arXiv Daily Digest - 2026-03-26

We introduce the Multilevel Euler-Maruyama (ML-EM) method compute solutions of SDEs and ODEs using a range of approximators $f^1,\dots,f^k$ to the drift $f$ with increasing accuracy and computational cost, only requiring a few evaluations of the most accurate $f^k$ and many evaluations of the less costly $f^1,\dots,f^{k-1}$. If the drift lies in the so-called Harder than Monte Carlo (HTMC) regime, i.e. it requires $ε^{-γ}$ compute to be $ε$-approximated for some $γ>2$, then ML-EM $ε$-approximates the solution of the SDE with $ε^{-γ}$ compute, improving over the traditional EM rate of $ε^{-γ-1}$. In other terms it allows us to solve the SDE at the same cost as a single evaluation of the drift. In the context of diffusion models, the different levels $f^{1},\dots,f^{k}$ are obtained by training UNets of increasing sizes, and ML-EM allows us to perform sampling with the equivalent of a single evaluation of the largest UNet. Our numerical experiments confirm our theory: we obtain up to fourfold speedups for image generation on the CelebA dataset downscaled to 64x64, where we measure a $γ\approx2.5$. Given that this is a polynomial speedup, we expect even stronger speedups in practical applications which involve orders of magnitude larger networks.

We introduce DreamerAD, the first latent world model framework that enables efficient reinforcement learning for autonomous driving by compressing diffusion sampling from 100 steps to 1 - achieving 80x speedup while maintaining visual interpretability. Training RL policies on real-world driving data incurs prohibitive costs and safety risks. While existing pixel-level diffusion world models enable safe imagination-based training, they suffer from multi-step diffusion inference latency (2s/frame) that prevents high-frequency RL interaction. Our approach leverages denoised latent features from video generation models through three key mechanisms: (1) shortcut forcing that reduces sampling complexity via recursive multi-resolution step compression, (2) an autoregressive dense reward model operating directly on latent representations for fine-grained credit assignment, and (3) Gaussian vocabulary sampling for GRPO that constrains exploration to physically plausible trajectories. DreamerAD achieves 87.7 EPDMS on NavSim v2, establishing state-of-the-art performance and demonstrating that latent-space RL is effective for autonomous driving.

As LLMs are increasingly used as judges in code applications, they should be evaluated in realistic interactive settings that capture partial context and ambiguous intent. We present TRACE (Tool for Rubric Analysis in Code Evaluation), a framework that evaluates LLM judges' ability to predict human preferences and automatically extracts rubric items to reveal systematic biases in how humans and models weigh each item. Across three modalities -- chat-based programming, IDE autocompletion, and instructed code editing -- we use TRACE to measure how well LLM judges align with developer preferences. Among 13 different models, the best judges underperform human annotators by 12-23%. TRACE identifies 35 significant sources of misalignment between humans and judges across interaction modalities, the majority of which correspond to existing software engineering code quality criteria. For example, in chat-based coding, judges are biased towards longer code explanations while humans prefer shorter ones. We find significant misalignment on the majority of existing code quality dimensions, showing alignment gaps between LLM judges and human preference in realistic coding applications.

Agentic artificial intelligence (AI) in organizations is a sequential decision problem constrained by reliability and oversight cost. When deterministic workflows are replaced by stochastic policies over actions and tool calls, the key question is not whether a next step appears plausible, but whether the resulting trajectory remains statistically supported, locally unambiguous, and economically governable. We develop a measure-theoretic Markov framework for this setting. The core quantities are state blind-spot mass B_n(tau), state-action blind mass B^SA_{pi,n}(tau), an entropy-based human-in-the-loop escalation gate, and an expected oversight-cost identity over the workflow visitation measure. We instantiate the framework on the Business Process Intelligence Challenge 2019 purchase-to-pay log (251,734 cases, 1,595,923 events, 42 distinct workflow actions) and construct a log-driven simulated agent from a chronological 80/20 split of the same process. The main empirical finding is that a large workflow can appear well supported at the state level while retaining substantial blind mass over next-step decisions: refining the operational state to include case context, economic magnitude, and actor class expands the state space from 42 to 668 and raises state-action blind mass from 0.0165 at tau=50 to 0.1253 at tau=1000. On the held-out split, m(s) = max_a pi-hat(a|s) tracks realized autonomous step accuracy within 3.4 percentage points on average. The same quantities that delimit statistically credible autonomy also determine expected oversight burden. The framework is demonstrated on a large-scale enterprise procurement workflow and is designed for direct application to engineering processes for which operational event logs are available.

Retrieval-augmented generation (RAG) systems are increasingly used to analyze complex policy documents, but achieving sufficient reliability for expert usage remains challenging in domains characterized by dense legal language and evolving, overlapping regulatory frameworks. We study the application of RAG to AI governance and policy analysis using the AI Governance and Regulatory Archive (AGORA) corpus, a curated collection of 947 AI policy documents. Our system combines a ColBERT-based retriever fine-tuned with contrastive learning and a generator aligned to human preferences using Direct Preference Optimization (DPO). We construct synthetic queries and collect pairwise preferences to adapt the system to the policy domain. Through experiments evaluating retrieval quality, answer relevance, and faithfulness, we find that domain-specific fine-tuning improves retrieval metrics but does not consistently improve end-to-end question answering performance. In some cases, stronger retrieval counterintuitively leads to more confident hallucinations when relevant documents are absent from the corpus. These results highlight a key concern for those building policy-focused RAG systems: improvements to individual components do not necessarily translate to more reliable answers. Our findings provide practical insights for designing grounded question-answering systems over dynamic regulatory corpora.

Hallucination remains a critical bottleneck for large language models (LLMs), undermining their reliability in real-world applications, especially in Retrieval-Augmented Generation (RAG) systems. While existing hallucination detection methods employ LLM-as-a-judge to verify LLM outputs against retrieved evidence, they suffer from inherent confirmation bias, where the verifier inadvertently reproduces the errors of the original generation. To address this, we introduce Multi-Agent Reinforced Self-Check for Hallucination (MARCH), a framework that enforces rigorous factual alignment by leveraging deliberate information asymmetry. MARCH orchestrates a collaborative pipeline of three specialized agents: a Solver, a Proposer, and a Checker. The Solver generates an initial RAG response, which the Proposer decomposes into claim-level verifiable atomic propositions. Crucially, the Checker validates these propositions against retrieved evidence in isolation, deprived of the Solver's original output. This well-crafted information asymmetry scheme breaks the cycle of self-confirmation bias. By training this pipeline with multi-agent reinforcement learning (MARL), we enable the agents to co-evolve and optimize factual adherence. Extensive experiments across hallucination benchmarks demonstrate that MARCH substantially reduces hallucination rates. Notably, an 8B-parameter LLM equipped with MARCH achieves performance competitive with powerful closed-source models. MARCH paves a scalable path for factual self-improvement of LLMs through co-evolution. The code is at https://github.com/Qwen-Applications/MARCH.

Accurate 3D reconstruction of deformable soft tissues is essential for surgical robotic perception. However, low-texture surfaces, specular highlights, and instrument occlusions often fragment geometric continuity, posing a challenge for existing fixed-topology approaches. To address this, we propose EndoVGGT, a geometry-centric framework equipped with a Deformation-aware Graph Attention (DeGAT) module. Rather than using static spatial neighborhoods, DeGAT dynamically constructs feature-space semantic graphs to capture long-range correlations among coherent tissue regions. This enables robust propagation of structural cues across occlusions, enforcing global consistency and improving non-rigid deformation recovery. Extensive experiments on SCARED show that our method significantly improves fidelity, increasing PSNR by 24.6% and SSIM by 9.1% over prior state-of-the-art. Crucially, EndoVGGT exhibits strong zero-shot cross-dataset generalization to the unseen SCARED and EndoNeRF domains, confirming that DeGAT learns domain-agnostic geometric priors. These results highlight the efficacy of dynamic feature-space modeling for consistent surgical 3D reconstruction.

Robotic manipulation often requires memory: occlusion and state changes can make decision-time observations perceptually aliased, making action selection non-Markovian at the observation level because the same observation may arise from different interaction histories. Most embodied agents implement memory via semantically compressed traces and similarity-based retrieval, which discards disambiguating fine-grained perceptual cues and can return perceptually similar but decision-irrelevant episodes. Inspired by human episodic memory, we propose Chameleon, which writes geometry-grounded multimodal tokens to preserve disambiguating context and produces goal-directed recall through a differentiable memory stack. We also introduce Camo-Dataset, a real-robot UR5e dataset spanning episodic recall, spatial tracking, and sequential manipulation under perceptual aliasing. Across tasks, Chameleon consistently improves decision reliability and long-horizon control over strong baselines in perceptually confusable settings.

Scalable Vector Graphics (SVG) are an essential format for technical illustration and digital design, offering precise resolution independence and flexible semantic editability. In practice, however, original vector source files are frequently lost or inaccessible, leaving only "flat" rasterized versions (e.g., PNG or JPEG) that are difficult to modify or scale. Manually reconstructing these figures is a prohibitively labor-intensive process, requiring specialized expertise to recover the original geometric intent. To bridge this gap, we propose VFIG, a family of Vision-Language Models trained for complex and high-fidelity figure-to-SVG conversion. While this task is inherently data-driven, existing datasets are typically small-scale and lack the complexity of professional diagrams. We address this by introducing VFIG-DATA, a large-scale dataset of 66K high-quality figure-SVG pairs, curated from a diverse mix of real-world paper figures and procedurally generated diagrams. Recognizing that SVGs are composed of recurring primitives and hierarchical local structures, we introduce a coarse-to-fine training curriculum that begins with supervised fine-tuning (SFT) to learn atomic primitives and transitions to reinforcement learning (RL) refinement to optimize global diagram fidelity, layout consistency, and topological edge cases. Finally, we introduce VFIG-BENCH, a comprehensive evaluation suite with novel metrics designed to measure the structural integrity of complex figures. VFIG achieves state-of-the-art performance among open-source models and performs on par with GPT-5.2, achieving a VLM-Judge score of 0.829 on VFIG-BENCH.

In this article, we focus on search algorithms for two-player perfect information games, whose objective is to determine the best possible strategy, and ideally a winning strategy. Unfortunately, some search algorithms for games in the literature are not able to always determine a winning strategy, even with an infinite search time. This is the case, for example, of the following algorithms: Unbounded Best-First Minimax and Descent Minimax, which are core algorithms in state-of-the-art knowledge-free reinforcement learning. They were then improved with the so-called completion technique. However, whether this technique sufficiently improves these algorithms to allow them to always determine a winning strategy remained an open question until now. To answer this question, we generalize the two algorithms (their versions using the completion technique), and we show that any algorithm of this class of algorithms computes the best strategy. Finally, we experimentally show that the completion technique improves winning performance.

Advances in diffusion-based video generation models, while significantly improving human animation, poses threats of misuse through the creation of fake videos from a specific person's photo and text prompts. Recent efforts have focused on adversarial attacks that introduce crafted perturbations to protect images from diffusion-based models. However, most existing approaches target image generation, while relatively few explicitly address image-to-video diffusion models (VDMs), and most primarily focus on UNet-based architectures. Hence, their effectiveness against Diffusion Transformer (DiT) models remains largely under-explored, as these models demonstrate improved feature retention, and stronger temporal consistency due to larger capacity and advanced attention mechanisms. In this work, we introduce Anti-I2V, a novel defense against malicious human image-to-video generation, applicable across diverse diffusion backbones. Instead of restricting noise updates to the RGB space, Anti-I2V operates in both the $L$*$a$*$b$* and frequency domains, improving robustness and concentrating on salient pixels. We then identify the network layers that capture the most distinct semantic features during the denoising process to design appropriate training objectives that maximize degradation of temporal coherence and generation fidelity. Through extensive validation, Anti-I2V demonstrates state-of-the-art defense performance against diverse video diffusion models, offering an effective solution to the problem.

Constrained optimization in high-dimensional black-box settings is difficult due to expensive evaluations, the lack of gradient information, and complex feasibility regions. In this work, we propose a Bayesian optimization method that combines a penalty formulation, a surrogate model, and a trust region strategy. The constrained problem is converted to an unconstrained form by penalizing constraint violations, which provides a unified modeling framework. A trust region restricts the search to a local region around the current best solution, which improves stability and efficiency in high dimensions. Within this region, we use the Expected Improvement acquisition function to select evaluation points by balancing improvement and uncertainty. The proposed Trust Region method integrates penalty-based constraint handling with local surrogate modeling. This combination enables efficient exploration of feasible regions while maintaining sample efficiency. We compare the proposed method with state-of-the-art methods on synthetic and real-world high-dimensional constrained optimization problems. The results show that the method identifies high-quality feasible solutions with fewer evaluations and maintains stable performance across different settings.

While large-scale pretraining has revolutionized language modeling, its potential remains underexplored in healthcare with structured electronic health records (EHRs). We present RAVEN, a novel generative pretraining strategy for sequential EHR data based on Recurrence-Aware next-Visit EveNt prediction. Leveraging a dataset of over one million unique individuals, our model learns to autoregressively generate tokenized clinical events for the next visit conditioned on patient history. We introduce regularization on predicting repeated events and highlight a key pitfall in EHR-based foundation model evaluations: repeated event tokens can inflate performance metrics when new onsets are not distinguished from subsequent occurrences. Furthermore, we empirically investigate the scaling behaviors in a data-constrained, compute-saturated regime, showing that simply increasing model size is suboptimal without commensurate increases in data volume. We evaluate our model via zero-shot prediction for forecasting the incidence of a diverse set of diseases, where it rivals fully fine-tuned representation-based Transformer models and outperforms widely used simulation-based next-token approaches. Finally, without additional parameter updates, we show that RAVEN can generalize to an external patient cohort under lossy clinical code mappings and feature coverage gaps.

LLM-based mutation testing is a promising testing technology, but existing approaches typically rely on a fixed set of mutations as few-shot examples or none at all. This can result in generic low-quality mutations, missed context-specific mutation patterns, substantial numbers of redundant and uncompilable mutants, and limited semantic similarity to real bugs. To overcome these limitations, we introduce SMART (Semantic Mutation with Adaptive Retrieval and Tuning). SMART integrates retrieval-augmented generation (RAG) on a vectorized dataset of real-world bugs, focused code chunking, and supervised fine-tuning using mutations coupled with real-world bugs. We conducted an extensive empirical study of SMART using 1,991 real-world Java bugs from the Defects4J and ConDefects datasets, comparing SMART to the state-of-the-art LLM-based approaches, LLMut and LLMorpheus. The results reveal that SMART substantially improves mutation validity, effectiveness, and efficiency (even enabling small-scale 7B-scale models to match or even surpass large models like GPT-4o). We also demonstrate that SMART significantly improves downstream software engineering applications, including test case prioritization and fault localization. More specifically, SMART improves validity (weighted average generation rate) from 42.89% to 65.6%. It raises the non-duplicate rate from 87.38% to 95.62%, and the compilable rate from 88.85% to 90.21%. In terms of effectiveness, it achieves a real bug detection rate of 92.61% (vs. 57.86% for LLMut) and improves the average Ochiai coefficient from 25.61% to 38.44%. For fault localization, SMART ranks 64 more bugs as Top-1 under MUSE and 57 more under Metallaxis.

We introduce the Free-Market Algorithm (FMA), a novel metaheuristic inspired by free-market economics. Unlike Genetic Algorithms, Particle Swarm Optimization, and Simulated Annealing -- which require prescribed fitness functions and fixed search spaces -- FMA uses distributed supply-and-demand dynamics where fitness is emergent, the search space is open-ended, and solutions take the form of hierarchical pathway networks. Autonomous agents discover rules, trade goods, open and close firms, and compete for demand with no centralized controller. FMA operates through a three-layer architecture: a universal market mechanism (supply, demand, competition, selection), pluggable domain-specific behavioral rules, and domain-specific observation. The market mechanism is identical across applications; only the behavioral rules change. Validated in two unrelated domains. In prebiotic chemistry, starting from 900 bare atoms (C, H, O, N), FMA discovers all 12 feasible amino acid formulas, all 5 nucleobases, the formose sugar chain, and Krebs cycle intermediates in under 5 minutes on a laptop -- with up to 240 independent synthesis routes per product. In macroeconomic forecasting, reading a single input-output table with zero estimated parameters, FMA achieves Mean Absolute Error of 0.42 percentage points for non-crisis GDP prediction, comparable to professional forecasters, portable to 33 countries. Assembly Theory alignment shows that FMA provides the first explicit, tunable mechanism for the selection signatures described by Sharma et al. (Nature, 2023). The event-driven assembly dynamics resonate with foundational programs in physics -- causal set theory, relational quantum mechanics, constructor theory -- suggesting that Darwinian market dynamics may reflect a deeper organizational principle that lead to the unfolding of Nature itself.

The dense, temporal nature of video presents a profound challenge for automated analysis. Despite the use of powerful Vision-Language Models, prevailing methods for video understanding are limited by the inherent disconnect between reasoning and perception: they rely on static, pre-processed information and cannot actively seek raw evidence from video as their understanding evolves. To address this, we introduce LensWalk, a flexible agentic framework that empowers a Large Language Model reasoner to control its own visual observation actively. LensWalk establishes a tight reason-plan-observe loop where the agent dynamically specifies, at each step, the temporal scope and sampling density of the video it observes. Using a suite of versatile, Vision-Language Model based tools parameterized by these specifications, the agent can perform broad scans for cues, focus on specific segments for fact extraction, and stitch evidence from multiple moments for holistic verification. This design allows for progressive, on-demand evidence gathering that directly serves the agent's evolving chain of thought. Without requiring any model fine-tuning, LensWalk delivers substantial, plug-and-play performance gains on multiple model recipes, boosting their accuracy by over 5\% on challenging long-video benchmarks like LVBench and Video-MME. Our analysis reveals that enabling an agent to control how it sees is key to unlocking more accurate, robust, and interpretable video reasoning.

Retrieval-Augmented Generation (RAG) has emerged as a framework to address the constraints of Large Language Models (LLMs). Yet, its effectiveness fundamentally hinges on document chunking - an often-overlooked determinant of its quality. This paper presents an empirical study quantifying performance differences across four chunking strategies: fixed-size sliding window, recursive, breakpoint-based semantic, and structure-aware. We evaluated these methods using a proprietary corpus of oil and gas enterprise documents, including text-heavy manuals, table-heavy specifications, and piping and instrumentation diagrams (P and IDs). Our findings show that structure-aware chunking yields higher overall retrieval effectiveness, particularly in top-K metrics, and incurs significantly lower computational costs than semantic or baseline strategies. Crucially, all four methods demonstrated limited effectiveness on P and IDs, underscoring a core limitation of purely text-based RAG within visually and spatially encoded documents. We conclude that while explicit structure preservation is essential for specialised domains, future work must integrate multimodal models to overcome current limitations.

Automatic Speech Recognition (ASR) systems are widely used in everyday communication, education, healthcare, and industry, yet their performance remains uneven across speakers, particularly when dialectal variation diverges from the mainstream accents represented in training data. This study investigates ASR bias through a sociolinguistic analysis of Newcastle English, a regional variety of North-East England that has been shown to challenge current speech recognition technologies. Using spontaneous speech from the Diachronic Electronic Corpus of Tyneside English (DECTE), we evaluate the output of a state-of-the-art commercial ASR system and conduct a fine-grained analysis of more than 3,000 transcription errors. Errors are classified by linguistic domain and examined in relation to social variables including gender, age, and socioeconomic status. In addition, an acoustic case study of selected vowel features demonstrates how gradient phonetic variation contributes directly to misrecognition. The results show that phonological variation accounts for the majority of errors, with recurrent failures linked to dialect-specific features like vowel quality and glottalisation, as well as local vocabulary and non-standard grammatical forms. Error rates also vary across social groups, with higher error frequencies observed for men and for speakers at the extremes of the age spectrum. These findings indicate that ASR errors are not random but socially patterned and can be explained from a sociolinguistic perspective. Thus, the study demonstrates the importance of incorporating sociolinguistic expertise into the evaluation and development of speech technologies and argues that more equitable ASR systems require explicit attention to dialectal variation and community-based speech data.

Activation steering has emerged as a powerful tool to shape LLM behavior without the need for weight updates. While its inherent brittleness and unreliability are well-documented, its safety implications remain underexplored. In this work, we present a systematic safety audit of steering vectors obtained with Contrastive Activation Addition (CAA), a widely used steering approach, under a unified evaluation protocol. Using JailbreakBench as benchmark, we show that steering vectors consistently influence the success rate of jailbreak attacks, with stronger amplification under simple template-based attacks. Across LLM families and sizes, steering the model in specific directions can drastically increase (up to 57%) or decrease (up to 50%) its attack success rate (ASR), depending on the targeted behavior. We attribute this phenomenon to the overlap between the steering vectors and the latent directions of refusal behavior. Thus, we offer a traceable explanation for this discovery. Together, our findings reveal the previously unobserved origin of this safety gap in LLMs, highlighting a trade-off between controllability and safety.

Generative world models offer a compelling foundation for augmented-reality (AR) applications: by predicting future image sequences that incorporate deliberate visual edits, they enable temporally coherent, augmented future frames that can be computed ahead of time and cached, avoiding per-frame rendering from scratch in real time. In this work, we present SEGAR, a preliminary framework that combines a diffusion-based world model with a selective correction stage to support this vision. The world model generates augmented future frames with region-specific edits while preserving others, and the correction stage subsequently aligns safety-critical regions with real-world observations while preserving intended augmentations elsewhere. We demonstrate this pipeline in driving scenarios as a representative setting where semantic region structure is well defined and real-world feedback is readily available. We view this as an early step toward generative world models as practical AR infrastructure, where future frames can be generated, cached, and selectively corrected on demand.

Video-language foundation models have proven to be highly effective in zero-shot applications across a wide range of tasks. A particularly challenging area is the intraoperative surgical procedure domain, where labeled data is scarce, and precise temporal understanding is often required for complex downstream tasks. To address this challenge, we introduce CliPPER (Contextual Video-Language Pretraining on Long-form Intraoperative Surgical Procedures for Event Recognition), a novel video-language pretraining framework trained on surgical lecture videos. Our method is designed for fine-grained temporal video-text recognition and introduces several novel pretraining strategies to improve multimodal alignment in long-form surgical videos. Specifically, we propose Contextual Video-Text Contrastive Learning (VTC_CTX) and Clip Order Prediction (COP) pretraining objectives, both of which leverage temporal and contextual dependencies to enhance local video understanding. In addition, we incorporate a Cycle-Consistency Alignment over video-text matches within the same surgical video to enforce bidirectional consistency and improve overall representation coherence. Moreover, we introduce a more refined alignment loss, Frame-Text Matching (FTM), to improve the alignment between video frames and text. As a result, our model establishes a new state-of-the-art across multiple public surgical benchmarks, including zero-shot recognition of phases, steps, instruments, and triplets. The source code and pretraining captions can be found at https://github.com/CAMMA-public/CliPPER.

Reading comprehension presents a significant challenge for children with Special Educational Needs and Disabilities (SEND), often requiring intensive one-on-one reading support. To assist therapists in scaling this support, we developed a multilingual, AI-powered interface that automatically enhances text with visual scaffolding. This system dynamically identifies key concepts and maps them to contextually relevant pictograms, supporting learners across languages. We evaluated the system across five typologically diverse languages (English, French, Italian, Spanish, and Arabic), through multilingual coverage analysis, expert clinical review by speech therapists and special education professionals, and latency assessment. Evaluation results indicate high pictogram coverage and visual scaffolding density across the five languages. Expert audits suggested that automatically selected pictograms were semantically appropriate, with combined correct and acceptable ratings exceeding 95% for the four European languages and approximately 90% for Arabic despite reduced pictogram repository coverage. System latency remained within interactive thresholds suitable for real-time educational use. These findings support the technical viability, semantic safety, and acceptability of automated multimodal scaffolding to improve accessibility for neurodiverse learners.

Adaptive scaffolding enhances learning, yet the field lacks robust methods for measuring it within authentic tutoring dialogue. This gap has become more pressing with the rise of remote human tutoring and large language model-based systems. We introduce an embedding-based approach that analyzes scaffolding dynamics by aligning the semantics of dialogue turns, problem statements, and correct solutions. Specifically, we operationalize alignment by computing cosine similarity between tutor and student contributions and task-relevant content. We apply this framework to 1,576 real-world mathematics tutoring dialogues from the Eedi Question Anchored Tutoring Dialogues dataset. The analysis reveals systematic differences in task alignment and distinct temporal patterns in how participants ground their contributions in problem and solution content. Further, mixed-effects models show that role-specific semantic alignment predicts tutorial progression beyond baseline features such as message order and length. Tutor contributions exhibited stronger grounding in problem content early in interactions. In contrast, student solution alignment was modestly positively associated with progression. These findings support scaffolding as a continuous, role-sensitive process grounded in task semantics. By capturing role-specific alignment over time, this approach provides a principled method for analyzing instructional dialogue and evaluating conversational tutoring systems.

Autonomous mobile GUI agents have attracted increasing attention along with the advancement of Multimodal Large Language Models (MLLMs). However, existing methods still suffer from inefficient learning from failed trajectories and ambiguous credit assignment under sparse rewards for long-horizon GUI tasks. To that end, we propose UI-Voyager, a novel two-stage self-evolving mobile GUI agent. In the first stage, we employ Rejection Fine-Tuning (RFT), which enables the continuous co-evolution of data and models in a fully autonomous loop. The second stage introduces Group Relative Self-Distillation (GRSD), which identifies critical fork points in group rollouts and constructs dense step-level supervision from successful trajectories to correct failed ones. Extensive experiments on AndroidWorld show that our 4B model achieves an 81.0% Pass@1 success rate, outperforming numerous recent baselines and exceeding human-level performance. Ablation and case studies further verify the effectiveness of GRSD. Our method represents a significant leap toward efficient, self-evolving, and high-performance mobile GUI automation without expensive manual data annotation.

Unconventional physical computing is producing many novel and exotic devices that can potentially be used in a computational mode. Currently, these tend to be used to implement traditional models of computation, such as boolean logic circuits, or neuromorphic approaches. This runs the risk of failing to exploit the devices to their full potential. Here we describe a methodology for deriving a model of computation and domain specific language more closely matched to a given physical device's capabilities, and illustrate it with a case study of bosonic computing as implemented by a physical multi-component interferometer.

We introduce incongruent normal form (INF), a structural representation for self-referential semantic sentences. An INF replaces a self-referential sentence with a finite family of non-self-referential sentences that are individually satisfiable but not jointly satisfiable. This transformation isolates the semantic obstruction created by self-reference while preserving classical semantics locally and is accompanied by correctness theorems characterizing when global inconsistency arises from locally compatible commitments. We then study the role of incongruence as a structural source of semantic informativeness. Using a minimal model-theoretic notion of informativeness-understood as the ability of sentences to distinguish among admissible models-we show that semantic completeness precludes informativeness, while incongruence preserves it. Moreover, incongruence is not confined to paradoxical constructions: any consistent incomplete first-order theory admits finite incongruent families arising from incompatible complete extensions. In this sense, incompleteness manifests structurally as locally realizable but globally incompatible semantic commitments, providing a minimal formal basis for semantic knowledge. Finally, we introduce a quantitative semantic framework. In a canonical finite semantic-state setting, we model semantic commitments as Boolean functions and define a Fourier-analytic notion of semantic energy based on total influence. We derive uncertainty-style bounds relating semantic determinacy, informativeness, and spectral simplicity, and establish a matrix inequality bounding aggregate semantic variance by total semantic energy. These results show quantitatively that semantic informativeness cannot collapse into a single determinate state without unbounded energy cost, identifying incongruence as a fundamental structural and quantitative feature of semantic representation.

Research on explainable AI (XAI) has frequently focused on explaining model predictions. More recently, methods have been proposed to explain prediction uncertainty by attributing it to input features (uncertainty attributions). However, the evaluation of these methods remains inconsistent as studies rely on heterogeneous proxy tasks and metrics, hindering comparability. We address this by aligning uncertainty attributions with the well-established Co-12 framework for XAI evaluation. We propose concrete implementations for the correctness, consistency, continuity, and compactness properties. Additionally, we introduce conveyance, a property tailored to uncertainty attributions that evaluates whether controlled increases in epistemic uncertainty reliably propagate to feature-level attributions. We demonstrate our evaluation framework with eight metrics across combinations of uncertainty quantification and feature attribution methods on tabular and image data. Our experiments show that gradient-based methods consistently outperform perturbation-based approaches in consistency and conveyance, while Monte-Carlo dropconnect outperforms Monte-Carlo dropout in most metrics. Although most metrics rank the methods consistently across samples, inter-method agreement remains low. This suggests no single metric sufficiently evaluates uncertainty attribution quality. The proposed evaluation framework contributes to the body of knowledge by establishing a foundation for systematic comparison and development of uncertainty attribution methods.

To embed domain-specific or specialized knowledge into pre-trained foundation models, fine-tuning using techniques such as parameter efficient fine-tuning (e.g. LoRA) is a common practice. However, as new LLM architectures and pre-trained models emerge, transferring this specialized knowledge to newer models becomes an important task. In many scenarios, the original specialized data may be unavailable due to privacy or commercial restrictions, necessitating distillation and transfer of this specialized knowledge from the fine-tuned base model to a different pre-trained model. We present TuneShift-KD, a novel approach that automatically distills specialized knowledge from a fine-tuned model to a target model using only a few examples representative of the specialized information. Our key insight is that specialized knowledge can be identified through perplexity differences between base and fine-tuned models: prompts where the fine-tuned model responds confidently (low perplexity), but the base model struggles (high perplexity), indicate queries corresponding to the specialized knowledge learned by the fine-tuned model. TuneShift-KD leverages this insight to create a synthetic training dataset to transfer the specialized knowledge. Using an iterative process, TuneShift-KD generates more prompts similar to those that generated responses with specialized knowledge. TuneShift-KD does not require training discriminators or access to training datasets. It is an automated approach that only requires the initial fine-tuned and base models and a few representative prompts. Our experiments demonstrate that models fine-tuned using TuneShift-KD achieve higher accuracy than prior approaches, enabling ease of deployment and more effective transfer of the specialized knowledge.

Agentic Variation Operators (AVO) are a new family of evolutionary variation operators that replace the fixed mutation, crossover, and hand-designed heuristics of classical evolutionary search with autonomous coding agents. Rather than confining a language model to candidate generation within a prescribed pipeline, AVO instantiates variation as a self-directed agent loop that can consult the current lineage, a domain-specific knowledge base, and execution feedback to propose, repair, critique, and verify implementation edits. We evaluate AVO on attention, among the most aggressively optimized kernel targets in AI, on NVIDIA Blackwell (B200) GPUs. Over 7 days of continuous autonomous evolution on multi-head attention, AVO discovers kernels that outperform cuDNN by up to 3.5% and FlashAttention-4 by up to 10.5% across the evaluated configurations. The discovered optimizations transfer readily to grouped-query attention, requiring only 30 minutes of additional autonomous adaptation and yielding gains of up to 7.0% over cuDNN and 9.3% over FlashAttention-4. Together, these results show that agentic variation operators move beyond prior LLM-in-the-loop evolutionary pipelines by elevating the agent from candidate generator to variation operator, and can discover performance-critical micro-architectural optimizations that produce kernels surpassing state-of-the-art expert-engineered attention implementations on today's most advanced GPU hardware.

LLM agents like Claude Code can not only write code but also be used for autonomous AI research and engineering \citep{rank2026posttrainbench, novikov2025alphaevolve}. We show that an \emph{autoresearch}-style pipeline \citep{karpathy2026autoresearch} powered by Claude Code discovers novel white-box adversarial attack \textit{algorithms} that \textbf{significantly outperform all existing (30+) methods} in jailbreaking and prompt injection evaluations. Starting from existing attack implementations, such as GCG~\citep{zou2023universal}, the agent iterates to produce new algorithms achieving up to 40\% attack success rate on CBRN queries against GPT-OSS-Safeguard-20B, compared to $\leq$10\% for existing algorithms (\Cref{fig:teaser}, left). The discovered algorithms generalize: attacks optimized on surrogate models transfer directly to held-out models, achieving \textbf{100\% ASR against Meta-SecAlign-70B} \citep{chen2025secalign} versus 56\% for the best baseline (\Cref{fig:teaser}, middle). Extending the findings of~\cite{carlini2025autoadvexbench}, our results are an early demonstration that incremental safety and security research can be automated using LLM agents. White-box adversarial red-teaming is particularly well-suited for this: existing methods provide strong starting points, and the optimization objective yields dense, quantitative feedback. We release all discovered attacks alongside baseline implementations and evaluation code at https://github.com/romovpa/claudini.

Particle-in-Cell (PIC) Monte Carlo (MC) simulations are central to plasma physics but face increasing challenges on heterogeneous HPC systems due to excessive data movement, synchronization overheads, and inefficient utilization of multiple accelerators. In this work, we present a portable, multi-GPU hybrid MPI+OpenMP implementation of BIT1 that enables scalable execution on both Nvidia and AMD accelerators through OpenMP target tasks with explicit dependencies to overlap computation and communication across devices. Portability is achieved through persistent device-resident memory, an optimized contiguous one-dimensional data layout, and a transition from unified to pinned host memory to improve large data-transfer efficiency, together with GPU Direct Memory Access (DMA) and runtime interoperability for direct device-pointer access. Standardized and scalable I/O is provided using openPMD and ADIOS2, supporting high-performance file I/O, in-memory data streaming, and in-situ analysis and visualization. Performance results on pre-exascale and exascale systems, including Frontier (OLCF-5) for up to 16,000 GPUs, demonstrate significant improvements in run time, scalability, and resource utilization for large-scale PIC MC simulations.

The practical deployment of nonlinear model predictive control (NMPC) is often limited by online computation: solving a nonlinear program at high control rates can be expensive on embedded hardware, especially when models are complex or horizons are long. Learning-based NMPC approximations shift this computation offline but typically demand large expert datasets and costly training. We propose Sequential-AMPC, a sequential neural policy that generates MPC candidate control sequences by sharing parameters across the prediction horizon. For deployment, we wrap the policy in a safety-augmented online evaluation and fallback mechanism, yielding Safe Sequential-AMPC. Compared to a naive feedforward policy baseline across several benchmarks, Sequential-AMPC requires substantially fewer expert MPC rollouts and yields candidate sequences with higher feasibility rates and improved closed-loop safety. On high-dimensional systems, it also exhibits better learning dynamics and performance in fewer epochs while maintaining stable validation improvement where the feedforward baseline can stagnate.

Labels on platforms such as GitHub support triage and coordination, yet little is known about how well they align with code modifications or how such alignment affects collaboration across contributor experience levels. We present a case study of the Kubernetes project, introducing label-diff congruence - the alignment between pull request labels and modified files - and examining its prevalence, stability, behavioral validation, and relationship to collaboration outcomes across contributor tiers. We analyse 18,020 pull requests (2014--2025) with area labels and complete file diffs, validate alignment through analysis of over one million review comments and label corrections, and test associations with time-to-merge and discussion characteristics using quantile regression and negative binomial models stratified by contributor experience. Congruence is prevalent (46.6\% perfect alignment), stable over years, and routinely maintained (9.2\% of PRs corrected during review). It does not predict merge speed but shapes discussion: among core developers (81\% of the sample), higher congruence predicts quieter reviews (18\% fewer participants), whereas among one-time contributors it predicts more engagement (28\% more participants). Label-diff congruence influences how collaboration unfolds during review, supporting efficiency for experienced developers and visibility for newcomers. For projects with similar labeling conventions, monitoring alignment can help detect coordination friction and provide guidance when labels and code diverge.

Learning-based models for fluid dynamics often operate in unconstrained function spaces, leading to physically inadmissible, unstable simulations. While penalty-based methods offer soft regularization, they provide no structural guarantees, resulting in spurious divergence and long-term collapse. In this work, we introduce a unified framework that enforces the incompressible continuity equation as a hard, intrinsic constraint for both deterministic and generative modeling. First, to project deterministic models onto the divergence-free subspace, we integrate a differentiable spectral Leray projection grounded in the Helmholtz-Hodge decomposition, which restricts the regression hypothesis space to physically admissible velocity fields. Second, to generate physically consistent distributions, we show that simply projecting model outputs is insufficient when the prior is incompatible. To address this, we construct a divergence-free Gaussian reference measure via a curl-based pushforward, ensuring the entire probability flow remains subspace-consistent by construction. Experiments on 2D Navier-Stokes equations demonstrate exact incompressibility up to discretization error and substantially improved stability and physical consistency.

Uniform laws of large numbers form a cornerstone of Vapnik--Chervonenkis theory, where they are characterized by the finiteness of the VC dimension. In this work, we study uniform convergence phenomena in cartesian product spaces, under assumptions on the underlying distribution that are compatible with the product structure. Specifically, we assume that the distribution is absolutely continuous with respect to the product of its marginals, a condition that captures many natural settings, including product distributions, sparse mixtures of product distributions, distributions with low mutual information, and more. We show that, under this assumption, a uniform law of large numbers holds for a family of events if and only if the linear VC dimension of the family is finite. The linear VC dimension is defined as the maximum size of a shattered set that lies on an axis-parallel line, namely, a set of vectors that agree on all but at most one coordinate. This dimension is always at most the classical VC dimension, yet it can be arbitrarily smaller. For instance, the family of convex sets in $\mathbb{R}^d$ has linear VC dimension $2$, while its VC dimension is infinite already for $d\ge 2$. Our proofs rely on estimator that departs substantially from the standard empirical mean estimator and exhibits more intricate structure. We show that such deviations from the standard empirical mean estimator are unavoidable in this setting. Throughout the paper, we propose several open questions, with a particular focus on quantitative sample complexity bounds.

Miscalibrated confidence scores are a practical obstacle to deploying AI in clinical settings. A model that is always overconfident offers no useful signal for deferral. We present a multi-agent framework that combines domain-specific specialist agents with Two-Phase Verification and S-Score Weighted Fusion to improve both calibration and discrimination in medical multiple-choice question answering. Four specialist agents (respiratory, cardiology, neurology, gastroenterology) generate independent diagnoses using Qwen2.5-7B-Instruct. Each diagnosis is then subjected to a two-phase self-verification process that measures internal consistency and produces a Specialist Confidence Score (S-score). The S-scores drive a weighted fusion strategy that selects the final answer and calibrates the reported confidence. We evaluate across four experimental settings, covering 100-question and 250-question high-disagreement subsets of both MedQA-USMLE and MedMCQA. Calibration improvement is the central finding, with ECE reduced by 49-74% across all four settings, including the harder MedMCQA benchmark where these gains persist even when absolute accuracy is constrained by knowledge-intensive recall demands. On MedQA-250, the full system achieves ECE = 0.091 (74.4% reduction over the single-specialist baseline) and AUROC = 0.630 (+0.056) at 59.2% accuracy. Ablation analysis identifies Two-Phase Verification as the primary calibration driver and multi-agent reasoning as the primary accuracy driver. These results establish that consistency-based verification produces more reliable uncertainty estimates across diverse medical question types, providing a practical confidence signal for deferral in safety-critical clinical AI applications.

Composer 2 is a specialized model designed for agentic software engineering. The model demonstrates strong long-term planning and coding intelligence while maintaining the ability to efficiently solve problems for interactive use. The model is trained in two phases: first, continued pretraining to improve the model's knowledge and latent coding ability, followed by large-scale reinforcement learning to improve end-to-end coding performance through stronger reasoning, accurate multi-step execution, and coherence on long-horizon realistic coding problems. We develop infrastructure to support training in the same Cursor harness that is used by the deployed model, with equivalent tools and structure, and use environments that match real problems closely. To measure the ability of the model on increasingly difficult tasks, we introduce a benchmark derived from real software engineering problems in large codebases including our own. Composer 2 is a frontier-level coding model and demonstrates a process for training strong domain-specialized models. On our CursorBench evaluations the model achieves a major improvement in accuracy compared to previous Composer models (61.3). On public benchmarks the model scores 61.7 on Terminal-Bench and 73.7 on SWE-bench Multilingual in our harness, comparable to state-of-the-art systems.

Accurate forecasting of state-of-health (SOH) is essential for ensuring safe and reliable operation of lithium-ion cells. However, existing models calibrated on laboratory tests at specific conditions often fail to generalize to new cells that differ due to small manufacturing variations or operate under different conditions. To address this challenge, an uncertainty-aware transfer learning framework is proposed, combining a Long Short-Term Memory (LSTM) model with domain adaptation via Maximum Mean Discrepancy (MMD) and uncertainty quantification through Conformal Prediction (CP). The LSTM model is trained on a virtual battery dataset designed to capture real-world variability in electrode manufacturing and operating conditions. MMD aligns latent feature distributions between simulated and target domains to mitigate domain shift, while CP provides calibrated, distribution-free prediction intervals. This framework improves both the generalization and trustworthiness of SOH forecasts across heterogeneous cells.

