not exhibit a clear energy dependence, and the transverse mass dependence of source size for kaons does not align with the trend observed for pions. Parameters extracted from UrQMD transport model calculations qualitatively capture the measured values.

The theory of nuclear excitations involving nucleon resonances is revisited and significantly extended to asymmetric nuclear matter and higher P- and S-wave $N^*$ resonances. Excited states of are described as superpositions of particle-hole configurations including $NN^{'-1}$ and $N^*N^{-1}$ configurations. Configuration mixing is taken into account on the one-loop level by solving the generalized $N^*RPA$ Dyson equation. The underlying coupled channels formalism is derived and response functions is discussed. Applications of the approach are illustrated for charge-exchange modes of asymmetric nuclear matter and finite nuclei. The spectral gross structures of corresponding excitations in finite nuclei are investigated in local density approximation. Applications of the approach to resonance studies by high-energy heavy ion reactions are recapitulated.

In this paper, we prove that simple affine vertex operator algebras with positive integral levels admit only trivial first-order deformations. Therefore, the deformation rigidity conjecture of strongly rational vertex operator algebras holds for these cases. We also show that the same holds simple affine vertex operator algebra of $\mathfrak{sl}_2$ at the non-integral admissible level $-4/3$. Therefore, neither $C_2$-cofiniteness nor rationality is a necessary condition for deformation rigidity of VOAs. We conjecture that the same should hold for every simple affine VOA that does not coincide with the corresponding universal affine VOA.

We investigate the discovery potential of the $T_{bc}$ states with $J^P = 0^+$ in proton-proton ($pp$) collisions at LHCb at a center-of-mass energy of $\sqrt{s} = 13~\mathrm{TeV}$. The study focuses on the decay channel $T_{bc} \to B^- D^+$. A phenomenological approach is employed to construct the background model based on the associated production of $B$ and $D$ mesons, incorporating previously published LHCb results. Background processes are simulated using $\texttt{MadGraph5\_aMC@NLO}$ and $\texttt{Pythia8.3}$. We explore the parameter space of the $T_{bc}$ mass, width, production cross section, and the effective double-parton-scattering cross section ($σ_{\mathrm{eff}}$) relevant for the $B D$ meson background. The integrated luminosity required for a $5σ$ discovery at LHCb is evaluated under various assumptions. We find that a $5σ$ observation is achievable for a production cross section of $103~\mathrm{nb}$, which is expected to be within reach during Run~4. In addition, we estimate the minimum observable $σ(T_{bc}) \times \mathrm{BR}(T_{bc} \to B^- D^+)$ for a $5σ$ discovery under different luminosity scenarios, providing guidance for future experimental searches at LHCb.

We present a comprehensive theoretical assessment of the masses of possible baryonic configurations characterized by the presence of two heavy top quarks, including $Ξ_{ttu}$, $Ξ_{ttd}$, $Ω_{tts}$, $Ω_{ttc}$, and $Ω_{ttb}$ systems. This analysis is rigorously executed within the specialized framework of two-point $\mathrm{QCD}$ sum rules, focusing on their predicted ground state masses. Our interest in these systems arises from recent CMS and ATLAS reports indicating a pseudoscalar excess close to $t\bar{t}$ threshold. Our evaluation incorporates both perturbative terms and nonperturbative effects, including condensate contributions up to dimension eight. Based on our results, the extracted central masses for all channels are slightly above the sum of the constituent quark masses, which is consistent with the inherent uncertainties of the method. These quantitative predictions provide a useful first-principle theoretical reference, which may help future experimental searches for such heavy configurations at the LHC and inform sensitivity studies at next-generation facilities such as the FCC.

Neutrons produced by cosmic ray muons constitute a significant background for underground experiments investigating neutron oscillations, neutrinoless double beta decay, dark matter, and other rare event signals. This work benchmarks measured neutron yields and neutron multiplicities--with a focus on data from the Daya Bay Reactor Neutrino Experiment--against comprehensive simulations using three GEANT4 hadronic physics lists. These simulations are further refined via a TALYS-based adjustment of hadronic cross sections. For the BERT-based models, the adjustment reduces the discrepancy in the total neutron yield from about 20% to approximately 6%, while for the BIC-based models it improves the agreement from roughly 13% to the sub-percent level (~0.3%), indicating a markedly better consistency of the BIC-based models with the experimental data. Nevertheless, a clear tension persists: simulations systematically underproduce single-neutron events while overproducing multi-neutron events. The study establishes a reproducible benchmark for cosmogenic neutron modeling and proposes a targeted refinement strategy--including channel-specific reweighting and intranuclear cascade parameter tuning--to guide future model development.

The discrepancy on the muon anomalous magnetic moment values obtained via a direct measurement and via a data-driven theory determination that uses the experimentally measured hadronic cross section, is among the long standing and most significant deviations from the Standard Model predictions. The recently presented final result of the direct measurement performed at the experiment at Fermilab, with an impressive accuracy of 127 parts-per-billion, further stresses the need for a theory estimate of comparable accuracy. The $e^+e^-$ hadronic cross section is the experimental input to the dispersive integral for the calculation of the hadronic contribution to the $g-2$, which is largely dominated by the $e^+e^-\to π^+π^-$ channel. Precise measurements of the $e^+e^-\to π^+π^-$ cross section with a sub-percent accuracy, in the energy region of the $ρ$-resonance peak, have been performed by several experiments, but the results differ way more than the published accuracies limiting the comparison with the Fermilab direct measurement. New results are expected to become available in the near future from KLOE, CMD-3, SND and BESIII experiments, while a new measurement from the BABAR collaboration is presented today and its impact on the present situation is discussed. This measurement makes use of the entire BABAR data set and adopts a different analysis strategy, with results (still preliminary) largely independent of the results published in the 2009 BABAR publication.

In this work, we focus on the possible linear relation between short-range correlations (SRCs) and the EMC effect for partons in nuclei. First, we test a linear relationship pertaining to gluons in bound nuclei; it is manifested as a correlation between the slope of the reduced cross section ratio in deep inelastic scattering (DIS) and the cross section of sub-threshold $J/ψ$ photoproduction. For comparison, the results from four different global analyses groups of nuclear parton distribution functions (nPDFs) are utilized. These results show a good linear correlation between the gluons in bound nuclei and the slope of the reduced cross section ratio, consistent with the possible presence of nuclear effects in the gluon distributions. Second, we investigate the linear relationship of quarks in the proton-induced Drell-Yan process. The corresponding results for quarks show strong sensitivity to the parameterization forms adopted by the different groups. These findings enhance our understanding of the substructure in bound nuclei and provide valuable reference for future global fitting of nPDFs.

A measurement of $t\bar{t}$ production is presented in the invariant-mass region near the pair production threshold, $m_{t\bar{t}} \sim 345$ GeV, in final states with two charged leptons and multiple jets. The measurement is based on $140\,\mathrm{fb}^{-1}$ of proton-proton collision data collected at $\sqrt{s} = 13$ TeV with the ATLAS detector at the Large Hadron Collider. The data are compared to two models of $t\bar{t}$ production: a baseline model including only perturbative QCD predictions for the hard process, and an extended model that, in addition, incorporates non-relativistic QCD simulations of colour-singlet quasi-bound-state formation near the $t\bar{t}$ threshold. The agreement between the data and the models is quantified via a profile-likelihood fit to the reconstructed $m_{t\bar{t}}$ distributions, in bins of two angular observables sensitive to spin-correlations in the $t\bar{t}$ system. An excess of events is observed over the baseline perturbative QCD prediction, with an observed significance over $8$ standard deviations. This excess is consistent with the formation of colour-singlet and spin-singlet $S$-wave quasi-bound $t\bar{t}$ states, as predicted by non-relativistic QCD, and corresponds to an observed cross-section of $9.3^{+1.4}_{-1.3}$ pb.

We explore the connection between the four-dimensional Minkowski-space Bethe-Salpeter equation and its light-front projection, emphasizing the implications for bound-state dynamics. Our approach incorporates dressed particles, such as quarks, via the integral representation of the corresponding propagator. We analyze the light-front dynamics of the valence component of the physical state using a hierarchical set of Green's functions, which reveals its coupling to higher Fock components when dressed particles are considered. We also present the light-front Faddeev-Bethe-Salpeter equations for three-body systems with dressed constituents. Furthermore, we discuss formal developments that are central to connecting the three-dimensional light-front dynamics onto the null-plane and the four-dimensional Minkowski-space framework, based on the Nakanishi integral representation. Selected applications to hadron structure are also reviewed.

The problem of solving non-linear equations would be considerably simplified by a possibility to convert known solutions into the new ones. This could seem an element of art, but in the context of ADHM-like equations describing quiver varieties there is a systematic approach. In this note we study moduli spaces and dualities of quiver gauge theories associated to effective dynamics of D-branes compactified on Calabi-Yau resolutions. We concentrate on a subfamily of quivers $\mathfrak{Q}_{\mathfrak{g}}$ covering Dynkin diagrams for simple Lie algebras $\mathfrak{g}$, where the respective BPS algebra is expected to be the Yangian algebra $Y(\mathfrak{g})$. For Yangians labeled by quivers their representations are described by solutions of ADHM-like equations. As quivers substitute Dynkin diagrams a generalization of the Weyl group $\mathcal{W}_{\mathfrak{g}}$ acts on the ADHM solutions. Here we work with the case $\mathfrak{g}=\mathfrak{sl}_{n+1}$ and treat this group as a group of electro-magnetic Seiberg-like dualities (we call them Weyl mutations) on the respective quiver gauge theories. We lift it to the case of higher representations associated to rectangular Young diagrams. An action of Weyl mutations on the BPS Yangian algebra is also discussed.

Self-supervised learning is a powerful pre-training method for learning feature representations without labels, which often capture generic underlying semantics from the data and can later be fine-tuned for downstream tasks. In this work, we introduce jBOT, a pre-training method based on self-distillation for jet data from the CERN Large Hadron Collider, which combines local particle-level distillation with global jet-level distillation to learn jet representations that support downstream tasks such as anomaly detection and classification. We observe that pre-training on unlabeled jets leads to emergent semantic class clustering in the representation space. The clustering in the frozen embedding, when pre-trained on background jets only, enables anomaly detection via simple distance-based metrics, and the learned embedding can be fine-tuned for classification with improved performance compared to supervised models trained from scratch.

Accurate and efficient detector simulation is essential for modern collider experiments. To reduce the high computational cost, various fast machine learning surrogate models have been proposed. Traditional surrogate models for calorimeter shower modeling train separate networks for each particle species, limiting scalability and reuse. We introduce AllShowers, a unified generative model that simulates calorimeter showers across multiple particle types using a single generative model. AllShowers is a continuous normalizing flow model with a Transformer architecture, enabling it to generate complex spatial and energy correlations in variable-length point cloud representations of showers. Trained on a diverse dataset of simulated showers in the highly granular ILD detector, the model demonstrates the ability to generate realistic showers for electrons, photons, and charged and neutral hadrons across a wide range of incident energies and angles without retraining. In addition to unifying shower generation for multiple particle types, AllShowers surpasses the fidelity of previous single-particle-type models for hadronic showers. Key innovations include the use of a layer embedding, allowing the model to learn all relevant calorimeter layer properties; a custom attention masking scheme to reduce computational demands and introduce a helpful inductive bias; and a shower- and layer-wise optimal transport mapping to improve training convergence and sample quality. AllShowers marks a significant step towards a universal model for calorimeter shower simulations in collider experiments.

We explore the space of non-gravitational, maximally supersymmetric, planar 4d effective field theories (EFTs) that have $\mathcal{N}=4$ super Yang-Mills (SYM) at leading order. We show that in the weakly-coupled regime, highly non-trivial nonlinear constraints on the 4-point Wilson coefficients follow from enforcing $\mathcal{N}=4$ supersymmetry and $SU(4)$ R-symmetry together with the requirement of standard tree-level factorization on the massless poles of the 4-, 5-, and 6-point EFT scattering amplitudes. Additionally, when these novel constraints are combined with positivity, the resulting bounds on the 4-point Wilson coefficients converge to the values of the open string Veneziano amplitude. Our results strongly suggest that supersymmetry, R-symmetry, and positivity are sufficient to single out this unique UV completion at tree level. Our findings, moreover, highlight the power of higher-point amplitudes in constraining EFT data and imply that the space of consistent quantum field theories is even more restricted than previously suggested by causality or swampland-based approaches.

We refine the constraints on the lifetime of decaying super-heavy dark matter particles (SHDM), with masses ranging from $10^7$ to $10^{15}$ GeV, by analyzing ultra-high-energy (UHE) gamma-ray data. Our approach involves an accurate comparison of the primary gamma-ray emissions resulting from prompt SHDM decays in the galactic halo with the most recent upper limits on isotropic UHE gamma-ray fluxes provided by various extensive air shower experiments. We demonstrate that a precise consideration of the field of view and the geometric acceptance of different UHE gamma-ray observatories has significant implications for the inferred limits of dark matter lifetime. In addition, we examine the influence of uncertainties linked to the current models of the galactic dark matter distribution, employing diverse halo density profiles while varying both their radial extent and the local dark matter density. Our findings indicate that the newly established UHE gamma-ray constraints are marginally less stringent than earlier evaluations, thereby revisiting the SHDM parameter space and allowing for observable neutrino fluxes.

Correlations between gamma-ray and X-ray spectral properties of pulsars are investigated in order to provide observational hints on physics involved in pulsars' high-energy emissions. Using a sample of 43 pulsars detected in both X-ray and gamma-ray bands, we find that pulsars' gamma-ray luminosity, $L_γ$, clearly correlates with the luminosity of non-thermal X-ray emission, $L_{\rm p}$, and anti-correlates with non-thermal X-ray photon index. Other gamma-ray spectral parameters show weaker or negligible correlations. The found relation that $L_γ\propto L_{\rm p}^{0.49\pm 0.05}$ implies a certain connection between radiation mechanisms and energy distributions of radiating particles for these high-energy emissions. Pulsars with and without detected thermal emissions seem to show different dependencies in those correlations, suggesting the possible existence of two different kinds of pulsars. The ones without detected thermal emissions may represent a population of pulsars with low surface temperature. The origin and energetics of high-energy emitting electron-positron pairs for this group of pulsars probably do not depend on their surface thermal emissions, while that of the other group do. The low surface temperatures might be evidence for the working of some exotic processes of neutron-star cooling. Similar to $L_{\rm p}$, some tempting relationships are found among each gamma-ray spectral parameter, surface temperature and thermally radiating area radius. It again strengthens the connection between gamma-ray and X-ray emissions from pulsars.

One of the primary goals of next-generation gravitational lensing surveys is to measure the large-scale distribution of dark matter, which requires accurate mass inversion to convert weak-lensing shear maps into convergence (kappa) fields. This work develops a mass inversion method tailored for upcoming space missions such as CSST and Euclid, aiming to recover both the mass distribution and the convergence power spectrum with high fidelity. We introduce MIU2Net, a versatile deep-learning framework for kappa-map reconstruction based on the U2-Net architecture. A new loss function is constructed to jointly estimate the convergence field and its frequency-domain energy distribution, effectively balancing optimal mean squared error and optimal power-spectrum recovery. The method incorporates realistic observational effects into shear fields, including shape noise, reduced shear, and complex masks. Under noise levels anticipated for future space-based lensing surveys, MIU2Net recovers the convergence power spectrum with 4% uncertainties up to l approximately 500, significantly outperforming Wiener filtering and MCALens. Beyond two-point statistics, the method accurately reconstructs the convergence distribution, peak centroid, and peak amplitude. Compared to other learning-based approaches such as DeepMass, MIU2Net reduces the root-mean-square error by 5% without smoothing and by 38% with a 1-arcmin smoothing scale. MIU2Net represents a substantial advancement in mass inversion methodology, offering improved accuracy in both RMSE and power-spectrum reconstruction. It provides a promising tool for mapping dark matter environments and large-scale structures in the era of next-generation space lensing surveys.