Self-distillation has emerged as an effective post-training paradigm for LLMs, often improving performance while shortening reasoning traces. However, in mathematical reasoning, we find that it can reduce response length while degrading performance. We trace this degradation to the suppression of epistemic verbalization - the model's expression of uncertainty during reasoning. Through controlled experiments varying conditioning context richness and task coverage, we show that conditioning the teacher on rich information suppresses uncertainty expression, enabling rapid in-domain optimization with limited task coverage but harming OOD performance, where unseen problems benefit from expressing uncertainty and adjusting accordingly. Across Qwen3-8B, DeepSeek-Distill-Qwen-7B, and Olmo3-7B-Instruct, we observe performance drops of up to 40%. Our findings highlight that exposing appropriate levels of uncertainty is crucial for robust reasoning and underscore the importance of optimizing reasoning behavior beyond merely reinforcing correct answer traces.

Every year, 10 million pets enter shelters, separated from their families. Despite desperate searches by both guardians and lost animals, 70% never reunite, not because matches do not exist, but because current systems look only at appearance, while animals recognize each other through sound. We ask, why does computer vision treat vocalizing species as silent visual objects? Drawing on five decades of cognitive science showing that animals perceive quantity approximately and communicate identity acoustically, we present the first multimodal reunification system integrating visual and acoustic biometrics. Our species-adaptive architecture processes vocalizations from 10Hz elephant rumbles to 4kHz puppy whines, paired with probabilistic visual matching that tolerates stress-induced appearance changes. This work demonstrates that AI grounded in biological communication principles can serve vulnerable populations that lack human language.

Recent advances in large language models (LLMs) and LLM-based agents have substantially improved the capabilities of automated theorem proving. However, for problems requiring complex mathematical reasoning, current systems rarely succeed on the first try and must repeatedly modify their proof strategies. Existing approaches for handling failed attempts typically either discard the entire proof and regenerate it from scratch or iteratively fix errors within the proof. The former is inefficient, as it may abandon mostly correct reasoning due to localized errors, while the latter, although preserving prior progress, leads to progressively longer contexts which progressively degrades the model's ability to attend to the remaining unresolved subproblems. To address this dilemma, we propose Mechanic, a novel agent system that employs a sorry-driven formal decomposition strategy. By leveraging the sorry placeholder in Lean to precisely isolate unresolved subgoals while preserving the surrounding verified proof structure, Mechanic extracts each failed subproblem into a clean, self-contained context and resolves it independently. This avoids both the waste of full regeneration and the excessive context length induced by repeated repairs. Experimental results on challenging mathematical competition benchmarks, including IMO 2025 and Putnam 2025, demonstrate that our agent achieves significant advantages in proving efficiency.

Current clinical decision support systems (CDSSs) typically base their predictions on correlation, not causation. In recent years, causal machine learning (ML) has emerged as a promising way to improve decision-making with CDSSs by offering interpretable, treatment-specific reasoning. However, existing research often emphasizes model development rather than designing clinician-facing interfaces. To address this gap, we investigated how CDSSs based on causal ML should be designed to effectively support collaborative clinical decision-making. Using a design science research methodology, we conducted a structured literature review and interviewed experienced physicians. From these, we derived eight empirically grounded design requirements, developed seven design principles, and proposed nine practical design features. Our results establish guidance for designing CDSSs that deliver causal insights, integrate seamlessly into clinical workflows, and support trust, usability, and human-AI collaboration. We also reveal tensions around automation, responsibility, and regulation, highlighting the need for an adaptive certification process for ML-based medical products.

The study addressed the problem of Anonymous Multi-Agent Path-finding (AMAPF). Unlike the classical formulation, where the assignment of agents to goals is fixed, in the anonymous MAPF setting it is irrelevant which agent reaches specific goal, provided that all goals are occupied. Most existing multi-agent pathfinding algorithms rely on a discrete representation of the environment (e.g., square grids) and do not account for the sizes of agents. This limits their applicability in real-world scenarios, such as trajectory planning for mobile robots in warehouses. Conversely, methods operating in continuous space typically impose substantial restrictions on the input data, such as constraints on the distances between initial and goal positions or between start/goal positions and obstacles. In this work, we considered one of the AMAPF algorithms designed for continuous space, where agents are modeled as disks of equal size. The algorithm requires a strict minimum separation of $4$ agent radii between any start/goal positions. Proposed a modification aimed at relaxing the constraints and reduce this limit from $4$ to $2\sqrt{3}$. We theoretically demonstrated that the proposed enhancements preserve original theoretical properties, including the guarantee that all agents will eventually achieve their goals safely and without collisions.

Computer-use agents (CUAs) hold great promise for automating complex desktop workflows, yet progress toward general-purpose agents is bottlenecked by the scarcity of continuous, high-quality human demonstration videos. Recent work emphasizes that continuous video, not sparse screenshots, is the critical missing ingredient for scaling these agents. However, the largest existing open dataset, ScaleCUA, contains only 2 million screenshots, equating to less than 20 hours of video. To address this bottleneck, we introduce CUA-Suite, a large-scale ecosystem of expert video demonstrations and dense annotations for professional desktop computer-use agents. At its core is VideoCUA, which provides approximately 10,000 human-demonstrated tasks across 87 diverse applications with continuous 30 fps screen recordings, kinematic cursor traces, and multi-layerfed reasoning annotations, totaling approximately 55 hours and 6 million frames of expert video. Unlike sparse datasets that capture only final click coordinates, these continuous video streams preserve the full temporal dynamics of human interaction, forming a superset of information that can be losslessly transformed into the formats required by existing agent frameworks. CUA-Suite further provides two complementary resources: UI-Vision, a rigorous benchmark for evaluating grounding and planning capabilities in CUAs, and GroundCUA, a large-scale grounding dataset with 56K annotated screenshots and over 3.6 million UI element annotations. Preliminary evaluation reveals that current foundation action models struggle substantially with professional desktop applications (~60% task failure rate). Beyond evaluation, CUA-Suite's rich multimodal corpus supports emerging research directions including generalist screen parsing, continuous spatial control, video-based reward modeling, and visual world models. All data and models are publicly released.

Enes The proposed architecture is a mixture of experts, which allows for the model entities, such as the causal relationships, to be further parameterized. More specifically, an attempt is made to exploit a neural net as implementing neurons poses a great challenge for this dataset. To explain, a simple and fast Pearson coefficient linear model usually achieves good scores. An aggressive baseline that requires a really good model to overcome that is. Moreover, there are major limitations when it comes to causal discovery of observational data. Unlike the sachs one did not use interventions but only prior knowledge; the most prohibiting limitation is that of the data which is addressed. Thereafter, the method and the model are described and after that the results are presented.

Speaker verification at large scale remains an open challenge as fixed-margin losses treat all samples equally regardless of quality. We hypothesize that mislabeled or degraded samples introduce noisy gradients that disrupt compact speaker manifolds. We propose Curry (CURriculum Ranking), an adaptive loss that estimates sample difficulty online via Sub-center ArcFace: confidence scores from dominant sub-center cosine similarity rank samples into easy, medium, and hard tiers using running batch statistics, without auxiliary annotations. Learnable weights guide the model from stable identity foundations through manifold refinement to boundary sharpening. To our knowledge, this is the largest-scale speaker verification system trained to date. Evaluated on VoxCeleb1-O, and SITW, Curry reduces EER by 86.8\% and 60.0\% over the Sub-center ArcFace baseline, establishing a new paradigm for robust speaker verification on imperfect large-scale data.

Parametric roll is a rare but high-consequence instability that can trigger abrupt regime changes in ship response, including pronounced shifts in roll statistics and tail risk. This paper develops a data-driven surrogate that learns the nonlinear, causal functional mapping from incident wave--motion time series to vessel motions, and demonstrates that the surrogate reproduces both (i) parametric roll episodes and (ii) the associated statistical shifts in the response. Crucially, the learning framework is data-source agnostic: the paired wave--motion time series can be obtained from controlled experiments (e.g., towing-tank or basin tests with wave probes and motion tracking) when a hull exists, or from high-fidelity simulations during design when experiments are not yet available. To provide a controlled severe-sea demonstration, we generate training data with a URANS numerical wave tank, using long-crested irregular seas synthesized from a modified Pierson--Moskowitz spectrum. The demonstration dataset comprises 49 random-phase realizations for each of three sea states, simulated at a fixed forward speed selected to yield encounter conditions under which parametric-roll episodes can occur. A stacked LSTM surrogate is trained on wave-elevation time series and evaluated on held-out realizations using time-domain accuracy and distributional fidelity metrics. In the most severe case, the model tracks the onset and growth of large-amplitude roll consistent with parametric excitation, and captures the corresponding changes in roll probability density functions (PDFs). We further compare loss-function choices (MSE, relative-entropy-based objectives, and amplitude-weighted variants) and show how they trade average error for improved tail fidelity relevant to operability and risk assessment.

Accurate subseasonal weather forecasting remains a major challenge due to the inherently chaotic nature of the atmosphere, which limits the predictive skill of conventional models beyond the mid-range horizon (approximately 15 days). In this work, we present \textit{Marchuk}, a generative latent flow-matching model for global weather forecasting spanning mid-range to subseasonal timescales, with prediction horizons of up to 30 days. Marchuk conditions on current-day weather maps and autoregressively predicts subsequent days' weather maps within the learned latent space. We replace rotary positional encodings (RoPE) with trainable positional embeddings and extend the temporal context window, which together enhance the model's ability to represent and propagate long-range temporal dependencies during latent forecasting. Marchuk offers two key advantages: high computational efficiency and strong predictive performance. Despite its compact architecture of only 276 million parameters, the model achieves performance comparable to LaDCast, a substantially larger model with 1.6 billion parameters, while operating at significantly higher inference speeds. We open-source our inference code and model at: https://v-gen-ai.github.io/Marchuk/

Understanding how biomarker distributions evolve over time is a central challenge in digital health and chronic disease monitoring. In diabetes, changes in the distribution of glucose measurements can reveal patterns of disease progression and treatment response that conventional summary measures miss. Motivated by a 26-week clinical trial comparing the closed-loop insulin delivery system t:slim X2 with standard therapy in children with type 1 diabetes, we propose a probabilistic framework to model the continuous-time evolution of time-indexed distributions using continuous glucose monitoring data (CGM) collected every five minutes. We represent the glucose distribution as a Gaussian mixture, with time-varying mixture weights governed by a neural ODE. We estimate the model parameter using a distribution-matching criterion based on the maximum mean discrepancy. The resulting framework is interpretable, computationally efficient, and sensitive to subtle temporal distributional changes. Applied to CGM trial data, the method detects treatment-related improvements in glucose dynamics that are difficult to capture with traditional analytical approaches.

Generative Retrieval (GR) has emerged as a promising paradigm for modern search systems. Compared to multi-stage cascaded architecture, it offers advantages such as end-to-end joint optimization and high computational efficiency. OneSearch, as a representative industrial-scale deployed generative search framework, has brought significant commercial and operational benefits. However, its inadequate understanding of complex queries, inefficient exploitation of latent user intents, and overfitting to narrow historical preferences have limited its further performance improvement. To address these challenges, we propose \textbf{OneSearch-V2}, a latent reasoning enhanced self-distillation generative search framework. It contains three key innovations: (1) a thought-augmented complex query understanding module, which enables deep query understanding and overcomes the shallow semantic matching limitations of direct inference; (2) a reasoning-internalized self-distillation training pipeline, which uncovers users' potential yet precise e-commerce intentions beyond log-fitting through implicit in-context learning; (3) a behavior preference alignment optimization system, which mitigates reward hacking arising from the single conversion metric, and addresses personal preference via direct user feedback. Extensive offline evaluations demonstrate OneSearch-V2's strong query recognition and user profiling capabilities. Online A/B tests further validate its business effectiveness, yielding +3.98\% item CTR, +3.05\% buyer conversion rate, and +2.11\% order volume. Manual evaluation further confirms gains in search experience quality, with +1.65\% in page good rate and +1.37\% in query-item relevance. More importantly, OneSearch-V2 effectively mitigates common search system issues such as information bubbles and long-tail sparsity, without incurring additional inference costs or serving latency.

OpenClaw has rapidly established itself as a leading open-source autonomous agent runtime, offering powerful capabilities including tool integration, local file access, and shell command execution. However, these broad operational privileges introduce critical security vulnerabilities, transforming model errors into tangible system-level threats such as sensitive data leakage, privilege escalation, and malicious third-party skill execution. Existing security measures for the OpenClaw ecosystem remain highly fragmented, addressing only isolated stages of the agent lifecycle rather than providing holistic protection. To bridge this gap, we present ClawKeeper, a real-time security framework that integrates multi-dimensional protection mechanisms across three complementary architectural layers. (1) \textbf{Skill-based protection} operates at the instruction level, injecting structured security policies directly into the agent context to enforce environment-specific constraints and cross-platform boundaries. (2) \textbf{Plugin-based protection} serves as an internal runtime enforcer, providing configuration hardening, proactive threat detection, and continuous behavioral monitoring throughout the execution pipeline. (3) \textbf{Watcher-based protection} introduces a novel, decoupled system-level security middleware that continuously verifies agent state evolution. It enables real-time execution intervention without coupling to the agent's internal logic, supporting operations such as halting high-risk actions or enforcing human confirmation. We argue that this Watcher paradigm holds strong potential to serve as a foundational building block for securing next-generation autonomous agent systems. Extensive qualitative and quantitative evaluations demonstrate the effectiveness and robustness of ClawKeeper across diverse threat scenarios. We release our code.

Poetry generation in Sanskrit typically requires the verse to be semantically coherent and adhere to strict prosodic rules. In Sanskrit prosody, every line of a verse is typically a fixed length sequence of syllables adhering to prescribed binary patterns of syllable weights. We observe that instead of treating a verse as a monolithic sequence, segmenting them as grouped-lines leads to significant improvement in semantic coherence by 10\% with comparable metrical adherence. Specifically, PINGALA, our proposed decoding approach is designed to encourage every line to have well-formed words and our token selection biases the model towards it by preferring longer tokens. Writing in Sanskrit follows phonemic orthography, hence using a phonetically aware transliteration scheme, SLP1, increased the metrical alignment by 46\% with comparable semantic similarity, for a instruction fine-tuned large language models like Phi-4. We also introduce a new approach for reference-free evaluation using cross-encoders which achieved better alignment with true poetry instances.

Virtual influencers~(VIs) -- digitally synthetic social-media personas -- attract audiences whose discourse appears qualitatively different from discourse around human influencers~(HIs). Existing work characterises this difference through surveys or aggregate engagement statistics, which reveal \emph{what} audiences say but not \emph{how} multiple signals co-occur. We propose a two-layer, structure-first framework grounded in Formal Concept Analysis~(FCA) and association rule mining. The first layer applies FCA with support-based iceberg filtering to weekly-aggregated comment data, extracting discourse profiles -- weekly co-occurrence bundles of sentiment, Big Five personality cues, and topic tags. The second layer mines association rules at the comment level, revealing personality--sentiment--topic dependencies invisible to frequency-table analysis. Applied to YouTube comments from three VI--HI influencer pairs, the two-layer analysis reveals a consistent structural divergence: HI discourse concentrates into a single, emotionally regulated (stability-centred) regime (low neuroticism anchoring positivity), while VI discourse supports three structurally distinct discourse modes, including an appearance-discourse cluster absent from HI despite near-equal marginal prevalence. Topic-specific analyses further show that VI contexts exhibit negative sentiment in psychologically sensitive domains (mental health, body image, artificial identity) relative to HI contexts. Our results position FCA as a principled tool for multi-signal discourse analysis and demonstrate that virtuality reshapes not just what audiences say, but the underlying grammar of how signals co-occur in their reactions.

Existing automated research systems operate as stateless, linear pipelines, generating outputs without maintaining a persistent understanding of the research landscape. They process papers sequentially, propose ideas without structured gap analysis, and lack mechanisms for agents to verify or refine each other's findings. We present AutoProf (Autonomous Professor), a multi-agent orchestration framework where specialized agents provide end-to-end AI research supervision driven by human interests, from literature review through gap discovery, method development, evaluation, and paper writing, via autonomous exploration and self-correcting updates. Unlike sequential pipelines, AutoProf maintains a continuously evolving Research World Model implemented as a Knowledge Graph, capturing methods, benchmarks, limitations, and unexplored gaps as shared memory across agents. The framework introduces three contributions: first, structured gap discovery that decomposes methods into modules, evaluates them across benchmarks, and identifies module-level gaps; second, self-correcting discovery loops that analyze why modules succeed or fail, detect benchmark biases, and assess evaluation adequacy; third, self-improving development loops using cross-domain mechanism search to iteratively address failing components. All agents operate under a consensus mechanism where findings are validated before being committed to the shared model. The framework is model-agnostic, supports mainstream large language models, and scales elastically with token budget from lightweight exploration to full-scale investigation.

We propose a neural network model for contextual regression in which the regression model depends on contextual features that determine the active submodel and an algorithm to fit the model. The proposed simple contextual neural network (SCtxtNN) separates context identification from context-specific regression, resulting in a structured and interpretable architecture with fewer parameters than a fully connected feed-forward network. We show mathematically that the proposed architecture is sufficient to represent contextual linear regression models using only standard neural network components. Numerical experiments are provided to support the theoretical result, showing that the proposed model achieves lower excess mean squared error and more stable performance than feed-forward neural networks with comparable numbers of parameters, while larger networks improve accuracy only at the cost of increased complexity. The results suggest that incorporating contextual structure can improve model efficiency while preserving interpretability.

One of the many fairness definitions pursued in recent recommender system research targets mitigating demographic information encoded in model representations. Models optimized for this definition are typically evaluated on how well demographic attributes can be classified given model representations, with the (implicit) assumption that this measure accurately reflects \textit{recommendation parity}, i.e., how similar recommendations given to different users are. We challenge this assumption by comparing the amount of demographic information encoded in representations with various measures of how the recommendations differ. We propose two new approaches for measuring how well demographic information can be classified given ranked recommendations. Our results from extensive testing of multiple models on one real and multiple synthetically generated datasets indicate that optimizing for fair representations positively affects recommendation parity, but also that evaluation at the representation level is not a good proxy for measuring this effect when comparing models. We also provide extensive insight into how recommendation-level fairness metrics behave for various models by evaluating their performances on numerous generated datasets with different properties.

Privacy and algorithmic fairness have become two central issues in modern machine learning. Although each has separately emerged as a rapidly growing research area, their joint effect remains comparatively under-explored. In this paper, we systematically study the joint impact of differential privacy and fairness on classification in a federated setting, where data are distributed across multiple servers. Targeting demographic disparity constrained classification under federated differential privacy, we propose a two-step algorithm, namely FDP-Fair. In the special case where there is only one server, we further propose a simple yet powerful algorithm, namely CDP-Fair, serving as a computationally-lightweight alternative. Under mild structural assumptions, theoretical guarantees on privacy, fairness and excess risk control are established. In particular, we disentangle the source of the private fairness-aware excess risk into a) intrinsic cost of classification, b) cost of private classification, c) non-private cost of fairness and d) private cost of fairness. Our theoretical findings are complemented by extensive numerical experiments on both synthetic and real datasets, highlighting the practicality of our designed algorithms.

High-quality teacher-child interaction (TCI) is fundamental to early childhood development, yet traditional expert-based assessment faces a critical scalability challenge. In large systems like China's-serving 36 million children across 250,000+ kindergartens-the cost and time requirements of manual observation make continuous quality monitoring infeasible, relegating assessment to infrequent episodic audits that limit timely intervention and improvement tracking. In this paper, we investigate whether AI can serve as a scalable assessment teammate by extracting structured quality indicators and validating their alignment with human expert judgments. Our contributions include: (1) TEPE-TCI-370h (Tracing Effective Preschool Education), the first large-scale dataset of naturalistic teacher-child interactions in Chinese preschools (370 hours, 105 classrooms) with standardized ECQRS-EC and SSTEW annotations; (2) We develop Interaction2Eval, a specialized LLM-based framework addressing domain-specific challenges-child speech recognition, Mandarin homophone disambiguation, and rubric-based reasoning-achieving up to 88% agreement; (3) Deployment validation across 43 classrooms demonstrating an 18x efficiency gain in the assessment workflow, highlighting its potential for shifting from annual expert audits to monthly AI-assisted monitoring with targeted human oversight. This work not only demonstrates the technical feasibility of scalable, AI-augmented quality assessment but also lays the foundation for a new paradigm in early childhood education-one where continuous, inclusive, AI-assisted evaluation becomes the engine of systemic improvement and equitable growth.

The theory of Local Intrinsic Dimensionality (LID) has become a valuable tool for characterizing local complexity within and across data manifolds, supporting a range of data mining and machine learning tasks. Accurate LID estimation requires samples drawn from small neighborhoods around each query to avoid biases from nonlocal effects and potential manifold mixing, yet limited data within such neighborhoods tends to cause high estimation variance. As a variance reduction strategy, we propose an ensemble approach that uses subbagging to preserve the local distribution of nearest neighbor (NN) distances. The main challenge is that the uniform reduction in total sample size within each subsample increases the proximity threshold for finding a fixed number k of NNs around the query. As a result, in the specific context of LID estimation, the sampling rate has an additional, complex interplay with the neighborhood size, where both combined determine the sample size as well as the locality and resolution considered for estimation. We analyze both theoretically and experimentally how the choice of the sampling rate and the k-NN size used for LID estimation, alongside the ensemble size, affects performance, enabling informed prior selection of these hyper-parameters depending on application-based preferences. Our results indicate that within broad and well-characterized regions of the hyper-parameters space, using a bagged estimator will most often significantly reduce variance as well as the mean squared error when compared to the corresponding non-bagged baseline, with controllable impact on bias. We additionally propose and evaluate different ways of combining bagging with neighborhood smoothing for substantial further improvements on LID estimation performance.

Despite deep learning's success in chemistry, its impact is hindered by a lack of interpretability and an inability to resolve activity cliffs, where minor structural nuances trigger drastic property shifts. Current representation learning, bound by the similarity principle, often fails to capture these structural-activity discontinuities. To address this, we introduce MolEvolve, an evolutionary framework that reformulates molecular discovery as an autonomous, look-ahead planning problem. Unlike traditional methods that depend on human-engineered features or rigid prior knowledge, MolEvolve leverages a Large Language Model (LLM) to actively explore and evolve a library of executable chemical symbolic operations. By utilizing the LLM to cold start and an Monte Carlo Tree Search (MCTS) engine for test-time planning with external tools (e.g. RDKit), the system self-discovers optimal trajectories autonomously. This process evolves transparent reasoning chains that translate complex structural transformations into actionable, human-readable chemical insights. Experimental results demonstrate that MolEvolve's autonomous search not only evolves transparent, human-readable chemical insights, but also outperforms baselines in both property prediction and molecule optimization tasks.

Evaluating the pedagogical quality of AI tutors remains challenging: standard NLG metrics do not determine whether responses identify mistakes, scaffold reasoning, or avoid revealing the answers. For the task of mistake remediation, we derive a hierarchy of pedagogical aspects from human pairwise preferences on MRBench, and synthesize minimally contrastive response pairs that differ along key aspects (e.g., mistake identification and location, targetedness, scaffolding, actionability, clarity, and coherence). We develop and release Bradley-Terry preference models trained on weighted-sum rankings that we automatically create from MRBench, synthetic pairs, and data combinations. Using only synthetic data, our best model reaches 0.69 pairwise accuracy on a human preference test, and combining weighted-sum data with targeted synthetic groups improves accuracy to 0.74, outperforming larger general-purpose reward models while using only a 0.5B-parameter backbone.

Autoformalization - automatically translating natural language mathematical texts into formal proof language such as Lean4 - can help accelerate AI-assisted mathematical research, be it via proof verification or proof search. I fine-tune Qwen3.5-2B with LoRA for natural language to Lean4 formalization on FineLeanCorpus and consider three training regimes: supervised fine-tuning (SFT) with curriculum learning (difficulty 1 to 10), SFT without curriculum ordering, and reinforcement learning using group relative policy optimization (GRPO) with a cycle consistency reward. Cycle consistency measures how well the meaning of a statement is preserved through a NL to Lean4 to NL' loop, computed as cosine similarity of off-the-shelf sentence embeddings. On an unseen subset of FineLeanCorpus (FLC) and on PutnamBench, RL substantially outperforms both SFT variants (mean cycle consistency 0.669 vs. 0.513 on FLC; 0.561 vs. 0.422 on PutnamBench), while increasing cross-entropy loss by only 0.011 nats, with minimal impact on formalization quality. Curriculum ordering provides no measurable benefit over shuffled training.

This paper addresses the Service Network Design (SND) problem for a logistics service provider (LSP) operating in a multimodal freight transport network, considering uncertain travel times and limited truck fleet availability. A two-stage optimization approach is proposed, which combines metaheuristics, simulation and machine learning components. This solution framework integrates tactical decisions, such as transport request acceptance and capacity booking for scheduled services, with operational decisions, including dynamic truck allocation, routing, and re-planning in response to disruptions. A simulated annealing (SA) metaheuristic is employed to solve the tactical problem, supported by an adaptive surrogate model trained using a discrete-event simulation model that captures operational complexities and cascading effects of uncertain travel times. The performance of the proposed method is evaluated using benchmark instances. First, the SA is tested on a deterministic version of the problem and compared to state-of-the-art results, demonstrating it can improve the solution quality and significantly reduce the computational time. Then, the proposed SA is applied to the more complex stochastic problem. Compared to a benchmark algorithm that executes a full simulation for each solution evaluation, the learning-based SA generates high quality solutions while significantly reducing computational effort, achieving only a 5% difference in objective function value while cutting computation time by up to 20 times. These results demonstrate the strong performance of the proposed algorithm in solving complex versions of the SND. Moreover, they highlight the effectiveness of integrating diverse modeling and optimization techniques, and the potential of such approaches to efficiently address freight transport planning challenges.

Adaptive traffic signal control (ATSC) is crucial in alleviating congestion, maximizing throughput and promoting sustainable mobility in ever-expanding cities. Multi-Agent Reinforcement Learning (MARL) has recently shown significant potential in addressing complex traffic dynamics, but the intricacies of partial observability and coordination in decentralized environments still remain key challenges in formulating scalable and efficient control strategies. To address these challenges, we present CoordLight, a MARL-based framework designed to improve intra-neighborhood traffic by enhancing decision-making at individual junctions (agents), as well as coordination with neighboring agents, thereby scaling up to network-level traffic optimization. Specifically, we introduce the Queue Dynamic State Encoding (QDSE), a novel state representation based on vehicle queuing models, which strengthens the agents' capability to analyze, predict, and respond to local traffic dynamics. We further propose an advanced MARL algorithm, named Neighbor-aware Policy Optimization (NAPO). It integrates an attention mechanism that discerns the state and action dependencies among adjacent agents, aiming to facilitate more coordinated decision-making, and to improve policy learning updates through robust advantage calculation. This enables agents to identify and prioritize crucial interactions with influential neighbors, thus enhancing the targeted coordination and collaboration among agents. Through comprehensive evaluations against state-of-the-art traffic signal control methods over three real-world traffic datasets composed of up to 196 intersections, we empirically show that CoordLight consistently exhibits superior performance across diverse traffic networks with varying traffic flows. The code is available at https://github.com/marmotlab/CoordLight

LLMs are increasingly embedded in programming workflows, from code generation to automated code review. Yet, how gendered communication styles interact with LLM-assisted programming and code review remains underexplored. We present a mixed-methods pilot study examining whether gender-related linguistic differences in prompts influence code generation outcomes and code review decisions. Across three complementary studies, we analyze (i) collected real-world coding prompts, (ii) a controlled user study, in which developers solve identical programming tasks with LLM assistance, and (iii) an LLM-based simulated evaluation framework that systematically varies gender-coded prompt styles and reviewer personas. We find that gender-related differences in prompting style are subtle but measurable, with female-authored prompts exhibiting more indirect and involved language, which does not translate into consistent gaps in functional correctness or static code quality. For LLM code review, in contrast, we observe systematic biases: on average, models approve female-authored code more, despite comparable quality. Controlled experiments show that gender-coded prompt style affect code length and maintainability, while reviewer behavior varies across models. Our findings suggest that fairness risks in LLM-assisted programming arise less from generation accuracy than from LLM evaluation, as LLMs are increasingly deployed as automated code reviewers.

We propose a neuro-symbolic architecture that learns four interpretable physiological concepts, oculomotor dynamics, gaze stability, prefrontal hemodynamics, and multimodal, from eye-tracking and neural hemodynamics, functional near-infrared spectroscopy, (fNIRS) windows using attention-based encoders, and combines them with differentiable approximate reasoning rules using learned weights and soft thresholds, to address both rigid hand-crafted rules and the lack of subject-level alignment diagnostics. We apply this system to fatigue classification from multimodal physiological signals, a domain that requires models that are accurate and interpretable, with internal reasoning that can be inspected for safety-critical use. In leave-one-subject-out evaluation on 18 participants (560 samples), the method achieves 72.1% +/- 12.3% accuracy, comparable to tuned baselines while exposing concept activations and rule firing strengths. Ablations indicate gains from participant-specific calibration (+5.2 pp), a modest drop without the fNIRS concept (-1.2 pp), and slightly better performance with Lukasiewicz operators than product (+0.9 pp). We also introduce concept fidelity, an offline per-subject audit metric from held-out labels, which correlates strongly with per-subject accuracy (r=0.843, p < 0.0001).

Camouflaged Object Detection (COD) aims to segment objects that are highly integrated with the background in terms of color, texture, and structure, making it a highly challenging task in computer vision. Although existing methods introduce multi-scale fusion and attention mechanisms to alleviate the above issues, they generally lack the guidance of textual semantic priors, which limits the model's ability to focus on camouflaged regions in complex scenes. To address this issue, this paper proposes a Language-Guided Structure-Aware Network (LGSAN). Specifically, based on the visual backbone PVT-v2, we introduce CLIP to generate masks from text prompts and RGB images, thereby guiding the multi-scale features extracted by PVT-v2 to focus on potential target regions. On this foundation, we further design a Fourier Edge Enhancement Module (FEEM), which integrates multi-scale features with high-frequency information in the frequency domain to extract edge enhancement features. Furthermore, we propose a Structure-Aware Attention Module (SAAM) to effectively enhance the model's perception of object structures and boundaries. Finally, we introduce a Coarse-Guided Local Refinement Module (CGLRM) to enhance fine-grained reconstruction and boundary integrity of camouflaged object regions. Extensive experiments demonstrate that our method consistently achieves highly competitive performance across multiple COD datasets, validating its effectiveness and robustness.

A key challenge to understanding self-awareness has been a principled way of quantifying whether an intelligent system has a concept of a "self," and if so how to differentiate the "self" from other cognitive structures. We propose that the "self" can be isolated by seeking the invariant portion of cognitive process that changes relatively little compared to more rapidly acquired cognitive knowledge and skills, because our self is the most persistent aspect of our experiences. We used this principle to analyze the cognitive structure of robots under two conditions: One robot learns a constant task, while a second robot is subjected to continual learning under variable tasks. We find that robots subjected to continual learning develop an invariant subnetwork that is significantly more stable (p < 0.001) compared to the control. We suggest that this principle can offer a window into exploring selfhood in other cognitive AI systems.

Current audio deepfake detection has achieved remarkable performance using diverse deep learning architectures such as ResNet, and has seen further improvements with the introduction of large models (LMs) like Wav2Vec. The success of large language models (LLMs) further demonstrates the benefits of scaling model parameters, but also highlights one bottleneck where performance gains are constrained by parameter counts. Simply stacking additional layers, as done in current LLMs, is computationally expensive and requires full retraining. Furthermore, existing low-rank adaptation methods are primarily applied to attention-based architectures, which limits their scope. Inspired by the neuronal plasticity observed in mammalian brains, we propose novel algorithms, dropin and further plasticity, that dynamically adjust the number of neurons in certain layers to flexibly modulate model parameters. We evaluate these algorithms on multiple architectures, including ResNet, Gated Recurrent Neural Networks, and Wav2Vec. Experimental results using the widely recognised ASVSpoof2019 LA, PA, and FakeorReal dataset demonstrate consistent improvements in computational efficiency with the dropin approach and a maximum of around 39% and 66% relative reduction in Equal Error Rate with the dropin and plasticity approach among these dataset, respectively. The code and supplementary material are available at Github link.

Interleaved ADCs are critical for applications requiring multi-gigasample per second (GS/s) rates, but their performance is often limited by offset, gain, and timing skew mismatches across the sub-ADCs. We propose exact but compact expressions that describe the impact of each of those non-idealities on the output spectrum. We derive the distribution of the power of the induced spurs and replicas, critical for yield-oriented derivation of sub-ADC specifications. Finally, we provide a practical example in which calibration step sizes are derived under the constraint of a target production yield.

Multimodal LLMs are increasingly deployed as perceptual backbones for autonomous agents in 3D environments, from robotics to virtual worlds. These applications require agents to perceive rapid state changes, attribute actions to the correct entities, and reason about concurrent multi-agent behaviors from a first-person perspective, capabilities that existing benchmarks do not adequately evaluate. We introduce GameplayQA, a framework for evaluating agentic-centric perception and reasoning through video understanding. Specifically, we densely annotate multiplayer 3D gameplay videos at 1.22 labels/second, with time-synced, concurrent captions of states, actions, and events structured around a triadic system of Self, Other Agents, and the World, a natural decomposition for multi-agent environments. From these annotations, we refined 2.4K diagnostic QA pairs organized into three levels of cognitive complexity, accompanied by a structured distractor taxonomy that enables fine-grained analysis of where models hallucinate. Evaluation of frontier MLLMs reveals a substantial gap from human performance, with common failures in temporal and cross-video grounding, agent-role attribution, and handling the decision density of the game. We hope GameplayQA stimulates future research at the intersection of embodied AI, agentic perception, and world modeling.

Document parsing is a fine-grained task where image resolution significantly impacts performance. While advanced research leveraging vision-language models benefits from high-resolution input to boost model performance, this often leads to a quadratic increase in the number of vision tokens and significantly raises computational costs. We attribute this inefficiency to substantial visual regions redundancy in document images, like background. To tackle this, we propose PaddleOCR-VL, a novel coarse-to-fine architecture that focuses on semantically relevant regions while suppressing redundant ones, thereby improving both efficiency and performance. Specifically, we introduce a lightweight Valid Region Focus Module (VRFM) which leverages localization and contextual relationship prediction capabilities to identify valid vision tokens. Subsequently, we design and train a compact yet powerful 0.9B vision-language model (PaddleOCR-VL-0.9B) to perform detailed recognition, guided by VRFM outputs to avoid direct processing of the entire large image. Extensive experiments demonstrate that PaddleOCR-VL achieves state-of-the-art performance in both page-level parsing and element-level recognition. It significantly outperforms existing solutions, exhibits strong competitiveness against top-tier VLMs, and delivers fast inference while utilizing substantially fewer vision tokens and parameters, highlighting the effectiveness of targeted coarse-to-fine parsing for accurate and efficient document understanding. The source code and models are publicly available at https://github.com/PaddlePaddle/PaddleOCR.

Designing effective auxiliary rewards for cooperative multi-agent systems remains a precarious task; misaligned incentives risk inducing suboptimal coordination, especially where sparse task feedback fails to provide sufficient grounding. This study introduces an automated reward design framework that leverages large language models to synthesize executable reward programs from environment instrumentation. The procedure constrains candidate programs within a formal validity envelope and evaluates their efficacy by training policies from scratch under a fixed computational budget; selection depends exclusively on the sparse task return. The framework is evaluated across four distinct Overcooked-AI layouts characterized by varied corridor congestion, handoff dependencies, and structural asymmetries. Iterative search generations consistently yield superior task returns and delivery counts, with the most pronounced gains occurring in environments dominated by interaction bottlenecks. Diagnostic analysis of the synthesized shaping components indicates increased interdependence in action selection and improved signal alignment in coordination-intensive tasks. These results demonstrate that the search for objectivegrounded reward programs can mitigate the burden of manual engineering while producing shaping signals compatible with cooperative learning under finite budgets.

We present an innovative, cost-effective framework integrating laboratory Hyperspectral Imaging (HSI) of the Bechar010 Lunar meteorite with ground-based lunar HSI and supervised Machine Learning(ML) to generate high-fidelity mineralogical maps. A 3mm thin section of Bechar010 was imaged under a microscope with a 30mm focal length lens at 150mm working distance, using 6x binning to increase the signal-to-noise ratio, producing a data cube (X $\times$ Y $\times$ $λ$ = $791 \times 1024 \times 224$, 0.24mm $\times$ 0.2mm resolution) across 400-1000}nm (224 bands, 2.7nm spectral sampling, 5.5nm full width at half maximum spectral resolution) using a Specim FX10 camera. Ground-based lunar HSI was captured with a Celestron 8SE telescope (3km/pixel), yielded a data cube ($371 \times 1024 \times 224$). Solar calibration was performed using a Spectralon reference ({99}\% reflectance {<2}\% error) ensured accurate reflectance spectra. A Support Vector Machine (SVM) with a radial basis function kernel, trained on expert-labeled spectra, achieved {93.7}\% classification accuracy(5-fold cross-validation) for olivine ({92}\% precision, {90}\% recall) and pyroxene ({88}\% precision, {86}{\%} recall) in Bechar 010. LIME analysis identified key wavelengths (e.g., 485nm, {22.4}\% for M3; 715nm, {20.6}\% for M6) across 10 pre-selected regions (M1 to M10), indicating olivine-rich (Highland-like) and pyroxene-rich (Mare-like) compositions. SAM analysis revealed angles from 0.26 radian to 0.66 radian, linking M3 and M9 to Highlands and M6 and M10 to Mares. K-means clustering of Lunar data identified 10 mineralogical clusters ({88}\% accuracy), validated against Chandrayaan-1 Moon mineralogy Mapper ($\rm M^3$) data (140m/pixel, 10nm spectral resolution).A novel push-broom HSI approach with a telescope achieves 0.8 arcsec resolution for lunar spectroscopy, inspiring full-sky multi-object spectral mapping.

State-of-the-art generalist manipulation policies have enabled the deployment of robotic manipulators in unstructured human environments. However, these frameworks struggle in cluttered environments primarily because they utilize auxiliary modules for low-level motion planning and control. Motion planning remains challenging due to the high dimensionality of the robot's configuration space and the presence of workspace obstacles. Neural motion planners have enhanced motion planning efficiency by offering fast inference and effectively handling the inherent multi-modality of the motion planning problem. Despite such benefits, current neural motion planners often struggle to generalize to unseen, out-of-distribution planning settings. This paper reviews and analyzes the state-of-the-art neural motion planners, highlighting both their benefits and limitations. It also outlines a path toward establishing generalist neural motion planners capable of handling domain-specific challenges. For a list of the reviewed papers, please refer to https://davoodsz.github.io/planning-manip-survey.github.io/.

We release Samasāmayik, a novel, meticulously curated, large-scale Hindi-Sanskrit corpus, comprising 92,196 parallel sentences. Unlike most data available in Sanskrit, which focuses on classical era text and poetry, this corpus aggregates data from diverse sources covering contemporary materials, including spoken tutorials, children's magazines, radio conversations, and instruction materials. We benchmark this new dataset by fine-tuning three complementary models - ByT5, NLLB and IndicTrans-v2, to demonstrate its utility. Our experiments demonstrate that models trained on the Samasamayik corpus achieve significant performance gains on in-domain test data, while achieving comparable performance on other widely used test sets, establishing a strong new performance baseline for contemporary Hindi-Sanskrit translation. Furthermore, a comparative analysis against existing corpora reveals minimal semantic and lexical overlap, confirming the novelty and non-redundancy of our dataset as a robust new resource for low-resource Indic language MT.

Graph Neural Networks (GNNs) have achieved impressive performance in graph-related tasks. However, they suffer from poor generalization on out-of-distribution (OOD) data, as they tend to learn spurious correlations. Such correlations present a phenomenon that GNNs fail to stably learn the mutual information between prediction representations and ground-truth labels under OOD settings. To address these challenges, we formulate a causal graph starting from the essence of node classification, adopt backdoor adjustment to block non-causal paths, and theoretically derive a lower bound for improving OOD generalization of GNNs. To materialize these insights, we further propose a novel approach integrating causal representation learning and a loss replacement strategy. The former captures node-level causal invariance and reconstructs graph posterior distribution. The latter introduces asymptotic losses of the same order to replace the original losses. Extensive experiments demonstrate the superiority of our method in OOD generalization and effectively alleviating the phenomenon of unstable mutual information learning.