Deuterium fractionation becomes highly efficient in cold, dense cores where CO is frozen out. Cyclopropenylidene (c-C3H2), an early-formed carbon ring, and its deuterated isotopologues trace gas-phase deuteration in these environments. We present a statistical study of c-C3H2 deuteration in starless and prestellar cores of the Perseus Molecular Cloud using observations of c-C3H2, c-C3HD and c-C3D2 obtained with the Yebes 40 m, ARO 12 m and IRAM 30 m telescopes towards 16 cores. Gaussian fits and RADEX modeling yield column densities for the detected species. c-C3H2 is detected in 14/15 covered cores, c-C3HD in 15/16, and c-C3D2 in 9/16. Derived column densities range from 0.5-8.1 x 10^{13} cm^{-2} for c-C3H2, 0.2-2.1 x 10^{12} cm^{-2} for c-C3HD, and 0.6-1.6 x 10^{11} cm^{-2} for c-C3D2. The ortho-to-para ratio of c-C3H2 is obtained for all but one core, with a median value of 3.5\pm0.4. Statistically corrected D/H ratios span 0.5-9.2% (median 1.5\pm0.2%), and D2/D ratios 9-55% (median 25.9\pm4.3%). No trend is found between the c-C3H2 ortho-to-para ratio and core evolutionary stage traced by n(H2). The median D/H ratio in Perseus appears lower than values reported for Taurus and Chamaeleon, while the D2/D ratio agrees with Taurus within uncertainties. A positive correlation between D/H and n(H2) supports the use of D/H as an evolutionary tracer. D2/D does not correlate with n(H2), but shows a positive correlation with T_{kin}, suggesting that its formation is influenced by a mildly endothermic pathway.

We investigate the redshift 1.98 galaxy cluster XLSSC 122 using the Hubble Space Telescope (HST) from the core of the cluster out to 3 Mpc, a scale equivalent to 10 times the R500 = 295 kpc radius. We present an expanded photometric and spectroscopic catalogue of the cluster, bringing the total number of spectroscopically classified member galaxies to 74, with 35 new member galaxies added in the outer regions of the cluster. We compute the radial galaxy number density profile in the cluster, and observe no clear evidence of infalling groups or cosmic filaments. We observe a clear bimodal colour relation in member galaxies, with red fraction increasing towards the cluster centre. This rapid increase of red fraction upon infall is indicative of a fast quenching mechanism, such as ram pressure stripping, as galaxies enter the cluster centre. We fit a luminosity function to the cluster members, finding a similar low mass slope but fainter scale magnitude than z = 1 clusters of similar temperature, implying a similar galaxy evolution rate to clusters at lower redshift.

Magnetic fields and turbulence may play a key role in the evolution of protoclusters, influencing the formation of dense cores and stars. Here, we examine the morphology of the magnetic fields in the G327.29 protocluster using both the velocity gradient technique (VGT) extracted from molecular line emissions and linear polarization in the dust continuum emission. The VGT analysis is performed using four molecular tracers: DCN (3-2), C18O (2-1), HN13C (3-2), and H13CO+ (3-2) - which probe gas across different density regimes, observed with the ALMA 12 m array. Owing to its sensitivity to gas dynamics, a comparison between VGT and dust polarization provides a powerful probe of the evolutionary processes in massive star-forming regions. From our analysis we reveal a complex magnetic-field structure, shaped by the combined influence of turbulence and gravity. In addition, it also appears that there is a large-scale (beyond the core scale) gravitational infall from the surrounding medium on to the filament and the central densest region. Furthermore, we observe that cores are dominated by a mix of turbulence and gravity. Overall, this work presents, likely for the first time, the application of VGT to a massive protocluster, G327.29, using high-resolution ALMA observations.

We present results from a coordinated, multi-epoch near-infrared and centimeter radio survey of young stellar objects (YSOs) in the Coronet, aimed at probing the connection between mass accretion and ionised mass loss. Using VLT-KMOS, we detect Br$γ$ emission in 5 of the 26 targets, which also exhibit 3.3-cm continuum emission in VLA images, consistent with partially ionised jets. For seven additional sources, stringent flux upper limits were obtained. The derived accretion and ionised mass-loss rates for class I and class II YSOs follow a sublinear correlation $\dot{M}_{\mathrm{ion}} \propto \dot{M}_{\mathrm{acc}}^{0.3}$, consistent with previous results for class II YSOs but extended here to earlier stages. Multi-epoch observations reveal modest variability in both tracers but no clear temporal correlation between accretion and ejection within timescales of a few months. The ratio $\dot{M}_{\mathrm{ion}}/\dot{M}_{\mathrm{acc}}$ shows an anti-correlation with $\dot{M}_{\mathrm{acc}}$, increasing with time from class I YSOs to class II YSOs, suggesting an increase in jet-launching efficiency or ionisation fraction with evolution. These findings support a direct connection between accretion and outflow across the $\sim$ Myr timescale of YSO evolution, while highlighting the complexity of their short-term interplay.

We present the analysis of four microlensing events, KMT-2020-BLG-0202, KMT-2022-BLG-1551, KMT-2023-BLG-0466, and KMT-2025-BLG-0121, which exhibit extended and complex anomalies in their light curves. These events were identified through a systematic reanalysis of KMTNet data aimed at detecting planetary signals that deviate from the typical short-term anomaly morphology. Detailed modeling indicates that all four anomalies were produced by planetary companions to low-mass stellar hosts. The events have mass ratios of $q \sim (5$--$14)\times10^{-3}$ and Einstein timescales of $t_{\rm E} \sim 20$--$43$ days. Bayesian analyses based on Galactic models show that the companions are super-Jupiters with masses of a few to approximately 10 $M_{\rm J}$, orbiting sub-solar-mass hosts located at distances of $D_{\rm L} \sim 4$--$7$~kpc. All planets lie well beyond the snow line of their hosts, placing them in the regime of cold giant planets. These detections demonstrate that extended and complex microlensing anomalies, which are often challenging to recognize as planetary in origin, can nonetheless contain planetary signals. This work underscores the unique sensitivity of microlensing to cold, massive planets beyond the snow line and highlights the importance of systematic reanalyses of survey data for achieving a more complete and unbiased census of exoplanets in the Galaxy.

We study the local evolution of the Universe around galaxy formation sites in the EAGLE50 large-volume reference simulation. Using the reduced inertia tensor (r-TOI), we followed the anisotropic evolution of initially spherical Lagrangian volumes (LVs) centred at galaxy formation sites, both in dark matter (DM) and in cold baryons (CB), from very high redshift $z=15$ onward. We describe LV deformation in terms of the r-TOI eigen-directions, principal axes, their derived shape parameters, and the timescales for the freezing-out of these principal directions and axes. Of particular interest are the age of the Universe, $t_{\rm U}$, when the local Cosmic Web (CW) spine emerges, and that when anisotropic DM mass arrangements (i.e., migrant mass flows) cease. We find that the shapes LVs acquire along their evolution affect the halo and stellar mass of their central galaxy: prolate-shaped LVs show a tendency to host low-mass galaxies at $z=0$, while massive galaxies tend to form within triaxial or oblate LVs. Also, the local CW spine tends to set in earlier on in LVs that are to host massive galaxies than in those harbouring less massive galaxies. In addition, anisotropic DM-mass rearrangements stop late on average, at $t_{\rm U}\sim 10.5\,$Gyr, and even slightly later for CB. Interestingly, $z=0$ LVs with either flattened configurations in CB or those that are highly prolate in DM, are more likely to host rotation-dominated galaxies. This effect increases from $z=1$ to $z=0$. Finally, the CB spine of LVs that are more likely to host rotation-dominated galaxies emerges at later times.

We report the detection of the polycyclic aromatic hydrocarbon (PAH) 1H-cyclopent[cd]indene (c-C11H8) in TMC-1 with the QUI- JOTE line survey. We detected 22 independent lines corresponding to 88 rotational transitions with quantum numbers ranging from J=19 up to J=24 and Ka <= 5 in the Q-band range. The identification of this new PAH was based on the agreement between the rotational parameters derived from the analysis of the lines and those obtained by quantum chemical calculations. The column density derived for 1H-cyclopent[cd]indene is (6.0 +- 0.5) x 10^12 cm-2, with a rotational temperature of 9 K. Its abundance is high, as is that of the rest of the PAHs, but it is the lowest of all those detected to date in TMC-1, being 2.66 times less abundant than indene and 4.66 times less than phenalene. This result will help us to better understand the growth of five- and six-membered rings in dark clouds. Chemical models explaining their formation through the bottom-up model are still very incomplete and require further experimental and theoretical effort. Even so, the most likely formation reactions would occur between the smallest rings with small hydrocarbons; the most probable reaction for the formation of cyclopentindene is that between indene and C2H, C2H3, and/or their cation.

We present a study of the gas kinematics within the Hestia project, a state-of-the-art set of simulations of the Local Group, with a particular focus on the velocity patterns of different ions and the large-scale motion of gas and galaxies towards the Local Group barycentre. Using two high-resolution Hestia runs, we examine the distribution and velocities of H I, C IV, Si III, O VI, O VII, and O VIII and their imprints on sightlines observed from the Sun's location in different reference frames. To mimic observational strategies, we assess the contribution of rotating disc gas, assuming simple kinematic and geometrical considerations. Our results indicate that local absorption features in observed sightlines most likely trace material in the circumgalactic medium of the Milky Way. Some sightlines, however, show that intragroup material could be more easily observed towards the barycentre, which defines a preferred direction in the sky. In particular, H I, Si III, and C IV roughly trace cold gas inside the Milky Way and Andromeda haloes, as most of their mass flux occurs within the virial region of each galaxy, while oxygen high ions mostly trace hot halo and intragroup gas, with comparable mass fluxes in the Local Group outskirts and the circumgalactic medium of the two main galaxies. Additionally, we find that pressures traced by different ionic species outside the Milky Way halo show systematically higher values towards the barycentre direction in contrast to its antipode in the sky. Kinematic imprints of the global motion towards the barycentre can be seen at larger distances for all ionic species as the Milky Way rams into material in the direction of Andromeda, with gas towards the anti-barycentre lagging behind.

Ultraluminous X-ray sources (ULXs) are point-like sources that exhibit apparent X-ray luminosities exceeding the Eddington limit for stellar-mass compact objects. A widely accepted interpretation is that these systems are X-ray binaries accreting matter possibly at super-Eddington rates. In this regime, photon trapping inflates the accretion disk, making it geometrically and optically thick. Radiation-driven winds launched from the supercritical disk form funnel-shaped walls along the symmetry axis. While the apparent X-ray luminosity can exceed the Eddington limit due to geometrical beaming within this funnel, a misalignment with the observer's line of sight strongly suppresses the X-ray emission, rendering the ULX electromagnetically obscured. This work explores the potential for high-energy neutrino production in black hole-hosting ULXs. We model proton acceleration via magnetic reconnection in the region above the super-accreting black hole. Although electromagnetic emission is efficiently absorbed by the dense wind and radiation fields, neutrinos generated from photomeson interactions can escape. Our model self-consistently accounts for energy losses of pions and muons in this environment. The results indicate that misaligned, electromagnetically obscured Galactic ULXs could produce a neutrino flux detectable by instruments like KM3NeT and IceCube within several years of observation.

We investigate the environmental parameters and gas fraction (f$_{gas}$) properties of type~2 AGN and non-AGN galaxies, utilizing a large sample of galaxies from SDSS DR7 with z $\le$ 0.3. We find that the environment affects type~2 AGN and non-AGN galaxies in similar ways and does not impact the strength of AGN-driven outflows. The f$_{gas}$ of type~2 AGN and non-AGN host galaxies show no variation between group and isolated environments, suggesting that host galaxy gas content is largely independent of large-scale environment. We find that type~2 AGN host galaxies possess systematically lower f$_{gas}$ than their non-AGN counterparts when matched in stellar mass and star formation rate (SFR). This suggests that AGN activity plays a significant role in regulating the molecular gas reservoir and, consequently, the star formation processes within galaxies. We find that Type~2 AGNs exhibiting strong outflows are associated with higher gas fractions, higher star-formation rates, and younger stellar populations than those with weak or no outflows. This may indicate either concurrent star formation in gas-rich systems hosting powerful outflows, or a time delay between AGN activity and its effect on star formation consistent with a delayed AGN feedback scenario.

We present the longest and the densest quasi-simultaneous radio, X-ray and optical campaign of the black hole low mass X-ray binary GX339-4, covering five years of weekly GX339-4 monitoring with MeerKAT, Swift-XRT and MeerLICHT, respectively. Complementary high frequency radio data with the Australia Telescope Compact Array are presented to track in more detail the evolution of GX339-4 and its transient ejecta. During the five years, GX339-4 has been through two "hard-only" outbursts and two "full" outbursts, allowing us to densely sample the rise, quenching and re-activation of the compact jets. Strong radio flares were also observed close to the transition between the hard and the soft states. Following the radio flare, a transient optically thin ejection was spatially resolved during the 2020 outburst, and was observed for a month. We also discuss the radio/X-ray correlation of GX339-4 during this five year period, which covers several states in detail from the rising phase to the quiescent state. This campaign allowed us to follow ejection events and provide information on the jet proper motion and its intrinsic velocity. With this work we publicly release the weekly MeerKAT L-band radio maps from data taken between September 2018 and October 2023.

In Very-Long Baseline Interferometric arrays, nearly co-located stations probe the largest scales and typically cannot resolve the observed source. In the absence of large-scale structure, closure phases constructed with these stations are zero and, since they are independent of station-based errors, they can be used to probe data issues. Here, we show with an expansion about co-located stations, how these trivial closure phases become non-zero with brightness distribution on smaller scales than their short baseline would suggest. When applied to sources that are made up of a bright compact and large-scale diffuse component, the trivial closure phases directly measure the centroid relative to the compact source and higher-order image moments. We present a technique to measure these image moments with minimal model assumptions and validate it on synthetic Event Horizon Telescope (EHT) data. We then apply this technique to 2017 and 2018 EHT observations of M87* and find a weak preference for extended emission in the direction of the large-scale jet. We also apply it to 2021 EHT data and measure the source centroid about 1 mas northwest of the compact ring, consistent with the jet observed at lower frequencies.

Dark galaxies - systems rich in neutral hydrogen (HI) gas but with no stars - are a common prediction of numerous theoretical models and cosmological simulations. However, the unequivocal identification of such sources in current HI surveys has proven challenging. In this work, we present interferometric follow-up observations with the VLA of a former dark galaxy candidate J0139+4328, originally detected with the single-dish FAST telescope. The improved spatial resolution of the VLA data allow us to identify a faint optical counterpart and characterize the galaxy. Located at a distance of about 31 Mpc, J0139+4328 has a stellar mass of 3 x 10^6 M_Sun and a relatively high gas richness of M_HI/M_star = 18. Despite its high ratio, the galaxy is consistent, within the scatter, with the stellar-to-HI mass relation of HI-selected samples in the literature and with the baryonic Tully-Fisher relation (BTFR), although its kinematic measurement is subject to large uncertainties. This case highlights the potential of modern high-sensitivity HI surveys for detecting low surface brightness, gas-rich galaxies, but underscores the need for careful interpretation of low-resolution HI data, with potentially large centroid errors, and for sufficiently deep optical imaging to ensure robust identification.

Reactions between atomic nuclei are measured in great detail in terrestrial laboratory experiments; transferring and extrapolating this knowledge to how the same reactions act within cosmic environments presents major challenges. Cross-disciplinary efforts are needed in view of the many nuclear reactions that govern the chemical evolution of the universe, and occur in a broad range of stellar plasma conditions that require astrophysical exploration. Since the early identification of 'processes' of nucleosynthesis, new insights have been obtained on the complexity of nuclear reaction mechanisms. We use 12C induced capture and fusion processes to illustrate the challenge of low-energy measurements and of using theoretical methods to extrapolate measurements towards energy regimes within cosmic sources. Particle beam experiments at accelerator facilities above and deep underground simulate stellar reactions, new experimental facilities and methods complement these, and this is further complemented by improved theoretical tools to calculate the quantum effects of nuclear reactions at the various cosmic conditions. Astronomical signatures of cosmic nuclear reactions are deduced from light curves characterizing cosmic explosions through gamma-ray lines and presolar grains to the detection of rare neutrino particles from our Sun to distant cosmic events. High resolution spectroscopy of stars has been expanded to objects measured in the X-ray and the gamma energy range of the electromagnetic spectrum. Astro-seismology and isotopic analysis of meteoritic inclusions provide new tools. Chemical-evolution models describe the complex dynamics during the evolution of galaxies. This article summarizes the experimental and theoretical work, and the broad range of observational tools that test the experimental data and the theoretical interpretation of nuclear processes in the cosmos.