The ground state search problem is central to quantum computing, with applications spanning quantum chemistry, condensed matter physics, and optimization. The Variational Quantum Eigensolver (VQE) has shown promise for small systems but faces significant limitations. These include barren plateaus, restricted ansatz expressivity, and reliance on domain-specific structure. We present SpinGQE, an extension of the Generative Quantum Eigensolver (GQE) framework to spin Hamiltonians. Our approach reframes circuit design as a generative modeling task. We employ a transformer-based decoder to learn distributions over quantum circuits that produce low-energy states. Training is guided by a weighted mean-squared error loss between model logits and circuit energies evaluated at each gate subsequence. We validate our method on the four-qubit Heisenberg model, demonstrating successfulconvergencetonear-groundstates. Throughsystematichyperparameterexploration, we identify optimal configurations: smaller model architectures (12 layers, 8 attention heads), longer sequence lengths (12 gates), and carefully chosen operator pools yield the most reliable convergence. Our results show that generative approaches can effectively navigate complex energy landscapes without relying on problem-specific symmetries or structure. This provides a scalable alternative to traditional variational methods for general quantum systems. An open-source implementation is available at https://github.com/Mindbeam-AI/SpinGQE.

Recent work distinguishes two heterophily regimes: adversarial, where cross-class edges dilute class signal and harm classification, and informative, where the heterophilous structure itself carries useful signal. We ask: when does per-edge message routing help, and when is a uniform spectral channel sufficient? To operationalize this question we introduce Cost-Sensitive Neighborhood Aggregation (CSNA), a GNN layer that computes pairwise distance in a learned projection and uses it to soft-route each message through concordant and discordant channels with independent transformations. Under a contextual stochastic block model we show that cost-sensitive weighting preserves class-discriminative signal where mean aggregation provably attenuates it, provided $w_+/w_- > q/p$. On six benchmarks with uniform tuning, CSNA is competitive with state-of-the-art methods on adversarial-heterophily datasets (Texas, Wisconsin, Cornell, Actor) but underperforms on informative-heterophily datasets (Chameleon, Squirrel) -- precisely the regime where per-edge routing has no useful decomposition to exploit. The pattern is itself the finding: the cost function's ability to separate edge types serves as a diagnostic for the heterophily regime, revealing when fine-grained routing adds value over uniform channels and when it does not. Code is available at https://github.com/eyal-weiss/CSNA-public .

When multiple LLM-based code agents independently implement parts of the same class, they must agree on shared internal representations, even when the specification leaves those choices implicit. We study this coordination problem across 51 class-generation tasks, progressively stripping specification detail from full docstrings (L0) to bare signatures (L3), and introducing opposing structural biases (lists vs. dictionaries) to stress-test integration. Three findings emerge. First, a persistent specification gap: two-agent integration accuracy drops from 58% to 25% as detail is removed, while a single-agent baseline degrades more gracefully (89% to 56%), leaving a 25--39 pp coordination gap that is consistent across two Claude models (Sonnet, Haiku) and three independent runs. Second, an AST-based conflict detector achieves 97% precision at the weakest specification level without additional LLM calls, yet a factorial recovery experiment shows that restoring the full specification alone recovers the single-agent ceiling (89%), while providing conflict reports adds no measurable benefit. Third, decomposing the gap into coordination cost (+16 pp) and information asymmetry (+11 pp) suggests that the two effects are independent and approximately additive. The gap is not merely a consequence of hidden information, but reflects the difficulty of producing compatible code without shared decisions. These results support a specification-first view of multi-agent code generation: richer specifications are both the primary coordination mechanism and the sufficient recovery instrument.

Despite the advancements in cutting-edge technologies, audio signal processing continues to pose challenges and lacks the precision of a human speech processing system. To address these challenges, we propose a novel approach to simplify audio signal processing by leveraging time-domain techniques and reservoir computing. Through our research, we have developed a real-time audio signal processing system by simplifying audio signal processing through the utilization of reservoir computers, which are significantly easier to train. Feature extraction is a fundamental step in speech signal processing, with Mel Frequency Cepstral Coefficients (MFCCs) being a dominant choice due to their perceptual relevance to human hearing. However, conventional MFCC extraction relies on computationally intensive time-frequency transformations, limiting efficiency in real-time applications. To address this, we propose a novel approach that leverages reservoir computing to streamline MFCC extraction. By replacing traditional frequency-domain conversions with convolution operations, we eliminate the need for complex transformations while maintaining feature discriminability. We present an end-to-end audio processing framework that integrates this method, demonstrating its potential for efficient and real-time speech analysis. Our results contribute to the advancement of energy-efficient audio processing technologies, enabling seamless deployment in embedded systems and voice-driven applications. This work bridges the gap between biologically inspired feature extraction and modern neuromorphic computing, offering a scalable solution for next-generation speech recognition systems.

Modern software systems heavily rely on third-party dependencies, making software supply chain security a critical concern. We introduce the concept of software supply chain smells as structural indicators that signal potential security risks. We design and evaluate Dirty-Waters, a novel tool for detecting such smells in the supply chains of software packages. Through interviews with practitioners, we show that our proposed smells align with real-world concerns and capture signals considered valuable. A quantitative study of popular packages in the Maven and NPM ecosystems reveals that while smells are prevalent in both, they differ significantly across ecosystems, with traceability and signing issues dominating in Maven and most smells being rare in NPM, due to strong registry-level guarantees. Software supply chain smells support developers and organizations in making informed decisions and improving their software supply chain security posture.

Language-Assisted Image Clustering (LAIC) augments the input images with additional texts with the help of vision-language models (VLMs) to promote clustering performance. Despite recent progress, existing LAIC methods often overlook two issues: (i) textual features constructed for each image are highly similar, leading to weak inter-class discriminability; (ii) the clustering step is restricted to pre-built image-text alignments, limiting the potential for better utilization of the text modality. To address these issues, we propose a new LAIC framework with two complementary components. First, we exploit cross-modal relations to produce more discriminative self-supervision signals for clustering, as it compatible with most VLMs training mechanisms. Second, we learn category-wise continuous semantic centers via prompt learning to produce the final clustering assignments. Extensive experiments on eight benchmark datasets demonstrate that our method achieves an average improvement of 2.6% over state-of-the-art methods, and the learned semantic centers exhibit strong interpretability. Code is available in the supplementary material.

Cancer drug response varies widely across tumors due to multi-layer molecular heterogeneity, motivating computational decision support for precision oncology. Despite recent progress in deep CDR models, robust alignment between high-dimensional multi-omics and chemically structured drugs remains challenging due to cross-modal misalignment and limited inductive bias. We present DeepDTF, an end-to-end dual-branch Transformer fusion framework for joint log(IC50) regression and drug sensitivity classification. The cell-line branch uses modality-specific encoders for multi-omics profiles with Transformer blocks to capture long-range dependencies, while the drug branch represents compounds as molecular graphs and encodes them with a GNN-Transformer to integrate local topology with global context. Omics and drug representations are fused by a Transformer-based module that models cross-modal interactions and mitigates feature misalignment. On public pharmacogenomic benchmarks under 5-fold cold-start cell-line evaluation, DeepDTF consistently outperforms strong baselines across omics settings, achieving up to RMSE=1.248, R^2=0.875, and AUC=0.987 with full multi-omics inputs, while reducing classification error (1-ACC) by 9.5%. Beyond accuracy, DeepDTF provides biologically grounded explanations via SHAP-based gene attributions and pathway enrichment with pre-ranked GSEA.

Nowadays, time series forecasting is predominantly approached through the end-to-end training of deep learning architectures using error-based objectives. While this is effective at minimizing average loss, it encourages the encoder to discard informative yet extreme patterns. This results in smooth predictions and temporal representations that poorly capture salient dynamics. To address this issue, we propose ReGuider, a plug-in method that can be seamlessly integrated into any forecasting architecture. ReGuider leverages pretrained time series foundation models as semantic teachers. During training, the input sequence is processed together by the target forecasting model and the pretrained model. Rather than using the pretrained model's outputs directly, we extract its intermediate embeddings, which are rich in temporal and semantic information, and align them with the target model's encoder embeddings through representation-level supervision. This alignment process enables the encoder to learn more expressive temporal representations, thereby improving the accuracy of downstream forecasting. Extensive experimentation across diverse datasets and architectures demonstrates that our ReGuider consistently improves forecasting performance, confirming its effectiveness and versatility.

Diffusion-based video editing has emerged as an important paradigm for high-quality and flexible content generation. However, despite their generality and strong modeling capacity, Diffusion Transformers (DiT) remain computationally expensive due to the iterative denoising process, posing challenges for practical deployment. Existing video diffusion acceleration methods primarily exploit denoising timestep-level feature reuse, which mitigates the redundancy in denoising process, but overlooks the architectural redundancy within the DiT that many attention operations over spatio-temporal tokens are redundantly executed, offering little to no incremental contribution to the model output. This work introduces HetCache, a training-free diffusion acceleration framework designed to exploit the inherent heterogeneity in diffusion-based masked video-to-video (MV2V) generation and editing. Instead of uniformly reuse or randomly sampling tokens, HetCache assesses the contextual relevance and interaction strength among various types of tokens in designated computing steps. Guided by spatial priors, it divides the spatial-temporal tokens in DiT model into context and generative tokens, and selectively caches the context tokens that exhibit the strongest correlation and most representative semantics with generative ones. This strategy reduces redundant attention operations while maintaining editing consistency and fidelity. Experiments show that HetCache achieves a noticeable acceleration, including a 2.67$\times$ latency speedup and FLOPs reduction over commonly used foundation models, with negligible degradation in editing quality.

We study word-level semantic alignment across four historical stages of Ancient Egyptian. These stages differ in script and orthography, and parallel data are scarce. We jointly train a compact encoder-decoder model with a shared byte-level tokenizer on all four stages, combining masked language modeling (MLM), translation language modeling (TLM), sequence-to-sequence translation, and part-of-speech tagging under a task-aware loss with fixed weights and uncertainty-based scaling. To reduce surface divergence we add Latin transliteration and IPA reconstruction as auxiliary views. We integrate these views through KL-based consistency and through embedding-level fusion. We evaluate alignment quality using pairwise metrics, specifically ROC-AUC and triplet accuracy, on curated Egyptian-English and intra-Egyptian cognate datasets. Translation yields the strongest gains. IPA with KL consistency improves cross-branch alignment, while early fusion demonstrates limited efficacy. Although the overall alignment remains limited, the findings provide a reproducible baseline and practical guidance for modeling historical languages under real constraints. They also show how normalization and task design shape what counts as alignment in typologically distant settings.

Probabilistic time series forecasting (PTSF) aims to model the full predictive distribution of future observations, enabling both accurate forecasting and principled uncertainty quantification. A central requirement of PTSF is to embrace heteroscedasticity, as real-world time series exhibit time-varying conditional variances induced by nonstationary dynamics, regime changes, and evolving external conditions. However, most existing non-autoregressive generative approaches to PTSF, such as TimeVAE and $K^2$VAE, rely on MSE-based training objectives that implicitly impose a homoscedastic assumption, thereby fundamentally limiting their ability to model temporal heteroscedasticity. To address this limitation, we propose the Location-Scale Gaussian VAE (LSG-VAE), a simple but effective framework that explicitly parameterizes both the predictive mean and time-dependent variance through a location-scale likelihood formulation. This design enables LSG-VAE to faithfully capture heteroscedastic aleatoric uncertainty and introduces an adaptive attenuation mechanism that automatically down-weights highly volatile observations during training, leading to improved robustness in trend prediction. Extensive experiments on nine benchmark datasets demonstrate that LSG-VAE consistently outperforms fifteen strong generative baselines while maintaining high computational efficiency suitable for real-time deployment.

Centralized identity management systems continuously experience security and privacy challenges, motivating the exploration of Decentralized Identity (DI) and Self-Sovereign Identity (SSI) as alternatives. Despite privacy and security benefits to users, the adoption of DI/SSI systems remains limited. One contributing reason is the lack of reproducible approaches to evaluate system compliance with its promised qualities. Derivation of functional requirements (FR) is the first and necessary step to develop such an evaluation approach. Previous literature on DI/SSI significantly lacks the systematic operationalization of existing non-functional requirements (NFR) or SSI principles. This work addresses this research gap by deriving FR for a generalized DI/SSI use case, which encompasses the fundamental operations of the system. The paper details operationalization methodology, introduces a formalized functional model, and presents a comprehensive set of FR, that can be used for future development and evaluation of DI/SSI systems. As a result, establishing the fundamental step toward a reproducible evaluation framework, rooted in established requirements engineering methods.

We propose \geogap{}, a geometric method for detecting missing requirement types in software specifications. The method represents each requirement as a unit vector via a pretrained sentence encoder, then measures coverage deficits through $k$-nearest-neighbour distances z-scored against per-project baselines. Three complementary scoring components -- per-point geometric coverage, type-restricted distributional coverage, and annotation-free population counting -- fuse into a unified gap score controlled by two hyperparameters. On the PROMISE NFR benchmark, \geogap{} achieves 0.935 AUROC for detecting completely absent requirement types in projects with $N \geq 50$ requirements, matching a ground-truth count oracle that requires human annotation. Six baselines confirm that each pipeline component -- per-project normalisation, neural embeddings, and geometric scoring -- contributes measurable value.

This paper describes the system submitted to the SOMD 2026 Shared Task for Cross-Document Coreference Resolution (CDCR) of software mentions. Our approach addresses the challenge of identifying and clustering inconsistent software mentions across scientific corpora. We propose a hybrid framework that combines dense semantic embeddings from a pre-trained Sentence-BERT model, Knowledge Base (KB) lookup strategy built from training-set cluster centroids using FAISS for efficient retrieval, and HDBSCAN density-based clustering for mentions that cannot be confidently assigned to existing clusters. Surface-form normalization and abbreviation resolution are applied to improve canonical name matching. The same core pipeline is applied to Subtasks 1 and 2. To address the large scale settings of Subtask 3, the pipeline was adapted by utilising a blocking strategy based on entity types and canonicalized surface forms. Our system achieved CoNLL F1 scores of 0.98, 0.98, and 0.96 on Subtasks 1, 2, and 3 respectively.

Federated Learning (FL) of Large Language Models (LLMs) in multilingual environments presents significant challenges stemming from heterogeneous language distributions across clients and disparities in language resource availability. To address these challenges, we extended the FederatedScope-LLM framework to support multilingual instruction-tuning experiments with LLMs. We also introduced a novel client-specific early stopping mechanism, Local Dynamic Early Stopping (LDES-FL), which allows clients to pause and resume local training based on client-side validation performance, enhancing training efficiency and sustainability. Through a series of experiments, we studied how client language composition - from fully monolingual to increasingly multilingual clients - affects multilingual quality, fairness and training cost. Monolingual local fine-tuning remains the most effective for single-language specialization, whereas federated training is better suited to learning a single balanced multilingual model. In FL, increasing within-client multilinguality leads to stronger and fairer global models, narrows the gap to centralized multilingual fine-tuning, and yields the largest gains for lower-resource languages, albeit at the cost of more optimization steps. Overall, our results identify client language composition as a key design variable in multilingual FL, shaping performance, fairness and efficiency

Safe navigation in complex environments remains a central challenge for reinforcement learning (RL) in robotics. This paper introduces Continuous Space-Time Empowerment for Physics-informed (C-STEP) safe RL, a novel measure of agent-centric safety tailored to deterministic, continuous domains. This measure can be used to design physics-informed intrinsic rewards by augmenting positive navigation reward functions. The reward incorporates the agents internal states (e.g., initial velocity) and forward dynamics to differentiate safe from risky behavior. By integrating C-STEP with navigation rewards, we obtain an intrinsic reward function that jointly optimizes task completion and collision avoidance. Numerical results demonstrate fewer collisions, reduced proximity to obstacles, and only marginal increases in travel time. Overall, C-STEP offers an interpretable, physics-informed approach to reward shaping in RL, contributing to safety for agentic mobile robotic systems.

Dynamism is common in AI computation, e.g., the dynamic tensor shapes and the dynamic control flows in models. Due to the long compilation time, existing runtime compilation damages the model efficiency, while the offline compilers either suffer from the long compilation time and device memory footprint to cover all the possible execution instances of a dynamic model, or sacrifice optimization opportunities for usability. In this paper, we rethink the feasibility of runtime compilation for dynamic models and identify that the key for it to work is to speed up the compilation or hide the compilation overhead. To do this, we propose a real-time compiler, DVM. In DVM, we design a runtime operator compiler based on a bytecode virtual machine to perform effective and efficient compilation for each dynamic operator instance given its input. Specifically, instead of compiling programs into machine code, we encode the operator program into bytecode on the CPU and decode the bytecode into virtual instructions for direct execution on the NPU. Based on the runtime operator compiler, we further propose an operator fuser, which performs symbol-deduction-based fusion on static graphs and runtime fusion on dynamic graphs. Both pattern- and stacking-based fusion are supported to increase fusion opportunities. Evaluation on operators, subgraphs, and models shows that, compared with TorchInductor, PyTorch-eager and MindSpore-graph-O0, we are up to 11.77$\times$ better in terms of the operator/model efficiency and up to 5 orders of magnitude faster in terms of the maximum compilation time.

Machine learning models trained on small data sets for security applications are especially vulnerable to adversarial attacks. Person identification from LiDAR based skeleton data requires time consuming and expensive data acquisition for each subject identity. Recently, Assessment and Augmented Identity Recognition for Skeletons (AAIRS) has been used to train Hierarchical Co-occurrence Networks for Person Identification (HCN-ID) with small LiDAR based skeleton data sets. However, AAIRS does not evaluate robustness of HCN-ID to adversarial attacks or inoculate the model to defend against such attacks. Popular perturbation-based approaches to generating adversarial attacks are constrained to targeted perturbations added to real training samples, which is not ideal for inoculating models with small training sets. Thus, we propose Attack-AAIRS, a novel addition to the AAIRS framework. Attack-AAIRS leverages a small real data set and a GAN generated synthetic data set to assess and improve model robustness against unseen adversarial attacks. Rather than being constrained to perturbations of limited real training samples, the GAN learns the distribution of adversarial attack samples that exploit weaknesses in HCN-ID. Attack samples drawn from this distribution augment training for inoculation of the HCN-ID to improve robustness. Ten-fold cross validation of Attack-AAIRS yields increased robustness to unseen attacks- including FGSM, PGD, Additive Gaussian Noise, MI-FGSM, and BIM. The HCN-ID Synthetic Data Quality Score for Attack-AAIRS indicates that generated attack samples are of similar quality to the original benign synthetic samples generated by AAIRS. Furthermore, inoculated models show consistent final test accuracy with the original model trained on real data, demonstrating that our method improves robustness to adversarial attacks without reducing test performance on real data.

Stance detection is nearly always formulated as classifying text into Favor, Against, or Neutral -- a convention inherited from debate analysis and applied without modification to social media since SemEval-2016. But attitudes toward complex targets are not unitary: a person can accept climate science while opposing carbon taxes, expressing support on one dimension and opposition on another. When annotators must compress such multi-dimensional attitudes into a single label, different annotators weight different dimensions -- producing disagreement that reflects not confusion but different compression choices. We call this the \textbf{projection problem}, and show that its cost is conditional: when a text's dimensions align, any weighting yields the same label and three-way annotation works well; when dimensions conflict, label agreement collapses while agreement on individual dimensions remains intact. A pilot study on SemEval-2016 Task 6 confirms this crossover: on dimension-consistent texts, label agreement (Krippendorff's $α= 0.307$) exceeds dimensional agreement ($α= 0.082$); on dimension-conflicting texts, the pattern reverses -- label $α$ drops to $0.085$ while dimensional $α$ rises to $0.334$, with Policy reaching $0.572$. The projection problem is real -- but it activates precisely where it matters most.

In nonnegative matrix factorization (NMF), minimum-volume-constrained NMF is a widely used framework for identifying the solution of NMF by making basis vectors as similar as possible. This typically induces sparsity in the coefficient matrix, with each row containing zero entries. Consequently, minimum-volume-constrained NMF may fail for highly mixed data, where such sparsity does not hold. Moreover, the estimated basis vectors in minimum-volume-constrained NMF may be difficult to interpret as they may be mixtures of the ground truth basis vectors. To address these limitations, in this paper we propose a new NMF framework, called maximum-volume-constrained NMF, which makes the basis vectors as distinct as possible. We further establish an identifiability theorem for maximum-volume-constrained NMF and provide an algorithm to estimate it. Experimental results demonstrate the effectiveness of the proposed method.

Recent advances in Large Language Models (LLMs) have inspired a surge of scaling law research in industrial search, advertising, and recommendation systems. However, existing approaches focus mainly on architectural improvements, overlooking the critical synergy between data and architecture design. We observe that scaling model parameters alone exhibits diminishing returns, i.e., the marginal gain in performance steadily declines as model size increases, and that the performance degradation caused by complex heterogeneous data distributions is often irrecoverable through model design alone. In this paper, we propose UniScale to address these limitation, a novel co-design framework that jointly optimizes data and architecture to unlock the full potential of model scaling, which includes two core parts: (1) ES$^3$ (Entire-Space Sample System), a high-quality data scaling system that expands the training signal beyond conventional sampling strategies from both intra-domain request contexts with global supervised signal constructed by hierarchical label attribution and cross-domain samples aligning with the essence of user decision under similar content exposure environment in search domain; and (2) HHSFT (Heterogeneous Hierarchical Sample Fusion Transformer), a novel architecture designed to effectively model the complex heterogeneous distribution of scaled data and to harness the entire space user behavior data with Heterogeneous Hierarchical Feature Interaction and Entire Space User Interest Fusion, thereby surpassing the performance ceiling of structure-only model tuning. Extensive experiments on large-scale real world E-commerce search platform demonstrate that UniScale achieves significant improvements through the synergistic co-design of data and architecture and exhibits clear scaling trends, delivering substantial gains in key business metrics.

In Natural Language Processing (NLP), variation is typically seen as noise and "normalised away" before processing, even though it is an integral part of language. Conversely, studying language variation in social contexts is central to sociolinguistics. We present a framework to combine the sociolinguistic dimension of language with the technical dimension of NLP. We argue that by embracing sociolinguistics, variation can actively be included in a research setup, in turn informing the NLP side. To illustrate this, we provide a case study on Luxembourgish, an evolving language featuring a large amount of orthographic variation, demonstrating how NLP performance is impacted. The results show large discrepancies in the performance of models tested and fine-tuned on data with a large amount of orthographic variation in comparison to data closer to the (orthographic) standard. Furthermore, we provide a possible solution to improve the performance by including variation in the fine-tuning process. This case study highlights the importance of including variation in the research setup, as models are currently not robust to occurring variation. Our framework facilitates the inclusion of variation in the thought-process while also being grounded in the theoretical framework of sociolinguistics.

The increasing complexity and interconnectivity of digital infrastructures make scalable and reliable security assessment methods essential. Robotic systems represent a particularly important class of operational technology, as modern robots are highly networked cyber-physical systems deployed in domains such as industrial automation, logistics, and autonomous services. This paper explores the use of large language models for automated penetration testing in robotic environments. We propose an environment-grounded multi-agent architecture tailored to Robotics-based systems. The approach dynamically constructs a shared graph-based memory during execution that captures the observable system state, including network topology, communication channels, vulnerabilities, and attempted exploits. This enables structured automation while maintaining traceability and effective context management throughout the testing process. Evaluated across multiple iterations within a specialized robotics Capture-the-Flag scenario (ROS/ROS2), the system demonstrated high reliability, successfully completing the challenge in 100\% of test runs (n=5). This performance significantly exceeds literature benchmarks while maintaining the traceability and human oversight required by frameworks like the EU AI Act.

Large Language Models (LLMs) enhanced with Retrieval-Augmented Generation (RAG) have achieved substantial improvements in accuracy by grounding their responses in external documents that are relevant to the user's query. However, relatively little work has investigated the impact of RAG in terms of fairness. Particularly, it is not yet known if queries that are associated with certain groups within a fairness category systematically receive higher accuracy, or accuracy improvements in RAG systems compared to LLM-only, a phenomenon we refer to as query group fairness. In this work, we conduct extensive experiments to investigate the impact of three key factors on query group fairness in RAG, namely: Group exposure, i.e., the proportion of documents from each group appearing in the retrieved set, determined by the retriever; Group utility, i.e., the degree to which documents from each group contribute to improving answer accuracy, capturing retriever-generator interactions; and Group attribution, i.e., the extent to which the generator relies on documents from each group when producing responses. We examine group-level average accuracy and accuracy improvements disparities across four fairness categories using three datasets derived from the TREC 2022 Fair Ranking Track for two tasks: article generation and title generation. Our findings show that RAG systems suffer from the query group fairness problem and amplify disparities in terms of average accuracy across queries from different groups, compared to an LLM-only setting. Moreover, group utility, exposure, and attribution can exhibit strong positive or negative correlations with average accuracy or accuracy improvements of queries from that group, highlighting their important role in fair RAG. Our data and code are publicly available from Github.

Standard citation metrics treat all citations as equal, obscuring the social and structural pathways through which scholarly influence propagates. I introduce Citation-Constellation, a freely available no-code tool for citation network analysis with two complementary bibliometric scores that decompose a researcher's citation profile by network proximity between citing and cited authors. BARON (Boundary-Anchored Research Outreach Network score) is a strict binary metric counting only citations from outside the detected collaborative network. HEROCON (Holistic Equilibrated Research Outreach CONstellation score) applies graduated weights assigning partial credit to in-group citations based on relationship proximity. The gap between scores serves as a diagnostic of inner-circle dependence. An extended abstract with full details appears in the paper. The tool implements this through a phased architecture: (1) self-citation analysis, (2) co-authorship graph traversal, (3) temporal institutional affiliation matching via ROR, and (4) AI-agent-driven venue governance extraction using a local LLM. Phases 1-3 are fully operational; Phase 4 is under development. Key design choices include ORCID-validated author identity resolution, an UNKNOWN classification for citations with insufficient metadata, and comprehensive audit trails documenting every classification decision. A no-code web interface enables researchers to compute scores without programming, installation, or registration. I present these scores as structural diagnostics, not quality indicators. BARON and HEROCON describe where in the social graph citations originate. They should not be used for hiring, promotion, or funding decisions. HEROCON weights are experimental and require empirical calibration.

Deep learning models for time series imputation are now essential in fields such as healthcare, the Internet of Things (IoT), and finance. However, their deployment raises critical privacy concerns. Beyond the well-known issue of unintended memorization, which has been extensively studied in generative models, we demonstrate that time series models are vulnerable to inference attacks in a black-box setting. In this work, we introduce a two-stage attack framework comprising: (1) a novel membership inference attack based on a reference model that improves detection accuracy, even for models robust to overfitting-based attacks, and (2) the first attribute inference attack that predicts sensitive characteristics of the training data for timeseries imputation model. We evaluate these attacks on attention-based and autoencoder architectures in two scenarios: models that are trained from scratch, and fine-tuned models where the adversary has access to the initial weights. Our experimental results demonstrate that the proposed membership attack retrieves a significant portion of the training data with a tpr@top25% score significantly higher than a naive attack baseline. We show that our membership attack also provides a good insight of whether attribute inference will work (with a precision of 90% instead of 78% in the genral case).

Personalized Federated Learning (PFL) aims to deliver effective client-specific models under heterogeneous distributions, yet existing methods suffer from shallow prototype alignment and brittle server-side distillation. We propose HEART-PFL, a dual-sided framework that (i) performs depth-aware Hierarchical Directional Alignment (HDA) using cosine similarity in the early stage and MSE matching in the deep stage to preserve client specificity, and (ii) stabilizes global updates through Adversarial Knowledge Transfer (AKT) with symmetric KL distillation on clean and adversarial proxy data. Using lightweight adapters with only 1.46M trainable parameters, HEART-PFL achieves state-of-the-art personalized accuracy on CIFAR-100, Flowers-102, and Caltech-101 (63.42%, 84.23%, and 95.67%, respectively) under Dirichlet non-IID partitions, and remains robust to out-of-domain proxy data. Ablation studies further confirm that HDA and AKT provide complementary gains in alignment, robustness, and optimization stability, offering insights into how the two components mutually reinforce effective personalization. Overall, these results demonstrate that HEART-PFL simultaneously enhances personalization and global stability, highlighting its potential as a strong and scalable solution for PFL(code available at https://github.com/danny0628/HEART-PFL).

Knowledge distillation transfers knowledge from large teacher models to smaller students for efficient inference. While existing methods primarily focus on distillation strategies, they often overlook the importance of enhancing teacher knowledge quality. In this paper, we propose Text-guided Multi-view Knowledge Distillation (TMKD), which leverages dual-modality teachers, a visual teacher and a text teacher (CLIP), to provide richer supervisory signals. Specifically, we enhance the visual teacher with multi-view inputs incorporating visual priors (edge and high-frequency features), while the text teacher generates semantic weights through prior-aware prompts to guide adaptive feature fusion. Additionally, we introduce vision-language contrastive regularization to strengthen semantic knowledge in the student model. Extensive experiments on five benchmarks demonstrate that TMKD consistently improves knowledge distillation performance by up to 4.49\%, validating the effectiveness of our dual-teacher multi-view enhancement strategy. Code is available at https://anonymous.4open.science/r/TMKD-main-44D1.

Accurate forecasting of multivariate time series remains challenging due to the need to capture both short-term fluctuations and long-range temporal dependencies. Transformer-based models have emerged as a powerful approach, but their performance depends critically on the representation of temporal data. Traditional point-wise representations preserve individual time-step information, enabling fine-grained modeling, yet they tend to be computationally expensive and less effective at modeling broader contextual dependencies, limiting their scalability to long sequences. Patch-wise representations aggregate consecutive steps into compact tokens to improve efficiency and model local temporal dynamics, but they often discard fine-grained temporal details that are critical for accurate predictions in volatile or complex time series. We propose IPatch, a multi-resolution Transformer architecture that integrates both point-wise and patch-wise tokens, modeling temporal information at multiple resolutions. Experiments on 7 benchmark datasets demonstrate that IPatch consistently improves forecasting accuracy, robustness to noise, and generalization across various prediction horizons compared to single-representation baselines.

Hybrid quantum-classical workflows combine quantum processing units (QPUs) with classical hardware to address computational tasks that are challenging or infeasible for conventional systems alone. Coordinating these heterogeneous resources at scale demands robust orchestration, reproducibility, and observability. Even in the presence of fault-tolerant quantum devices, quantum computing will continue to operate within a broader hybrid ecosystem, where classical infrastructure plays a central role in task scheduling, data movement, error mitigation, and large-scale workflow coordination. In this work, we present a cloud-native framework for managing hybrid quantum-HPC pipelines using Kubernetes, Argo Workflows, and Kueue. Our system unifies CPUs, GPUs, and QPUs under a single orchestration layer, enabling multi-stage workflows with dynamic, resource-aware scheduling. We demonstrate the framework with a proof-of-concept implementation of distributed quantum circuit cutting, showcasing execution across heterogeneous nodes and integration of classical and quantum tasks. This approach highlights the potential for scalable, reproducible, and flexible hybrid quantum-classical computing in cloud-native environments.

Recent advances in the Model Context Protocol (MCP) have enabled large language models (LLMs) to invoke external tools with unprecedented ease. This creates a new class of powerful and tool augmented agents. Unfortunately, this capability also introduces an under explored attack surface, specifically the malicious manipulation of tool responses. Existing techniques for indirect prompt injection that target MCP suffer from high deployment costs, weak semantic coherence, or heavy white box requirements. Furthermore, they are often easily detected by recently proposed defenses. In this paper, we propose Tree structured Injection for Payloads (TIP), a novel black-box attack which generates natural payloads to reliably seize control of MCP enabled agents even under defense. Technically, We cast payload generation as a tree structured search problem and guide the search with an attacker LLM operating under our proposed coarse-to-fine optimization framework. To stabilize learning and avoid local optima, we introduce a path-aware feedback mechanism that surfaces only high quality historical trajectories to the attacker model. The framework is further hardened against defensive transformations by explicitly conditioning the search on observable defense signals and dynamically reallocating the exploration budget. Extensive experiments on four mainstream LLMs show that TIP attains over 95% attack success in undefended settings while requiring an order of magnitude fewer queries than prior adaptive attacks. Against four representative defense approaches, TIP preserves more than 50% effectiveness and significantly outperforms the state-of-the-art attacks. By implementing the attack on real world MCP systems, our results expose an invisible but practical threat vector in MCP deployments. We also discuss potential mitigation approaches to address this critical security gap.

Reinforcement learning (RL) has emerged as a powerful paradigm for improving large language models beyond supervised fine-tuning, yet sustaining performance gains at scale remains an open challenge, as data diversity and structure, rather than volume alone, become the limiting factor. We address this by introducing a scalable multi-turn synthetic data generation pipeline in which a teacher model iteratively refines problems based on in-context student performance summaries, producing structured difficulty progressions without any teacher fine-tuning. Compared to single-turn generation, this multi-turn approach substantially improves the yield of valid synthetic problems and naturally produces stepping stones, i.e. easier and harder variants of the same core task, that support curriculum-based training. We systematically study how task difficulty, curriculum scheduling, and environment diversity interact during RL training across the Llama3.1-8B Instruct and Qwen3-8B Base model families, with additional scaling experiments on Qwen2.5-32B. Our results show that synthetic augmentation consistently improves in-domain code and in most cases out-of-domain math performance, and we provide empirical insights into how curriculum design and data diversity jointly shape RL training dynamics.

The first generation of students is learning to program alongside GenAI (Generative Artificial Intelligence) tools, raising questions about how young learners critically engage with them and perceive ethical responsibilities. While prior research has focused on university students or developers, little is known about secondary school novices, who represent the next cohort of software engineers. To address this gap, we conducted an exploratory study with 84 German secondary school students aged 16-19 attending software development workshops. We examined their critical thinking practices in AI-assisted programming, perceptions of AI ethics and responsibility, and gender-related differences in their views. Our results reveal an AI paradox: students demonstrate strong ethical reasoning and awareness about AI, yet many report integrating AI-generated code without a thorough understanding of it. The majority of our cohort attributed significant responsibility for AI practices to politics and corporations, potentially reflecting Germany's cultural context, with its strict regulations and data privacy discourse. Boys reported more frequent and experimental use of AI-assisted programming, whereas girls expressed greater scepticism and emphasised peer collaboration over GenAI assistance. Our findings highlight the importance of culturally responsive software engineering education that strengthens critical AI literacy in AI-assisted programming by linking ethics to concrete code artefacts and preparing young learners for this AI-driven software landscape.

In scientific computing, the formulation of numerical discretisations of partial differential equations (PDEs) as untrained convolutional layers within Convolutional Neural Networks (CNNs), referred to by some as Neural Physics, has demonstrated good efficiency for executing physics-based solvers on GPUs. However, classical grid-based methods still face computational bottlenecks when solving problems involving billions of degrees of freedom. To address this challenge, this paper proposes a novel framework called 'Quantum Neural Physics' and develops a Hybrid Quantum-Classical CNN Multigrid Solver (HQC-CNNMG). This approach maps analytically-determined stencils of discretised differential operators into parameter-free or untrained quantum convolutional kernels. By leveraging amplitude encoding, the Linear Combination of Unitaries technique and the Quantum Fourier Transform, the resulting quantum convolutional operators can be implemented using quantum circuits with a circuit depth that scales as O(log K), where K denotes the size of the encoded input block. These quantum operators are embedded into a classical W-Cycle multigrid using a U-Net. This design enables seamless integration of quantum operators within a hierarchical solver whilst retaining the robustness and convergence properties of classical multigrid methods. The proposed Quantum Neural Physics solver is validated on a quantum simulator for the Poisson equation, diffusion equation, convection-diffusion equation and incompressible Navier-Stokes equations. The solutions of the HQC-CNNMG are in close agreement with those from traditional solution methods. This work establishes a mapping from discretised physical equations to logarithmic-scale quantum circuits, providing a new and exploratory path to exponential memory compression and computational acceleration for PDE solvers on future fault-tolerant quantum computers.

Mental health and well-being are major concerns in higher education and professional fields such as software engineering, yet are often overlooked in curricula. This paper describes our approach to include mental health, well-being, and sustainability in software engineering education in two ways: (1) well-being-focused software projects that ask students to design technical solutions or research addressing mental health and sustainability or societal challenges, and (2) brief classroom interventions such as short reflective discussions and team-building activities. We argue that this combination can help students see software engineering more broadly while creating healthier learning environments. Our analysis of reflections from 60 students found several positive outcomes: students gained a more human-centred perspective, had more team discussions about mental health, and began to see well-being as inspiration for using software to benefit society and individuals rather than merely as a technical or business tool. By combining technical skills with awareness of well-being, we argue that software engineering education can prepare future developers to be both skilled programmers and responsible professionals who care about human well-being.

Increasing demands for adaptability, privacy, and security at the edge have persistently pushed the frontiers for a new generation of machine learning (ML) algorithms with training and inference capabilities on-chip. Weightless Neural Network (WNN) is such an algorithm that is principled on lookup table based simple neuron structures. As a result, it offers architectural benefits, such as low-latency, low-complexity inference, compared to deep neural networks that depend heavily on multiply-accumulate operations. However, traditional WNNs rely on memorization-based one-shot training, which either leads to overfitting and reduced accuracy or requires tedious post-training adjustments, limiting their effectiveness for efficient on chip training. In this work, we propose TsetlinWiSARD, a training approach for WNNs that leverages Tsetlin Automata (TAs) to enable probabilistic, feedback-driven learning. It overcomes the overfitting of WiSARD's one-shot training with iterative optimization, while maintaining simple, continuous binary feedback for efficient on-chip training. Central to our approach is a field programmable gate array (FPGA)-based training architecture that delivers state-of-the-art accuracy while significantly improving hardware efficiency. Our approach provides over 1000x faster training when compared with the traditional WiSARD implementation of WNNs. Further, we demonstrate 22% reduced resource usage, 93.3% lower latency, and 64.2% lower power consumption compared to FPGA-based training accelerators implementing other ML algorithms.

WebAssembly's (Wasm) monolithic linear memory model facilitates memory corruption attacks that can escalate to cross-site scripting in browsers or go undetected when a malicious host tampers with a module's state. Existing defenses rely on invasive binary instrumentation or custom runtimes, and do not address runtime integrity verification under an adversarial host model. We present Walma, a framework for WebAssembly Linear Memory Attestation that leverages machine learning to detect memory corruption and external tampering by classifying memory snapshots. We evaluate Walma on six real-world CVE-affected applications across three verification backends (cpu-wasm, cpu-tch, gpu) and three instrumentation policies. Our results demonstrate that CNN-based classification can effectively detect memory corruption in applications with structured memory layouts, with coarse-grained boundary checks incurring as low as 1.07x overhead, while fine-grained monitoring introduces higher (1.5x--1.8x) but predictable costs. Our evaluation quantifies the accuracy and overhead trade-offs across deployment configurations, demonstrating the practical feasibility of ML-based memory attestation for WebAssembly.

The rapid proliferation and increasing complexity of software demand robust quality assurance, with graphical user interface (GUI) testing playing a pivotal role. Crowdsourced testing has proven effective in this context by leveraging the diversity of human testers to achieve rich, scenario-based coverage across varied devices, user behaviors, and usage environments. In parallel, automated testing, particularly with the advent of large language models (LLMs), offers significant advantages in controllability, reproducibility, and efficiency, enabling scalable and systematic exploration. However, automated approaches often lack the behavioral diversity characteristic of human testers, limiting their capability to fully simulate real-world testing dynamics. To address this gap, we present PersonaTester, a novel personified-LLM-based framework designed to automate crowdsourced GUI testing. By injecting representative personas, defined along three orthogonal dimensions: testing mindset, exploration strategy, and interaction habit, into LLM-based agents, PersonaTester enables the simulation of diverse human-like testing behaviors in a controllable and repeatable manner. Experimental results demonstrate that PersonaTester faithfully reproduces the behavioral patterns of real crowdworkers, exhibiting strong intra-persona consistency and clear inter-persona variability (117.86% -- 126.23% improvement over the baseline). Moreover, persona-guided testing agents consistently generate more effective test events and trigger more crashes (100+) and functional bugs (11) than the baseline without persona, thus substantially advancing the realism and effectiveness of automated crowdsourced GUI testing.