The Flat Spectrum Radio Quasar OP 313 is a high-redshift (z = 0.997) blazar that entered an intense gamma-ray active phase from November 2023 to March 2024, as observed by the Large Area Telescope (LAT) on board the Fermi Gamma-ray Space Telescope. We present a multiwavelength analysis covering 15 years of data, from August 2008 to March 2024, to contextualize this period of extreme gamma-ray activity within the long-term emission of the source. We analyzed a long-term, comprehensive, multiwavelength dataset from different facilities and projects from radio to gamma-rays. We identified the 7 most intense gamma-ray flaring periods and performed a kinematic analysis of Very Long Baseline Array (VLBA) data to determine whether new jet components emerged before or during these flares. For 2 of these flaring periods, we performed the modeling of the spectral energy distribution (SED). The VLBA-BU-BLAZAR and MOJAVE datasets reveal a new jet component appearing in both visibility datasets prior to the onset of one of the strongest gamma-ray flares. By comparing the timing of the VLBA-BU-BLAZAR knots ejection with the gamma-ray flaring periods, we constrained the setup of the SED modeling. We also found that the first gamma-ray flaring period is less Compton-dominated than the others. Our results suggest that the recent activity of OP 313 is triggered by new jet components emerging from the core and interacting with a standing shock. The γ-ray emission likely arises from dusty torus photons upscattered via Inverse Compton (IC) by relativistic jet electrons. The SED modeling indicates that this component is less dominant during the first γ-ray flaring period than the later ones.

We present a method to test the impact of ad-hoc modifications of some of the generic parameters of hadronic interactions -- cross section, elasticity, and multiplicity -- on any observable quantity using full 3-dimensional simulations of extensive air showers induced by ultra-high-energy cosmic rays. Our approach not only extends the existing 1-dimensional tools to three dimensions, but also introduces more flexible features to better respond to the needs of both theory and experiment. We first thoroughly validate the \conexD framework for the simulation of both longitudinal and lateral features of air showers, in particular for a non-standard configuration of the framework in which different energy thresholds for modifications are applied. Moreover, we show that the implementations of the ad-hoc modifications in this configuration are consistent with the previous one-dimensional simulations. Lastly, we discuss the importance of studying the interaction modifications in three dimensions and the effects of parallel modifications of multiple parameters.

Recent work has demonstrated that widely used strong-line oxygen abundance indicators, such as O3N2, $\rm R23$, and $\widehat{\rm R}$, suffer from large uncertainties when applied to high-redshift galaxies. We show that this loss of precision primarily arises because, at fixed \Oabund, galaxies span a wide dynamic range in ionization parameter and nitrogen enrichment. Here we develop updated indicators that explicitly incorporate both effects via the proxies O32 and N2O2. We define ${\rm R}_{\rm u}\equiv \rm R23+α_1 O32+α_2 N2O2$, $\widehat{\rm R}_{\rm u}\equiv \rm \widehat{R}+β_1 O32+β_2 N2O2$, and ${\rm O}_{\rm u}\equiv \rm O3N2+γ_1 O32+γ_2 N2O2$, and calibrate \Oabund~as low-order polynomials in each composite indicator. Applied to a JWST sample with $T_{\rm e}$-method abundances, the updated indicators substantially tighten the correlations with \Oabund, boosting adjusted coefficients of determination from $\mathbb{R}^2\lesssim 0$ (classical indicators) to $\mathbb{R}^2\gtrsim 0.5$ for the full sample and to $\sim 0.7$ at $z>2$. The residuals reveal a redshift evolution in the mapping between \Oabund, strong lines, ionization, and nitrogen enrichment, with a pivotal turning point near the cosmic noon ($z\sim 2$). Our calibrations provide a practical, physically grounded path to precise metallicity measurements in the JWST era and a firmer basis for quantifying early chemical enrichment and feedback.

We investigate the circumstellar environment of the O-type bloated star candidate IRAS 19520+2759 (I19520) using high-resolution observations from the Atacama Large Millimeter/submillimeter Array (ALMA) and the Expanded Very Large Array (EVLA). Radio continuum emission traced by the EVLA (C, K, and Q bands) exhibits a spectral index of 0.5, consistent with a thermal jet. ALMA 1.3 mm continuum map reveals a compact source coincident with the optical counterpart of I19520, likely tracing the dense core hosting the central massive young stellar object. A prominent molecular outflow in the east-west direction, along with a possible secondary outflow oriented northeast-southwest, is identified in the $^{13}\mathrm{CO}$ emission. A hot molecular core and a Keplerian disk are detected in several $\mathrm{SO}_2$ transitions. Assuming an edge-on disk geometry, the dynamical mass of the central object is estimated to be in the range of $10$-$15~M_\odot$.

We present a comprehensive analysis of the dark matter (DM) content and its structural dependence in star-forming disk-like galaxies at intermediate redshifts ($0.1 \lesssim z \lesssim 0.85$), utilizing spatially resolved kinematic data from the MAGPI survey. We report the following: (1) Low stellar mass galaxies ($M_{\rm star} < 10^{9.5}\, M_\odot$) are strongly DM dominated across all radii, with average $\langle f_{_{\rm DM}} \rangle \sim 0.85$, while high-mass ($M_{\rm star} > 10^{10.5}\, M_\odot$) systems exhibit relatively low DM fractions in their inner regions ($\langle f_{_{\rm DM}} \rangle \sim 0.47$) which is equivalent to local massive disk galaxies (e.g., Milky Way and Andromeda). This suggests a mass-dependent structural dichotomy, most-likely governed by a combination of internal galactic processes and environmental influences. (2) A tight inverse correlation between $f_{_{\rm DM}}$ and baryon mass surface density ($Σ_{\rm bar}$), with intrinsic scatter of $\sim 0.11$ dex. This is consistent with an inside-out baryon assembly scenario and suggests that the fundamental structural correlations of galaxies were already established by $z\sim 0.85$. (3) No significant evolution in $f_{_{\rm DM}}$ with redshift across the MAGPI window, and when combined with higher-redshift ($0.6 \leq z \leq 1.5$) data from Sharma et al. 2025, we quantitatively show that the reported decline in $f_{_{\rm DM}}(z)$ is most-likely due to observational biases against low-mass systems at $z > 1$. These results offer empirical evidence for a scenario in which disk-like galaxies evolve through a co-regulated build-up of baryonic and DM components, preserving internal structural regularities (such as the total mass distribution and rotation-curve shape) throughout cosmic time.

Millisecond pulsars (MSPs) have been proposed as evolutionary products of low-mass X-ray binaries (LMXBs) through a stage in which they are spider pulsars (i.e., redbacks and black widows). However, recent work has found that the systemic kicks of observed MSPs are significantly lower than the kicks of LMXBs and spiders, which appears to be in tension with this evolutionary model. We argue that this tension can be relieved, at least to some degree, by considering the fact that the observed MSPs are located at relatively short distances, whereas spider pulsars are located at greater distances and LMXBs are situated even further away. We model the distance-dependent kinematic bias for dynamically old objects, which favors observing objects that have received low kicks at short distances and correct the observed systemic kicks for this bias. We find that this kinematic bias can be big enough to close the gap between the MSP and LMXB kicks, although the spider pulsars appear to come from a slightly different systemic kick distribution, but this difference is not necessarily physical. All corrected systemic kick distributions are consistent with predictions from binary population synthesis for progenitor systems with a post-supernova orbital period of $P_{\text{orb}}\leq10\,$d and a companion mass of $M_{c}\leq1\,M_{\odot}$, where the natal kicks are calibrated to the velocities of young isolated pulsars. We conclude that the difference in observed systemic kicks is not necessarily in tension with a common origin for MSPs, spider pulsars, and LMXBs.

In this Letter, we report the detection of soft X-ray time lags-i.e. variability in the softer photons lagging behind that in the harder photons-in seven XMM-Newton observations of the tidal disruption event (TDE) candidate AT2021ehb. We find correlated variability between the soft (0.3-0.7 keV) and hard (0.9-10 keV) bands on about 10^4 s time-scales, and measure a soft lag of about 500 s. This behaviour is broadly consistent with the disk-corona reverberation scenario established in active galactic nuclei (AGNs). Together with the previously reported strong hard X-ray emission and broad Fe K line, our results suggest the presence of a compact corona and prominent relativistic disk reflection in AT2021ehb. The unusually high blackbody temperature (peaking at about 200 eV) is difficult to reconcile with thermal emission from a standard accretion disk around a about 10^7 Msun black hole, and may instead be analogous to the soft excess commonly observed in AGNs, whose physical origin remains debated. Finally, the measured lags offer a possible explanation for the rapid X-ray flux decline that occurred only three days after the peak, pointing to a scenario in which the corona cools following a sudden loss of the magnetic support required to sustain it.

Pulsar glitches are generally viewed as stochastic events driven by sudden angular momentum transfer from the neutron star's superfluid interior to its crust. Except two peculiar pulsars with quasi-periodic glitches, this stochastic view has prevailed. Here, by clustering temporally proximate small glitches of the Crab pulsar, we uncover clear evidence of an underlying quasi-periodic modulation, challenging the paradigm of purely random behavior. Furthermore, our correlation analyses reveal a strong positive relationship between glitch cluster size and waiting time since the preceding clusters. These findings demonstrate the effect of angular momentum reservoir operating over long-term scales and enable the predictions of next glitching window. Remarkably, two minor glitches detected in July and August 2025, which align with our initial prediction made in June, should be confirmed as the onset of this predicted activity. Inspired by the initial success, we forecast the occurrence of a major glitch from now until August 2026, with possible glitch size up to a relative change in rotational frequency of $697.2 \times 10^{-9}$. Physically, the observed long-term quasi-periodicity and cluster size-waiting time correlations imply that each glitch event releases only a fraction of the stored superfluid angular momentum. This partial-release mechanism provides a unified framework for both stochastic and quasi-periodic glitch behaviors across different pulsars, underscoring the universality of the superfluid angular momentum reservoir effect. As the most intensively monitored object, the Crab pulsar serves as a natural laboratory for studying angular momentum inside neutron stars.

Intermediate polars are a class of cataclysmic variables in which a white dwarf accretes material from a companion star. The intermediate polar nature confirmation usually derives from the detection of two periods in both $X$-ray and optical photometry. In this respect, the high energy signal is often characterized by modulations on the white dwarf spin and the orbital period. However, noting that the periodograms may be characterized by strong features also at the synodic period and/or other sidebands, the timing analysis of the $X$-ray signal may offer the unique possibility to firmly discover an intermediate polar candidate. Here, we concentrate on a sample of five cataclysmic variable {binary} candidates: {i.e. SAXJ1748.2-2808, 1RXS J211336.1+542226, CXOGC J174622.7-285218, CXOGC J174517.4-290650, and V381 Vel, listed in the IPHome catalogue. Our main aim is to confirm if they belong to the intermediate polar class or not. The results of our analysis show that we can safely assess the intermediate polar nature of all the considered sources, apart for the case of V381 Vel which instead behaves like a cataclysmic variable of the polar subclass. Moreover, the source SAXJ1748.2-2808, previously classified as a HMXB, appears to be, most likely, an intermediate polar variable.

Coherent oscillations at kilohertz frequencies have recently been detected in a small number of gamma-ray bursts (GRBs), suggesting quasi-periodic dynamics in their central engines. A prominent example is GRB~230307A, which exhibited a brief, highly coherent, energy-dependent periodic signal interpreted as the possible spin signature of a nascent millisecond magnetar formed after a compact binary merger. Motivated by these developments, we conducted a comprehensive search for similar signals, accounting for both temporal and spectral dependencies, in 532 short GRBs with time-tagged event data recorded by the Burst and Transient Source Experiment (BATSE) onboard the \textit{Compton Gamma-Ray Observatory}. Within this sample, we identify a single statistically significant case: GRB~960616 (BATSE trigger~5502), in which the $\sim$30~ms main emission episode is coherently modulated at 1100~Hz, with the strongest modulation above 320~keV and a fractional amplitude of $\sim$47\%. Assuming the presence of a coherent periodic modulation, we use data-driven Monte Carlo simulations to place an upper limit of $\sim$8\% on the fraction of the total radiated energy that can be modulated by the QPO. This event, exhibiting a periodicity at $\sim$0.91~ms, further supports the possibility that millisecond periodicities can arise during GRBs in merger-driven scenarios.

Supernovae are observed to occur approximately 1-2 times per century in a galaxy like the Milky Way. Based on historical records, however, the last core-collapse galactic supernova observed by humans occurred almost 1,000 years ago. Luckily, we are well positioned to catch the next one with the advent of new neutrino detectors and astronomical observatories. Neutrino observatories can provide unprecedented triggers for a galactic supernova event as they are likely to see a supernova neutrino signal anywhere from minutes to days before the shock breakout causes the supernova to brighten in optical wavelengths. Given its large etendue, the Vera C. Rubin Observatory is ideally positioned to rapidly localize the optical counterpart based on the neutrino trigger. In this paper we simulate events to study the efficiency with which supernovae are optimally localized by the Vera C. Rubin Observatory. We find that the observatory is ideal for initial localization of nearly all observable supernova triggers and has a 57-97% chance of catching any supernova based on theoretical stellar mass density predictions and observations. We provide an analysis of optimal filter selection and exposure times and discuss observational caveats.

Herbig Ae/Be stars (HAeBes) are critical tracers of intermediate- and high-mass star formation, yet their census remains incomplete compared to low-mass young stellar objects like T-Tauri stars. To expand the known population, we systematically searched for HAeBes in LAMOST DR7 low-resolution spectra. Following Sun et al., we applied Uniform Manifold Approximation and Projection (UMAP) for dimensionality reduction and Support Vector Machine (SVM) classification, identifying $\sim$240,000 spectra with potential H$α$ emission. After removing contaminants (non-stellar objects, extragalactic sources, CVs, and Algol systems) and restricting to B/A-type stars, we obtained 1,835 candidates through 2MASS/WISE visual inspection. Spectral energy distribution analysis confirmed 143 sources with infrared excess ($J$-band or longer wavelengths), including 92 known HAeBes. From the remaining 51 candidates, we classified 26 with strong infrared excess as new HAeBes. Color-index analysis of confirmed HAeBes and classical Ae/Be stars (CAeBes) revealed that the $(K-W1)_0$ vs. $(W2-W3)_0$ diagram effectively separates these populations: CAeBes predominantly occupy $(K-W1)_0 \leq 0.5$ and $(W2-W3)_0 \leq 1.1$, while other regions trace transition disks ($(K-W1)_0 < 0.5$ and $(W2-W3)_0 > 1.1$), globally depleted disks ($(K-W1)_0 > 0.5$ and $(W2-W3)_0 < 1.1$), and Class I/Flat/II HAeBes ($(K-W1)_0 > 0.5$ and $(W2-W3)_0 > 1.1$). More importantly, the HAeBes exhibit a clear evolutionary gradient on this diagram, with those in the Class III, Class II, Flat-SED, and Class I evolutionary stages being effectively distinguished by concentric ellipses that are roughly centered at (0,0) with semi-major axes of $a$=1.5, $a$=3.0, and $a$=4.0, and a semi-major to semi-minor axis ratio of 1.6:1.

We present a detailed analysis of the old, extended stellar component of the Local Group dwarf galaxy IC 10 using deep resolved-star photometry in the VIS and NISP bands of the Euclid Early Release Observations. Leveraging Euclid's unique combination of wide field of view and high spatial resolution, we trace red giant branch (RGB) stars out to $\sim$8 kpc from the galaxy centre, reaching azimuthally-averaged surface brightness levels as faint as $μ_{HE}\sim$29 mag arcsec$^{-2}$. Our analysis reveals that IC 10's stellar distribution is significantly more extended than previously thought. After correcting for foreground extinction and subtracting contamination from Milky Way stars and background galaxies, we derive a radial stellar density profile from RGB star counts. The profile shows a marked flattening beyond $\sim$5 kpc, and is best fit by a two-component (Sersic + exponential) model, yielding a total stellar mass in old (age $>$1 Gyr) stars of $M_{\star}=(6.7-8.1)\times10^8 M_{\odot}$. The origin of the outer stellar component is unclear. It may be accreted, even possibly associated with the counter-rotating HI gas in the outer regions of IC 10, or it may represent an ancient in-situ stellar halo. We tentatively detect two symmetric stellar overdensities at the edge of our imagery. These roughly align with the direction of IC 10's orbit around M31, suggesting that they may be signatures of tidal stripping. As part of our analysis, we derive a new distance to IC 10 based on the RGB tip, finding $D=762\pm 20$ kpc and the distance modulus is $(m-M)_0=24.41\pm 0.05$.