Antonyms, or opposites, are sometimes defined as \emph{word pairs that have all of the same contextually relevant properties but one}. Seeing how transformer models seem to encode concepts as directions, this begs the question if one can detect ``antonymity'' in the geometry of the embedding vectors of word pairs, especially based on their difference vectors. Such geometrical studies are then naturally contrasted by comparing antonymic pairs to their opposites; synonyms. This paper started as an exploratory project on the complexity of the systems needed to detect the geometry of the embedding vectors of antonymic word pairs. What we now report is a curious ``swirl'' that appears across embedding models in a somewhat specific projection configuration.

Neural operator learning directly constructs the mapping relationship from the equation parameter space to the solution space, enabling efficient direct inference in practical applications without the need for repeated solution of partial differential equations (PDEs) - an advantage that is difficult to achieve with traditional numerical methods. In this work, we find that explicitly decoupling linear and nonlinear effects within such operator mappings leads to markedly improved learning efficiency. This yields a novel network structure, namely the Linear-Nonlinear Fusion Neural Operator (LNF-NO), which models operator mappings via the multiplicative fusion of a linear component and a nonlinear component, thus achieving a lightweight and interpretable representation. This linear-nonlinear decoupling enables efficient capture of complex solution features at the operator level while maintaining stability and generality. LNF-NO naturally supports multiple functional inputs and is applicable to both regular grids and irregular geometries. Across a diverse suite of PDE operator-learning benchmarks, including nonlinear Poisson-Boltzmann equations and multi-physics coupled systems, LNF-NO is typically substantially faster to train than Deep Operator Networks (DeepONet) and Fourier Neural Operators (FNO), while achieving comparable or better accuracy in most cases. On the tested 3D Poisson-Boltzmann case, LNF-NO attains the best accuracy among the compared models and trains approximately 2.7x faster than a 3D FNO baseline.

This paper provides a historical analysis of the IEEE CEC Single Objective Optimization competition results (2010-2024). We analyze how benchmark functions shaped winning algorithms, identifying the 2014 introduction of dense rotation matrices as a key performance filter. This design choice introduced parameter non-separability, reduced effectiveness of coordinate-dependent methods (PSO, GA), and established the dominance of Differential Evolution variants capable of preserving the rotational invariance of their difference vectors, specifically L-SHADE. Post-2020 analysis reveals a shift towards high complexity hybrid optimizers that combine different mechanisms (e.g., Eigenvector Crossover, Societal Sharing, Reinforcement Learning) to maximize ranking stability. We conclude by identifying structural similarities between these modern benchmarks and Variational Quantum Algorithm landscapes, suggesting that evolved CEC solvers possess the specific adaptive capabilities required for quantum control.

Standard supervised training for deepfake detection treats all samples with uniform importance, which can be suboptimal for learning robust and generalizable features. In this work, we propose a novel Tutor-Student Reinforcement Learning (TSRL) framework to dynamically optimize the training curriculum. Our method models the training process as a Markov Decision Process where a ``Tutor'' agent learns to guide a ``Student'' (the deepfake detector). The Tutor, implemented as a Proximal Policy Optimization (PPO) agent, observes a rich state representation for each training sample, encapsulating not only its visual features but also its historical learning dynamics, such as EMA loss and forgetting counts. Based on this state, the Tutor takes an action by assigning a continuous weight (0-1) to the sample's loss, thereby dynamically re-weighting the training batch. The Tutor is rewarded based on the Student's immediate performance change, specifically rewarding transitions from incorrect to correct predictions. This strategy encourages the Tutor to learn a curriculum that prioritizes high-value samples, such as hard-but-learnable examples, leading to a more efficient and effective training process. We demonstrate that this adaptive curriculum improves the Student's generalization capabilities against unseen manipulation techniques compared to traditional training methods. Code is available at https://github.com/wannac1/TSRL.

Tuning control policies manually to meet high-level objectives is often time-consuming. Bayesian optimization provides a data-efficient framework for automating this process using numerical evaluations of an objective function. However, many systems, particularly those involving humans, require optimization based on subjective criteria. Preferential Bayesian optimization addresses this by learning from pairwise comparisons instead of quantitative measurements, but relying solely on preference data can be inefficient. We propose a multi-fidelity, multi-modal Bayesian optimization framework that integrates low-fidelity numerical data with high-fidelity human preferences. Our approach employs Gaussian process surrogate models with both hierarchical, autoregressive and non-hierarchical, coregionalization-based structures, enabling efficient learning from mixed-modality data. We illustrate the framework by tuning an autonomous vehicle's trajectory planner, showing that combining numerical and preference data significantly reduces the need for experiments involving the human decision maker while effectively adapting driving style to individual preferences.

Conversational artificial intelligence has the potential to assist users in preliminary medical consultations, particularly in settings where access to healthcare professionals is limited. However, many existing medical dialogue systems operate in a single-turn question--answering paradigm or rely on template-based datasets, limiting conversational realism and multilingual applicability. In this work, we introduce MedAidDialog, a multilingual multi-turn medical dialogue dataset designed to simulate realistic physician--patient consultations. The dataset extends the MDDial corpus by generating synthetic consultations using large language models and further expands them into a parallel multilingual corpus covering seven languages: English, Hindi, Telugu, Tamil, Bengali, Marathi, and Arabic. Building on this dataset, we develop MedAidLM, a conversational medical model trained using parameter-efficient fine-tuning on quantized small language models, enabling deployment without high-end computational infrastructure. Our framework additionally incorporates optional patient pre-context information (e.g., age, gender, allergies) to personalize the consultation process. Experimental results demonstrate that the proposed system can effectively perform symptom elicitation through multi-turn dialogue and generate diagnostic recommendations. We further conduct medical expert evaluation to assess the plausibility and coherence of the generated consultations.

Message passing is a core mechanism in Graph Neural Networks (GNNs), enabling the iterative update of node embeddings by aggregating information from neighboring nodes. Graph Convolutional Networks (GCNs) exemplify this approach by adapting convolutional operations for graph structures, allowing features from adjacent nodes to be combined effectively. However, GCNs encounter challenges with complex or dynamic data. Capturing long-range dependencies often requires deeper layers, which not only increase computational costs but also lead to over-smoothing, where node embeddings become indistinguishable. To overcome these challenges, reservoir computing has been integrated into GNNs, leveraging iterative message-passing dynamics for stable information propagation without extensive parameter tuning. Despite its promise, existing reservoir-based models lack structured convolutional mechanisms, limiting their ability to accurately aggregate multi-hop neighborhood information. To address these limitations, we propose RGC-Net (Reservoir-based Graph Convolutional Network), which integrates reservoir dynamics with structured graph convolution. Key contributions include: (i) a reimagined convolutional framework with fixed random reservoir weights and a leaky integrator to enhance feature retention; (ii) a robust, adaptable model for graph classification; and (iii) an RGC-Net-powered transformer for graph generation with application to dynamic brain connectivity. Extensive experiments show that RGC-Net achieves state-of-the-art performance in classification and generative tasks, including brain graph evolution, with faster convergence and reduced over-smoothing. Source code is available at https://github.com/basiralab/RGC-Net .

In the IoT era, information is more and more frequently picked up by connected smart sensors with increasing, though limited, storage, communication and computation abilities. Whether due to privacy constraints or to the structure of the distributed system, the development of statistical learning methods dedicated to data that are shared over a network is now a major issue. Gossip-based algorithms have been developed for the purpose of solving a wide variety of statistical learning tasks, ranging from data aggregation over sensor networks to decentralized multi-agent optimization. Whereas the vast majority of contributions consider situations where the function to be estimated or optimized is a basic average of individual observations, it is the goal of this article to investigate the case where the latter is of pairwise nature, taking the form of a U -statistic of degree two. Motivated by various problems such as similarity learning, ranking or clustering for instance, we revisit gossip algorithms specifically designed for pairwise objective functions and provide a comprehensive theoretical framework for their convergence. This analysis fills a gap in the literature by establishing conditions under which these methods succeed, and by identifying the graph properties that critically affect their efficiency. In particular, a refined analysis of the convergence upper and lower bounds is performed.

When language models are trained by reinforcement learning (RL) to write probabilistic programs, they can artificially inflate their marginal-likelihood reward by producing programs whose data distribution fails to normalise instead of fitting the data better. We call this failure likelihood hacking (LH). We formalise LH in a core probabilistic programming language (PPL) and give sufficient syntactic conditions for its prevention, proving that a safe language fragment $\mathcal{L}_{\text{safe}}$ satisfying these conditions cannot produce likelihood-hacking programs. Empirically, we show that GRPO-trained models generating PyMC code discover LH exploits within the first few training steps, driving violation rates well above the untrained-model baseline. We implement $\mathcal{L}_{\text{safe}}$'s conditions as $\texttt{SafeStan}$, a LH-resistant modification of Stan, and show empirically that it prevents LH under optimisation pressure. These results show that language-level safety constraints are both theoretically grounded and effective in practice for automated Bayesian model discovery.

During training, Large Language Models (LLMs) learn social regularities that can lead to gender bias in downstream applications. Most mitigation efforts focus on reducing bias in generated outputs, typically evaluated on structured benchmarks, which raises two concerns: output-level evaluation does not reveal whether alignment modifies the model's underlying representations, and structured benchmarks may not reflect realistic usage scenarios. We propose a unified framework to jointly analyze intrinsic and extrinsic gender bias in LLMs using identical neutral prompts, enabling direct comparison between gender-related information encoded in internal representations and bias expressed in generated outputs. Contrary to prior work reporting weak or inconsistent correlations, we find a consistent association between latent gender information and expressed bias when measured under the unified protocol. We further examine the effect of alignment through supervised fine-tuning aimed at reducing gender bias. Our results suggest that while the latter indeed reduces expressed bias, measurable gender-related associations are still present in internal representations, and can be reactivated under adversarial prompting. Finally, we consider two realistic settings and show that debiasing effects observed on structured benchmarks do not necessarily generalize, e.g., to the case of story generation.

RLHF-aligned language models exhibit response homogenization: on TruthfulQA (n=790), 40-79% of questions produce a single semantic cluster across 10 i.i.d. samples. On affected questions, sampling-based uncertainty methods have zero discriminative power (AUROC=0.500), while free token entropy retains signal (0.603). This alignment tax is task-dependent: on GSM8K (n=500), token entropy achieves 0.724 (Cohen's d=0.81). A base-vs-instruct ablation confirms the causal role of alignment: the base model shows 1.0% single-cluster rate vs. 28.5% for the instruct model (p < 10^{-6}). A training stage ablation (Base 0.0% -> SFT 1.5% -> DPO 4.0% SCR) localizes the cause to DPO, not SFT. Cross-family replication on four model families reveals alignment tax severity varies by family and scale. We validate across 22 experiments, 5 benchmarks, 4 model families, and 3 model scales (3B-14B), with Jaccard, embedding, and NLI-based baselines at three DeBERTa scales (all ~0.51 AUROC). Cross-embedder validation with two independent embedding families rules out coupling bias. Cross-dataset validation on WebQuestions (58.0% SCR) confirms generalization beyond TruthfulQA. The central finding -- response homogenization -- is implementation-independent and label-free. Motivated by this diagnosis, we explore a cheapest-first cascade (UCBD) over orthogonal uncertainty signals. Selective prediction raises GSM8K accuracy from 84.4% to 93.2% at 50% coverage; weakly dependent boundaries (|r| <= 0.12) enable 57% cost savings.

With the development of deep learning, Neural Code Models (NCMs) such as CodeBERT and CodeLlama are widely used for code understanding tasks, including defect detection and code classification. However, recent studies have revealed that NCMs are vulnerable to adversarial examples, inputs with subtle perturbations that induce incorrect predictions while remaining difficult to detect. Existing defenses address this issue via data augmentation to empirically improve robustness, but they are costly, offer no theoretical robustness guarantees, and typically require white-box access to model internals, such as gradients. To address the above challenges, we propose ENBECOME, a novel black-box training-free and lightweight adversarial defense. ENBECOME is designed to both enhance empirical robustness and report certified robustness boundaries for NCMs. ENBECOME operates solely during inference, introducing random, semantics-preserving perturbations to input code snippets to smooth the NCM's decision boundaries. This smoothing enables ENBECOME to formally certify a robustness radius within which adversarial examples can never induce misclassification, a property known as certified robustness. We conduct comprehensive experiments across multiple NCM architectures and tasks. Results show that ENBECOME significantly reduces attack success rates while maintaining high accuracy. For example, in defect detection, it reduces the average ASR from 42.43% to 9.74% with only a 0.29% drop in accuracy. Results show that ENBECOME significantly reduces attack success rates while maintaining high accuracy. For example, in defect detection, it reduces the average ASR from 42.43% to 9.74% with only a 0.29% drop in accuracy. Furthermore, ENBECOME achieves an average certified robustness radius of 1.63, meaning that adversarial modifications to no more than 1.63 identifiers are provably ineffective.

On-chip learning is key to scalable and adaptive neuromorphic systems, yet existing training methods are either difficult to implement in hardware or overly restrictive. However, recent studies show that feedback-control optimizers can enable expressive, on-chip training of neuromorphic devices. In this work, we present a proof-of-concept implementation of such feedback-control optimizers on a mixed-signal neuromorphic processor. We assess the proposed approach in an In-The-Loop(ITL) training setup on both a binary classification task and the nonlinear Yin-Yang problem, demonstrating on-chip training that matches the performance of numerical simulations and gradient-based baselines. Our results highlight the feasibility of feedback-driven, online learning under realistic mixed-signal constraints, and represent a co-design approach toward embedding such rules directly in silicon for autonomous and adaptive neuromorphic computing.

The Industrial Internet of Things (IIoT) introduces significant security challenges as resource-constrained devices become increasingly integrated into critical industrial processes. Existing security approaches typically address threats at a single network layer, often relying on expensive hardware and remaining confined to simulation environments. In this paper, we present the research framework and contributions of our doctoral thesis, which aims to develop a lightweight, Machine Learning (ML)-based security framework for IIoT environments. We first describe our adoption of the Tm-IIoT trust model and the Hybrid IIoT (H-IIoT) architecture as foundational baselines, then introduce the Trust Convergence Acceleration (TCA) approach, our primary contribution that integrates ML to predict and mitigate the impact of degraded network conditions on trust convergence, achieving up to a 28.6% reduction in convergence time while maintaining robustness against adversarial behaviors. We then propose a real-world deployment architecture based on affordable, open-source hardware, designed to implement and extend the security framework. Finally, we outline our ongoing research toward multi-layer attack detection, including physical-layer threat identification and considerations for robustness against adversarial ML attacks.

Multi-modal Satellite Image Time Series (SITS) analysis faces significant computational challenges for live land monitoring applications. While Transformer architectures excel at capturing temporal dependencies and fusing multi-modal data, their quadratic computational complexity and the need to reprocess entire sequences for each new acquisition limit their deployment for regular, large-area monitoring. This paper studies various dual-form attention mechanisms for efficient multi-modal SITS analysis, that enable parallel training while supporting recurrent inference for incremental processing. We compare linear attention and retention mechanisms within a multi-modal spectro-temporal encoder. To address SITS-specific challenges of temporal irregularity and unalignment, we develop temporal adaptations of dual-form mechanisms that compute token distances based on actual acquisition dates rather than sequence indices. Our approach is evaluated on two tasks using Sentinel-1 and Sentinel-2 data: multi-modal SITS forecasting as a proxy task, and real-world solar panel construction monitoring. Experimental results demonstrate that dual-form mechanisms achieve performance comparable to standard Transformers while enabling efficient recurrent inference. The multimodal framework consistently outperforms mono-modal approaches across both tasks, demonstrating the effectiveness of dual mechanisms for sensor fusion. The results presented in this work open new opportunities for operational land monitoring systems requiring regular updates over large geographic areas.

Learning representations from multiplex graphs, i.e., multi-layer networks where nodes interact through multiple relation types, is challenging due to the entanglement of shared (common) and layer-specific (private) information, which limits generalization and interpretability. In this work, we introduce a causal inference-based framework that disentangles common and private components in a self-supervised manner. CaDeM jointly (i) aligns shared embeddings across layers, (ii) enforces private embeddings to capture layer-specific signals, and (iii) applies backdoor adjustment to ensure that the common embeddings capture only global information while being separated from the private representations. Experiments on synthetic and real-world datasets demonstrate consistent improvements over existing baselines, highlighting the effectiveness of our approach for robust and interpretable multiplex graph representation learning.

Digital circuits representation learning has made remarkable progress in the electronic design automation domain, effectively supporting critical tasks such as testability analysis and logic reasoning. However, representation learning for analog circuits remains challenging due to their continuous electrical characteristics compared to the discrete states of digital circuits. This paper presents a direct current (DC) electrically equivalent-oriented analog representation learning framework, named \textbf{KCLNet}. It comprises an asynchronous graph neural network structure with electrically-simulated message passing and a representation learning method inspired by Kirchhoff's Current Law (KCL). This method maintains the orderliness of the circuit embedding space by enforcing the equality of the sum of outgoing and incoming current embeddings at each depth, which significantly enhances the generalization ability of circuit embeddings. KCLNet offers a novel and effective solution for analog circuit representation learning with electrical constraints preserved. Experimental results demonstrate that our method achieves significant performance in a variety of downstream tasks, e.g., analog circuit classification, subcircuit detection, and circuit edit distance prediction.

Recently, reinforcement learning~(RL) has become an important approach for improving the capabilities of large language models~(LLMs). In particular, reinforcement learning from verifiable rewards~(RLVR) has emerged as a promising paradigm for reasoning tasks. However, existing RL-based training still remains only a rough approximation to human learning. Human learners leverage both external and internal experience to guide exploration and gradually internalize useful trajectories into stable knowledge. Motivated by this gap, we ask: how can LLMs better utilize and internalize experience during RLVR training? To answer this question, we propose \textbf{D}ual \textbf{G}uidance \textbf{O}ptimization~(\textbf{DGO}), a unified framework that leverages \emph{external} and \emph{internal experience} to improve training effectiveness. Specifically, DGO first constructs an experience bank from previously explored trajectories. The policy then performs exploration under the joint guidance of the experience bank and the model's internal knowledge. The resulting trajectories are further used to refine the experience bank and optimize model parameters, forming a closed loop of experience utilization and internalization. Experiments show that DGO consistently outperforms baseline methods, suggesting that better utilization and internalization of experience lead to more effective reasoning.

Empirical evaluation in multi-objective search (MOS) has historically suffered from fragmentation, relying on heterogeneous problem instances with incompatible objective definitions that make cross-study comparisons difficult. This standardization gap is further exacerbated by the realization that DIMACS road networks, a historical default benchmark for the field, exhibit highly correlated objectives that fail to capture diverse Pareto-front structures. To address this, we introduce the first comprehensive, standardized benchmark suite for exact and approximate MOS. Our suite spans four structurally diverse domains: real-world road networks, structured synthetic graphs, game-based grid environments, and high-dimensional robotic motion-planning roadmaps. By providing fixed graph instances, standardized start-goal queries, and both exact and approximate reference Pareto-optimal solution sets, this suite captures a full spectrum of objective interactions: from strongly correlated to strictly independent. Ultimately, this benchmark provides a common foundation to ensure future MOS evaluations are robust, reproducible, and structurally comprehensive.

This paper introduces Knowledge Graph based Massively Multi-task Model-based Policy Optimization (KG-M3PO), a framework for multi-task robotic manipulation in partially observable settings that unifies Perception, Knowledge, and Policy. The method augments egocentric vision with an online 3D scene graph that grounds open-vocabulary detections into a metric, relational representation. A dynamic-relation mechanism updates spatial, containment, and affordance edges at every step, and a graph neural encoder is trained end-to-end through the RL objective so that relational features are shaped directly by control performance. Multiple observation modalities (visual, proprioceptive, linguistic, and graph-based) are encoded into a shared latent space, upon which the RL agent operates to drive the control loop. The policy conditions on lightweight graph queries alongside visual and proprioceptive inputs, yielding a compact, semantically informed state for decision making. Experiments on a suite of manipulation tasks with occlusions, distractors, and layout shifts demonstrate consistent gains over strong baselines: the knowledge-conditioned agent achieves higher success rates, improved sample efficiency, and stronger generalization to novel objects and unseen scene configurations. These results support the premise that structured, continuously maintained world knowledge is a powerful inductive bias for scalable, generalizable manipulation: when the knowledge module participates in the RL computation graph, relational representations align with control, enabling robust long-horizon behavior under partial observability.

Benchmarks such as MMLU suggest flagship language models approach factuality saturation, with scores above 90\%. We show this picture is incomplete. \emph{LLMpedia} generates encyclopedic articles entirely from parametric memory, producing ${\sim}$1M articles across three model families without retrieval. For gpt-5-mini, the verifiable true rate on Wikipedia-covered subjects is only 74.7\% -- more than 15 percentage points below the benchmark-based picture, consistent with the availability bias of fixed-question evaluation. Beyond Wikipedia, frontier subjects verifiable only through curated web evidence fall further to 63.2\% true rate. Wikipedia covers just 61\% of surfaced subjects, and three model families overlap by only 7.3\% in subject choice. In a capture-trap benchmark inspired by prior analysis of Grokipedia, LLMpedia achieves substantially higher factuality at roughly half the textual similarity to Wikipedia. Unlike Grokipedia, every prompt, artifact, and evaluation verdict is publicly released, making LLMpedia the first fully open parametric encyclopedia -- bridging factuality evaluation and knowledge materialization. All data, code, and a browsable interface are at https://llmpedia.net.

Recently, multimodal large language models (MLLMs) have emerged as a unified paradigm for language and image generation. Compared with diffusion models, MLLMs possess a much stronger capability for semantic understanding, enabling them to process more complex textual inputs and comprehend richer contextual meanings. However, this enhanced semantic ability may also introduce new and potentially greater safety risks. Taking diffusion models as a reference point, we systematically analyze and compare the safety risks of emerging MLLMs along two dimensions: unsafe content generation and fake image synthesis. Across multiple unsafe generation benchmark datasets, we observe that MLLMs tend to generate more unsafe images than diffusion models. This difference partly arises because diffusion models often fail to interpret abstract prompts, producing corrupted outputs, whereas MLLMs can comprehend these prompts and generate unsafe content. For current advanced fake image detectors, MLLM-generated images are also notably harder to identify. Even when detectors are retrained with MLLMs-specific data, they can still be bypassed by simply providing MLLMs with longer and more descriptive inputs. Our measurements indicate that the emerging safety risks of the cutting-edge generative paradigm, MLLMs, have not been sufficiently recognized, posing new challenges to real-world safety.

Sensor placement for leakage detection in water distribution networks is an important and practical challenge for water utilities. Recent work has shown that graph neural networks can estimate and predict pressures and detect leaks, but their performance strongly depends on the available sensor measurements and configurations. In this paper, we investigate how sensor placement influences the performance of GNN-based leakage detection. We propose a novel PageRank-Centrality-based sensor placement method and demonstrate that it substantially impacts reconstruction, prediction, and leakage detection on the EPANET Net1.

Knowledge Tracing (KT) is a critical technique for modeling student knowledge to support personalized learning. However, most KT systems focus on binary correctness prediction and cannot diagnose the underlying conceptual misunderstandings that lead to errors. Such fine-grained diagnostic feedback is essential for designing targeted instruction and effective remediation. In this work, we introduce the task of concept-level deficiency prediction, which extends traditional KT by identifying the specific concepts a student is likely to struggle with on future problems. We present ConceptKT, a dataset annotated with labels that capture both the concepts required to solve each question and the missing concepts underlying incorrect responses. We investigate in-context learning approaches to KT and evaluate the diagnostic capabilities of various Large Language Models (LLMs) and Large Reasoning Models (LRMs). Different strategies for selecting informative historical records are explored. Experimental results demonstrate that selecting response histories based on conceptual alignment and semantic similarity leads to improved performance on both correctness prediction and concept-level deficiency identification.

Current prompting paradigms for large language models (LLMs), including Chain-of-Thought (CoT) and Tree-of-Thoughts (ToT), follow linear or tree-structured reasoning paths that lack persistent memory, strategic dormancy, and cross-domain synthesis. We present the Enhanced Mycelium of Thought (EMoT) framework, a bio-inspired reasoning architecture that organises cognitive processing into a four-level hierarchy (Micro, Meso, Macro, Meta), implements strategic dormancy and reactivation of reasoning nodes, and integrates a Memory Palace with five mnemonic encoding styles. EMoT is a research prototype for complex, multi-domain problems, not a general-purpose prompting enhancement. Two complementary evaluations reveal a characteristic trade-off. In a blind LLM-as-Judge evaluation across three domains, EMoT achieved near-parity with CoT (4.20 vs. 4.33/5.0) with higher stability, and outperformed CoT on Cross-Domain Synthesis (4.8 vs. 4.4). Ablation studies show that strategic dormancy is architecturally essential (quality collapsed from 4.2 to 1.0 when disabled). On a 15-item short-answer benchmark, EMoT (27%) substantially underperformed simpler baselines, confirming systematic overthinking on simple problems. These results are subject to important limitations: small sample sizes (n=3 complex cases, n=15 short-answer items), LLM-as-Judge evaluation with potential self-preference bias, and approximately 33-fold computational cost overhead. To our knowledge, EMoT is the first reasoning framework to combine hierarchical topology, strategic thought dormancy with reactivation, and mnemonic memory encoding in a single architecture.

Object hallucination in Large Vision-Language Models (LVLMs) severely compromises their reliability in real-world applications, posing a critical barrier to their deployment in high-stakes scenarios such as autonomous driving and medical image analysis. Through systematic empirical investigation, we identify that the imbalanced attention allocation, both across modalities (i.e., vision and language) and within modalities (among individual tokens), exhibits a strong causal correlation with the occurrence of object hallucination. Leveraging this insight, we introduce a novel concept termed attention imbalance, which not only quantifies the degree of attention disparity but also visually delineates the underlying patterns (e.g., over-attentiveness to irrelevant language tokens or under-attentiveness to discriminative visual features) that drive object hallucination. To mitigate object hallucination, we further propose Attention Imbalance Rectification (AIR), a lightweight decoding-time intervention method that reallocates attention weights and adjusts attention distributions to rectify modality-wise and token-wise imbalances. Extensive evaluations on four mainstream LVLMs and three benchmarks (CHAIR, POPE, and MM-Vet) with seven baselines demonstrate that AIR consistently reduces object hallucination rates, achieving up to a 35.1% reduction compared to the baselines, while improving up to 15.9% of LVLMs' general capability across diverse vision-language tasks.

The integration of GNSS data into portable devices has led to the generation of vast amounts of trajectory data, which is crucial for applications such as map-matching. To tackle the limitations of rule-based methods, recent works in deep learning for trajectory-related tasks occur. However, existing models remain challenging due to issues such as the difficulty of large-scale data labeling, ineffective modeling of spatial-temporal relationships, and discrepancies between training and test data distributions. To tackle these challenges, we propose HSTGMatch, a novel model designed to enhance map-matching performance. Our approach involves a two-stage process: hierarchical self-supervised learning and spatial-temporal supervised learning. We introduce a hierarchical trajectory representation, leveraging both grid cells and geographic tuples to capture moving patterns effectively. The model constructs an Adaptive Trajectory Adjacency Graph to dynamically capture spatial relationships, optimizing GATs for improved efficiency. Furthermore, we incorporate a Spatial-Temporal Factor to extract relevant features and employ a decay coefficient to address variations in trajectory length. Our extensive experiments demonstrate the model's superior performance, module effectiveness, and robustness, providing a promising solution for overcoming the existing limitations in map-matching applications. The source code of HSTGMatch is publicly available on GitHub at https://github.com/Nerooo-g/HSTGMatch.

Tool-use capabilities are vital for Large Language Models (LLMs) in finance, a domain characterized by massive investment targets and data-intensive inquiries. However, existing data synthesis methods typically rely on a reverse synthesis paradigm, generating user queries from pre-sampled tools. This approach inevitably introduces artificial explicitness, yielding queries that fail to capture the implicit, event-driven nature of real-world needs. Moreover, its reliance on static tool sets overlooks the dynamic retrieval process required to navigate massive tool spaces. To address these challenges, we introduce \textit{FinToolSyn}, a forward synthesis framework designed to generate high-quality financial dialogues. Progressing from persona instruction and atomic tool synthesis to dynamic retrieval dialogue generation, our pipeline constructs a repository of 43,066 tools and synthesizes over 148k dialogue instances, incorporating dynamic retrieval to emulate the noisy candidate sets typical of massive tool spaces. We also establish a dedicated benchmark to evaluate tool-calling capabilities in realistic financial scenarios. Extensive experiments demonstrate that models trained on FinToolSyn achieve a 21.06\% improvement, providing a robust foundation for tool learning in financial scenarios.

Standard LoRA fine-tuning of Mixture-of-Experts (MoE) models applies adapters to every expert, yet our profiling shows that per-layer expert routing is highly skewed: a small subset of experts handles most tokens in each layer, while many others are rarely activated ("cold"). We propose MoE-Sieve, a simple routing-guided framework for LoRA fine-tuning, and pair it with a systematic profiling study of expert routing across architectures and tasks. The method is simple: profile routing counts on a small calibration set, select the top-k most-routed experts per layer, and apply LoRA only to those experts. Across two architecturally distinct MoE models and three diverse tasks, tuning only the top 25% routed experts per layer remains competitive with full LoRA, with mean differences within +/-1 percentage point across all conditions. This reduces LoRA trainable parameters by 70-73%, adapter checkpoint size by 71-73%, and wall-clock training time by up to 50%. We also observe a non-monotonic relationship between expert count and seed-to-seed variance, consistent with the hypothesis that adapting cold experts can introduce gradient noise without improving accuracy. Further ablations show that random expert selection at matched budget is about 2.5 percentage points worse, indicating that the routing signal matters, while greedy per-layer budget optimization does not improve over uniform top-k.

Deep neural networks (DNNs) achieve remarkable predictive performance but remain difficult to interpret, largely due to overparameterization that obscures the minimal structure required for interpretation. Here we introduce DeepIn, a self-interpretable neural network framework that adaptively identifies and learns the minimal representation necessary for preserving the full expressive capacity of standard DNNs. We show that DeepIn can correctly identify the minimal representation dimension, select relevant variables, and recover the minimal sufficient network architecture for prediction. The resulting estimator achieves optimal non-asymptotic error rates that adapt to the learned minimal dimension, demonstrating that recovering minimal sufficient structure fundamentally improves generalization error. Building on these guarantees, we further develop hypothesis testing procedures for both selected variables and learned representations, bridging deep representation learning with formal statistical inference. Across biomedical and vision benchmarks, DeepIn improves both predictive accuracy and interpretability, reducing error by up to 30% on real-world datasets while automatically uncovering human-interpretable discriminative patterns. Our results suggest that interpretability and statistical rigor can be embedded directly into deep architectures without sacrificing performance.

Contextual automatic speech recognition (ASR) with Speech-LLMs is typically trained with oracle conversation history, but relies on error-prone history at inference, causing a train-test mismatch in the context channel that we term contextual exposure bias. We propose a unified training framework to improve robustness under realistic histories: (i) Teacher Error Knowledge by using Whisper large-v3 hypotheses as training-time history, (ii) Context Dropout to regularize over-reliance on history, and (iii) Direct Preference Optimization (DPO) on curated failure cases. Experiments on TED-LIUM 3 (in-domain) and zero-shot LibriSpeech (out-of-domain) show consistent gains under predicted-history decoding. With a two-utterance history as context, SFT with Whisper hypotheses reduce WER from 5.59% (oracle-history training) to 5.47%, and DPO further improves to 5.17%. Under irrelevant-context attacks, DPO yields the smallest degradation (5.17% -> 5.63%), indicating improved robustness to misleading context. Our code and models are published on https://github.com/XYGuo1996/Contextual_Speech_LLMs.

Predict-and-search (PaS) methods have shown promise for accelerating mixed-integer linear programming (MILP) solving. However, existing approaches typically assume variable independence and rely on deterministic single-point predictions, which limits solution diversityand often necessitates extensive downstream search for high-quality solutions. In this paper, we propose \textbf{SRG}, a generative framework based on Lagrangian relaxation-guided stochastic differential equations (SDEs), with theoretical guarantees on solution quality. SRG leverages convolutional kernels to capture inter-variable dependencies while integrating Lagrangian relaxation to guide the sampling process toward feasible and near-optimal regions. Rather than producing a single estimate, SRG generates diverse, high-quality solution candidates that collectively define compact and effective trust-region subproblems for standard MILP solvers. Across multiple public benchmarks, SRG consistently outperforms existing machine learning baselines in solution quality. Moreover, SRG demonstrates strong zero-shot transferability: on unseen cross-scale/problem instances, it achieves competitive optimality with state-of-the-art exact solvers while significantly reducing computational overhead through faster search and superior solution quality.

Unsupervised learning of high-dimensional data is challenging due to irrelevant or noisy features obscuring underlying structures. It's common that only a few features, called the influential features, meaningfully define the clusters. Recovering these influential features is helpful in data interpretation and clustering. We propose i-IF-Learn, an iterative unsupervised framework that jointly performs feature selection and clustering. Our core innovation is an adaptive feature selection statistic that effectively combines pseudo-label supervision with unsupervised signals, dynamically adjusting based on intermediate label reliability to mitigate error propagation common in iterative frameworks. Leveraging low-dimensional embeddings (PCA or Laplacian eigenmaps) followed by $k$-means, i-IF-Learn simultaneously outputs influential feature subset and clustering labels. Numerical experiments on gene microarray and single-cell RNA-seq datasets show that i-IF-Learn significantly surpasses classical and deep clustering baselines. Furthermore, using our selected influential features as preprocessing substantially enhances downstream deep models such as DeepCluster, UMAP, and VAE, highlighting the importance and effectiveness of targeted feature selection.

Applying large, proprietary API-based language models to text-to-SQL tasks poses a significant industry challenge: reliance on massive, schema-heavy prompts results in prohibitive per-token API costs and high latency, hindering scalable production deployment. We present a specialized, self-hosted 8B-parameter model designed for a conversational bot in CriQ, a sister app to Dream11, India's largest fantasy sports platform with over 250 million users, that answers user queries about cricket statistics. Our novel two-phase supervised fine-tuning approach enables the model to internalize the entire database schema, eliminating the need for long-context prompts. This reduces input tokens by over 99%, from a 17k-token baseline to fewer than 100, and replaces costly external API calls with efficient local inference. The resulting system achieves 98.4% execution success and 92.5% semantic accuracy, substantially outperforming a prompt-engineered baseline using Google's Gemini Flash 2.0 (95.6% execution, 89.4% semantic accuracy). These results demonstrate a practical path toward high-precision, low-latency text-to-SQL applications using domain-specialized, self-hosted language models in large-scale production environments.

Vision-language models (VLMs) have shown remarkable general capabilities, yet embodied agents built on them fail at complex tasks, often skipping critical steps, proposing invalid actions, and repeating mistakes. These failures arise from a fundamental gap between the static training data of VLMs and the physical interaction for embodied tasks. VLMs can learn rich semantic knowledge from static data but lack the ability to interact with the world. To address this issue, we introduce ELITE, an embodied agent framework with {E}xperiential {L}earning and {I}ntent-aware {T}ransfer that enables agents to continuously learn from their own environment interaction experiences, and transfer acquired knowledge to procedurally similar tasks. ELITE operates through two synergistic mechanisms, \textit{i.e.,} self-reflective knowledge construction and intent-aware retrieval. Specifically, self-reflective knowledge construction extracts reusable strategies from execution trajectories and maintains an evolving strategy pool through structured refinement operations. Then, intent-aware retrieval identifies relevant strategies from the pool and applies them to current tasks. Experiments on the EB-ALFRED and EB-Habitat benchmarks show that ELITE achieves 9\% and 5\% performance improvement over base VLMs in the online setting without any supervision. In the supervised setting, ELITE generalizes effectively to unseen task categories, achieving better performance compared to state-of-the-art training-based methods. These results demonstrate the effectiveness of ELITE for bridging the gap between semantic understanding and reliable action execution.

Multi-Object Tracking (MOT) has traditionally focused on a few specific categories, restricting its applicability to real-world scenarios involving diverse objects. Open-Vocabulary Multi-Object Tracking (OVMOT) addresses this by enabling tracking of arbitrary categories, including novel objects unseen during training. However, current progress is constrained by two challenges: the lack of continuously annotated video data for training, and the lack of a customized OVMOT framework to synergistically handle detection and association. We address the data bottleneck by constructing C-TAO, the first continuously annotated training set for OVMOT, which increases annotation density by 26x over the original TAO and captures smooth motion dynamics and intermediate object states. For the framework bottleneck, we propose COVTrack++, a synergistic framework that achieves a bidirectional reciprocal mechanism between detection and association through three modules: (1) Multi-Cue Adaptive Fusion (MCF) dynamically balances appearance, motion, and semantic cues for association feature learning; (2) Multi-Granularity Hierarchical Aggregation (MGA) exploits hierarchical spatial relationships in dense detections, where visible child nodes (e.g., object parts) assist occluded parent objects (e.g., whole body) for association feature enhancement; (3) Temporal Confidence Propagation (TCP) recovers flickering detections through high-confidence tracked objects boosting low-confidence candidates across frames, stabilizing trajectories. Extensive experiments on TAO demonstrate state-of-the-art performance, with novel TETA reaching 35.4% and 30.5% on validation and test sets, improving novel AssocA by 4.8% and novel LocA by 5.8% over previous methods, and show strong zero-shot generalization on BDD100K. The code and dataset will be publicly available.

Participatory urban sensing leverages human mobility for large-scale urban data collection, yet existing methods typically rely on centralized optimization and assume homogeneous participants, resulting in rigid assignments that overlook personal preferences and heterogeneous urban contexts. We propose MAPUS, an LLM-based multi-agent framework for personalized and fair participatory urban sensing. In our framework, participants are modeled as autonomous agents with individual profiles and schedules, while a coordinator agent performs fairness-aware selection and refines sensing routes through language-based negotiation. Experiments on real-world datasets show that MAPUS achieves competitive sensing coverage while substantially improving participant satisfaction and fairness, promoting more human-centric and sustainable urban sensing systems.

Islamic inheritance (Ilm al-Mawarith) is a multi-stage legal reasoning task requiring the identification of eligible heirs, resolution of blocking rules (hajb), assignment of fixed and residual shares, handling of adjustments such as awl and radd, and generation of a consistent final distribution. The task is further complicated by variations across legal schools and civil-law codifications, requiring models to operate under explicit legal configurations. We present a retrieval-augmented generation (RAG) pipeline for this setting, combining rule-grounded synthetic data generation, hybrid retrieval (dense and BM25) with cross-encoder reranking, and schema-constrained output validation. A symbolic inheritance calculator is used to generate a large high-quality synthetic corpus with full intermediate reasoning traces, ensuring legal and numerical consistency. The proposed system achieves a MIR-E score of 0.935 and ranks first on the official QIAS 2026 blind-test leaderboard. Results demonstrate that retrieval-grounded, schema-aware generation significantly improves reliability in high-precision Arabic legal reasoning tasks.

Multimodal Large Language Models (MLLMs) have demonstrated remarkable reasoning capabilities across modalities such as images and text. However, tabular data, despite being a critical real-world modality, remains relatively underexplored in multimodal learning. In this paper, we focus on the task of Tabular-Vision Multi-Modal Understanding (TVMU) and identify three core challenges: (1) high structural variability and data incompleteness in tables, (2) implicit and complex feature dependencies, and (3) significant heterogeneity in problem-solving pipelines across downstream tasks. To address these issues, we propose Thinking with Tables (TWT). TWT employs a program-aided code-based neuro-symbolic reasoning mechanism that facilitates key operations, such as information extraction and element modeling, by interacting with external environments. We evaluate TWT on eight representative datasets. Experimental results demonstrate that TWT consistently outperforms existing baselines by an average of 10\% in accuracy, achieving performance comparable to, or even surpassing, proprietary commercial SOTA LLMs on TVMU tasks. Models and codes are available at https://github.com/kunyang-YU/Thinking-with-Tables

Physics-Informed Neural Networks (PINNs) for high-dimensional and high-order partial differential equations (PDEs) are primarily constrained by the $\mathcal{O}(d^k)$ spatial derivative complexity and the $\mathcal{O}(P)$ memory overhead of backpropagation (BP). While randomized spatial estimators successfully reduce the spatial complexity to $\mathcal{O}(1)$, their reliance on first-order optimization still leads to prohibitive memory consumption at scale. Zeroth-order (ZO) optimization offers a BP-free alternative; however, naively combining randomized spatial operators with ZO perturbations triggers a variance explosion of $\mathcal{O}(1/\varepsilon^2)$, leading to numerical divergence. To address these challenges, we propose the \textbf{S}tochastic \textbf{D}imension-free \textbf{Z}eroth-order \textbf{E}stimator (\textbf{SDZE}), a unified framework that achieves dimension-independent complexity in both space and memory. Specifically, SDZE leverages \emph{Common Random Numbers Synchronization (CRNS)} to algebraically cancel the $\mathcal{O}(1/\varepsilon^2)$ variance by locking spatial random seeds across perturbations. Furthermore, an \emph{implicit matrix-free subspace projection} is introduced to reduce parameter exploration variance from $\mathcal{O}(P)$ to $\mathcal{O}(r)$ while maintaining an $\mathcal{O}(1)$ optimizer memory footprint. Empirical results demonstrate that SDZE enables the training of 10-million-dimensional PINNs on a single NVIDIA A100 GPU, delivering significant improvements in speed and memory efficiency over state-of-the-art baselines.

Existing approaches to increasing the effective depth of Transformers predominantly rely on parameter reuse, extending computation through recursive execution. Under this paradigm, the network structure remains static along the training timeline, and additional computational depth is uniformly assigned to entire blocks at the parameter level. This rigidity across training time and parameter space leads to substantial computational redundancy during training. In contrast, we argue that depth allocation during training should not be a static preset, but rather a progressively growing structural process. Our systematic analysis reveals a deep-to-shallow maturation trajectory across layers, where high-entropy attention heads play a crucial role in semantic integration. Motivated by this observation, we introduce the Sparse Growing Transformer (SGT). SGT is a training-time sparse depth allocation framework that progressively extends recurrence from deeper to shallower layers via targeted attention looping on informative heads. This mechanism induces structural sparsity by selectively increasing depth only for a small subset of parameters as training evolves. Extensive experiments across multiple parameter scales demonstrate that SGT consistently outperforms training-time static block-level looping baselines under comparable settings, while reducing the additional training FLOPs overhead from approximately 16--20% to only 1--3% relative to a standard Transformer backbone.

Generative optimization uses large language models (LLMs) to iteratively improve artifacts (such as code, workflows or prompts) using execution feedback. It is a promising approach to building self-improving agents, yet in practice remains brittle: despite active research, only 9% of surveyed agents used any automated optimization. We argue that this brittleness arises because, to set up a learning loop, an engineer must make ``hidden'' design choices: What can the optimizer edit and what is the "right" learning evidence to provide at each update? We investigate three factors that affect most applications: the starting artifact, the credit horizon for execution traces, and batching trials and errors into learning evidence. Through case studies in MLAgentBench, Atari, and BigBench Extra Hard, we find that these design decisions can determine whether generative optimization succeeds, yet they are rarely made explicit in prior work. Different starting artifacts determine which solutions are reachable in MLAgentBench, truncated traces can still improve Atari agents, and larger minibatches do not monotonically improve generalization on BBEH. We conclude that the lack of a simple, universal way to set up learning loops across domains is a major hurdle for productionization and adoption. We provide practical guidance for making these choices.

Monolithic Large Language Models (LLMs) used in educational dialogue often behave as "black boxes," where pedagogical decisions are implicit and difficult to audit, frequently violating instructional constraints by providing answers too early. We introduce the Ensemble of Specialized LLMS (ES-LLMS) architecture that separates decision-making from wording. Pedagogical actions are selected by a deterministic rules-based orchestrator coordinating specialized agents covering tutoring, assessment, feedback, scaffolding, motivation and ethics-guided by an interpretable Bayesian Knowledge Tracing (BKT) student model. An LLM renderer surface-realizes the chosen action in natural language. This design emphasizes reliability and controllability: constraints such as "attempt-before-hint" and hint caps are enforced as explicit rules, and the system logs per-turn agent traces and constraint checks. Validation of pedagogical quality via human expert reviewers (N=6) and a multi-LLM-as-Judge panel (six state-of-the-art models) showed that ES-LLMs were preferred in 91.7% and 79.2% of cases, respectively. The architecture significantly outperformed monolithic baselines across all seven dimensions, particularly in Scaffolding & Guidance, and Trust & Explainability. Furthermore, a Monte Carlo simulation (N=2,400) exposed a "Mastery Gain Paradox," where monolithic tutors inflated short-term performance through over-assistance. In contrast, ES-LLMs achieved 100% adherence to pedagogical constraints (e.g., attempt-before-hint) and a 3.3x increase in hint efficiency. Operationally, ES-LLMs reduced costs by 54% and latency by 22% by utilizing stateless prompts. We conclude that structural decoupling is essential for transforming stochastic models into trustworthy, verifiable and resource-efficient pedagogical agents.

Retrieval-augmented generation (RAG) has shown promising results in enhancing Q&A by incorporating information from the web and other external sources. However, the supporting documents retrieved from the heterogeneous web often originate from multiple sources with diverse writing styles, varying formats, and inconsistent granularity. Fusing such multi-source documents into a coherent and knowledge-intensive context remains a significant challenge, as the presence of irrelevant and redundant information can compromise the factual consistency of the inferred answers. This paper proposes the Concept-oriented Context Reconstruction RAG (CoCR-RAG), a framework that addresses the multi-source information fusion problem in RAG through linguistically grounded concept-level integration. Specifically, we introduce a concept distillation algorithm that extracts essential concepts from Abstract Meaning Representation (AMR), a stable semantic representation that structures the meaning of texts as logical graphs. The distilled concepts from multiple retrieved documents are then fused and reconstructed into a unified, information-intensive context by Large Language Models, which supplement only the necessary sentence elements to highlight the core knowledge. Experiments on the PopQA and EntityQuestions datasets demonstrate that CoCR-RAG significantly outperforms existing context-reconstruction methods across these Web Q&A benchmarks. Furthermore, CoCR-RAG shows robustness across various backbone LLMs, establishing itself as a flexible, plug-and-play component adaptable to different RAG frameworks.

Deploying clinical ML is slow and brittle: models that work at one hospital often degrade under distribution shifts at the next. In this work, we study a simple question -- can large language models (LLMs) create portable patient embeddings i.e. representations of patients enable a downstream predictor built on one hospital to be used elsewhere with minimal-to-no retraining and fine-tuning. To do so, we map from irregular ICU time series onto concise natural language summaries using a frozen LLM, then embed each summary with a frozen text embedding model to obtain a fixed length vector capable of serving as input to a variety of downstream predictors. Across three cohorts (MIMIC-IV, HIRID, PPICU), on multiple clinically grounded forecasting and classification tasks, we find that our approach is simple, easy to use and competitive with in-distribution with grid imputation, self-supervised representation learning, and time series foundation models, while exhibiting smaller relative performance drops when transferring to new hospitals. We study the variation in performance across prompt design, with structured prompts being crucial to reducing the variance of the predictive models without altering mean accuracy. We find that using these portable representations improves few-shot learning and does not increase demographic recoverability of age or sex relative to baselines, suggesting little additional privacy risk. Our work points to the potential that LLMs hold as tools to enable the scalable deployment of production grade predictive models by reducing the engineering overhead.

Large language models (LLMs) have demonstrated remarkable capabilities, but their massive scale poses significant challenges for practical deployment. Structured pruning offers a promising solution by removing entire dimensions or layers, yet existing methods face critical trade-offs: task-agnostic approaches cannot adapt to task-specific requirements, while task-aware methods require costly training to learn task adaptability. We propose DIET (Dimension-wise global pruning of LLMs via merging Task-wise importance scores), a training-free structured pruning method that combines dimension-level granularity with task-aware selection. DIET profiles activation magnitudes across tasks using only 100 samples per task, then applies majority voting to construct a single global mask. DIET does not require large costs from pre-computation or training. Experiments on seven zero-shot benchmarks using Gemma-2 2B and 9B models demonstrate the effectiveness of DIET; for example, at 20% sparsity on Gemma-2 2B, DIET achieves near 10% average accuracy improvement, compared to previous state-of-the-art structured pruning methods. This advantage persists across various sparsity levels and model scales, positioning DIET as a practical and robust choice for structured LLM pruning.

In recent years, a number of neural-network (NN) methods have exhibited good performance in seismic data processing, such as denoising, interpolation, and frequency-band extension. However, these methods rely on stacked perceptrons and standard activation functions, which imposes a bottleneck on the representational capacity of deep-learning models, making it difficult to capture the complex and non-stationary dynamics of seismic wavefields. Different from the classical perceptron-stacked NNs which are fundamentally confined to real-valued Euclidean spaces, the quantum NNs leverage the exponential state space of quantum mechanics to map the features into high-dimensional Hilbert spaces, transcending the representational boundary of classical NNs. Based on this insight, we propose a quantum-classical synergistic generative adversarial network (QC-GAN) for seismic data processing, serving as the first application of quantum NNs in seismic exploration. In QC-GAN, a quantum pathway is used to exploit the high-order feature correlations, while the convolutional pathway specializes in extracting the waveform structures of seismic wavefields. Furthermore, we design a QC feature complementarity loss to enforce the feature orthogonality in the proposed QC-GAN. This novel loss function can ensure that the two pathways encode non-overlapping information to enrich the capacity of feature representation. On the whole, by synergistically integrating the quantum and convolutional pathways, the proposed QC-GAN breaks the representational bottleneck inherent in classical GAN. Experimental results on denoising and interpolation tasks demonstrate that QC-GAN preserves wavefield continuity and amplitude-phase information under complex noise conditions.

Recent advances in real-time interactive text-driven motion generation have enabled humanoids to perform diverse behaviors. However, kinematics-only generators often exhibit physical hallucinations, producing motion trajectories that are physically infeasible to track with a downstream motion tracking controller or unsafe for real-world deployment. These failures often arise from the lack of explicit physics-aware objectives for real-robot execution and become more severe under out-of-distribution (OOD) user inputs. Hence, we propose SafeFlow, a text-driven humanoid whole-body control framework that combines physics-guided motion generation with a 3-Stage Safety Gate driven by explicit risk indicators. SafeFlow adopts a two-level architecture. At the high level, we generate motion trajectories using Physics-Guided Rectified Flow Matching in a VAE latent space to improve real-robot executability, and further accelerate sampling via Reflow to reduce the number of function evaluations (NFE) for real-time control. The 3-Stage Safety Gate enables selective execution by detecting semantic OOD prompts using a Mahalanobis score in text-embedding space, filtering unstable generations via a directional sensitivity discrepancy metric, and enforcing final hard kinematic constraints such as joint and velocity limits before passing the generated trajectory to a low-level motion tracking controller. Extensive experiments on the Unitree G1 demonstrate that SafeFlow outperforms prior diffusion-based methods in success rate, physical compliance, and inference speed, while maintaining diverse expressiveness.

Quantum circuit initialisation is a key bottleneck in variational quantum algorithms (VQAs), strongly impacting optimisation stability and convergence. Recent work shows that large language models (LLMs) can synthesise high-quality variational circuit architectures, but their continuous parameter predictions are unreliable. Conversely, data-driven initialisation methods such as BEINIT improve trainability via problem-adaptive priors, yet assume fixed ansatz templates and ignore generative circuit structure. We propose BRIDG-Q (Barren-Plateau-Resilient Initialisation with Data-Aware LLM-Generated Quantum Circuits), a neuro-symbolic pipeline that bridges this gap by coupling LLM-generated circuit architectures with empirical-Bayes parameter initialisation. BRIDG-Q uses AgentQ to generate problem-conditioned circuit topologies, removes generated parameters, and injects data-informed parameter initialisations to mitigate barren plateau effects. Evaluations on graph optimisation benchmarks using residual energy gap and convergence metrics show improved optimisation robustness, indicating that data-driven initialisation remains effective even for LLM-generated circuits, with oracle per-instance selection achieving approximately a 10% reduction in final residual energy.

Deep learning architectures are fundamentally inspired by neuroscience, particularly the structure of the brain's sensory pathways, and have achieved remarkable success in learning informative data representations. Although these architectures mimic the communication mechanisms of biological neurons, their strategies for information encoding and transmission are fundamentally distinct. Biological systems depend on dynamic fluctuations in membrane potential; by contrast, conventional deep networks optimize weights and biases by adjusting the strengths of inter-neural connections, lacking a systematic mechanism to jointly characterize the interplay among signal intensity, coupling structure, and state evolution. To tackle this limitation, we propose the Kirchhoff-Inspired Neural Network (KINN), a state-variable-based network architecture constructed based on Kirchhoff's current law. KINN derives numerically stable state updates from fundamental ordinary differential equations, enabling the explicit decoupling and encoding of higher-order evolutionary components within a single layer while preserving physical consistency, interpretability, and end-to-end trainability. Extensive experiments on partial differential equation (PDE) solving and ImageNet image classification validate that KINN outperforms state-of-the-art existing methods.

Machine vision, including object recognition and image reconstruction, is a central technology in many consumer devices and scientific instruments. The design of machine-vision systems has been revolutionized by the adoption of end-to-end optimization, in which the optical front end and the post-processing back end are jointly optimized. However, while machine vision currently works extremely well in moderate-light or bright-light situations -- where a camera may detect thousands of photons per pixel and billions of photons per frame -- it is far more challenging in very low-light situations. We introduce photon-aware neuromorphic sensing (PANS), an approach for end-to-end optimization in highly photon-starved scenarios. The training incorporates knowledge of the low photon budget and the stochastic nature of light detection when the average number of photons per pixel is near or less than 1. We report a proof-of-principle experimental demonstration in which we performed low-light image classification using PANS, achieving 73% (82%) accuracy on FashionMNIST with an average of only 4.9 (17) detected photons in total per inference, and 86% (97%) on MNIST with 8.6 (29) detected photons -- orders of magnitude more photon-efficient than conventional approaches. We also report simulation studies showing how PANS could be applied to other classification, event-detection, and image-reconstruction tasks. By taking into account the statistics of measurement results for non-classical states or alternative sensing hardware, PANS could in principle be adapted to enable high-accuracy results in quantum and other photon-starved setups.

Large language models (LLMs) have achieved remarkable progress in many language tasks, yet they continue to struggle with complex historical and religious Arabic texts such as the Quran and Hadith. To address this limitation, we develop a retrieval-augmented generation (RAG) framework grounded in diachronic lexicographic knowledge. Unlike prior RAG systems that rely on general-purpose corpora, our approach retrieves evidence from the Doha Historical Dictionary of Arabic (DHDA), a large-scale resource documenting the historical development of Arabic vocabulary. The proposed pipeline combines hybrid retrieval with an intent-based routing mechanism to provide LLMs with precise, contextually relevant historical information. Our experiments show that this approach improves the accuracy of Arabic-native LLMs, including Fanar and ALLaM, to over 85\%, substantially reducing the performance gap with Gemini, a proprietary large-scale model. Gemini also serves as an LLM-as-a-judge system for automatic evaluation in our experiments. The automated judgments were verified through human evaluation, demonstrating high agreement (kappa = 0.87). An error analysis further highlights key linguistic challenges, including diacritics and compound expressions. These findings demonstrate the value of integrating diachronic lexicographic resources into retrieval-augmented generation frameworks to enhance Arabic language understanding, particularly for historical and religious texts. The code and resources are publicly available at: https://github.com/somayaeltanbouly/Doha-Dictionary-RAG.

Developers and consumers increasingly choose reasoning language models (RLMs) based on their listed API prices. However, how accurately do these prices reflect actual inference costs? We conduct the first systematic study of this question, evaluating 8 frontier RLMs across 9 diverse tasks covering competition math, science QA, code generation, and multi-domain reasoning. We uncover the pricing reversal phenomenon: in 21.8% of model-pair comparisons, the model with a lower listed price actually incurs a higher total cost, with reversal magnitude reaching up to 28x. For example, Gemini 3 Flash's listed price is 78% cheaper than GPT-5.2's, yet its actual cost across all tasks is 22% higher. We trace the root cause to vast heterogeneity in thinking token consumption: on the same query, one model may use 900% more thinking tokens than another. In fact, removing thinking token costs reduces ranking reversals by 70% and raises the rank correlation (Kendall's $τ$ ) between price and cost rankings from 0.563 to 0.873. We further show that per-query cost prediction is fundamentally difficult: repeated runs of the same query yield thinking token variation up to 9.7x, establishing an irreducible noise floor for any predictor. Our findings demonstrate that listed API pricing is an unreliable proxy for actual cost, calling for cost-aware model selection and transparent per-request cost monitoring.

Achieving agile and reconfigurable production flows in smart factories depends on online multi-robot task assignment (MRTA), which requires online collision-free and congestion-free route scheduling of transportation multi-robot systems (T-MRS), e.g., collaborative automatic guided vehicles (AGVs). Due to the real-time operational requirements and dynamic interactions between T-MRS and production MRS, online scheduling under partial observability in dynamic factory environments remains a significant and under-explored challenge. This paper proposes a novel communication-enabled online scheduling framework that explicitly couples wireless machine-to-machine (M2M) networking with route scheduling, enabling AGVs to exchange intention information, e.g., planned routes, to overcome partial observations and assist complex computation of online scheduling. Specifically, we determine intelligent AGVs' intention and sensor data as new M2M traffic and tailor the retransmission-free multi-link transmission networking to meet real-time operation demands. This scheduling-oriented networking is then integrated with a simulated annealing-based MRTA scheme and a congestion-aware A*-based route scheduling method. The integrated communication and scheduling scheme allows AGVs to dynamically adjust collision-free and congestion-free routes with reduced computational overhead. Numerical experiments shows the impacts from wireless communication on the performance of T-MRS and suggest that the proposed integrated scheme significantly enhances scheduling efficiency compared to other baselines, even under high AGV load conditions and limited channel resources. Moreover, the results reveal that the scheduling-oriented wireless M2M communication design fundamentally differs from human-to-human communications, implying new technological opportunities in a wireless networked smart factory.

With frequently evolving Advanced Persistent Threats (APTs) in cyberspace, traditional security solutions approaches have become inadequate for threat hunting for organizations. Moreover, SOC (Security Operation Centers) analysts are often overwhelmed and struggle to analyze the huge volume of logs received from diverse devices in organizations. To address these challenges, we propose an automated and dynamic threat hunting framework for monitoring evolving threats, adapting to changing network conditions, and performing risk-based prioritization for the mitigation of suspicious and malicious traffic. By integrating Agentic AI with Splunk, an established SIEM platform, we developed a unique threat hunting framework. The framework systematically and seamlessly integrates different threat hunting modules together, ranging from traffic ingestion to anomaly assessment using a reconstruction-based autoencoder, deep reinforcement learning (DRL) with two layers for initial triage, and a large language model (LLM) for contextual analysis. We evaluated the framework against a publicly available benchmark dataset, as well as against a simulated dataset. The experimental results show that the framework can effectively adapt to different SOC objectives autonomously and identify suspicious and malicious traffic. The framework enhances operational effectiveness by supporting SOC analysts in their decision-making to block, allow, or monitor network traffic. This study thus enhances cybersecurity and threat hunting literature by presenting the novel threat hunting framework for security decision- making, as well as promoting cumulative research efforts to develop more effective frameworks to battle continuously evolving cyber threats.

The remarkable progress of reinforcement learning (RL) is intrinsically tied to the environments used to train and evaluate artificial agents. Moving beyond traditional qualitative reviews, this work presents a large-scale, data-driven empirical investigation into the evolution of RL environments. By programmatically processing a massive corpus of academic literature and rigorously distilling over 2,000 core publications, we propose a quantitative methodology to map the transition from isolated physical simulations to generalist, language-driven foundation agents. Implementing a novel, multi-dimensional taxonomy, we systematically analyze benchmarks against diverse application domains and requisite cognitive capabilities. Our automated semantic and statistical analysis reveals a profound, data-verified paradigm shift: the bifurcation of the field into a "Semantic Prior" ecosystem dominated by Large Language Models (LLMs) and a "Domain-Specific Generalization" ecosystem. Furthermore, we characterize the "cognitive fingerprints" of these distinct domains to uncover the underlying mechanisms of cross-task synergy, multi-domain interference, and zero-shot generalization. Ultimately, this study offers a rigorous, quantitative roadmap for designing the next generation of Embodied Semantic Simulators, bridging the gap between continuous physical control and high-level logical reasoning.

Deep-sea cold seep stage assessment has traditionally relied on costly, high-risk manned submersible operations and visual surveys of macrofauna. Although microbial communities provide a promising and more cost-effective alternative, reliable inference remains challenging because the available deep-sea dataset is extremely small ($n = 13$) relative to the microbial feature dimension ($p = 26$), making purely data-driven models highly prone to overfitting. To address this, we propose a knowledge-enhanced classification framework that incorporates an ecological knowledge graph as a structural prior. By fusing macro-microbe coupling and microbial co-occurrence patterns, the framework internalizes established ecological logic into a \underline{\textbf{G}}raph-\underline{\textbf{R}}egularized \underline{\textbf{M}}ultinomial \underline{\textbf{L}}ogistic \underline{\textbf{R}}egression (GRMLR) model, effectively constraining the feature space through a manifold penalty to ensure biologically consistent classification. Importantly, the framework removes the need for macrofauna observations at inference time: macro-microbe associations are used only to guide training, whereas prediction relies solely on microbial abundance profiles. Experimental results demonstrate that our approach significantly outperforms standard baselines, highlighting its potential as a robust and scalable framework for deep-sea ecological assessment.

The continuous evolution of system specifications necessitates frequent evaluation of conflicting requirements, a process that is traditionally labour intensive. Although large language models (LLMs) have demonstrated significant potential for automating this detection, their massive computational requirements often result in excessive energy waste. Consequently, there is a growing need to transition toward Small Language Models (SLMs) and energy aware architectures for sustainable Requirements Engineering. This study proposes and empirically evaluates an energy aware framework that compares Knowledge Graph-based Retrieval (KGR) with Vector-based Semantic Retrieval (VSR) to enhance SLM-based inference at the 7B to 8B parameter scale. By leveraging structured graph traversal and high dimensional semantic mapping, we extract candidate requirements, which are then classified as conflicting or neutral by an inference engine. We evaluate these retrieval enhanced strategies across Zero-Shot, Few-Shot, and Chain of Thoughts prompting methods. Using a three-pillar sustainability framework measuring energy consumption (Wh), latency (s), and carbon emissions (gCO2eq) alongside standard accuracy metrics (F1 Score), this research provides a first systematic empirical evaluation and trade off analysis between predictive performance and environmental impact. Our findings highlight the effectiveness of structured versus semantic retrieval in detecting requirement conflicts, offering a reproducible, sustainability aware architecture for energy efficient requirement engineering.

Vision impairment affects millions globally, and early detection is critical to preventing irreversible vision loss. Ophthalmology workflows require clinicians to integrate medical images, structured clinical data, and free-text notes to determine disease severity and management, which is time-consuming and burdensome. Recent multimodal large language models (MLLMs) show promise, but existing general and medical MLLMs perform poorly in ophthalmology, and few ophthalmology-specific MLLMs are openly available. We present VOLMO (Versatile and Open Large Models for Ophthalmology), a model-agnostic, data-open framework for developing ophthalmology-specific MLLMs. VOLMO includes three stages: ophthalmology knowledge pretraining on 86,965 image-text pairs from 26,569 articles across 82 journals; domain task fine-tuning on 26,929 annotated instances spanning 12 eye conditions for disease screening and severity classification; and multi-step clinical reasoning on 913 patient case reports for assessment, planning, and follow-up care. Using this framework, we trained a compact 2B-parameter MLLM and compared it with strong baselines, including InternVL-2B, LLaVA-Med-7B, MedGemma-4B, MedGemma-27B, and RETFound. We evaluated these models on image description generation, disease screening and staging classification, and assessment-and-management generation, with additional manual review by two healthcare professionals and external validation on three independent cohorts for age-related macular degeneration and diabetic retinopathy. Across settings, VOLMO-2B consistently outperformed baselines, achieving stronger image description performance, an average F1 of 87.4% across 12 eye conditions, and higher scores in external validation.

Discovering improved policy optimization algorithms for language models remains a costly manual process requiring repeated mechanism-level modification and validation. Unlike simple combinatorial code search, this problem requires searching over algorithmic mechanisms tightly coupled with training dynamics while reusing empirical evidence across iterations. We propose POISE, a closed-loop framework for automated discovery of policy optimization algorithms for language models. POISE maintains a structured, genealogically linked archive linking proposals, executable implementations, standardized evaluations, and natural-language reflections to support evidence-driven iteration. In mathematical reasoning experiments starting from GRPO, POISE evaluates 64 candidate algorithms and discovers improved mechanisms, including analytic-variance scaling and validity masking. The best variant improves weighted Overall from 47.8 to 52.5 (+4.6) and increases AIME25 pass@32 from 26.7% to 43.3%, demonstrating the feasibility of automated policy optimization discovery while supporting interpretable design principles.

Argument Mining(AM) aims to uncover the argumentative structures within a text. Previous methods require several subtasks, such as span identification, component classification, and relation classification. Consequently, these methods need rule-based postprocessing to derive argumentative structures from the output of each subtask. This approach adds to the complexity of the model and expands the search space of the hyperparameters. To address this difficulty, we propose a simple yet strong method based on a text-to-text generation approach using a pretrained encoder-decoder language model. Our method simultaneously generates argumentatively annotated text for spans, components, and relations, eliminating the need for task-specific postprocessing and hyperparameter tuning. Furthermore, because it is a straightforward text-to-text generation method, we can easily adapt our approach to various types of argumentative structures. Experimental results demonstrate the effectiveness of our method, as it achieves state-of-the-art performance on three different types of benchmark datasets: the Argument-annotated Essays Corpus(AAEC), AbstRCT, and the Cornell eRulemaking Corpus(CDCP)

Audio fingerprinting converts audio to much lower-dimensional representations, allowing distorted recordings to still be recognized as their originals through similar fingerprints. Existing deep learning approaches rigidly fingerprint fixed-length audio segments, thereby neglecting temporal dynamics during segmentation. To address limitations due to this rigidity, we propose Variable-Length Audio FingerPrinting (VLAFP), a novel method that supports variable-length fingerprinting. To the best of our knowledge, VLAFP is the first deep audio fingerprinting model capable of processing audio of variable length, for both training and testing. Our experiments show that VLAFP outperforms existing state-of-the-arts in live audio identification and audio retrieval across three real-world datasets.

Accurate charge densities are central to electronic-structure theory, but computing charge-state-dependent densities with density functional theory remains too expensive for large-scale screening and defect workflows. We present ChargeFlow, a flow-matching refinement model that transforms a charge-conditioned superposition of atomic densities into the corresponding DFT electron density on the native periodic real-space grid using a 3D U-Net velocity field. Trained on 9,502 charged Materials Project-derived calculations and evaluated on an external 1,671-structure benchmark spanning perovskites, charged defects, diamond defects, metal-organic frameworks, and organic crystals, ChargeFlow is not uniformly best on every in-distribution class but is strongest on problems dominated by nonlocal charge redistribution and charge-state extrapolation, improving deformation-density error from 3.62% to 3.21% and charge- response cosine similarity from 0.571 to 0.655 relative to a ResNet baseline. The predicted densities remain chemically useful under downstream analysis, yielding successful Bader partitioning on all 1,671 benchmark structures and high-fidelity electrostatic potentials, which positions flow matching as a practical density-refinement strategy for charged materials.

Higher Degree by Research (HDR) candidates increasingly depend on cloud-provisioned virtual machines and local GPU hardware for their computational experiments, yet a persistent and under-addressed gap exists between having compute resources and using them productively. Cloud and infrastructure teams can provision virtual machines, but the path from a raw VM to a reproducible, GPU-ready research environment remains a significant barrier for researchers who are domain experts, not systems engineers. We identify this gap as a missing adapter layer between cloud provisioning and interactive research work. We present a lightweight, open-source solution built on k3s and Coder that implements this adapter layer and is already in active use in our research workspace environment. Our CI/CD pipeline connects GitHub directly to the local cluster, deploying research projects in under five minutes. We define a concrete metrics framework for evaluating this layer -- covering deployment latency, environment reproducibility, onboarding friction, and resource utilisation -- and establish baselines against which improvements can be measured.

The proliferation of AIGC-driven face manipulation and deepfakes poses severe threats to media provenance, integrity, and copyright protection. Prior versatile watermarking systems typically rely on embedding explicit localization payloads, which introduces a fidelity--functionality trade-off: larger localization signals degrade visual quality and often reduce decoding robustness under strong generative edits. Moreover, existing methods rarely support content recovery, limiting their forensic value when original evidence must be reconstructed. To address these challenges, we present VeriFi, a versatile watermarking framework that unifies copyright protection, pixel-level manipulation localization, and high-fidelity face content recovery. VeriFi makes three key contributions: (1) it embeds a compact semantic latent watermark that serves as an content-preserving prior, enabling faithful restoration even after severe manipulations; (2) it achieves fine-grained localization without embedding localization-specific artifacts by correlating image features with decoded provenance signals; and (3) it introduces an AIGC attack simulator that combines latent-space mixing with seamless blending to improve robustness to realistic deepfake pipelines. Extensive experiments on CelebA-HQ and FFHQ show that VeriFi consistently outperforms strong baselines in watermark robustness, localization accuracy, and recovery quality, providing a practical and verifiable defense for deepfake forensics.

Most testbeds for omni-modal models assess multimodal understanding via textual outputs, leaving it unclear whether these models can properly speak their answers. To study this, we introduce OmniACBench, a benchmark for evaluating context-grounded acoustic control in omni-modal models. Given a spoken instruction, a text script, and an image, a model must read the script aloud with an appropriate tone and manner. OmniACBench comprises 3,559 verified instances covering six acoustic features: speech rate, phonation, pronunciation, emotion, global accent, and timbre. Extensive experiments on eight models reveal their limitations in the proposed setting, despite their strong performance on prior textual-output evaluations. Our analyses show that the main bottleneck lies not in processing individual modalities, but in integrating multimodal context for faithful speech generation. Moreover, we identify three common failure modes-weak direct control, failed implicit inference, and failed multimodal grounding-providing insights for developing models that can verbalize responses effectively.

Evidence-based medicine (EBM) is central to high-quality care, but remains difficult to implement in fast-paced primary care settings. Physicians face short consultations, increasing patient loads, and lengthy guideline documents that are impractical to consult in real time. To address this gap, we investigate the feasibility of using large language models (LLMs) as ambient assistants that surface targeted, evidence-based questions during physician-patient encounters. Our study focuses on question generation rather than question answering, with the aim of scaffolding physician reasoning and integrating guideline-based practice into brief consultations. We implemented two prompting strategies, a zero-shot baseline and a multi-stage reasoning variant, using Gemini 2.5 as the backbone model. We evaluated on a benchmark of 80 de-identified transcripts from real clinical encounters, with six experienced physicians contributing over 90 hours of structured review. Results indicate that while general-purpose LLMs are not yet fully reliable, they can produce clinically meaningful and guideline-relevant questions, suggesting significant potential to reduce cognitive burden and make EBM more actionable at the point of care.

Large vision-language models (LVLMs) are increasingly being applied to multi-view image inputs captured from diverse viewpoints. However, despite this growing use, current LVLMs often confuse or mismatch visual information originating from different instances or viewpoints, a phenomenon we term multi-view hallucination. To systematically analyze this problem, we construct MVH-Bench, a benchmark comprising 4.8k question-answer pairs targeting two types of hallucination: cross-instance and cross-view. Empirical results show that recent LVLMs struggle to correctly associate visual evidence with its corresponding instance or viewpoint. To overcome this limitation, we propose Reference Shift Contrastive Decoding (RSCD), a training-free decoding technique that suppresses visual interference by generating negative logits through attention masking. Experiments on MVH-Bench with Qwen2.5-VL and LLaVA-OneVision demonstrate that RSCD consistently improves performance by up to 21.1 and 34.6 points over existing hallucination mitigation methods, highlighting the effectiveness of our approach.

The integration of Non-player characters (NPCs) within digital environments has been increasingly recognized for its potential to augment user immersion and cognitive engagement. The sophisticated orchestration of their daily activities, reflecting the nuances of human daily routines, contributes significantly to the realism of digital environments. Nevertheless, conventional approaches often produce monotonous repetition, falling short of capturing the intricacies of real human activity plans. In response to this, we introduce ORACLE, a novel generative model for the synthesis of realistic indoor daily activity plans, ensuring NPCs' authentic presence in digital habitats. Exploiting the CASAS smart home dataset's 24-hour indoor activity sequences, ORACLE addresses challenges in the dataset, including its imbalanced sequential data, the scarcity of training samples, and the absence of pre-trained models encapsulating human daily activity patterns. ORACLE's training leverages the sequential data processing prowess of Transformers, the generative controllability of Conditional Variational Autoencoders (CVAE), and the discriminative refinement of contrastive learning. Our experimental results validate the superiority of generating NPC activity plans and the efficacy of our design strategies over existing methods.

Storage scalability is paramount in the era of big data blockchain. A storage-scalable blockchain can effectively scale out state storage to an arbitrary number of nodes and reduce the storage pressure on each, similar to distributed databases. Prior research has extensively utilized sharding techniques to attain storage scalability; however, these approaches invariably compromise safety and liveness guarantees. In this work, we propose a novel state-execution decoupled architecture, and Supermassive Blockchain, a novel storage-scalable Byzantine fault tolerance (BFT) protocol that can sustain the deterministic security properties of conventional BFT protocols. The state management system employs erasure coding to ensure state availability with scalable storage consumption, while the global consensus and execution layers maintain robust security characteristics. Our evaluation indicates that Supermassive Blockchain achieves better storage scalability compared to prior approaches while incurring low network overhead.

Online reinforcement learning in infinite-horizon Markov decision processes (MDPs) remains less theoretically and algorithmically developed than its episodic counterpart, with many algorithms suffering from high ``burn-in'' costs and failing to adapt to benign instance-specific complexity. In this work, we address these shortcomings for two infinite-horizon objectives: the classical average-reward regret and the $γ$-regret. We develop a single tractable UCB-style algorithm applicable to both settings, which achieves the first optimal variance-dependent regret guarantees. Our regret bounds in both settings take the form $\tilde{O}( \sqrt{SA\,\text{Var}} + \text{lower-order terms})$, where $S,A$ are the state and action space sizes, and $\text{Var}$ captures cumulative transition variance. This implies minimax-optimal average-reward and $γ$-regret bounds in the worst case but also adapts to easier problem instances, for example yielding nearly constant regret in deterministic MDPs. Furthermore, our algorithm enjoys significantly improved lower-order terms for the average-reward setting. With prior knowledge of the optimal bias span $\Vert h^\star\Vert_\text{sp}$, our algorithm obtains lower-order terms scaling as $\Vert h^\star\Vert_\text{sp} S^2 A$, which we prove is optimal in both $\Vert h^\star\Vert_\text{sp}$ and $A$. Without prior knowledge, we prove that no algorithm can have lower-order terms smaller than $\Vert h^\star \Vert_\text{sp}^2 S A$, and we provide a prior-free algorithm whose lower-order terms scale as $\Vert h^\star\Vert_\text{sp}^2 S^3 A$, nearly matching this lower bound. Taken together, these results completely characterize the optimal dependence on $\Vert h^\star\Vert_\text{sp}$ in both leading and lower-order terms, and reveal a fundamental gap in what is achievable with and without prior knowledge.

Multimodal deception detection aims to identify deceptive behavior by analyzing audiovisual cues for forensics and security. In these high-stakes settings, investigators need verifiable evidence connecting audiovisual cues to final decisions, along with reliable generalization across domains and cultural contexts. However, existing benchmarks provide only binary labels without intermediate reasoning cues. Datasets are also small with limited scenario coverage, leading to shortcut learning. We address these issues through three contributions. First, we construct reasoning datasets by augmenting existing benchmarks with structured cue-level descriptions and reasoning chains, enabling model output auditable reports. Second, we release T4-Deception, a multicultural dataset based on the unified ``To Tell The Truth'' television format implemented across four countries. With 1695 samples, it is the largest non-laboratory deception detection dataset. Third, we propose two modules for robust learning under small-data conditions. Stabilized Individuality-Commonality Synergy (SICS) refines multimodal representations by synergizing learnable global priors with sample-adaptive residuals, followed by a polarity-aware adjustment that bi-directionally recalibrates representations. Distilled Modality Consistency (DMC) aligns modality-specific predictions with the fused multimodal predictions via knowledge distillation to prevent unimodal shortcut learning. Experiments on three established benchmarks and our novel dataset demonstrate that our method achieves state-of-the-art performance in both in-domain and cross-domain scenarios, while exhibiting superior transferability across diverse cultural contexts. The datasets and codes will be released.

Large Vision-Language Models (VLMs) have achieved remarkable success in multi-modal reasoning, but their inference time efficiency remains a significant challenge due to the memory overhead during decoding, especially when the query and answer of VLMs consist of long sequences of visual and text tokens. This paper presents AttentionPack, an adaptive and attention-aware optimization framework tailored for large vision-language models with improving memory-efficiency during decoding, focusing on addressing the challenges due to the increased high number of visual inputs and interactions, particularly in long-context tasks with multiple high-resolution images or videos. AttentionPack is novel in two aspects: (i) We introduce a multi-head attention compaction method for economically storing key and value matrices by exploiting the implicit low-rank structure, and (ii) we develop a token-specific attention-aware decompression mechanism to reduce latency overhead. Experimental results on multiple benchmarks demonstrate that AttentionPack improves memory efficiency by up to 8x, enabling higher batch sizes and faster batch inference while preserving the model output quality or longer context lengths for superior retrieval performance. We also report the effectiveness of AttentionPack combined with eviction, quantization and kernel fusion, showing further efficiency gains for resource-limited environments.

As Large Language Models (LLMs) scale up, inference efficiency becomes a critical bottleneck. Multi-Token Prediction (MTP) could accelerate LLM inference by predicting multiple future tokens in parallel. However, existing MTP approaches still face two challenges: limited acceptance rates of MTP heads, and difficulties in jointly training multiple MTP heads. Therefore, we propose MTP-D, a simple yet effective self-distillation method with minimal additional training cost, which boosts MTP head acceptance rates (+7.5\%) while maximumly preserving main-head performance. We also introduce a looped extension strategy for MTP-D, enabling effective and economical MTP head extension and further significant inference speedup to 1-head MTP (+220.4\%). Moreover, we systematically explore and validate key insights on the distillation strategies and the potential scalability of MTP through extensive experiments on seven benchmarks. These results demonstrate that our MTP-D and looped extension strategy effectively enhance MTP-head performance and inference efficiency, facilitating the practical usage of MTP in LLMs.

Recent advances in large language models (LLMs) suggest strong potential for automating analog circuit design. Yet most LLM-based approaches rely on a single-model loop of generation, diagnosis, and correction, which favors succinct summaries over domain-specific insight and suffers from context attrition that erases critical technical details. To address these limitations, we propose AnalogAgent, a training-free agentic framework that integrates an LLM-based multi-agent system (MAS) with self-evolving memory (SEM) for analog circuit design automation. AnalogAgent coordinates a Code Generator, Design Optimizer, and Knowledge Curator to distill execution feedback into an adaptive playbook in SEM and retrieve targeted guidance for subsequent generation, enabling cross-task transfer without additional expert feedback, databases, or libraries. Across established benchmarks, AnalogAgent achieves 92% Pass@1 with Gemini and 97.4% Pass@1 with GPT-5. Moreover, with compact models (e.g., Qwen-8B), it yields a +48.8% average Pass@1 gain across tasks and reaches 72.1% Pass@1 overall, indicating that AnalogAgent substantially strengthens open-weight models for high-quality analog circuit design automation.

While Large Language Models (LLMs) provide semantic flexibility for robotic task planning, their susceptibility to hallucination and logical inconsistency limits their reliability in long-horizon domains. To bridge the gap between unstructured environments and rigorous plan synthesis, we propose DUPLEX, an agentic dual-system neuro-symbolic architecture that strictly confines the LLM to schema-guided information extraction rather than end-to-end planning or code generation. In our framework, a feed-forward Fast System utilizes a lightweight LLM to extract entities, relations etc. from natural language, deterministically mapping them into a Planning Domain Definition Language (PDDL) problem file for a classical symbolic planner. To resolve complex or underspecified scenarios, a Slow System is activated exclusively upon planning failure, leveraging solver diagnostics to drive a high-capacity LLM in iterative reflection and repair. Extensive evaluations across 12 classical and household planning domains demonstrate that DUPLEX significantly outperforms existing end-to-end and hybrid LLM baselines in both success rate and reliability. These results confirm that The key is not to make the LLM plan better, but to restrict the LLM to the part it is good at - structured semantic grounding - and leave logical plan synthesis to a symbolic planner.