A transient (LPA J0108+13) with repeated bursts was detected on the Large Phased Array (LPA) radio telescope at a central frequency of 110.4 MHz in the direction of the radio galaxy 3C 33. The flux density of bursts ranges from tens to hundreds of Jy, and the duration of the bursts is \approx 1^m - 4^m. In daily observations, the total duration of which at the location of the transient exceeds 200 hours in the observation interval 2013-2025, 6 bursts were found. The nature of the source could not be determined. We believe that a new type of transients has been discovered.

A small body orbiting around an accreting massive object and periodically crossing its accretion disk is a common configuration in astrophysics. In this work, we investigate the secular evolution of extreme mass-ratio inspirals (EMRIs), in which a stellar-mass object (SMO), e.g., a star or a stellar-mass black hole (sBH), collides with the accretion disk of a central supermassive black hole (SMBH), within a fully relativistic framework. We find (1) the disk always tends to align the SMO no matter what the initial orbital inclination $ι$ relative to the disk is, (2) the final orbital eccentricity of the SMO captured by the disk is always low though the orbital eccentricity may temporarily grow when the orbital inclination $ι$ is large and the SMO is an sBH, and (3) via collisions with the accretion disk only, only a small fraction of sBHs that are initially close to the SMBH and close to the disk can be captured by the disk within typical disk lifetime of active galactic nuclei. Two-body scatterings between SMOs in the nuclear stellar cluster play an essential role in randomly kicking sBHs towards the disk and boosting the capture rate.

The neutron-star X-ray source 1E 161348-5055, associated with the supernova remnant RCW 103, exhibits clear intensity variations with a period of 6.67 hr. To clarify the nature of this object and its long periodicity, detailed timing studies were applied to its archival X-ray data, taken with ASCA (in 1993), XMM-Newton (in 2001, 2005, and 2016), and NuSTAR (2016 and 2017). It was assumed that the 6.67 hr period arises due to the beat between the rotation and free precession periods of the star that is slightly aspherical. By removing timing perturbations to be caused by this long periodicity, the six data sets consistently yielded evidence for pulsations at periods of P~1.01 s, to be interpreted as the objects' spin period, although the optimum energy range differed among the data sets. The measured six periods accurately line up on a linear spin-down trend of dP/dt = 1.097x 10^{-12} s/s. The object is implied to have a characteristic age of 14.7 kyr, a spin-down luminosity of 4.2x10^{34} erg/s, which is insufficient to power the X-ray luminosity, a dipole magnetic field of ~4.6x10^{13} G, and a toroidal field of ~7 x10^{15} G. Its similarity and dissimilarity to magnetars are discussed. An emission geometry, which crudely explain these results, is presented.

Fast Radio Bursts (FRBs) are millisecond-duration radio transients of mysterious origin, with growing evidence linking at least some of them to magnetars. While FRBs are primarily observed in the radio band, their potential multi-wavelength afterglows remain largely unexplored. We investigate the possible afterglow of FRB 20171020A, a rare nearby and bright FRB localized in a galaxy at only 37 Mpc. Assuming that this source produces a future bright burst, we model the expected afterglow emission in the radio, optical, and X-ray bands under both uniform and wind-like ambient media, within the framework of the magnetar model. Our results show that the optical afterglow is the most promising for detection, but it fades rapidly and requires follow-up within a few hundred seconds post-burst. The radio afterglow may be detectable under favorable conditions in a dense stellar wind, whereas the X-ray counterpart is too faint for current telescopes. These findings suggest that rapid optical follow-up offers the best opportunity to detect the afterglow of the next bright burst from FRB 20171020A, providing unique insights into the progenitor and its environment. To assess observational feasibility, we estimate the event rate of nearby FRBs with sufficient energy to power detectable afterglows, finding a rate of $\sim$0.3 per year for CHIME surveys. Although this rate is low and the optical detection timescale is short, coordinated fast-response strategies using global telescope networks could significantly improve the chance of success. As more nearby FRBs are discovered, multi-wavelength observations will be essential in unveiling the physical nature of these enigmatic events.

Fast radio bursts (FRBs) are mysterious radio transients with uncertain origins and environments. Recent studies suggest that some active FRBs may originate from compact objects in binary systems. In this work, we develop a unified theoretical framework to model the multi-wavelength afterglows of FRBs resided in binary systems and apply it to two representative repeaters, FRB 20200120E and FRB 20201124A. By solving the dynamics and radiation processes of FRB ejecta interacting with the surrounding medium, we compute afterglow light curves in the radio, optical, and X-ray bands. Our results show that radio afterglows offer the best prospects for detection, with their brightness highly sensitive to ejecta kinetic energy and ambient density. Future high-sensitivity radio telescopes, such as the Square Kilometre Array (SKA), could detect these signals. Optical afterglows, though short-lived and challenging to observe, may be significantly enhanced in dense environments, potentially making them detectable with facilities like the Large Synoptic Survey Telescope (LSST). In contrast, X-ray afterglows are predicted to be too faint for detection with current instruments. Our study highlights the potential of multi-wavelength afterglows as probes of FRB progenitors and their surrounding environments, offering crucial insights into the nature of these mysterious transients.

Observations of molecular structures on scales of $\sim 0.1-50$ pc show that the specific angular momentum ($j$) scales with radius ($R$) as $j\sim R^{3/2}$. We study the effects of turbulence, gravity, and the magnetic field in shaping this scaling, by measuring clump size and specific angular momentum in three SPH simulations of the formation of giant molecular clouds, progressively adding these three ingredients. In each simulation, we define ``full'' and ``reduced'' clump samples, the latter restricted to aspect ratios $A<3$. We find that, in the non-magnetic runs, elongated clumps deviate the most from the \jR\ relation, which is best reproduced by the reduced sample in the gravity+turbulence run. In the purely hydrodynamic case, no dense elongated structures form, suggesting that turbulence alone is insufficient to generate dense filaments, although clumps have $j$ magnitudes consistent with observations. In the gravity+turbulence+magnetic field run, most of the clumps are filamentary, yet the full sample appears to follow the observed \jR\ relation. This result, rather than being a real trend, could be the combination of the increase in $j$ by the filamentary geometry, and its reduction by turbulence inhibition by the magnetic field. Finally, we measure the gravitational, magnetic, pressure-gradient, and hydrodynamic torques (which involve turbulent viscosity) in our clump samples. We find that, in magnitude, the hydrodynamic torques tend to be larger than the rest. This result is consistent with our previous work, where we proposed that gravity drives cloud formation and contraction, while turbulence redistributes angular momentum through fluid-parcel exchanges.

EP J171159.4-333253 is a new neutron-star low-mass X-ray binary discovered in outburst by the Einstein Probe (EP) on 2025 June 23, exhibiting clocked type-I X-ray bursts, eclipses and dips. In this paper, we report on the results of the X-ray spectral and timing analyses for EP J171159.4-333253 using data collected by EP and NuSTAR during the first 21 days of the outburst. The X-ray burst recurrence time can be characterized over a subset of nine bursts spanning 1.6 days around the NuSTAR observation, and the result is $t_{\rm rec}=8196 \pm 177\,$s with indications of a possible decreasing trend. From the X-ray eclipse events, the binary orbital period and the eclipse duration are estimated to be $P_{\rm orb}=6.48301 \pm 0.00003\,$hr and $D_{\star,X} = 1245.5^{+6.9}_{-6.5}\,$s, respectively. These enable an estimate of the mass and radius of the companion star and the binary inclination, which are $M_2\approx0.6-0.8\,M_\odot$, $R_2\approx0.7-0.8\,R_\odot$ and $i\approx73-75^\circ$, respectively. We also report on joint ULTRACAM and EP observations on 2025 July 21--22, detecting the source optical counterpart and covering an eclipse in both X-ray and optical bands. The optical eclipse is wavelength-dependent and broader than in X-rays, indicating that part of the optical emission arises from an extended region in the accretion flow. Despite a moderate variation in the source flux, the properties of the persistent X-ray emission are typical of a hard spectral state. We further evaluated the ratio of the accretion energy to the thermonuclear energy to be 120--130, implying helium bursts with the accreted hydrogen being depleted in-between bursts.