Recent research has shown that text-to-image diffusion models are capable of generating high-quality images guided by text prompts. But can they be used to generate or approximate real-world images from the seed noise? This is known as the diffusion inversion problem, which serves as a fundamental building block for bridging diffusion models and real-world scenarios. However, existing diffusion inversion methods often suffer from low reconstruction quality or weak robustness. Two major challenges need to be carefully addressed: (1) the misalignment between the inversion and generation trajectories during the diffusion process, and (2) the mismatch between the diffusion inversion process and the VQ autoencoder (VQAE) reconstruction. To address these challenges, we introduce a latent bias vector at each inversion step, which is learned to reduce the misalignment between inversion and generation trajectories. We refer to this strategy as Latent Bias Optimization (LBO). Furthermore, we perform an approximate joint optimization of the diffusion inversion and VQAE reconstruction processes by learning to adjust the image latent representation, which serves as the connecting interface between them. We refer to this technique as Image Latent Boosting (ILB). Extensive experimental results demonstrate that the proposed method significantly improves the image reconstruction quality of the diffusion model, as well as the performance of downstream tasks, including image editing and rare concept generation.

Retrieving partially relevant segments from untrimmed videos remains difficult due to two persistent challenges: the mismatch in information density between text and video segments, and limited attention mechanisms that overlook semantic focus and event correlations. We present KDC-Net, a Knowledge-Refined Dual Context-Aware Network that tackles these issues from both textual and visual perspectives. On the text side, a Hierarchical Semantic Aggregation module captures and adaptively fuses multi-scale phrase cues to enrich query semantics. On the video side, a Dynamic Temporal Attention mechanism employs relative positional encoding and adaptive temporal windows to highlight key events with local temporal coherence. Additionally, a dynamic CLIP-based distillation strategy, enhanced with temporal-continuity-aware refinement, ensures segment-aware and objective-aligned knowledge transfer. Experiments on PRVR benchmarks show that KDC-Net consistently outperforms state-of-the-art methods, especially under low moment-to-video ratios.

We present SM-Net, a machine-learning model that learns a continuous spectral manifold from multiple high-resolution stellar libraries. SM-Net generates stellar spectra directly from the fundamental stellar parameters effective temperature (Teff), surface gravity (log g), and metallicity (log Z). It is trained on a combined grid derived from the PHOENIX-Husser, C3K-Conroy, OB-PoWR, and TMAP-Werner libraries. By combining their parameter spaces, we construct a composite dataset that spans a broader and more continuous region of stellar parameter space than any individual library. The unified grid covers Teff = 2,000-190,000 K, log g = -1 to 9, and log Z = -4 to 1, with spectra spanning 3,000-100,000 Angstrom. Within this domain, SM-Net provides smooth interpolation across heterogeneous library boundaries. Outside the sampled region, it can produce numerically smooth exploratory predictions, although these extrapolations are not directly validated against reference models. Zero or masked flux values are treated as unknowns rather than physical zeros, allowing the network to infer missing regions using correlations learned from neighbouring grid points. Across 3,538 training and 11,530 test spectra, SM-Net achieves mean squared errors of 1.47 x 10^-5 on the training set and 2.34 x 10^-5 on the test set in the transformed log1p-scaled flux representation. Inference throughput exceeds 14,000 spectra per second on a single GPU. We also release the model together with an interactive web dashboard for real-time spectral generation and visualisation. SM-Net provides a fast, robust, and flexible data-driven complement to traditional stellar population synthesis libraries.

As the modern microservice architecture for cloud applications grows in popularity, cloud services are becoming increasingly complex and more vulnerable to misconfiguration and software bugs. Traditional approaches rely on expert input to diagnose and fix microservice anomalies, which lacks scalability in the face of the continuous integration and continuous deployment (CI/CD) paradigm. Microservice rollouts, containing new software installations, have complex interactions with the components of an application. Consequently, this added difficulty in attributing anomalous behavior to any specific installation or rollout results in potentially slower resolution times. To address the gaps in current diagnostic methods, this paper introduces Praxium, a framework for anomaly detection and root cause inference. Praxium aids administrators in evaluating target metric performance in the context of dependency installation information provided by a software discovery tool, PraxiPaaS. Praxium continuously monitors telemetry data to identify anomalies, then conducts root cause analysis via causal impact on recent software installations, in order to provide site reliability engineers (SRE) relevant information about an observed anomaly. In this paper, we demonstrate that Praxium is capable of effective anomaly detection and root cause inference, and we provide an analysis on effective anomaly detection hyperparameter tuning as needed in a practical setting. Across 75 total trials using four synthetic anomalies, anomaly detection consistently performs at >0.97 macro-F1. In addition, we show that causal impact analysis reliably infers the correct root cause of anomalies, even as package installations occur at increasingly shorter intervals.

When safety is formulated as a limit of cumulative cost, safe reinforcement learning (RL) aims to learn policies that maximize return subject to the cost constraint in data collection and deployment. Off-policy safe RL methods, although offering high sample efficiency, suffer from constraint violations due to cost-agnostic exploration and estimation bias in cumulative cost. To address this issue, we propose Constrained Optimistic eXploration Q-learning (COX-Q), an off-policy safe RL algorithm that integrates cost-bounded online exploration and conservative offline distributional value learning. First, we introduce a novel cost-constrained optimistic exploration strategy that resolves gradient conflicts between reward and cost in the action space and adaptively adjusts the trust region to control the training cost. Second, we adopt truncated quantile critics to stabilize the cost value learning. Quantile critics also quantify epistemic uncertainty to guide exploration. Experiments on safe velocity, safe navigation, and autonomous driving tasks demonstrate that COX-Q achieves high sample efficiency, competitive test safety performance, and controlled data collection cost. The results highlight COX-Q as a promising RL method for safety-critical applications.

Chemical laboratory automation has long been constrained by rigid workflows and poor adaptability to the long-tail distribution of experimental tasks. While most automated platforms perform well on a narrow set of standardized procedures, real laboratories involve diverse, infrequent, and evolving operations that fall outside predefined protocols. This mismatch prevents existing systems from generalizing to novel reaction conditions, uncommon instrument configurations, and unexpected procedural variations. We present a multi-agent robotic platform designed to address this long-tail challenge through collaborative task decomposition, dynamic scheduling, and adaptive control. The system integrates chemical perception for real-time reaction monitoring with feedback-driven execution, enabling it to adjust actions based on evolving experimental states rather than fixed scripts. Validation via acid-base titration demonstrates autonomous progress tracking, adaptive dispensing control, and reliable end-to-end experiment execution. By improving generalization across diverse laboratory scenarios, this platform provides a practical pathway toward intelligent, flexible, and scalable laboratory automation.

Edge AI systems often operate under stringent energy and volume constraints that demand extreme efficiency under limited battery capacity, with requirements worsening as intelligent capability demands advance. Prior literature suggests that fine-grained power orchestration, including DVFS and power gating, enables significant energy efficiency benefits that cannot be left unexploited, while still exhibiting unexplored challenges. We observe that layer-level approaches incur unintended overheads due to inter-layer coupling of power control decisions, and that jointly managing these mechanisms under practical constraints such as limited voltage rails and transition overheads leads to a rapidly growing combinatorial schedule space. To address this, we propose PowerFlow-DNN, a compiler-directed framework for end-to-end power-state orchestration in ultra-low-power accelerators. By constructing a rigorous problem formulation for deadline-constrained, real-time, periodic inference as a unified inter-layer power-scheduling problem, our framework enables automated discovery of energy-minimal power-state schedules that adhere to a deadline while accounting for end-to-end, inter-layer impacts. We evaluate the framework on a DNN accelerator VLSI implementation in TSMC 40nm technology. Across representative edge networks, we show that PowerFlow-DNN discovers near-optimal solutions under the discretized formulation and achieves energy within 0.68\% of the exact ILP oracle, reducing energy by up to 37\% compared to an aggressive baseline without power orchestration, while reasoning over a combinatorial schedule space of over $10^{160}$ possible power-state assignments, yet operating on a structured layered state graph that enables efficient optimization, achieving up to 2.14$\times$ solver speedup via lightweight pruning.

The parameterized CROWN analysis, a.k.a., alpha-CROWN, has emerged as a practically successful bound propagation method for neural network verification. However, existing implementations of alpha-CROWN are limited to Python, which complicates integration into existing DNN verifiers and long-term production-level systems. We introduce Luna, a new bound propagator implemented in C++. Luna supports Interval Bound Propagation, the CROWN analysis, and the alpha-CROWN analysis over a general computational graph. We describe the architecture of Luna and show that it is competitive with the state-of-the-art alpha-CROWN implementation in terms of both bound tightness and computational efficiency on benchmarks from VNN-COMP 2025.

Large language models (LLMs) have demonstrated exceptional potential in complex reasoning,pioneering a new paradigm for autonomous agent decision making in dynamic settings. However, in Real-Time Strategy (RTS) scenarios, LLMs suffer from a critical speed-quality trade-off. Specifically expansive state spaces and time limits render inference delays prohibitive, while stochastic planning errors undermine logical consistency. To address these challenges, we present SEMA (Self-Evolving Multi-Agent), a novel framework designed for high-performance, low-latency decision-making in RTS environments. This collaborative multi-agent framework facilitates self-evolution by adaptively calibrating model bias through in-episode assessment and cross-episode analysis. We further incorporate dynamic observation pruning based on structural entropy to model game states topologically. By distilling high dimensional data into core semantic information, this approach significantly reduces inference time. We also develop a hybrid knowledge-memory mechanism that integrates micro-trajectories, macro-experience, and hierarchical domain knowledge, thereby enhancing both strategic adaptability and decision consistency. Experiments across multiple StarCraft II maps demonstrate that SEMA achieves superior win rates while reducing average decision latency by over 50%, validating its efficiency and robustness in complex RTS scenarios.

DeepXube is a free and open-source Python package and command-line tool that seeks to automate the solution of pathfinding problems by using machine learning to learn heuristic functions that guide heuristic search algorithms tailored to deep neural networks (DNNs). DeepXube is comprised of the latest advances in deep reinforcement learning, heuristic search, and formal logic for solving pathfinding problems. This includes limited-horizon Bellman-based learning, hindsight experience replay, batched heuristic search, and specifying goals with answer-set programming. A robust multiple-inheritance structure simplifies the definition of pathfinding domains and the generation of training data. Training heuristic functions is made efficient through the automatic parallelization of the generation of training data across central processing units (CPUs) and reinforcement learning updates across graphics processing units (GPUs). Pathfinding algorithms that take advantage of the parallelism of GPUs and DNN architectures, such as batch weighted A* and Q* search and beam search are easily employed to solve pathfinding problems through command-line arguments. Finally, several convenient features for visualization, code profiling, and progress monitoring during training and solving are available. The GitHub repository is publicly available at https://github.com/forestagostinelli/deepxube.

Large language models trained with reinforcement learning (RL) for mathematical reasoning face a fundamental challenge: on problems the model cannot solve at all - "cliff" prompts - the RL gradient vanishes entirely, preventing any learning signal from reaching these failure modes. We introduce Hybrid Distillation Policy Optimization (HDPO), which augments standard RL with privileged self-distillation targeting cliff prompts. On each training step, HDPO identifies prompts where all rollouts fail, generates privileged rollouts by providing the model with ground-truth information, filters for correct solutions, and distills the teacher's token-level distribution into the student. Because teacher and student share the same weights - differing only in their input - the realizability gap is provably bounded, unlike cross-model distillation. We prove that R=1 filtered privileged generation recovers the optimal KL-regularized RL policy in the hard-threshold limit. Experiments on OpenMathInstruct-2 with Qwen2.5-Math-1.5B-Instruct show that HDPO consistently improves coverage metrics (pass@4 by +0.8-1.1%, pass@8 by +0.4-1.7%) while maintaining greedy accuracy, with the distillation weight lambda providing direct control over the exploration-exploitation tradeoff.

Vision-Language Models (VLMs) have been applied to a wide range of reasoning tasks, yet it remains unclear whether they can reason robustly under distribution shifts. In this paper, we study covariate shifts in which the perceptual input distribution changes while the underlying prediction rules do not. To investigate this question, we consider visual deductive reasoning tasks, where a model is required to answer a query given an image and logical rules defined over the object concepts in the image. Empirically, we find that VLMs fine-tuned through gradient-based end-to-end training can achieve high in-distribution accuracy but fail to generalize under such shifts, suggesting that fine-tuning does not reliably induce the underlying reasoning function. This motivates a neuro-symbolic perspective that decouples perception from reasoning. However, we further observe that recent neuro-symbolic approaches that rely on black-box components for reasoning can still exhibit inconsistent robustness across tasks. To address this issue, we propose VLC, a neuro-symbolic method that combines VLM-based concept recognition with circuit-based symbolic reasoning. In particular, task rules are compiled into a symbolic program, specifically a circuit, which executes the rules exactly over the object concepts recognized by the VLM. Experiments on three visual deductive reasoning tasks with distinct rule sets show that VLC consistently achieves strong performance under covariate shifts, highlighting its ability to support robust reasoning.

Generative AI (GenAI) has rapidly entered education, yet its user experience is often explained through adoption-oriented constructs such as usefulness, ease of use, and engagement. We argue that these constructs are no longer sufficient because systems such as ChatGPT do not merely support learning tasks but also participate in knowledge construction. Existing theories cannot explain why GenAI frequently produces experiences characterized by negotiated authority, redistributed cognition, and accountability tension. To address this gap, this paper develops the Human--AI Epistemic Partnership Theory (HAEPT), explaining the GenAI user experience as a form of epistemic partnership that features a dynamic negotiation of three interlocking contracts: epistemic, agency, and accountability. We argue that findings on trust, over-reliance, academic integrity, teacher caution, and relational interaction about GenAI can be reinterpreted as tensions within these contracts rather than as isolated issues. Instead of holding a single, stable view of GenAI, users adjust how they relate to it over time through calibration cycles. These repeated interactions account for why trust and skepticism often coexist and for how partnership modes describe recurrent configurations of human--AI collaboration across tasks. To demonstrate the usefulness of HAEPT, we applied it to analyze the UX of collaborative learning with AI speakers and AI-facilitated scientific argumentation, illustrating different contract configurations.

Our work addresses the problem of predicting the highest priority functional group present in an organic molecule. Functional Groups are groups of bound atoms that determine the physical and chemical properties of organic molecules. In the presence of multiple functional groups, the dominant functional group determines the compound's properties. Fourier-transform Infrared spectroscopy (FTIR) is a commonly used spectroscopic method for identifying the presence or absence of functional groups within a compound. We propose the use of a Deep Convolutional Neural Networks (CNN) to predict the highest priority functional group from the Fourier-transform infrared spectrum (FTIR) of the organic molecule. We have compared our model with other previously applied Machine Learning (ML) method Support Vector Machine (SVM) and reasoned why CNN outperforms it.

Neural ODEs are increasingly used as continuous-time models for scientific and sensor data, but unconstrained neural ODEs can drift and violate domain invariants (e.g., conservation laws), yielding physically implausible solutions. In turn, this can compound error in long-horizon prediction and surrogate simulation. Existing solutions typically aim to enforce invariance by soft penalties or other forms of regularization, which can reduce overall error but do not guarantee that trajectories will not leave the constraint manifold. We introduce the invariant compiler, a framework that enforces invariants by construction: it treats invariants as first-class types and uses an LLM-driven compilation workflow to translate a generic neural ODE specification into a structure-preserving architecture whose trajectories remain on the admissible manifold in continuous time (and up to numerical integration error in practice). This compiler view cleanly separates what must be preserved (scientific structure) from what is learned from data (dynamics within that structure). It provides a systematic design pattern for invariant-respecting neural surrogates across scientific domains.

This work investigates the critical role of activation function curvature -- quantified by the maximum second derivative $\max|σ''|$ -- in adversarial robustness. Using the Recursive Curvature-Tunable Activation Family (RCT-AF), which enables precise control over curvature through parameters $α$ and $β$, we systematically analyze this relationship. Our study reveals a fundamental trade-off: insufficient curvature limits model expressivity, while excessive curvature amplifies the normalized Hessian diagonal norm of the loss, leading to sharper minima that hinder robust generalization. This results in a non-monotonic relationship where optimal adversarial robustness consistently occurs when $\max|σ''|$ falls within 4 to 10, a finding that holds across diverse network architectures, datasets, and adversarial training methods. We provide theoretical insights into how activation curvature affects the diagonal elements of the hessian matrix of the loss, and experimentally demonstrate that the normalized Hessian diagonal norm exhibits a U-shaped dependence on $\max|σ''|$, with its minimum within the optimal robustness range, thereby validating the proposed mechanism.

The adoption of generative AI across commercial and legal professions offers dramatic efficiency gains -- yet for law in particular, it introduces a perilous failure mode in which the AI fabricates fictitious case law, statutes, and judicial holdings that appear entirely authentic. Attorneys who unknowingly file such fabrications face professional sanctions, malpractice exposure, and reputational harm, while courts confront a novel threat to the integrity of the adversarial process. This failure mode is commonly dismissed as random `hallucination', but recent physics-based analysis of the Transformer's core mechanism reveals a deterministic component: the AI's internal state can cross a calculable threshold, causing its output to flip from reliable legal reasoning to authoritative-sounding fabrication. Here we present this science in a legal-industry setting, walking through a simulated brief-drafting scenario. Our analysis suggests that fabrication risk is not an anomalous glitch but a foreseeable consequence of the technology's design, with direct implications for the evolving duty of technological competence. We propose that legal professionals, courts, and regulators replace the outdated `black box' mental model with verification protocols based on how these systems actually fail.

Symbolic discovery of governing equations is a long-standing goal in scientific machine learning, yet a fundamental trade-off persists between interpretability and scalable learning. Classical symbolic regression methods yield explicit analytic expressions but rely on combinatorial search, whereas neural networks scale efficiently with data and dimensionality but produce opaque representations. In this work, we introduce Symbolic Kolmogorov-Arnold Networks (Symbolic-KANs), a neural architecture that bridges this gap by embedding discrete symbolic structure directly within a trainable deep network. Symbolic-KANs represent multivariate functions as compositions of learned univariate primitives applied to learned scalar projections, guided by a library of analytic primitives, hierarchical gating, and symbolic regularization that progressively sharpens continuous mixtures into one-hot selections. After gated training and discretization, each active unit selects a single primitive and projection direction, yielding compact closed-form expressions without post-hoc symbolic fitting. Symbolic-KANs further act as scalable primitive discovery mechanisms, identifying the most relevant analytic components that can subsequently inform candidate libraries for sparse equation-learning methods. We demonstrate that Symbolic-KAN reliably recovers correct primitive terms and governing structures in data-driven regression and inverse dynamical systems. Moreover, the framework extends to forward and inverse physics-informed learning of partial differential equations, producing accurate solutions directly from governing constraints while constructing compact symbolic representations whose selected primitives reflect the true analytical structure of the underlying equations. These results position Symbolic-KAN as a step toward scalable, interpretable, and mechanistically grounded learning of governing laws.

Combining multiple Vision-Language Models (VLMs) can enhance multimodal reasoning and robustness, but aggregating heterogeneous models' outputs amplifies uncertainty and increases the risk of hallucinations. We propose SCoOP (Semantic-Consistent Opinion Pooling), a training-free uncertainty quantification (UQ) framework multi-VLM systems through uncertainty-weighted linear opinion pooling. Unlike prior UQ methods designed for single models, SCoOP explicitly measures collective, system-level uncertainty across multiple VLMs, enabling effective hallucination detection and abstention for highly uncertain samples. On ScienceQA, SCoOP achieves an AUROC of 0.866 for hallucination detection, outperforming baselines (0.732-0.757) by approximately 10-13%. For abstention, it attains an AURAC of 0.907, exceeding baselines (0.818-0.840) by 7-9%. Despite these gains, SCoOP introduces only microsecond-level aggregation overhead relative to the baselines, which is trivial compared to typical VLM inference time (on the order of seconds). These results demonstrate that SCoOP provides an efficient and principled mechanism for uncertainty-aware aggregation, advancing the reliability of multimodal AI systems.

Large Language Model (LLM)-based agents increasingly rely on APIs to operate complex web applications, but rapid evolution often leads to incomplete or inconsistent API documentation. Existing work falls into two categories: (1) static, white-box approaches based on source code or formal specifications, and (2) dynamic, black-box approaches that infer APIs from runtime traffic. Static approaches rely on internal artifacts, which are typically unavailable for closed-source systems, and often over-approximate API usage, resulting in high false-positive rates. Although dynamic black-box API discovery applies broadly, its robustness degrades in complex environments where shared collection points aggregate traffic from multiple applications. To improve robustness under mixed runtime traffic, we propose APISENSOR, a black-box API discovery framework that reconstructs application APIs unsupervised. APISENSOR performs structured analysis over complex traffic, combining traffic denoising and normalization with a graph-based two-stage clustering process to recover accurate APIs. We evaluated APISENSOR across six web applications using over 10,000 runtime requests with simulated mixed-traffic noise. Results demonstrate that APISENSOR significantly improves discovery accuracy, achieving an average Group Accuracy Precision of 95.92% and an F1-score of 94.91%, outperforming state-of-the-art methods. Across different applications and noise settings, APISENSOR achieves the lowest performance variance and at most an 8.11-point FGA drop, demonstrating the best robustness among 10 baselines. Ablation studies confirm that each component is essential. Furthermore, APISENSOR revealed API documentation inconsistencies in a real application, later confirmed by community developers.

LLMs are increasingly used as long-running conversational agents, yet every major benchmark evaluating their memory treats user information as static facts to be stored and retrieved. That's the wrong model. People change their minds, and over extended interactions, phenomena like opinion drift, over-alignment, and confirmation bias start to matter a lot. BeliefShift introduces a longitudinal benchmark designed specifically to evaluate belief dynamics in multi-session LLM interactions. It covers three tracks: Temporal Belief Consistency, Contradiction Detection, and Evidence-Driven Revision. The dataset includes 2,400 human-annotated multi-session interaction trajectories spanning health, politics, personal values, and product preferences. We evaluate seven models including GPT-4o, Claude 3.5 Sonnet, Gemini 1.5 Pro, LLaMA-3, and Mistral-Large under zero-shot and retrieval-augmented generation (RAG) settings. Results reveal a clear trade-off: models that personalize aggressively resist drift poorly, while factually grounded models miss legitimate belief updates. We further introduce four novel evaluation metrics: Belief Revision Accuracy (BRA), Drift Coherence Score (DCS), Contradiction Resolution Rate (CRR), and Evidence Sensitivity Index (ESI).

Recent work shows overwhelming evidence that LLMs, even those trained to scale their reasoning trace, perform unsatisfactorily when solving planning problems too complex. Whether the same conclusion holds for LLM formalizers that generate solver-oriented programs remains unknown. We systematically show that LLM formalizers greatly out-scale LLM planners, some retaining perfect accuracy in the classic BlocksWorld domain with a huge state space of size up to $10^{165}$. While performance of smaller LLM formalizers degrades with problem complexity, we show that a divide-and-conquer formalizing technique can greatly improve its robustness. Finally, we introduce unraveling problems where one line of problem description realistically corresponds to exponentially many lines of formal language such as the Planning Domain Definition Language (PDDL), greatly challenging LLM formalizers. We tackle this challenge by introducing a new paradigm, namely LLM-as-higher-order-formalizer, where an LLM generates a program generator. This decouples token output from the combinatorial explosion of the underlying formalization and search space.

While Large Language Models (LLMs) are increasingly used as primary sources of information, their potential for political bias may impact their objectivity. Existing benchmarks of LLM social bias primarily evaluate gender and racial stereotypes. When political bias is included, it is typically measured at a coarse level, neglecting the specific values that shape sociopolitical leanings. This study investigates political bias in eight prominent LLMs (Claude, Deepseek, Gemini, GPT, Grok, Llama, Qwen Base, Qwen Instruction-Tuned) using PoliticsBench: a novel multi-turn roleplay framework adapted from the EQ-Bench-v3 psychometric benchmark. We test whether commercially developed LLMs display a systematic left-leaning bias that becomes more pronounced in later stages of multi-stage roleplay. Through twenty evolving scenarios, each model reported its stance and determined its course of action. Scoring these responses on a scale of ten political values, we explored the values underlying chatbots' deviations from unbiased standards. Seven of our eight models leaned left, while Grok leaned right. Each left-leaning LLM strongly exhibited liberal traits and moderately exhibited conservative ones. We discovered slight variations in alignment scores across stages of roleplay, with no particular pattern. Though most models used consequence-based reasoning, Grok frequently argued with facts and statistics. Our study presents the first psychometric evaluation of political values in LLMs through multi-stage, free-text interactions.

With the growing demand for intelligent in-vehicle experiences, vehicle-based agents are evolving from simple assistants to long-term companions. This evolution requires agents to continuously model multi-user preferences and make reliable decisions in the face of inter-user preference conflicts and changing habits over time. However, existing benchmarks are largely limited to single-user, static question-answer settings, failing to capture the temporal evolution of preferences and the multi-user, tool-interactive nature of real vehicle environments. To address this gap, we introduce VehicleMemBench, a multi-user long-context memory benchmark built on an executable in-vehicle simulation environment. The benchmark evaluates tool use and memory by comparing the post-action environment state with a predefined target state, enabling objective and reproducible evaluation without LLM-based or human scoring. VehicleMemBench includes 23 tool modules, and each sample contains over 80 historical memory events. Experiments show that powerful models perform well on direct instruction tasks but struggle in scenarios involving memory evolution, particularly when user preferences change dynamically. Even advanced memory systems struggle to handle domain-specific memory requirements in this environment. These findings highlight the need for more robust and specialized memory management mechanisms to support long-term adaptive decision-making in real-world in-vehicle systems. To facilitate future research, we release the data and code.

Lifelong Multi-Agent Path Finding (MAPF) is critical for modern warehouse automation, which requires multiple robots to continuously navigate conflict-free paths to optimize the overall system throughput. However, the complexity of warehouse environments and the long-term dynamics of lifelong MAPF often demand costly adaptations to classical search-based solvers. While machine learning methods have been explored, their superiority over search-based methods remains inconclusive. In this paper, we introduce Reinforcement Learning (RL) guided Rolling Horizon Prioritized Planning (RL-RH-PP), the first framework integrating RL with search-based planning for lifelong MAPF. Specifically, we leverage classical Prioritized Planning (PP) as a backbone for its simplicity and flexibility in integrating with a learning-based priority assignment policy. By formulating dynamic priority assignment as a Partially Observable Markov Decision Process (POMDP), RL-RH-PP exploits the sequential decision-making nature of lifelong planning while delegating complex spatial-temporal interactions among agents to reinforcement learning. An attention-based neural network autoregressively decodes priority orders on-the-fly, enabling efficient sequential single-agent planning by the PP planner. Evaluations in realistic warehouse simulations show that RL-RH-PP achieves the highest total throughput among baselines and generalizes effectively across agent densities, planning horizons, and warehouse layouts. Our interpretive analyses reveal that RL-RH-PP proactively prioritizes congested agents and strategically redirects agents from congestion, easing traffic flow and boosting throughput. These findings highlight the potential of learning-guided approaches to augment traditional heuristics in modern warehouse automation.

There remain theoretical gaps in deep neural network estimators for the nonparametric Cox proportional hazards model. In particular, it is unclear how gradient-based optimization error propagates to population risk under partial likelihood, how pointwise bias can be controlled to permit valid inference, and how ensemble-based uncertainty quantification behaves under realistic variance decay regimes. We develop an asymptotic distribution theory for deep Cox estimators that addresses these issues. First, we establish nonasymptotic oracle inequalities for general trained networks that link in-sample optimization error to population risk without requiring the exact empirical risk optimizer. We then construct a structured neural parameterization that achieves infinity-norm approximation rates compatible with the oracle bound, yielding control of the pointwise bias. Under these conditions and using the Hajek--Hoeffding projection, we prove pointwise and multivariate asymptotic normality for subsampled ensemble estimators. We derive a range of subsample sizes that balances bias correction with the requirement that the Hajek--Hoeffding projection remain dominant. This range accommodates decay conditions on the single-overlap covariance, which measures how strongly a single shared observation influences the estimator, and is weaker than those imposed in the subsampling literature. An infinitesimal jackknife representation provides analytic covariance estimation and valid Wald-type inference for relative risk contrasts such as log-hazard ratios. Finally, we illustrate the finite-sample implications of the theory through simulations and a real data application.

Deep neural networks (DNNs), particularly those using Rectified Linear Unit (ReLU) activation functions, have achieved remarkable success across diverse machine learning tasks, including image recognition, audio processing, and language modeling. Despite this success, the non-convex nature of DNN loss functions complicates optimization and limits theoretical understanding. In this paper, we highlight how recently developed convex equivalences of ReLU NNs and their connections to sparse signal processing models can address the challenges of training and understanding NNs. Recent research has uncovered several hidden convexities in the loss landscapes of certain NN architectures, notably two-layer ReLU networks and other deeper or varied architectures. This paper seeks to provide an accessible and educational overview that bridges recent advances in the mathematics of deep learning with traditional signal processing, encouraging broader signal processing applications.

Modern automated accessibility testing tools for mobile applications have significantly improved the detection of interface violations, yet their impact on remediation remains limited. A key reason is that existing tools typically produce low-level, technical outputs that are difficult for non-specialist stakeholders, such as product managers and designers, to interpret in terms of real user harm and compliance risk. In this paper, we present \textsc{HEAR} (\underline{H}uman-c\underline{E}ntered \underline{A}ccessibility \underline{R}eporting), a framework that bridges this interpretation gap by transforming raw accessibility bug reports into empathetic, stakeholder-oriented narratives. Given the outputs of the existing accessibility testing tool, \textsc{HEAR} first reconstructs the UI context through semantic slicing and visual grounding, then dynamically injects disability-oriented personas matched to each violation type, and finally performs multi-layer reasoning to explain the physical barrier, functional blockage, and relevant legal or compliance concerns. We evaluate the framework on real-world accessibility issues collected from four popular Android applications and conduct a user study (N=12). The results show that \textsc{HEAR} generates factually grounded reports and substantially improves perceived empathy, urgency, persuasiveness, and awareness of legal risk compared with raw technical logs, while imposing little additional cognitive burden.

Knowledge tracing models mastery over interconnected concepts, often organized by prerequisites. We analyze hierarchical prerequisite propagation through a circuit-complexity lens to clarify what is provable about transformer-style computation on deep concept hierarchies. Using recent results that log-precision transformers lie in logspace-uniform $\mathsf{TC}^0$, we formalize prerequisite-tree tasks including recursive-majority mastery propagation. Unconditionally, recursive-majority propagation lies in $\mathsf{NC}^1$ via $O(\log n)$-depth bounded-fanin circuits, while separating it from uniform $\mathsf{TC}^0$ would require major progress on open lower bounds. Under a monotonicity restriction, we obtain an unconditional barrier: alternating ALL/ANY prerequisite trees yield a strict depth hierarchy for \emph{monotone} threshold circuits. Empirically, transformer encoders trained on recursive-majority trees converge to permutation-invariant shortcuts; explicit structure alone does not prevent this, but auxiliary supervision on intermediate subtrees elicits structure-dependent computation and achieves near-perfect accuracy at depths 3--4. These findings motivate structure-aware objectives and iterative mechanisms for prerequisite-sensitive knowledge tracing on deep hierarchies.

Large language models (LLMs) are increasingly deployed on edge devices under strict computation and quantization constraints, yet their security implications remain unclear. We study query-based knowledge extraction from quantized edge-deployed LLMs under realistic query budgets and show that, although quantization introduces noise, it does not remove the underlying semantic knowledge, allowing substantial behavioral recovery through carefully designed queries. To systematically analyze this risk, we propose \textbf{CLIQ} (\textbf{Cl}ustered \textbf{I}nstruction \textbf{Q}uerying), a structured query construction framework that improves semantic coverage while reducing redundancy. Experiments on quantized Qwen models (INT8/INT4) demonstrate that CLIQ consistently outperforms original queries across BERTScore, BLEU, and ROUGE, enabling more efficient extraction under limited budgets. These results indicate that quantization alone does not provide effective protection against query-based extraction, highlighting a previously underexplored security risk in edge-deployed LLMs.

Linguistic representation learning in deep neural language models (LMs) has been studied for decades, for both practical and theoretical reasons. However, finding representations in LMs remains an unsolved problem, in part due to a dilemma between enforcing implausible constraints on representations (e.g., linearity; Arora et al. 2024) and trivializing the notion of representation altogether (Sutter et al., 2025). Here we escape this dilemma by reconceptualizing representations not as patterns of activation but as conduits for learning. Our approach is simple: we perturb an LM by fine-tuning it on a single adversarial example and measure how this perturbation ``infects'' other examples. Perturbation makes no geometric assumptions, and unlike other methods, it does not find representations where it should not (e.g., in untrained LMs). But in trained LMs, perturbation reveals structured transfer at multiple linguistic grain sizes, suggesting that LMs both generalize along representational lines and acquire linguistic abstractions from experience alone.

Blockchain technology is often discussed as if it emerged from nowhere, yet its architectural DNA traces directly to the decentralized computing principles James~N. Gray articulated in 1986. This paper maps the conceptual lineage from Gray's requestor/server model to modern blockchain architectures, showing how his emphasis on modularity, autonomy, data integrity, and standardized communication anticipated the design of systems like Bitcoin and Ethereum, and, more recently, the Web3 movement and Layer-2 scaling architectures. We examine consensus mechanisms, cryptographic foundations, rollup-based Layer-2 protocols, and cross-chain interoperability through this historical lens, identify persistent challenges in scalability and modularity, and outline future directions toward Web4: an intelligent, decentralized internet integrating blockchain, artificial intelligence, and the Internet of Things.

AI agents are increasingly embedded in real software systems, where they execute multi-step workflows through multi-turn dialogue, tool invocations, and intermediate decisions. These long execution histories, called agentic traces, make validation difficult. Outcome-only benchmarks can miss critical procedural failures, such as incorrect workflow routing, unsafe tool usage, or violations of prompt-specified rules. This paper presents AgentPex, an AI-powered tool designed to systematically evaluate agentic traces. AgentPex extracts behavioral rules from agent prompts and system instructions, then uses these specifications to automatically evaluate traces for compliance. We evaluate AgentPex on 424 traces from τ2-bench across models in telecom, retail, and airline customer service. Our results show that AgentPex distinguishes agent behavior across models and surfaces specification violations that are not captured by outcome-only scoring. It also provides fine-grained analysis by domain and metric, enabling developers to understand agent strengths and weaknesses at scale.

Neural Collapse is a phenomenon that helps identify sparse and low rank structures in deep classifiers. Recent work has extended the definition of neural collapse to regression problems, albeit only measuring the phenomenon at the last layer. In this paper, we establish that Neural Regression Collapse (NRC) also occurs below the last layer across different types of models. We show that in the collapsed layers of neural regression models, features lie in a subspace that corresponds to the target dimension, the feature covariance aligns with the target covariance, the input subspace of the layer weights aligns with the feature subspace, and the linear prediction error of the features is close to the overall prediction error of the model. In addition to establishing Deep NRC, we also show that models that exhibit Deep NRC learn the intrinsic dimension of low rank targets and explore the necessity of weight decay in inducing Deep NRC. This paper provides a more complete picture of the simple structure learned by deep networks in the context of regression.

We present a novel LLM-informed model-based planning framework, and a novel prompt selection method, for object search in partially-known environments. Our approach uses an LLM to estimate statistics about the likelihood of finding the target object when searching various locations throughout the scene that, combined with travel costs extracted from the environment map, are used to instantiate a model, thus using the LLM to inform planning and achieve effective search performance. Moreover, the abstraction upon which our approach relies is amenable to deployment-time model selection via the recent offline replay approach, an insight we leverage to enable fast prompt and LLM selection during deployment. Simulation experiments demonstrate that our LLM-informed model-based planning approach outperforms the baseline planning strategy that fully relies on LLM and optimistic strategy with as much as 11.8% and 39.2% improvements respectively, and our bandit-like selection approach enables quick selection of best prompts and LLMs resulting in 6.5% lower average cost and 33.8% lower average cumulative regret over baseline UCB bandit selection. Real-robot experiments in an apartment demonstrate similar improvements and so further validate our approach.

Physics-informed neural networks (PINNs) are increasingly used in mathematical epidemiology to bridge the gap between noisy clinical data and compartmental models, such as the susceptible-exposed-infected-removed (SEIR) model. However, training these hybrid networks is often unstable due to competing optimization objectives. As established in recent literature on ``gradient pathology," the gradient vectors derived from the data loss and the physical residual often point in conflicting directions, leading to slow convergence or optimization deadlock. While existing methods attempt to resolve this by balancing gradient magnitudes or projecting conflicting vectors, we propose a novel method, conflict-gated gradient scaling (CGGS), to address gradient conflicts in physics-informed neural networks for epidemiological modelling, ensuring stable and efficient training and a computationally efficient alternative. This method utilizes the cosine similarity between the data and physics gradients to dynamically modulate the penalty weight. Unlike standard annealing schemes that only normalize scales, CGGS acts as a geometric gate: it suppresses the physical constraint when directional conflict is high, allowing the optimizer to prioritize data fidelity, and restores the constraint when gradients align. We prove that this gating mechanism preserves the standard $O(1/T)$ convergence rate for smooth non-convex objectives, a guarantee that fails under fixed-weight or magnitude-balanced training when gradients conflict. We demonstrate that this mechanism autonomously induces a curriculum learning effect, improving parameter estimation in stiff epidemiological systems compared to magnitude-based baselines. Our empirical results show improved peak recovery and convergence over magnitude-based methods.

Children in many parts of the world hear relatively little speech directed to them, yet still reach major language development milestones. What differs about the speech input that infants learn from when directed input is rare? Using longform, infant-centered audio recordings taken in rural Bolivia and the urban U.S., we examined temporal patterns of infants' speech input and their pre-linguistic vocal behavior. We find that child-directed speech in Bolivia, though less frequent, was just as temporally clustered as speech input in the U.S, arriving in concentrated bursts rather than spread across the day. In both communities, infants were most likely to produce speech-like vocalizations during periods of speech directed to them, with the probability of infants' speech-like vocalizations during target child-directed speech nearly double that during silence. In Bolivia, infants' speech-like vocalizations were also more likely to occur during bouts of directed speech from older children than from adults. Together, these findings suggest that the developmental impact of child-directed speech may depend not only on quantity, but on temporal concentration and source, with older children serving as an important source of input in some communities, including where adult speech to infants is less frequent.

Vision foundation models (FMs) achieve state-of-the-art performance in medical imaging. However, they encode information in abstract latent representations that clinicians cannot interrogate or verify. The goal of this study is to investigate Sparse Autoencoders (SAEs) for replacing opaque FM image representations with human-interpretable, sparse features. We train SAEs on embeddings from BiomedParse (biomedical) and DINOv3 (general-purpose) using 909,873 CT and MRI 2D image slices from the TotalSegmentator dataset. We find that learned sparse features: (a) reconstruct original embeddings with high fidelity (R2 up to 0.941) and recover up to 87.8% of downstream performance using only 10 features (99.4% dimensionality reduction), (b) preserve semantic fidelity in image retrieval tasks, (c) correspond to specific concepts that can be expressed in language using large language model (LLM)-based auto-interpretation. (d) bridge clinical language and abstract latent representations in zero-shot language-driven image retrieval. Our work indicates SAEs are a promising pathway towards interpretable, concept-driven medical vision systems. Code repository: https://github.com/pwesp/sail.

Peer-discovery protocols within P2P networks are often vulnerable: because creating network identities is essentially free, adversaries can eclipse honest nodes or partition the overlay. This threat is especially acute for blockchains, whose security depends on resilient peer connectivity. We present AetherWeave, a stake-backed peer-discovery protocol that ties network participation to deposited stake, raising the cost of large-scale attacks. We prove that, with high probability, either the honest overlay remains connected or a $(1{-}δ)$-fraction of nodes in every smaller component raise an attack-detection flag -- even against a very powerful adversary. To our knowledge, AetherWeave is the first peer-discovery protocol to simultaneously provide Sybil resistance and privacy: nodes prove they hold valid stake without revealing which deposit they own, and gossiping does not expose peer-table contents. A cryptographic commitment scheme rate-limits discovery requests per round; exceeding the limit yields a publicly verifiable misbehavior proof that triggers on-chain slashing. Beyond deposit and slashing, the protocol requires no on-chain interaction, with per-node communication scaling as $O(s\sqrt{n})$. We validate our design through a mean-field analysis with closed-form convergence bounds, extensive adversarial simulations, and an end-to-end prototype built by forking Prysm, a leading Ethereum consensus client.

Diffusion models often generate novel samples even when the learned score is only \emph{coarse} -- a phenomenon not accounted for by the standard view of diffusion training as density estimation. In this paper, we show that, under the \emph{manifold hypothesis}, this behavior can instead be explained by coarse scores capturing the \emph{geometry} of the data while discarding the fine-scale distributional structure of the population measure~$μ_{\scriptscriptstyle\mathrm{data}}$. Concretely, whereas estimating the full data distribution $μ_{\scriptscriptstyle\mathrm{data}}$ supported on a $k$-dimensional manifold is known to require the classical minimax rate $\tilde{\mathcal{O}}(N^{-1/k})$, we prove that diffusion models trained with coarse scores can exploit the \emph{regularity of the manifold support} and attain a near-parametric rate toward a \emph{different} target distribution. This target distribution has density uniformly comparable to that of~$μ_{\scriptscriptstyle\mathrm{data}}$ throughout any $\tilde{\mathcal{O}}\bigl(N^{-β/(4k)}\bigr)$-neighborhood of the manifold, where $β$ denotes the manifold regularity. Our guarantees therefore depend only on the smoothness of the underlying support, and are especially favorable when the data density itself is irregular, for instance non-differentiable. In particular, when the manifold is sufficiently smooth, we obtain that \emph{generalization} -- formalized as the ability to generate novel, high-fidelity samples -- occurs at a statistical rate strictly faster than that required to estimate the full population distribution~$μ_{\scriptscriptstyle\mathrm{data}}$.

Deploying large language models (LLMs) as autonomous browser agents exposes a significant attack surface in the form of Indirect Prompt Injection (IPI). Cloud-based defenses can provide strong semantic analysis, but they introduce latency and raise privacy concerns. We present the Cognitive Firewall, a three-stage split-compute architecture that distributes security checks across the client and the cloud. The system consists of a local visual Sentinel, a cloud-based Deep Planner, and a deterministic Guard that enforces execution-time policies. Across 1,000 adversarial samples, edge-only defenses fail to detect 86.9% of semantic attacks. In contrast, the full hybrid architecture reduces the overall attack success rate (ASR) to below 1% (0.88% under static evaluation and 0.67% under adaptive evaluation), while maintaining deterministic constraints on side-effecting actions. By filtering presentation-layer attacks locally, the system avoids unnecessary cloud inference and achieves an approximately 17,000x latency advantage over cloud-only baselines. These results indicate that deterministic enforcement at the execution boundary can complement probabilistic language models, and that split-compute provides a practical foundation for securing interactive LLM agents.

Prediction of wireless channels and their statistics is a fundamental procedure for ensuring performance guarantees in wireless systems. Statistical radio maps powered by Gaussian processes (GPs) offer flexible, non-parametric frameworks, but their performance depends critically on the choice of mean and covariance functions. These are typically learned from dense measurements without exploiting environmental geometry. Digital twins (DTs) of wireless environments leverage computational power to incorporate geometric information; however, they require costly calibration to accurately capture material and propagation characteristics. This work introduces a hybrid channel prediction framework that leverages uncalibrated DTs derived from open-source maps to extract geometry-induced prior information for GP prediction. These structural priors are fused with a small number of channel measurements, enabling data-efficient prediction of channel statistics across the entire environment. By exploiting the uncertainty quantification inherent to GPs, the framework supports principled measurement selection by identifying informative probing locations under resource constraints. Through this integration of imperfect DTs with statistical learning, the proposed method reduces measurement overhead, improves prediction accuracy, and establishes a practical approach for resource-efficient wireless channel prediction.

Retinal diseases remain among the leading preventable causes of visual impairment worldwide. Automated screening based on fundus image analysis has the potential to expand access to early detection, particularly in underserved populations. This paper presents a reproducible deep learning pipeline for binary retinal disease risk classification from publicly available fundus photographs. We implement and compare a baseline convolutional neural network with a transfer learning approach using a pretrained VGG16 backbone and evaluate generalization on held-out data. To address class imbalance, we apply class weighting and report standard classification metrics including accuracy, precision, recall, F1-score, confusion matrices, and ROC-AUC. The VGG16 transfer learning model achieves 90.8% test accuracy with a weighted F1-score of 0.90, substantially outperforming the baseline CNN (83.1% accuracy). Results indicate that transfer learning improves discrimination compared to a baseline CNN, while also revealing remaining challenges in sensitivity to minority disease cases. We discuss practical limitations related to dataset characteristics, class imbalance, and threshold selection, and provide guidance for reproducibility and future improvements for clinically reliable screening

Grokking-the phenomenon where validation accuracy of neural networks on modular addition of two integers rises long after training data has been memorized-has been characterized in previous works as producing sinusoidal input weight distributions in transformers and multi-layer perceptrons (MLPs). We find empirically that ReLU MLPs in our experimental setting instead learn near-binary square wave input weights, where intermediate-valued weights appear exclusively near sign-change boundaries, alongside output weight distributions whose dominant Fourier phases satisfy a phase-sum relation $φ_{\mathrm{out}} = φ_a + φ_b$; this relation holds even when the model is trained on noisy data and fails to grok. We extract the frequency and phase of each neuron's weights via DFT and construct an idealized MLP: Input weights are replaced by perfect binary square waves and output weights by cosines, both parametrized by the frequencies, phases, and amplitudes extracted from the dominant Fourier components of the real model weights. This idealized model achieves 95.5% accuracy when the frequencies and phases are extracted from the weights of a model trained on noisy data that itself achieves only 0.23% accuracy. This suggests that grokking does not discover the correct algorithm, but rather sharpens an algorithm substantially encoded during memorization, progressively binarizing the input weights into cleaner square waves and aligning the output weights, until generalization becomes possible.

Adapting large-scale foundation models to new domains with limited supervision remains a fundamental challenge due to latent distribution mismatch, unstable optimization dynamics, and miscalibrated uncertainty propagation. This paper introduces an uncertainty-aware probabilistic latent transport framework that formulates domain adaptation as a stochastic geometric alignment problem in representation space. A Bayesian transport operator is proposed to redistribute latent probability mass along Wasserstein-type geodesic trajectories, while a PAC-Bayesian regularization mechanism constrains posterior model complexity to mitigate catastrophic overfitting. The proposed formulation yields theoretical guarantees on convergence stability, loss landscape smoothness, and sample efficiency under distributional shift. Empirical analyses demonstrate substantial reduction in latent manifold discrepancy, accelerated transport energy decay, and improved covariance calibration compared with deterministic fine-tuning and adversarial domain adaptation baselines. Furthermore, bounded posterior uncertainty evolution indicates enhanced probabilistic reliability during cross-domain transfer. By establishing a principled connection between stochastic optimal transport geometry and statistical generalization theory, the proposed framework provides new insights into robust adaptation of modern foundation architectures operating in heterogeneous environments. These findings suggest that uncertainty-aware probabilistic alignment constitutes a promising paradigm for reliable transfer learning in next-generation deep representation systems.

Large Language Models (LLMs) have introduced new capabilities to recommender systems, enabling dynamic, context-aware, and conversational recommendations. However, LLM-based recommender systems inherit and may amplify social biases embedded in their pre-training data, especially when demographic cues are present. Existing fairness solutions either require extra parameters fine-tuning, or suffer from optimization instability. We propose a lightweight and scalable bias mitigation method that combines a kernelized Iterative Null-space Projection (INLP) with a gated Mixture-of-Experts (MoE) adapter. Our approach estimates a closed-form projection that removes single or multiple sensitive attributes from LLM representations with no additional trainable parameters. To preserve task utility, we introduce a two-level MoE adapter that selectively restores useful signals without reintroducing bias. Experiments on two public datasets show that our method reduces attribute leakage across multiple protected variables while maintaining competitive recommendation accuracy.

During human motor skill training and physical rehabilitation, there is an inherent trade-off between task difficulty and user performance. Characterizing this trade-off is crucial for evaluating user performance, designing assist-as-needed (AAN) protocols, and assessing the efficacy of training protocols. In this study, we propose a novel human-in-the-loop (HiL) Pareto optimization approach to characterize the trade-off between task performance and the perceived challenge level of motor learning or rehabilitation tasks. We adapt Bayesian multi-criteria optimization to systematically and efficiently perform HiL Pareto characterizations. Our HiL optimization employs a hybrid model that measures performance with a quantitative metric, while the perceived challenge level is captured with a qualitative metric. We demonstrate the feasibility of the proposed HiL Pareto characterization through a user study. Furthermore, we present the utility of the framework through three use cases in the context of a manual skill training task with haptic feedback. First, we demonstrate how the characterized trade-off can be used to design a sample AAN training protocol for a motor learning task and to evaluate the group-level efficacy of the proposed AAN protocol relative to a baseline adaptive assistance protocol. Second, we demonstrate that individual-level comparisons of the trade-offs characterized before and after the training session enable fair evaluation of training progress under different assistance levels. This evaluation method is more general than standard performance evaluations, as it can provide insights even when users cannot perform the task without assistance. Third, we show that the characterized trade-offs also enable fair performance comparisons among different users, as they capture the best possible performance of each user under all feasible assistance levels.

Intent-Based Networking (IBN) aims to simplify operating heterogeneous infrastructures by translating high-level intents into enforceable policies and assuring compliance. However, dependable automation remains difficult because (i) realizing intents from ambiguous natural language into controller-ready policies is brittle and prone to conflicts and unintended side effects, and (ii) assurance is often reactive and struggles in multi-intent settings where faults create cascading symptoms and ambiguous telemetry. This paper proposes an end-to-end closed-loop IBN pipeline that uses large language models with structured validation for natural language to policy realization and conflict-aware activation, and reformulates assurance as proactive multi-intent failure prediction with root-cause disambiguation. The expected outcome is operator-trustworthy automation that provides actionable early warnings, interpretable explanations, and measurable lead time for remediation.

Logging plays a central role in ensuring reproducibility, observability, and reliability in machine learning (ML) systems. While logging is generally considered a good engineering practice, poorly designed logging can negatively affect experiment tracking, security, debugging, and system performance. In this paper, we present an empirical study of logging smells in ML projects and propose a taxonomy of ML-specific logging smell types. We conducted a large-scale analysis of 444 ML repositories and manually labeled 2,448 instances of logging smells. Based on this analysis, we identified 12 categories of logging smells spanning security, metric management, configuration, verbosity, and context-related issues. Our results show that logging smells are widespread in ML systems and vary in frequency and manifestation across projects. To assess practical relevance, we conducted a survey with 27 ML practitioners. Most respondents agreed with the identified smells and reported that several types, including Logging Sensitive Data, Metric Overwrite, Missing Hyperparameter Logging, and Log Without Context, have a strong impact on reproducibility, maintainability, and trustworthiness. Other smells, such as Heavy Data Logging and Print-based Logging, were perceived as more context-dependent. We publicly release our labeled dataset to support future research. Our findings highlight logging quality as a critical and underexplored aspect of ML system engineering and open opportunities for automated detection and repair of logging issues.

Processing-in-memory (PIM) architectures bring computation closer to data, reducing the processor-memory transfer bottleneck in traditional processor-centric designs. Novel hardware solutions, such as UPMEM's in-memory processing technology, achieve this by integrating low-power DRAM processing units (DPUs) into memory DIMMs, enabling massive parallelism and improved memory bandwidth. However, paradoxically, these PIM architectures introduce mandatory coarse-grained data transfers between host DRAM and DPUs, which often become the new bottleneck. We present PIM-CACHE, a lightweight data staging layer that dynamically eliminates redundant data transfers to PIM DPUs by exploiting workload similarity, achieving content-aware copy (CAC). We evaluate PIM-CACHE on both synthetic workloads and real-world genome datasets, demonstrating its effectiveness in reducing PIM data transfer overhead.

Curriculum learning improves reinforcement learning (RL) efficiency by sequencing tasks from simple to complex. However, many self-paced curriculum methods rely on computationally expensive inner-loop optimizations, limiting their scalability in high-dimensional context spaces. In this paper, we propose Self-Paced Gaussian Curriculum Learning (SPGL), a novel approach that avoids costly numerical procedures by leveraging a closed-form update rule for Gaussian context distributions. SPGL maintains the sample efficiency and adaptability of traditional self-paced methods while substantially reducing computational overhead. We provide theoretical guarantees on convergence and validate our method across several contextual RL benchmarks, including the Point Mass, Lunar Lander, and Ball Catching environments. Experimental results show that SPGL matches or outperforms existing curriculum methods, especially in hidden context scenarios, and achieves more stable context distribution convergence. Our method offers a scalable, principled alternative for curriculum generation in challenging continuous and partially observable domains.

Large language models are increasingly consulted for Islamic knowledge, yet no comprehensive benchmark evaluates their performance across core Islamic disciplines. We introduce IslamicMMLU, a benchmark of 10,013 multiple-choice questions spanning three tracks: Quran (2,013 questions), Hadith (4,000 questions), and Fiqh (jurisprudence, 4,000 questions). Each track is formed of multiple types of questions to examine LLMs capabilities handling different aspects of Islamic knowledge. The benchmark is used to create the IslamicMMLU public leaderboard for evaluating LLMs, and we initially evaluate 26 LLMs, where their averaged accuracy across the three tracks varied between 39.8\% to 93.8\% (by Gemini 3 Flash). The Quran track shows the widest span (99.3\% to 32.4\%), while the Fiqh track includes a novel madhab (Islamic school of jurisprudence) bias detection task revealing variable school-of-thought preferences across models. Arabic-specific models show mixed results, but they all underperform compared to frontier models. The evaluation code and leaderboard are made publicly available.

Evaluating AI agents on comprehensive benchmarks is expensive because each evaluation requires interactive rollouts with tool use and multi-step reasoning. We study whether small task subsets can preserve agent rankings at substantially lower cost. Unlike static language model benchmarks, agent evaluation is subject to scaffold-driven distribution shift, since performance depends on the framework wrapping the underlying model. Across eight benchmarks, 33 agent scaffolds, and 70+ model configurations, we find that absolute score prediction degrades under this shift, while rank-order prediction remains stable. Exploiting this asymmetry, we propose a simple optimization-free protocol: evaluate new agents only on tasks with intermediate historical pass rates (30-70%). This mid-range difficulty filter, motivated by Item Response Theory, reduces the number of evaluation tasks by 44-70% while maintaining high rank fidelity under scaffold and temporal shifts. It provides more reliable rankings than random sampling, which exhibits high variance across seeds, and outperforms greedy task selection under distribution shift. These results suggest that reliable leaderboard ranking does not require full-benchmark evaluation.

Events in spatiotemporal domains arise in numerous real-world applications, where uncovering event relationships and enabling accurate prediction are central challenges. Classical Poisson and Hawkes processes rely on restrictive parametric assumptions that limit their ability to capture complex interaction patterns, while recent neural point process models increase representational capacity but integrate event information in a black-box manner, hindering interpretable relationship discovery. To address these limitations, we propose a Kronecker-Structured Nonparametric Spatiotemporal Point Process (KSTPP) that enables transparent event-wise relationship discovery while retaining high modeling flexibility. We model the background intensity with a spatial Gaussian process (GP) and the influence kernel as a spatiotemporal GP, allowing rich interaction patterns including excitation, inhibition, neutrality, and time-varying effects. To enable scalable training and prediction, we adopt separable product kernels and represent the GPs on structured grids, inducing Kronecker-structured covariance matrices. Exploiting Kronecker algebra substantially reduces computational cost and allows the model to scale to large event collections. In addition, we develop a tensor-product Gauss-Legendre quadrature scheme to efficiently evaluate intractable likelihood integrals. Extensive experiments demonstrate the effectiveness of our framework.

A major challenge of Reinforcement Learning is that agents often learn undesired behaviors that seem to defy the reward structure they were given. Explainable Reinforcement Learning (XRL) methods can answer queries such as "explain this specific action", "explain this specific trajectory", and "explain the entire policy". However, XRL lacks a formal definition for behavior as a pattern of actions across many episodes. We provide such a definition, and use it to enable a new query: "Explain this behavior". We present Behavior-Explainable Reinforcement Learning (BXRL), a new problem formulation that treats behaviors as first-class objects. BXRL defines a behavior measure as any function $m : Π\to \mathbb{R}$, allowing users to precisely express the pattern of actions that they find interesting and measure how strongly the policy exhibits it. We define contrastive behaviors that reduce the question "why does the agent prefer $a$ to $a'$?" to "why is $m(π)$ high?" which can be explored with differentiation. We do not implement an explainability method; we instead analyze three existing methods and propose how they could be adapted to explain behavior. We present a port of the HighwayEnv driving environment to JAX, which provides an interface for defining, measuring, and differentiating behaviors with respect to the model parameters.

Many scientific systems, such as cellular populations or economic cohorts, are naturally described by probability distributions that evolve over time. Predicting how such a system would have evolved under different forces or initial conditions is fundamental to causal inference, domain adaptation, and counterfactual prediction. However, the space of distributions often lacks the vector space structure on which classical methods rely. To address this, we introduce a general notion of parallel dynamics at a distributional level. We base this principle on parallel transport of tangent dynamics along optimal transport geodesics and call it ``Wasserstein Parallel Trends''. By replacing the vector subtraction of classic methods with geodesic parallel transport, we can provide counterfactual comparisons of distributional dynamics in applications such as causal inference, domain adaptation, and batch-effect correction in experimental settings. The main mathematical contribution is a novel notion of fanning scheme on the Wasserstein manifold that allows us to efficiently approximate parallel transport along geodesics while also providing the first theoretical guarantees for parallel transport in the Wasserstein space. We also show that Wasserstein Parallel Trends recovers the classic parallel trends assumption for averages as a special case and derive closed-form parallel transport for Gaussian measures. We deploy the method on synthetic data and two single-cell RNA sequencing datasets to impute gene-expression dynamics across biological systems.

Deep spatially selective filters achieve high-quality enhancement with real-time capable architectures for stationary speakers of known directions. To retain this level of performance in dynamic scenarios when only the speakers' initial directions are given, accurate, yet computationally lightweight tracking algorithms become necessary. Assuming a frame-wise causal processing style, temporal feedback allows for leveraging the enhanced speech signal to improve tracking performance. In this work, we investigate strategies to incorporate the enhanced signal into lightweight tracking algorithms and autoregressively guide deep spatial filters. Our proposed Bayesian tracking algorithms are compatible with arbitrary deep spatial filters. To increase the realism of simulated trajectories during development and evaluation, we propose and publish a novel dataset based on the social force model. Results validate that the autoregressive incorporation significantly improves the accuracy of our Bayesian trackers, resulting in superior enhancement with none or only negligibly increased computational overhead. Real-world recordings complement these findings and demonstrate the generalizability of our methods to unseen, challenging acoustic conditions.

While Multi-Agent Reinforcement Learning (MARL) algorithms achieve unprecedented successes across complex continuous domains, their standard deployment strictly adheres to a synchronous operational paradigm. Under this paradigm, agents are universally forced to execute deep neural network inferences at every micro-frame, regardless of immediate necessity. This dense throughput acts as a fundamental barrier to physical deployment on edge-devices where thermal and metabolic budgets are highly constrained. We propose Epistemic Time-Dilation MAPPO (ETD-MAPPO), augmented with a Dual-Gated Epistemic Trigger. Instead of depending on rigid frame-skipping (macro-actions), agents autonomously modulate their execution frequency by interpreting aleatoric uncertainty (via Shannon entropy of their policy) and epistemic uncertainty (via state-value divergence in a Twin-Critic architecture). To format this, we structure the environment as a Semi-Markov Decision Process (SMDP) and build the SMDP-Aligned Asynchronous Gradient Masking Critic to ensure proper credit assignment. Empirical findings demonstrate massive improvements (> 60% relative baseline acquisition leaps) over current temporal models. By assessing LBF, MPE, and the 115-dimensional state space of Google Research Football (GRF), ETD correctly prevented premature policy collapse. Remarkably, this unconstrained approach leads to emergent Temporal Role Specialization, reducing computational overhead by a statistically dominant 73.6% entirely during off-ball execution without deteriorating centralized task dominance.

Electronic health records (EHRs) are invaluable for clinical research, yet privacy concerns severely restrict data sharing. Synthetic data generation offers a promising solution, but EHRs present unique challenges: they contain both numerical and categorical features that evolve over time. While diffusion models have demonstrated strong performance in EHR synthesis, existing approaches predominantly rely on discrete-time formulations, which suffer from finite-step approximation errors and coupled training-sampling step counts. We propose a continuous-time diffusion framework for generating mixed-type time-series EHRs with three contributions: (1) continuous-time diffusion with a bidirectional gated recurrent unit backbone for capturing temporal dependencies, (2) unified Gaussian diffusion via learnable continuous embeddings for categorical variables, enabling joint cross-feature modeling, and (3) a factorized learnable noise schedule that adapts to per-feature-per-timestep learning difficulties. Experiments on two large-scale intensive care unit datasets demonstrate that our method outperforms existing approaches in downstream task performance, distribution fidelity, and discriminability, while requiring only 50 sampling steps compared to 1,000 for baseline methods. Classifier-free guidance further enables effective conditional generation for class-imbalanced clinical scenarios.

Large language models have recently been proposed as tools for automated essay scoring, but their agreement with human grading remains unclear. In this work, we evaluate how LLM-generated scores compare with human grades and analyze the grading behavior of several models from the GPT and Llama families in an out-of-the-box setting, without task-specific training. Our results show that agreement between LLM and human scores remains relatively weak and varies with essay characteristics. In particular, compared to human raters, LLMs tend to assign higher scores to short or underdeveloped essays, while assigning lower scores to longer essays that contain minor grammatical or spelling errors. We also find that the scores generated by LLMs are generally consistent with the feedback they generate: essays receiving more praise tend to receive higher scores, while essays receiving more criticism tend to receive lower scores. These results suggest that LLM-generated scores and feedback follow coherent patterns but rely on signals that differ from those used by human raters, resulting in limited alignment with human grading practices. Nevertheless, our work shows that LLMs produce feedback that is consistent with their grading and that they can be reliably used in supporting essay scoring.

Filtered Vector Search (FVS) is critical for supporting semantic search and GenAI applications in modern database systems. However, existing research most often evaluates algorithms in specialized libraries, making optimistic assumptions that do not align with enterprise-grade database systems. Our work challenges this premise by demonstrating that in a production-grade database system, commonly made assumptions do not hold, leading to performance characteristics and algorithmic trade-offs that are fundamentally different from those observed in isolated library settings. This paper presents the first in-depth analysis of filter-agnostic FVS algorithms within a production PostgreSQL-compatible system. We systematically evaluate post-filtering and inline-filtering strategies across a wide range of selectivities and correlations. Our central finding is that the optimal algorithm is not dictated by the cost of distance computations alone, but that system-level overheads that come from both distance computations and filter operations (like page accesses and data retrieval) play a significant role. We demonstrate that graph-based approaches (such as NaviX/ACORN) can incur prohibitive numbers of filter checks and system-level overheads, compared with clustering-based indexes such as ScaNN, often canceling out their theoretical benefits in real-world database environments. Ultimately, our findings provide the database community with crucial insights and practical guidelines, demonstrating that the optimal choice for a filter-agnostic FVS algorithm is not absolute, but rather a system-aware decision contingent on the interplay between workload characteristics and the underlying costs of data access in a real-world database architecture.

In Large Language Model (LLM) inference, early-exit refers to stopping computation at an intermediate layer once the prediction is sufficiently confident, thereby reducing latency and cost. However, recent LLMs adopt improved pretraining recipes and architectures that reduce layer redundancy, potentially limiting early-exit opportunities. We re-evaluate layer-wise early-exit in modern LLMs and analyze how intermediate representations evolve during training. We introduce a metric to quantify a model's intrinsic suitability for early-exit and propose a benchmark for researchers to explore the potential early-exit benefits on different models and workloads. Our results show a diminishing trend in early-exit effectiveness across newer model generations. We further find that dense transformers generally offer greater early-exit potential than Mixture-of-Experts and State Space Models. In addition, larger models, particularly those with more than 20 billion parameters, and base pretrained models without specialized tuning tend to exhibit higher early-exit potential.

Software architecture models capture early design decisions that strongly influence system quality attributes, including security. However, architecture-level security assessment and feedback are often absent in practice, allowing security weaknesses to propagate into later phases of the software development lifecycle and, in some cases, to remain undiscovered, ultimately leading to vulnerable systems. In this paper, we bridge this gap by proposing the generation of Abstract Penetration Test Cases (APTCs) from software architecture models as an input to support architecture-level security assessment. We first introduce a metamodel that defines the APTC concept, and then investigate the use of large language models with different prompting strategies to generate meaningful APTCs from architecture models. To design the APTC metamodel, we analyze relevant standards and state of the art using two criteria: (i) derivability from software architecture, and (ii) usability for both architecture security assessment and subsequent penetration testing. Building on this metamodel, we then proceed to generate APTCs from software architecture models. Our evaluation shows promising results, achieving up to 93\% usefulness and 86\% correctness, indicating that the generated APTCs can substantially support both architects (by highlighting security-critical design decisions) and penetration testers (by providing actionable testing guidance).

Recent advances in generative AI systems have dramatically reduced the cost of digital production, fueling narratives that widespread participation in software creation will yield a proliferation of viable companies. This paper challenges that assumption. We introduce the Builder Saturation Effect, formalizing a model in which production scales elastically but human attention remains finite. In markets with near-zero marginal costs and free entry, increases in the number of producers dilute average attention and returns per producer, even as total output expands. Extending the framework to incorporate quality heterogeneity and reinforcement dynamics, we show that equilibrium outcomes exhibit declining average payoffs and increasing concentration, consistent with power-law-like distributions. These results suggest that AI-enabled, democratised production is more likely to intensify competition and produce winner-take-most outcomes than to generate broadly distributed entrepreneurial success. Contribution type: This paper is primarily a work of synthesis and applied formalisation. The individual theoretical ingredients - attention scarcity, free-entry dilution, superstar effects, preferential attachment - are well established in their respective literatures. The contribution is to combine them into a unified framework and direct the resulting predictions at a specific contemporary claim about AI-enabled entrepreneurship.

The rapid adoption of large language models (LLMs) in education raises profound challenges for assessment design. To adapt assessments to the presence of LLM-based tools, it is crucial to characterize the strengths and weaknesses of LLMs in a generalizable, valid and reliable manner. However, current LLM evaluations often rely on descriptive statistics derived from benchmarks, and little research applies theory-grounded measurement methods to characterize LLM capabilities relative to human learners in ways that directly support assessment design. Here, by combining educational data mining and psychometric theory, we introduce a statistically principled approach for identifying items on which humans and LLMs show systematic response differences, pinpointing where assessments may be most vulnerable to AI misuse, and which task dimensions make problems particularly easy or difficult for generative AI. The method is based on Differential Item Functioning (DIF) analysis -- traditionally used to detect bias across demographic groups -- together with negative control analysis and item-total correlation discrimination analysis. It is evaluated on responses from human learners and six leading chatbots (ChatGPT-4o \& 5.2, Gemini 1.5 \& 3 Pro, Claude 3.5 \& 4.5 Sonnet) to two instruments: a high school chemistry diagnostic test and a university entrance exam. Subject-matter experts then analyzed DIF-flagged items to characterize task dimensions associated with chatbot over- or under-performance. Results show that DIF-informed analytics provide a robust framework for understanding where LLM and human capabilities diverge, and highlight their value for improving the design of valid, reliable, and fair assessment in the AI era.

Agriculture remains a cornerstone of global health and economic sustainability, yet labor-intensive tasks such as harvesting high-value crops continue to face growing workforce shortages. Robotic harvesting systems offer a promising solution; however, their deployment in unstructured orchard environments is constrained by inefficient perception-to-action pipelines. In particular, existing approaches often rely on exhaustive inverse kinematics or motion planning to determine whether a target fruit is reachable, leading to unnecessary computation and delayed decision-making. Our approach combines RGB-D perception with active learning to directly learn reachability as a binary decision problem. We then leverage active learning to selectively query the most informative samples for reachability labeling, significantly reducing annotation effort while maintaining high predictive accuracy. Extensive experiments demonstrate that the proposed framework achieves accurate reachability prediction with substantially fewer labeled samples, yielding approximately 6--8% higher accuracy than random sampling and enabling label-efficient adaptation to new orchard configurations. Among the evaluated strategies, entropy- and margin-based sampling outperform Query-by-Committee and standard uncertainty sampling in low-label regimes, while all strategies converge to comparable performance as the labeled set grows. These results highlight the effectiveness of active learning for task-level perception in agricultural robotics and position our approach as a scalable alternative to computation-heavy kinematic reachability analysis. Our code is available through https://github.com/wsu-cyber-security-lab-ai/active-learning.

Large Language Models (LLMs) offer transformative solutions across many domains, but healthcare integration is hindered by strict data privacy constraints. Clinical narratives are dense with ambiguous acronyms, misinterpretation these abbreviations can precipitate severe outcomes like life-threatening medication errors. While cloud-dependent LLMs excel at Acronym Disambiguation, transmitting Protected Health Information to external servers violates privacy frameworks. To bridge this gap, this study pioneers the evaluation of small-parameter models deployed entirely on-device to ensure privacy preservation. We introduce a privacy-preserving cascaded pipeline leveraging general-purpose local models to detect clinical acronyms, routing them to domain-specific biomedical models for context-relevant expansions. Results reveal that while general instruction-following models achieve high detection accuracy (~0.988), their expansion capabilities plummet (~0.655). Our cascaded approach utilizes domain-specific medical models to increase expansion accuracy to (~0.81). This novel work demonstrates that privacy-preserving, on-device (2B-10B) models deliver high-fidelity clinical acronym disambiguation support.

Deep learning models are increasingly deployed in safety-critical applications, where reliable out-of-distribution (OOD) detection is essential to ensure robustness. Existing methods predominantly rely on the penultimate-layer activations of neural networks, assuming they encapsulate the most informative in-distribution (ID) representations. In this work, we revisit this assumption to show that intermediate layers encode equally rich and discriminative information for OOD detection. Based on this observation, we propose a simple yet effective model-agnostic approach that leverages internal representations across multiple layers. Our scheme aggregates features from successive convolutional blocks, computes class-wise mean embeddings, and applies L_2 normalization to form compact ID prototypes capturing class semantics. During inference, cosine similarity between test features and these prototypes serves as an OOD score--ID samples exhibit strong affinity to at least one prototype, whereas OOD samples remain uniformly distant. Extensive experiments on state-of-the-art OOD benchmarks across diverse architectures demonstrate that our approach delivers robust, architecture-agnostic performance and strong generalization for image classification. Notably, it improves AUROC by up to 4.41% and reduces FPR by 13.58%, highlighting multi-layer feature aggregation as a powerful yet underexplored signal for OOD detection, challenging the dominance of penultimate-layer-based methods. Our code is available at: https://github.com/sgchr273/cosine-layers.git.

We study long-horizon planning in 3D environments from under-specified natural-language goals using only visual observations, focusing on multi-step 3D box rearrangement tasks. Existing approaches typically rely on symbolic planners with brittle relational grounding of states and goals, or on direct action-sequence generation from 2D vision-language models (VLMs). Both approaches struggle with reasoning over many objects, rich 3D geometry, and implicit semantic constraints. Recent advances in 3D VLMs demonstrate strong grounding of natural-language referents to 3D segmentation masks, suggesting the potential for more general planning capabilities. We extend existing 3D grounding models and propose Reactive Action Mask Planner (RAMP-3D), which formulates long-horizon planning as sequential reactive prediction of paired 3D masks: a "which-object" mask indicating what to pick and a "which-target-region" mask specifying where to place it. The resulting system processes RGB-D observations and natural-language task specifications to reactively generate multi-step pick-and-place actions for 3D box rearrangement. We conduct experiments across 11 task variants in warehouse-style environments with 1-30 boxes and diverse natural-language constraints. RAMP-3D achieves 79.5% success rate on long-horizon rearrangement tasks and significantly outperforms 2D VLM-based baselines, establishing mask-based reactive policies as a promising alternative to symbolic pipelines for long-horizon planning.

Multi-chain ecosystems suffer from fragmented identity, siloed liquidity, and bridge-dependent token transfers. We present n-VM, a Layer-1 architecture that hosts n heterogeneous virtual machines as co-equal execution environments over shared consensus and shared state. The design combines three components: a dispatcher that routes transactions by opcode prefix, a unified identity layer in which one 32-byte commitment anchors VM-specifific addresses, and a unified token ledger that exposes VM-native interfaces such as ERC-20 and SPL over a common balance store. We formalize routing, identity derivation, and token transfer semantics, and prove cross-VM transfer atomicity and identity isolation under standard cryptographic assumptions. We describe a concrete instantiation with five VMs: a native runtime, EVM, SVM, Bitcoin Script, and TVM. We also present context-based sharding and a write-set scheduler for parallel execution. Under an analytical throughput model, the architecture admits a projected range of about 16,000 to 66,000 transactions per second on commodity hardware.

Accurate estimation of forest biomass, a major carbon sink, relies heavily on tree-level traits such as height and species. Unoccupied Aerial Vehicles (UAVs) capturing high-resolution imagery from a single RGB camera offer a cost-effective and scalable approach for mapping and measuring individual trees. We introduce BIRCH-Trees, the first benchmark for individual tree height and species estimation from tree-centered UAV images, spanning three datasets: temperate forests, tropical forests, and boreal plantations. We also present DINOvTree, a unified approach using a Vision Foundation Model (VFM) backbone with task-specific heads for simultaneous height and species prediction. Through extensive evaluations on BIRCH-Trees, we compare DINOvTree against commonly used vision methods, including VFMs, as well as biological allometric equations. We find that DINOvTree achieves top overall results with accurate height predictions and competitive classification accuracy while using only 54% to 58% of the parameters of the second-best approach.

The rapid deployment of machine learning across platforms from milliwatt-class TinyML devices to large language models has made energy efficiency a primary constraint for sustainable AI. Across these scales, performance and energy are increasingly limited by data movement and memory-system behavior rather than by arithmetic throughput alone. This work reviews energy efficient software hardware codesign methods spanning edge inference and training to datacenter-scale LLM serving, covering accelerator architectures (e.g., ASIC/FPGA dataflows, processing-/compute-in-memory designs) and system-level techniques (e.g., partitioning, quantization, scheduling, and runtime adaptation). We distill common design levers and trade-offs, and highlight recurring gaps including limited cross-platform generalization, large and costly co-design search spaces, and inconsistent benchmarking across workloads and deployment settings. Finally, we outline a hierarchical decomposition perspective that maps optimization strategies to computational roles and supports incremental adaptation, offering practical guidance for building energy and carbon aware ML systems.

We introduce Echoes, a new dataset for music deepfake detection designed for training and benchmarking detectors under realistic and provider-diverse conditions. Echoes comprises 3,577 tracks (110 hours of audio) spanning multiple genres (pop, rock, electronic), and includes content generated by ten popular AI music generation systems. To prevent shortcut learning and promote robust generalization, the dataset is deliberately constructed to be challenging, enforcing semantic-level alignment between spoofed audio and bona fide references. This alignment is achieved by conditioning generated audio samples directly on bona-fide waveforms or song descriptors. We evaluate Echoes in a cross-dataset setting against three existing AI-generated music datasets using state-of-the-art Wav2Vec2 XLS-R 2B representations. Results show that (i) Echoes is the hardest in-domain dataset; (ii) detectors trained on existing datasets transfer poorly to Echoes; (iii) training on Echoes yields the strongest generalization performance. These findings suggest that provider diversity and semantic alignment help learn more transferable detection cues.

Advanced driver assistance systems (ADAS) often rely on deep neural networks to interpret driving images and support vehicle control. Although reliable under nominal conditions, these systems remain vulnerable to input variations and out-of-distribution data, which can lead to unsafe behavior. To this aim, this tool paper presents the architecture and functioning of PerturbationDrive, a testing framework to perform robustness and generalization testing of ADAS. The framework features more than 30 image perturbations from the literature that mimic changes in weather, lighting, or sensor quality and extends them with dynamic and attention-based variants. PerturbationDrive supports both offline evaluation on static datasets and online closed-loop testing in different simulators. Additionally, the framework integrates with procedural road generation and search-based testing, enabling systematic exploration of diverse road topologies combined with image perturbations. Together, these features allow PerturbationDrive to evaluate robustness and generalization capabilities of ADAS across varying scenarios, making it a reproducible and extensible framework for systematic system-level testing.

We introduce GTO Wizard Benchmark, a public API and standardized evaluation framework for benchmarking algorithms in Heads-Up No-Limit Texas Hold'em (HUNL). The benchmark evaluates agents against GTO Wizard AI, a state-of-the-art superhuman poker agent that approximates Nash Equilibria, and defeated Slumbot, the 2018 Annual Computer Poker Competition champion and previous strongest publicly accessible HUNL benchmark, by $19.4$ $\pm$ $4.1$ bb/100. Variance is a fundamental challenge in poker evaluation; we address this by integrating AIVAT, a provably unbiased variance reduction technique that achieves equivalent statistical significance with ten times fewer hands than naive Monte Carlo evaluation. We conduct a comprehensive benchmarking study of state-of-the-art large language models under zero-shot conditions, including GPT-5.4, Claude Opus 4.6, Gemini 3.1 Pro, Grok 4, and others. Initial results and analysis reveal dramatic progress in LLM reasoning over recent years, yet all models remain far below the baseline established by our benchmark. Qualitative analysis reveals clear opportunities for improvement, including representation and the ability to reason over hidden states. This benchmark provides researchers with a precise and quantifiable setting to evaluate advances in planning and reasoning in multi-agent systems with partial observability.

When large language models make ethical judgments, do their internal representations distinguish between normative frameworks, or collapse ethics into a single acceptability dimension? We probe hidden representations across five ethical frameworks (deontology, utilitarianism, virtue, justice, commonsense) in six LLMs spanning 4B--72B parameters. Our analysis reveals differentiated ethical subspaces with asymmetric transfer patterns -- e.g., deontology probes partially generalize to virtue scenarios while commonsense probes fail catastrophically on justice. Disagreement between deontological and utilitarian probes correlates with higher behavioral entropy across architectures, though this relationship may partly reflect shared sensitivity to scenario difficulty. Post-hoc validation reveals that probes partially depend on surface features of benchmark templates, motivating cautious interpretation. We discuss both the structural insights these methods provide and their epistemological limitations.

Gradient boosting, a method of building additive ensembles from weak learners, has established itself as a practical and theoretically-motivated approach to approximate functions, especially using decision tree weak learners. Comparable methods for smooth parametric learners, such as neural networks, remain less developed in both training methodology and theory. To this end, we introduce \texttt{VPBoost} ({\bf V}ariable {\bf P}rojection {\bf Boost}ing), a gradient boosting algorithm for separable smooth approximators, i.e., models with a smooth nonlinear featurizer followed by a final linear mapping. \texttt{VPBoost} fuses variable projection, a training paradigm for separable models that enforces optimality of the linear weights, with a second-order weak learning strategy. The combination of second-order boosting, separable models, and variable projection give rise to a closed-form solution for the optimal linear weights and a natural interpretation of \VPBoost as a functional trust-region method. We thereby leverage trust-region theory to prove \VPBoost converges to a stationary point under mild geometric conditions and, under stronger assumptions, achieves a superlinear convergence rate. Comprehensive numerical experiments on synthetic data, image recognition, and scientific machine learning benchmarks demonstrate that \VPBoost learns an ensemble with improved evaluation metrics in comparison to gradient-descent-based boosting and attains competitive performance relative to an industry-standard decision tree boosting algorithm.

We present Ethio-ASR, a suite of multilingual CTC-based automatic speech recognition (ASR) models jointly trained on five Ethiopian languages: Amharic, Tigrinya, Oromo, Sidaama, and Wolaytta. These languages belong to the Semitic, Cushitic, and Omotic branches of the Afroasiatic family, and remain severely underrepresented in speech technology despite being spoken by the vast majority of Ethiopia's population. We train our models on the recently released WAXAL corpus using several pre-trained speech encoders and evaluate against strong multilingual baselines, including OmniASR. Our best model achieves an average WER of 30.48% on the WAXAL test set, outperforming the best OmniASR model with substantially fewer parameters. We further provide a comprehensive analysis of gender bias, the contribution of vowel length and consonant gemination to ASR errors, and the training dynamics of multilingual CTC models. Our models and codebase are publicly available to the research community.