The chemistry of shielded molecular gas is primarily driven by energetic, charged particles dubbed cosmic rays (CRs), in particular those with energies under 1 GeV. CRs ionize molecular hydrogen and helium, the latter of which contributes greatly to the destruction of molecules. CR ionization initiates a wide range of gas-phase chemistry, including pathways important for the so-called "carbon cycle", C$^+$/C/CO. Therefore, the CR ionization rate, $ζ$, is fundamental in theoretical and observational astrochemistry. Although observational methods show a wide range of ionization rates -- varying with the environment, especially decreasing into dense clouds -- astrochemical models often assume a constant rate. To address this limitation, we employ a post-processed gas-phase chemical model of a simulated dense molecular cloud that incorporates CR energy losses within the cloud. This approach allows us to investigate changes in abundance profiles of important chemical tracers and gas temperature. Furthermore, we analyze analytical calibrators for estimating $ζ$ in dense molecular gas that are robust when tested against a full chemical network. Additionally, we provide improved estimations of the electron fraction in dense gas for better consistency with observational data and theoretical calibrations for UV-shielded regions.

We present Green's function solutions of the Newtonian time-dependent thin disk equations in the presence of outflows, showing that simple and exact analytical expressions exist in various natural limits of the problem. These Green's functions are mathematically very similar to the classical Lynden-Bell & Pringle solutions in the absence of outflows, but differ strongly in their precise physical details and observational implications. Solutions are presented for phenomenological radius-dependent outflows which both do and do not torque the local accretion flow, and for outflows which are launched proportional to the local accretion rate. Generically, outflows lead to a more rapid decay of the bolometric luminosity of the disk, flatten the radial dependence of the disk temperature, and suppress variability in the accretion rate at small radii and low frequencies (on long timescales). Observational implications of these four results are discussed in detail.

We present results from simultaneous X-ray polarimetric and spectroscopic observations of the bright neutron star low-mass X-ray binary Cyg X-2, performed by the Imaging X-ray Polarimetry Explorer (IXPE) and the Nuclear Spectroscopic Telescope Array (NuSTAR). IXPE detected significant polarization (15 sigma) from the source in the 2-8 keV energy band with an average polarization degree (PD) of 4.5% +/- 0.3% and a polarization angle (PA) of 128 +/- 2 degrees as the source moved along the horizontal branch of its Z-track. The PD increases with energy reaching 9.9% +/- 2.8% in the 7-8 keV band, with no evidence for energy-dependent variation in the PA. The PA is roughly consistent with previous measurements obtained during the normal and flaring branches and also with the known radio jet axis. From spectropolarimetric analysis, the main contribution to the polarized radiation is due to Comptonized photons, but the polarization is higher than predicted in typical spreading layer geometries. The observed high polarization may be due to a combination of a highly polarized reflected component and a moderately polarized spreading layer on the neutron star surface or produced by electron scattering in an equatorial wind.

We present a catalog of 643 diffuse galaxies identified through a citizen science search of the Fornax cluster, of which we estimate 21.8% are nucleated (139/637; 6 inconclusive). This marks the first crowd-sourced effort to construct a cluster-scale census of diffuse galaxies. These objects were visually identified using a combination of the Fornax Deep Survey and Dark Energy Camera Legacy Survey imaging across 26 deg$^2$. Over 1,400 volunteers cataloged the candidates within this sky area at a rate of 1.15 days/deg$^2$. Our catalog is highly complete relative to existing dwarf catalogs of Fornax ($> 80\%$ of objects recovered) down to an effective radius $r_{\mathrm{eff}} = 5^{\prime \prime}$, the minimum size we suggested volunteers classify, and to an effective r-band surface brightness as faint as $\langle μ_r \rangle \simeq26$ mag arcsec$^{-2}$. We detect 97 candidates that existing automated searches of Fornax did not find and three candidates not found by any prior search, automated or visual. The stellar mass distribution of our sample is consistent with similar dwarf studies of Fornax, with the nucleated fraction peaking at 80% for a host galaxy mass of $\sim$10$^{8.5}M_{\odot}$. The efficiency and completeness of our catalog thus establishes citizen science as a valuable tool for mapping diffuse galaxy populations in future sky surveys, such as the Legacy Survey of Space and Time.

CO gas is detected in a significant number of debris discs, but its origin and evolution remains unclear. Key constraints are its mass and spectro-spatial distribution, which are coupled through optical depth and have only been analysed at low to moderate resolution so far. The ALMA survey to Resolve exoKuiper belt Substructures (ARKS) is the first ALMA large program to target debris discs at high spectro-spatial resolution. We used $^{12}$CO and $^{13}$CO J=3-2 line data of 18 ARKS debris belts, 5 of which were already known to host gas, to analyse the spectro-spatial distribution of CO, constrain the gas masses, and to search for gas in the remaining systems. We developed a line-imaging pipeline and produced line cubes for each disc, with a spatial resolution down to $\sim$70 mas and spectral resolution of 26 m s$^{-1}$. Using spectro-spatial shifting and stacking, we produced high signal-to-noise maps, and radial and spectral profiles that reveal the distribution and kinematics of gas in 5 gas-bearing discs. For these discs, we constrained the inner radius of the $^{12}$CO, and found the radial brightness profile of CO peaked interior to the dust ring, but that CO was more radially extended than the dust. We present the first radially resolved $^{12}$CO/$^{13}$CO isotopologue flux ratios in gas-bearing debris discs, which are constant with radius for the majority of systems, indicating $^{12}$CO and $^{13}$CO are both optically thick or thin throughout the discs. We report CO line fluxes/upper limits for all systems and optical depth dependant masses for the 5 gas-bearing systems. Finally, we analysed the $^{12}$CO J=3-2 line luminosities for the ARKS debris discs and discs from the literature. We confirm that gas is mostly detected in young systems. However, the high scatter seen in young/high fractional luminosity systems indicates no trend within the systems with detected gas.

Resonant scattering of He I 1.0833$μ$m triplet photons by metastable He 2 $^3$S$_1$ is studied for optical depths characteristic of H II regions. Regions with large He 2 $^3$S$_1$ column densities are predicted to have unusually broad, multi-peaked 1.0833$μ$m emission profiles, with the centroid blue-shifted by up to $\sim$14 km/s relative to other lines. The feature FWHM can exceed 100 km/s for some regions. Resonant trapping enhances dust absorption and reduces the He I 1.0833$μ$m emission. Care must be taken when using the He I 1.0833$μ$m/H I 1.0941$μ$m (Pa$γ$) ratio to estimate the He$^+$/H$^+$ ratio. Predicted spectra are computed for examples, including M-17B and NGC3603 in the Galaxy, and a star-forming region in M51. Observations of the 1.0833$μ$m triplet with spectrometers such as NIRSPEC, CARMENES, or X-Shooter can confirm the predicted effects of resonant scattering in H II regions, and constrain the nebular conditions.

We present a new free-form lens model for the $z=0.348$ galaxy cluster AS1063, based on previously spectroscopically confirmed lensed galaxies and new images from the GLIMPSE program. We use the ultra-deep data to identify new counterimages for previously confirmed (spectroscopically) lensed systems. We use the full set of spectroscopically confirmed systems to derive a new lens model, which is later used to confirm many of the previous lensed system candidates and discover new lensed system candidates in the images. We compute the geometric redshifts, time delays, and magnification for all counterimages (confirmed and not confirmed). Among the new systems we find a peculiar multiply lensed galaxy with a strong emission line at $\approx 4\, μ$m that likely corresponds to H$-β$ and/or OIII at $z\approx 7.5$. This galaxy could be a little-red-dot or an extreme emission line galaxy. We also identify a yellow supergiant lensed star candidate at $z\approx 3.1$. This star shows some similarities with previous Kaiju stars and we nickname it "Hedorah", in honor of the famous yellow-eyed Kaiju. Previous lensed stars at $z>0.1$ are either blue supergiants or red supergiants, making Hedorah the first yellow supergiant discovered beyond $z=0.1$ and confirming that, despite their rarity, they can also be found at these redshifts. Since many Cepheid stars are yellow supergiants, we consider the possibility that Hedorah could also be the first Cepheid discovered at cosmological distances, but we conclude that Hedorah is more likely a hypergiant yellow star approaching the end of its life. Alternatively, Hedorah could be a small group of stars, although this is less likely based on Hedorah's peculiar colors and additionally may require the more exotic fuzzy dark matter to help explain the lack of counterimage.