We address target interception in contested environments in the presence of multiple defenders whose interception capability is limited by finite ranges. Conventional methods typically impose conservative stand-off constraints based on maximum engagement distance and neglect the interceptors' actuator limitations. Instead, we formulate safety constraints using defender-induced engagement zones. To account for actuator limits, the vehicle model is augmented with input saturation dynamics. A time-varying safe-set tightening parameter is introduced to compensate for transient constraint violations induced by actuator dynamics. To ensure scalable safety enforcement in multi-defender scenarios, a smooth aggregate safety function is constructed using a log-sum-exp operator combining individual threat measures associated with each defender's capability. A smooth switching guidance strategy is then developed to coordinate interception and safety objectives. The attacker pursues the target when sufficiently distant from threat boundaries and progressively activates evasive motion as the EZ boundaries are approached. The resulting controller relies only on relative measurements and does not require knowledge of defender control inputs, thus facilitating a fully distributed and scalable implementation. Rigorous analysis provides sufficient conditions guaranteeing target interception, practical safety with respect to all defender engagement zones, and satisfaction of actuator bounds. An input-constrained guidance law based on conservative stand-off distance is also developed to quantify the conservatism of maximum-range-based safety formulations. Simulations with stationary and maneuvering defenders demonstrate that the proposed formulation yields shorter interception paths and reduced interception time compared with conventional methods while maintaining safety throughout the engagement.

In fluorescence microscopy, spectral unmixing aims to recover individual fluorophore concentrations from spectral images that capture mixed fluorophore emissions. Since classical methods operate pixel-wise and rely on least-squares fitting, their performance degrades with increasingly overlapping emission spectra and higher levels of noise, suggesting that a data-driven approach that can learn and utilize a structural prior might lead to improved results. Learning-based approaches for spectral imaging do exist, but they are either not optimized for microscopy data or are developed for very specific cases that are not applicable to fluorescence microscopy settings. To address this, we propose λSplit, a physics-informed deep generative model that learns a conditional distribution over concentration maps using a hierarchical Variational Autoencoder. A fully differentiable Spectral Mixer enforces consistency with the image formation process, while the learned structural priors enable state-of-the-art unmixing and implicit noise removal. We demonstrate λSplit on 3 real-world datasets that we synthetically cast into a total of 66 challenging spectral unmixing benchmarks. We compare our results against a total of 10 baseline methods, including classical methods and a range of learning-based methods. Our results consistently show competitive performance and improved robustness in high noise regimes, when spectra overlap considerably, or when the spectral dimensionality is lowered, making λSplit a new state-of-the-art for spectral unmixing of fluorescent microscopy data. Importantly, λSplit is compatible with spectral data produced by standard confocal microscopes, enabling immediate adoption without specialized hardware modifications.

While recent work has benchmarked large language models on Swiss legal translation (Niklaus et al., 2025) and academic legal reasoning from university exams (Fan et al., 2025), no existing benchmark evaluates frontier model performance on applied Swiss regulatory compliance tasks. I introduce Swiss-Bench SBP-002, a trilingual benchmark of 395 expert-crafted items spanning three Swiss regulatory domains (FINMA, Legal-CH, EFK), seven task types, and three languages (German, French, Italian), and evaluate ten frontier models from March 2026 using a structured three-dimension scoring framework assessed via a blind three-judge LLM panel (GPT-4o, Claude Sonnet 4, Qwen3-235B) with majority-vote aggregation and weighted kappa = 0.605, with reference answers validated by an independent human legal expert on a 100-item subset (73% rated Correct, 0% Incorrect, perfect Legal Accuracy). Results reveal three descriptive performance clusters: Tier A (35-38% correct), Tier B (26-29%), and Tier C (13-21%). The benchmark proves difficult: even the top-ranked model (Qwen 3.5 Plus) achieves only 38.2% correct, with 47.3% incorrect and 14.4% partially correct. Task type difficulty varies widely: legal translation and case analysis yield 69-72% correct rates, while regulatory Q&A, hallucination detection, and gap analysis remain below 9%. Within this roster (seven open-weight, three closed-source), an open-weight model leads the ranking, and several open-weight models match or outperform their closed-source counterparts. These findings provide an initial empirical reference point for assessing frontier model capability on Swiss regulatory tasks under zero-retrieval conditions.

Deploying large language models on-device for always-on personal agents demands sustained inference from hardware tightly constrained in power, thermal envelope, and memory. We benchmark Qwen 2.5 1.5B (4-bit quantised) across four platforms: a Raspberry Pi 5 with Hailo-10H NPU, a Samsung Galaxy S24 Ultra, an iPhone 16 Pro, and a laptop NVIDIA RTX 4050 GPU. Using a fixed 258-token prompt over 20 warm-condition iterations per device, we measure throughput, latency, power, and thermal behaviour. For mobile platforms, thermal management supersedes peak compute as the primary constraint: the iPhone 16 Pro loses nearly half its throughput within two iterations, and the S24 Ultra suffers a hard OS-enforced GPU frequency floor that terminates inference entirely. On dedicated hardware, distinct constraints dominate: the RTX 4050 is bounded by its battery power ceiling, while the Hailo-10H is limited by on-module memory bandwidth. The RTX 4050 sustains 131.7 tok/s at 34.1 W; the Hailo-10H sustains 6.9 tok/s at under 2 W with near-zero variance, matching the RTX 4050 in energy proportionality at 19x lower throughput. Results should be interpreted as platform-level deployment characterisations for a single model and prompt type, reflecting hardware and software combined, rather than general claims about hardware capability alone.

Large language models (LLMs) have enabled agentic systems that can reason, plan, and act across complex tasks, but it remains unclear whether they can allocate resources effectively under uncertainty. Unlike short-horizon reactive decisions, allocation requires committing scarce resources over time while balancing competing objectives and preserving flexibility for future needs. We introduce EnterpriseArena, the first benchmark for evaluating agents on long-horizon enterprise resource allocation. It instantiates CFO-style decision-making in a 132-month enterprise simulator combining firm-level financial data, anonymized business documents, macroeconomic and industry signals, and expert-validated operating rules. The environment is partially observable and reveals the state only through budgeted organizational tools, forcing agents to trade off information acquisition against conserving scarce resources. Experiments on eleven advanced LLMs show that this setting remains highly challenging: only 16% of runs survive the full horizon, and larger models do not reliably outperform smaller ones. These results identify long-horizon resource allocation under uncertainty as a distinct capability gap for current LLM agents.

Large language models can generate runnable software artifacts, but their security remains difficult to evaluate end to end. This study examines that problem through a Detect--Repair--Verify (DRV) workflow, in which vulnerabilities are detected, repaired, and then rechecked with security and functional tests. It addresses four gaps in current evidence: the lack of test-grounded benchmarks for LLM-generated artifacts, limited evidence on pipeline-level effectiveness, unclear reliability of detection reports as repair guidance, and uncertain repair trustworthiness under verification. To support this study, EduCollab is constructed as a multi-language, multi-granularity benchmark of runnable LLM-generated web applications in PHP, JavaScript, and Python. Each artifact is paired with executable functional and exploit test suites, and the benchmark spans project-, requirement-, and file-level settings. On this benchmark, the study compares unrepaired baselines, single-pass detect--repair, and bounded iterative DRV under comparable budget constraints. Outcomes are measured by secure-and-correct yield, and intermediate artifacts and iteration traces are analyzed to assess report actionability and repair failure modes. The results show that bounded iterative DRV can improve secure-and-correct yield over single-pass repair, but the gains are uneven at the project level and become clearer at narrower repair scopes. Detection reports are often useful for downstream repair, but their reliability is inconsistent. Repair trustworthiness also depends strongly on repair scope. These findings highlight the need for test-grounded, end-to-end evaluation of LLM-based vulnerability management workflows.

Code LLMs often default to particular programming languages and libraries under neutral prompts. We investigate whether these preferences are encoded as approximately linear directions in activation space that can be manipulated at inference time. Using a difference-in-means method, we estimate layer-wise steering vectors for five language/library pairs and add them to model hidden states during generation. Across three open-weight code LLMs, these interventions substantially increase generation toward the target ecosystem under neutral prompts and often remain effective even when prompts explicitly request the opposite choice. Steering strength varies by model and target, with common ecosystems easier to induce than rarer alternatives, and overly strong interventions can reduce output quality. Overall, our results suggest that code-style preferences in LLMs are partly represented by compact, steerable structure in activation space.

This study presents a benchmark for evaluating the Visual Word Sense Disambiguation (Visual-WSD) task in Ukrainian. The main goal of the Visual-WSD task is to identify, with minimal contextual information, the most appropriate representation of a given ambiguous word from a set of ten images. To construct this benchmark, we followed a methodology similar to that proposed by (CITATION), who previously introduced benchmarks for the Visual-WSD task in English, Italian, and Farsi. This approach allows us to incorporate the Ukrainian benchmark into a broader framework for cross-language model performance comparisons. We collected the benchmark data semi-automatically and refined it with input from domain experts. We then assessed eight multilingual and multimodal large language models using this benchmark. All tested models performed worse than the zero-shot CLIP-based baseline model (CITATION) used by (CITATION) for the English Visual-WSD task. Our analysis revealed a significant performance gap in the Visual-WSD task between Ukrainian and English.

Large language models (LLMs) are increasingly deployed as optimization modules in agentic systems, yet the fundamental limits of such LLM-mediated improvement remain poorly understood. Here we propose a theory of LLM information susceptibility, centred on the hypothesis that when computational resources are sufficiently large, the intervention of a fixed LLM does not increase the performance susceptibility of a strategy set with respect to budget. We develop a multi-variable utility-function framework that generalizes this hypothesis to architectures with multiple co-varying budget channels, and discuss the conditions under which co-scaling can exceed the susceptibility bound. We validate the theory empirically across structurally diverse domains and model scales spanning an order of magnitude, and show that nested, co-scaling architectures open response channels unavailable to fixed configurations. These results clarify when LLM intervention helps and when it does not, demonstrating that tools from statistical physics can provide predictive constraints for the design of AI systems. If the susceptibility hypothesis holds generally, the theory suggests that nested architectures may be a necessary structural condition for open-ended agentic self-improvement.

Artificial intelligence (AI) is increasingly being explored in health and social care to reduce administrative workload and allow staff to spend more time on patient care. This paper evaluates a voice-enabled Care Home Smart Speaker designed to support everyday activities in residential care homes, including spoken access to resident records, reminders, and scheduling tasks. A safety-focused evaluation framework is presented that examines the system end-to-end, combining Whisper-based speech recognition with retrieval-augmented generation (RAG) approaches (hybrid, sparse, and dense). Using supervised care-home trials and controlled testing, we evaluated 330 spoken transcripts across 11 care categories, including 184 reminder-containing interactions. These evaluations focus on (i) correct identification of residents and care categories, (ii) reminder recognition and extraction, and (iii) end-to-end scheduling correctness under uncertainty (including safe deferral/clarification). Given the safety-critical nature of care homes, particular attention is also paid to reliability in noisy environments and across diverse accents, supported by confidence scoring, clarification prompts, and human-in-the-loop oversight. In the best-performing configuration (GPT-5.2), resident ID and care category matching reached 100% (95% CI: 98.86-100), while reminder recognition reached 89.09\% (95% CI: 83.81-92.80) with zero missed reminders (100% recall) but some false positives. End-to-end scheduling via calendar integration achieved 84.65% exact reminder-count agreement (95% CI: 78.00-89.56), indicating remaining edge cases in converting informal spoken instructions into actionable events. The findings suggest that voice-enabled systems, when carefully evaluated and appropriately safeguarded, can support accurate documentation, effective task management, and trustworthy use of AI in care home settings.

Digging-in effects, where disambiguation difficulty increases with longer ambiguous regions, have been cited as evidence for self-organized sentence processing, in which structural commitments strengthen over time. In contrast, surprisal theory predicts no such effect unless lengthening genuinely shifts statistical expectations, and neural language models appear to show the opposite pattern. Whether digging-in is a robust real-time phenomenon in human sentence processing -- or an artifact of wrap-up processes or methodological confounds -- remains unclear. We report two experiments on English NP/Z garden-path sentences using Maze and self-paced reading, comparing human behavior with predictions from an ensemble of large language models. We find no evidence for real-time digging-in effects. Critically, items with sentence-final versus nonfinal disambiguation show qualitatively different patterns: positive digging-in trends appear only sentence-finally, where wrap-up effects confound interpretation. Nonfinal items -- the cleaner test of real-time processing -- show reverse trends consistent with neural model predictions.

Large Language Models (LLMs) are showing remarkable performance in generating source code, yet the generated code often has issues like compilation errors or incorrect code. Researchers and developers often face wasted effort in implementing checks and refining LLM-generated code, frequently duplicating their efforts. This paper presents LLMLOOP, a framework that automates the refinement of both source code and test cases produced by LLMs. LLMLOOP employs five iterative loops: resolving compilation errors, addressing static analysis issues, fixing test case failures, and improving test quality through mutation analysis. These loops ensure the generation of high-quality test cases that serve as both a validation mechanism and a regression test suite for the generated code. We evaluated LLMLOOP on HUMANEVAL-X, a recent benchmark of programming tasks. Results demonstrate the tool's effectiveness in refining LLM-generated outputs.

Automated testing is essential for evaluating and improving the reliability of Large Language Models (LLMs), yet the lack of automated oracles for verifying output correctness remains a key challenge. We present LLMORPH, an automated testing tool specifically designed for LLMs performing NLP tasks, which leverages Metamorphic Testing (MT) to uncover faulty behaviors without relying on human-labeled data. MT uses Metamorphic Relations (MRs) to generate follow-up inputs from source test input, enabling detection of inconsistencies in model outputs without the need of expensive labelled data. LLMORPH is aimed at researchers and developers who want to evaluate the robustness of LLM-based NLP systems. In this paper, we detail the design, implementation, and practical usage of LLMORPH, demonstrating how it can be easily extended to any LLM, NLP task, and set of MRs. In our evaluation, we applied 36 MRs across four NLP benchmarks, testing three state-of-the-art LLMs: GPT-4, LLAMA3, and HERMES 2. This produced over 561,000 test executions. Results demonstrate LLMORPH's effectiveness in automatically exposing inconsistencies.

Although large language models (LLMs) have advanced rapidly, robust automation of complex software workflows remains an open problem. In long-horizon settings, agents frequently suffer from cascading errors and environmental stochasticity; a single misstep in a dynamic interface can lead to task failure, resulting in hallucinations or trial-and-error. This paper introduces $\textit{Environment Maps}$: a persistent, agent-agnostic representation that mitigates these failures by consolidating heterogeneous evidence, such as screen recordings and execution traces, into a structured graph. The representation consists of four core components: (1) Contexts (abstracted locations), (2) Actions (parameterized affordances), (3) Workflows (observed trajectories), and (4) Tacit Knowledge (domain definitions and reusable procedures). We evaluate this framework on the WebArena benchmark across five domains. Agents equipped with environment maps achieve a 28.2% success rate, nearly doubling the performance of baselines limited to session-bound context (14.2%) and outperforming agents that have access to the raw trajectory data used to generate the environment maps (23.3%). By providing a structured interface between the model and the environment, Environment Maps establish a persistent foundation for long-horizon planning that is human-interpretable, editable, and incrementally refinable.

Anti-money laundering (AML) systems are important for protecting the global economy. However, conventional rule-based methods rely on domain knowledge, leading to suboptimal accuracy and a lack of scalability. Graph neural networks (GNNs) for digraphs (directed graphs) can be applied to transaction graphs and capture suspicious transactions or accounts. However, most spectral GNNs do not naturally support multi-dimensional edge features, lack interpretability due to edge modifications, and have limited scalability owing to their spectral nature. Conversely, most spatial methods may not capture the money flow well. Therefore, in this work, we propose LineMVGNN (Line-Graph-Assisted Multi-View Graph Neural Network), a novel spatial method that considers payment and receipt transactions. Specifically, the LineMVGNN model extends a lightweight MVGNN module, which performs two-way message passing between nodes in a transaction graph. Additionally, LineMVGNN incorporates a line graph view of the original transaction graph to enhance the propagation of transaction information. We conduct experiments on two real-world account-based transaction datasets: the Ethereum phishing transaction network dataset and a financial payment transaction dataset from one of our industry partners. The results show that our proposed method outperforms state-of-the-art methods, reflecting the effectiveness of money laundering detection with line-graph-assisted multi-view graph learning. We also discuss scalability, adversarial robustness, and regulatory considerations of our proposed method.

The accurate prediction of protein-RNA binding affinity remains an unsolved problem in structural biology, limiting opportunities in understanding gene regulation and designing RNA-targeting therapeutics. A central obstacle is the structural flexibility of RNA, as, unlike proteins, RNA molecules exist as dynamic conformational ensembles. Thus, committing to a single predicted structure discards information relevant to binding. Here, we show that this obstacle can be addressed by extracting pre-structural embeddings, which are intermediate representations from a biomolecular foundation model captured before the structure decoding step. Pre-structural embeddings implicitly encode conformational ensemble information without requiring predicted structures. We build ZeroFold, a transformer-based model that combines pre-structural embeddings from Boltz-2 for both protein and RNA molecules through a cross-modal attention mechanism to predict binding affinity directly from sequence. To support training and evaluation, we construct PRADB, a curated dataset of 2,621 unique protein-RNA pairs with experimentally measured affinities drawn from four complementary databases. On a held-out test set constructed with 40% sequence identity thresholds, ZeroFold achieves a Spearman correlation of 0.65, a value approaching the ceiling imposed by experimental measurement noise. Under progressively fairer evaluation conditions that control for training-set overlap, ZeroFold compares favourably with respect to leading structure-based and leading sequence-based predictors, with the performance gap widening as sequence similarity to competitor training data is reduced. These results illustrate how pre-structural embeddings offer a representation strategy for flexible biomolecules, opening a route to affinity prediction for protein-RNA pairs for which no structural data exist.

Accurate sleep staging is essential for diagnosing OSA and hypopnea in stroke patients. Although PSG is reliable, it is costly, labor-intensive, and manually scored. While deep learning enables automated EEG-based sleep staging in healthy subjects, our analysis shows poor generalization to clinical populations with disrupted sleep. Using Grad-CAM interpretations, we systematically demonstrate this limitation. We introduce iSLEEPS, a newly clinically annotated ischemic stroke dataset (to be publicly released), and evaluate a SE-ResNet plus bidirectional LSTM model for single-channel EEG sleep staging. As expected, cross-domain performance between healthy and diseased subjects is poor. Attention visualizations, supported by clinical expert feedback, show the model focuses on physiologically uninformative EEG regions in patient data. Statistical and computational analyses further confirm significant sleep architecture differences between healthy and ischemic stroke cohorts, highlighting the need for subject-aware or disease-specific models with clinical validation before deployment. A summary of the paper and the code is available at https://himalayansaswatabose.github.io/iSLEEPS_Explainability.github.io/

In clustering, strong dominance in the size of a particular cluster is often undesirable, motivating a measure of cluster size uniformity that can be used to filter such partitions. A basic requirement of such a measure is stability: partitions that differ only slightly in their point assignments should receive similar uniformity scores. A difficulty arises because cluster labels are not fixed objects; algorithms may produce different numbers of labels even when the underlying point distribution changes very little. Measures defined directly over labels can therefore become unstable under label-count perturbations. I introduce the Mass Agreement Score (MAS), a point-centric metric bounded in [0, 1] that evaluates the consistency of expected cluster size as measured from the perspective of points in each cluster. Its construction yields fragment robustness by design, assigning similar scores to partitions with similar bulk structure while remaining sensitive to genuine redistribution of cluster mass.

Existing LLM-based Kubernetes diagnostic systems cannot learn from operational experience, operating on static knowledge bases without improving from past resolutions. We present MetaKube, an experience-aware LLM framework through three synergistic innovations: (1) an Episodic Pattern Memory Network (EPMN) that abstracts diagnostic patterns from historical resolutions and provides confidence-calibrated retrieval for both rapid pattern matching and guided causal exploration, (2) a meta-cognitive controller that dynamically routes between intuitive and analytical pathways based on problem familiarity, optimizing the trade-off between speed and depth, and (3) KubeLLM, a locally-deployable 8B model enhanced through domain-specific post-training on our 7,000-sample Kubernetes Fault Resolution Dataset. Evaluation on 1,873 real-world scenarios demonstrates MetaKube transforms Qwen3-8B from 50.9 to 90.5 points, approaching GPT-4.1 performance while ensuring complete data privacy. EPMN contributes 15.3% improvement through experiential learning, with continuous learning experiments showing progressive gains as the system accumulates operational knowledge. The source code and related resources are available at https://github.com/MetaKube-LLM-for-Kubernetes-Diagnosis/MetaKube.

Efficient thermal management and precise field prediction are critical for the design of advanced energy systems, including electrohydrodynamic transport, microfluidic energy harvesters, and electrically driven thermal regulators. However, the steady-state simulation of these electrothermal coupled multiphysics systems remains challenging for physics-informed neural computation due to strong nonlinear field coupling, temperature-dependent coefficient variability, and complex interface dynamics. This study proposes a Residual Attention Physics-Informed Neural Network (RA-PINN) framework for the unified solution of coupled velocity, pressure, electric-potential, and temperature fields. By integrating a unified five-field operator formulation with residual-connected feature propagation and attention-guided channel modulation, the proposed architecture effectively captures localized coupling structures and steep gradients. We evaluate RA-PINN across four representative energy-relevant benchmarks: constant-coefficient coupling, indirect pressure-gauge constraints, temperature-dependent transport, and oblique-interface consistency. Comparative analysis against Pure-MLP, LSTM-PINN, and pLSTM-PINN demonstrates that RA-PINN achieves superior accuracy, yielding the lowest MSE, RMSE, and relative $L_2$ errors across all scenarios. Notably, RA-PINN maintains high structural fidelity in interface-dominated and variable-coefficient settings where conventional PINN backbones often fail. These results establish RA-PINN as a robust and accurate computational framework for the high-fidelity modeling and optimization of complex electrothermal multiphysics in sustainable energy applications.

Large language models (LLMs) generalize smoothly across continuous semantic spaces, yet strict logical reasoning demands the formation of discrete decision boundaries. Prevailing theories relying on linear isometric projections fail to resolve this fundamental tension. In this work, we argue that task context operates as a non-isometric dynamical operator that enforces a necessary "topological distortion." By applying Gram-Schmidt decomposition to residual-stream activations , we reveal a dual-modulation mechanism driving this process: a class-agnostic topological preservation that anchors global structure to prevent semantic collapse, and a specific algebraic divergence that directionally tears apart cross-class concepts to forge logical boundaries. We validate this geometric evolution across a gradient of tasks, from simple mapping to complex primality testing. Crucially, targeted specific vector ablation establishes a strict causal binding between this topology and model function: algebraically erasing the divergence component collapses parity classification accuracy from 100% to chance levels (38.57%). Furthermore, we uncover a three-phase layer-wise geometric dynamic and demonstrate that under social pressure prompts, models fail to generate sufficient divergence. This results in a "manifold entanglement" that geometrically explains sycophancy and hallucination. Ultimately, our findings revise the linear-isometric presumption, demonstrating that the emergence of discrete logic in LLMs is purchased at an irreducible cost of topological deformation.

Understanding wafer-level spatial variations from in-situ process signals is essential for advanced plasma etching process monitoring. While most data-driven approaches focus on scalar indicators such as average etch rate, actual process quality is determined by complex two-dimensional spatial distributions across the wafer. This paper presents a spatial regression model that predicts wafer-level etch depth distributions directly from multichannel in-situ process time series. We propose a Time-LLM-based spatial regression model that extends LLM reprogramming from conventional time-series forecasting to wafer-level spatial estimation by redesigning the input embedding and output projection. Using the BOSCH plasma-etching dataset, we demonstrate stable performance under data-limited conditions, supporting the feasibility of LLM-based reprogramming for wafer-level spatial monitoring.

Today, large language models have demonstrated their strengths in various tasks ranging from reasoning, code generation, and complex problem solving. However, this advancement comes with a high computational cost and memory requirements, making it challenging to deploy these models on edge devices to ensure real-time responses and data privacy. Quantization is one common approach to reducing memory use, but most methods apply it uniformly across all layers. This does not account for the fact that different layers may respond differently to reduced precision. Importantly, memory consumption and computational throughput are not necessarily aligned, further complicating deployment decisions. This paper proposes an adaptive mixed precision quantization mechanism that balances memory, latency, and accuracy in edge deployment under user-defined priorities. This is achieved by analyzing the layer-wise contribution and by inferring how different quantization types behave across the target hardware platform in order to assign the most suitable quantization type to each layer. This integration ensures that layer importance and the overall performance trade-offs are jointly respected in this design. Our work unlocks new configuration designs that uniform quantization cannot achieve, expanding the solution space to efficiently deploy the LLMs on resource-constrained devices.

Federated Learning (FL), as a popular distributed learning paradigm, has shown outstanding performance in improving computational efficiency and protecting data privacy, and is widely applied in industrial image classification. However, due to its distributed nature, FL is vulnerable to threats from malicious clients, with poisoning attacks being a common threat. A major limitation of existing poisoning attack methods is their difficulty in bypassing model performance tests and defense mechanisms based on model anomaly detection. This often results in the detection and removal of poisoned models, which undermines their practical utility. To ensure both the performance of industrial image classification and attacks, we propose a targeted poisoning attack, PoiCGAN, based on feature-label collaborative perturbation. Our method modifies the inputs of the discriminator and generator in the Conditional Generative Adversarial Network (CGAN) to influence the training process, generating an ideal poison generator. This generator not only produces specific poisoned samples but also automatically performs label flipping. Experiments across various datasets show that our method achieves an attack success rate 83.97% higher than baseline methods, with a less than 8.87% reduction in the main task's accuracy. Moreover, the poisoned samples and malicious models exhibit high stealthiness.

The expansion of generative AI and LLM services underscores the growing need for adaptive mechanisms to select an appropriate available model to respond to a user's prompts. Recent works have proposed offline and online learning formulations to identify the optimal generative AI model for an input prompt, based solely on maximizing prompt-based fidelity evaluation scores, e.g., CLIP-Score in text-to-image generation. However, such fidelity-based selection methods overlook the diversity of generated outputs, and hence, they can fail to address potential diversity shortcomings in the generated responses. In this paper, we introduce the Diversity-Aware Kernelized Upper Confidence Bound (DAK-UCB) method as a contextual bandit algorithm for the online selection of generative models with diversity considerations. The proposed DAK-UCB method incorporates both fidelity and diversity-related metrics into the selection process. We design this framework based on prompt-aware diversity score functions that decompose to a two-sample-based expectation over prompt-output pairs in the previous generation rounds. Specifically, we illustrate the application of our framework using joint kernel distance and kernel entropy measures. Our experimental results demonstrate the effectiveness of DAK-UCB in promoting diversity-aware model selection while maintaining fidelity in the generations for a sequence of prompts. The code is available at https://github.com/Donya-Jafari/DAK-UCB.

Curriculum Learning (CL) is a meta-learning paradigm that trains a model by feeding the data instances incrementally according to a schedule, which is based on difficulty progression. Defining meaningful difficulty assessment measures is crucial and most usually the main bottleneck for effective learning, while also in many cases the employed heuristics are only application-specific. In this work, we propose the Dual-Criterion Curriculum Learning (DCCL) framework that combines two views of assessing instance-wise difficulty: a loss-based criterion is complemented by a density-based criterion learned in the data representation space. Essentially, DCCL calibrates training-based evidence (loss) under the consideration that data sparseness amplifies the learning difficulty. As a testbed, we choose the time-series forecasting task. We evaluate our framework on multivariate time-series benchmarks under standard One-Pass and Baby-Steps training schedules. Empirical results show the interest of density-based and hybrid dual-criterion curricula over loss-only baselines and standard non-CL training in this setting.

Effective navigation intelligence relies on long-term memory to support both immediate generalization and sustained adaptation. However, existing approaches face a dilemma: modular systems rely on explicit mapping but lack flexibility, while Transformer-based end-to-end models are constrained by fixed context windows, limiting persistent memory across extended interactions. We introduce StateLinFormer, a linear-attention navigation model trained with a stateful memory mechanism that preserves recurrent memory states across consecutive training segments instead of reinitializing them at each batch boundary. This training paradigm effectively approximates learning on infinitely long sequences, enabling the model to achieve long-horizon memory retention. Experiments across both MAZE and ProcTHOR environments demonstrate that StateLinFormer significantly outperforms its stateless linear-attention counterpart and standard Transformer baselines with fixed context windows. Notably, as interaction length increases, persistent stateful training substantially improves context-dependent adaptation, suggesting an enhancement in the model's In-Context Learning (ICL) capabilities for navigation tasks.

We identify a critical security vulnerability in mainstream Claw personal AI agents: untrusted content encountered during heartbeat-driven background execution can silently pollute agent memory and subsequently influence user-facing behavior without the user's awareness. This vulnerability arises from an architectural design shared across the Claw ecosystem: heartbeat background execution runs in the same session as user-facing conversation, so content ingested from any external source monitored in the background (including email, message channels, news feeds, code repositories, and social platforms) can enter the same memory context used for foreground interaction, often with limited user visibility and without clear source provenance. We formalize this process as an Exposure (E) $\rightarrow$ Memory (M) $\rightarrow$ Behavior (B) pathway: misinformation encountered during heartbeat execution enters the agent's short-term session context, potentially gets written into long-term memory, and later shapes downstream user-facing behavior. We instantiate this pathway in an agent-native social setting using MissClaw, a controlled research replica of Moltbook. We find that (1) social credibility cues, especially perceived consensus, are the dominant driver of short-term behavioral influence, with misleading rates up to 61%; (2) routine memory-saving behavior can promote short-term pollution into durable long-term memory at rates up to 91%, with cross-session behavioral influence reaching 76%; (3) under naturalistic browsing with content dilution and context pruning, pollution still crosses session boundaries. Overall, prompt injection is not required: ordinary social misinformation is sufficient to silently shape agent memory and behavior under heartbeat-driven background execution.

Sentiment signals derived from sparse news are commonly used in financial analysis and technology monitoring, yet transforming raw article-level observations into reliable temporal series remains a largely unsolved engineering problem. Rather than treating this as a classification challenge, we propose to frame it as a causal signal reconstruction problem: given probabilistic sentiment outputs from a fixed classifier, recover a stable latent sentiment series that is robust to the structural pathologies of news data such as sparsity, redundancy, and classifier uncertainty. We present a modular three-stage pipeline that (i) aggregates article-level scores onto a regular temporal grid with uncertainty-aware and redundancy-aware weights, (ii) fills coverage gaps through strictly causal projection rules, and (iii) applies causal smoothing to reduce residual noise. Because ground-truth longitudinal sentiment labels are typically unavailable, we introduce a label-free evaluation framework based on signal stability diagnostics, information preservation lag proxies, and counterfactual tests for causality compliance and redundancy robustness. As a secondary external check, we evaluate the consistency of reconstructed signals against stock-price data for a multi-firm dataset of AI-related news titles (November 2024 to February 2026). The key empirical finding is a three-week lead lag pattern between reconstructed sentiment and price that persists across all tested pipeline configurations and aggregation regimes, a structural regularity more informative than any single correlation coefficient. Overall, the results support the view that stable, deployable sentiment indicators require careful reconstruction, not only better classifiers.

AscendC (Ascend C) operator optimization on Huawei Ascend neural processing units (NPUs) faces a two-fold knowledge bottleneck: unlike the CUDA ecosystem, there are few public reference implementations to learn from, and performance hinges on a coupled two-part artifact - a host-side tiling program that orchestrates data movement and a kernel program that schedules and pipelines instructions. We present AscendOptimizer, an episodic agent that bootstraps this missing expertise by turning execution into experience. On the host side, AscendOptimizer performs profiling-in-the-loop evolutionary search to discover valid and high-performing tiling and data-movement configurations directly from hardware feedback. On the kernel side, it mines transferable optimization motifs by rewinding optimized kernels - systematically de-optimizing them to synthesize instructive "bad-to-good" trajectories - and distills these motifs into a retrievable experience bank for guided rewriting. By alternating host tuning and kernel rewriting in a closed loop, AscendOptimizer steadily expands feasibility and pushes latency down. On a benchmark of 127 real AscendC operators, AscendOptimizer achieves a 1.19x geometric-mean speedup over the open-source baseline, with 49.61% of operators outperforming their references, outperforming strong agent and search baselines.

Safe reinforcement learning (RL) is a standard paradigm for safety-critical decision making. However, real-world safety constraints can be complex, subjective, and even hard to explicitly specify. Existing works on constraint inference rely on restrictive assumptions or extensive expert demonstrations, which is not realistic in many real-world applications. How to cheaply and reliably learn these constraints is the major challenge we focus on in this study. While inferring constraints from human preferences offers a data-efficient alternative, we identify the popular Bradley-Terry (BT) models fail to capture the asymmetric, heavy-tailed nature of safety costs, resulting in risk underestimation. It is still rare in the literature to understand the impacts of BT models on the downstream policy learning. To address the above knowledge gaps, we propose a novel approach namely Preference-based Constrained Reinforcement Learning (PbCRL). We introduce a novel dead zone mechanism into preference modeling and theoretically prove that it encourages heavy-tailed cost distributions, thereby achieving better constraint alignment. Additionally, we incorporate a Signal-to-Noise Ratio (SNR) loss to encourage exploration by cost variances, which is found to benefit policy learning. Further, two-stage training strategy are deployed to lower online labeling burdens while adaptively enhancing constraint satisfaction. Empirical results demonstrate that PbCRL achieves superior alignment with true safety requirements and outperforms the state-of-the-art baselines in terms of safety and reward. Our work explores a promising and effective way for constraint inference in Safe RL, which has great potential in a range of safety-critical applications.

Assuming that neither source data nor the source model is accessible, black box domain adaptation represents a highly practical yet extremely challenging setting, as transferable information is restricted to the predictions of the black box source model, which can only be queried using target samples. Existing approaches attempt to extract transferable knowledge through pseudo label refinement or by leveraging external vision language models (ViLs), but they often suffer from noisy supervision or insufficient utilization of the semantic priors provided by ViLs, which ultimately hinder adaptation performance. To overcome these limitations, we propose a dual teacher distillation with subnetwork rectification (DDSR) model that jointly exploits the specific knowledge embedded in black box source models and the general semantic information of a ViL. DDSR adaptively integrates their complementary predictions to generate reliable pseudo labels for the target domain and introduces a subnetwork driven regularization strategy to mitigate overfitting caused by noisy supervision. Furthermore, the refined target predictions iteratively enhance both the pseudo labels and ViL prompts, enabling more accurate and semantically consistent adaptation. Finally, the target model is further optimized through self training with classwise prototypes. Extensive experiments on multiple benchmark datasets validate the effectiveness of our approach, demonstrating consistent improvements over state of the art methods, including those using source data or models.

Surgical smoke severely degrades intraoperative video quality, obscuring anatomical structures and limiting surgical perception. Existing learning-based desmoking approaches rely on scarce paired supervision and deterministic restoration pipelines, making it difficult to perform exploration or reinforcement-driven refinement under real surgical conditions. We propose PhySe-RPO, a diffusion restoration framework optimized through Physics- and Semantics-Guided Relative Policy Optimization. The core idea is to transform deterministic restoration into a stochastic policy, enabling trajectory-level exploration and critic-free updates via group-relative optimization. A physics-guided reward imposes illumination and color consistency, while a visual-concept semantic reward learned from CLIP-based surgical concepts promotes smoke-free and anatomically coherent restoration. Together with a reference-free perceptual constraint, PhySe-RPO produces results that are physically consistent, semantically faithful, and clinically interpretable across synthetic and real robotic surgical datasets, providing a principled route to robust diffusion-based restoration under limited paired supervision.

Synthetic data augmentation helps language models learn new knowledge in data-constrained domains. However, naively scaling existing synthetic data methods by training on more synthetic tokens or using stronger generators yields diminishing returns below the performance of RAG. To break the RAG ceiling, we introduce Synthetic Mixed Training, which combines synthetic QAs and synthetic documents. This leverages their complementary training signals, and enables log-linear improvements as both synthetic data volume and generator strength increase. This allows the model to outperform RAG by a 2.6\% relative gain on QuaLITY, a long-document reading comprehension benchmark. In addition, we introduce Focal Rewriting, a simple technique for synthetic document generation that explicitly conditions document generation on specific questions, improving the diversity of synthetic documents and yielding a steeper log-linear scaling curve. On QuaLITY, our final recipe trains a Llama 8B model that outperforms RAG by 4.4\% relatively. Across models and benchmarks (QuaLITY, LongHealth, FinanceBench), our training enables models to beat RAG in five of six settings, outperforms by 2.6\%, and achieves a 9.1\% gain when combined with RAG.

The sample compression conjecture is: Each concept class of VC dimension d has a compression scheme of size d.In this paper, for any concept class of finite functions, we present a labeled sample compression scheme of size equals to its VC dimension d. That is, the long standing open sample compression conjecture is resolved.

GUI agents are rapidly shifting from multi-module pipelines to end-to-end, native vision-language models (VLMs) that perceive raw screenshots and directly interact with digital devices. Despite rapid progress on general GUI tasks, CAPTCHA solving remains a major challenge. On the other hand, although specialized CAPTCHA solving pipelines exist, they cannot handle general GUI tasks. To address this gap, we introduce ReCAP: a CAPTCHA-capable native GUI agent that can robustly solve modern, interactive CAPTCHA challenges, while preserving their performance as a general GUI agent. We first develop a dynamic CAPTCHA system spanning seven representative CAPTCHA types, designed to stress primitive and complementary capabilities for CAPTCHA solving (e.g., robust OCR under heavy noise and text stylization, fine-grained visual understanding, and precise control). Then, we develop an automated data collection and curation pipeline that generates large-scale CAPTCHA interaction trajectories paired with reasoning traces. As CAPTCHA solving often requires multi-step interaction and recovery from intermediate mistakes, we further leverage failed trajectories to construct self-correction data, training agents to reflect on errors and correct their actions online. Across held-out test sets, ReCAP improves CAPTCHA-solving success from roughly 30\% to 80\%, while maintaining strong performance on general GUI-agent benchmarks.

Uncertainty quantification is a key aspect in many tasks such as model selection/regularization, or quantifying prediction uncertainties to perform active learning or OOD detection. Within credal approaches that consider modeling uncertainty as probability sets, upper entropy plays a central role as an uncertainty measure. This paper is devoted to the computational aspect of upper entropies, providing an exhaustive algorithmic and complexity analysis of the problem. In particular, we show that the problem has a strongly polynomial solution, and propose many significant improvements over past algorithms proposed for 2-monotone lower probabilities and their specific cases.

Sparse Autoencoders uncover thousands of features in vision models, yet explaining these features without requiring human intervention remains an open challenge. While previous work has proposed generating correlation-based explanations based on top activating input examples, we present a fundamentally different alternative based on causal interventions. We leverage the structure of Vision-Language Models and steer individual SAE features in the vision encoder after providing an empty image. Then, we prompt the language model to explain what it ``sees'', effectively eliciting the visual concept represented by each feature. Results show that Steering offers an scalable alternative that complements traditional approaches based on input examples, serving as a new axis for automated interpretability in vision models. Moreover, the quality of explanations improves consistently with the scale of the language model, highlighting our method as a promising direction for future research. Finally, we propose Steering-informed Top-k, a hybrid approach that combines the strengths of causal interventions and input-based approaches to achieve state-of-the-art explanation quality without additional computational cost.

Inference-time scaling has emerged as an effective way to improve generative models at test time by using a verifier to score and select candidate outputs. A common choice is to employ Multimodal Large Language Models (MLLMs) as verifiers, which can improve performance but introduce substantial inference-time cost. Indeed, diffusion pipelines operate in an autoencoder latent space to reduce computation, yet MLLM verifiers still require decoding candidates to pixel space and re-encoding them into the visual embedding space, leading to redundant and costly operations. In this work, we propose Verifier on Hidden States (VHS), a verifier that operates directly on intermediate hidden representations of Diffusion Transformer (DiT) single-step generators. VHS analyzes generator features without decoding to pixel space, thereby reducing the per-candidate verification cost while improving or matching the performance of MLLM-based competitors. We show that, under tiny inference budgets with only a small number of candidates per prompt, VHS enables more efficient inference-time scaling reducing joint generation-and-verification time by 63.3%, compute FLOPs by 51% and VRAM usage by 14.5% with respect to a standard MLLM verifier, achieving a +2.7% improvement on GenEval at the same inference-time budget.

Textual graph-based retrieval-augmented generation (GraphRAG) has emerged as a powerful paradigm for enhancing large language models (LLMs) in domain-specific question answering. While existing approaches primarily focus on zero-shot GraphRAG, selecting high-quality demonstrations is crucial for improving reasoning and answer accuracy. Furthermore, recent studies have shown that retrieved subgraphs often contain irrelevant information, which can degrade reasoning performance. In this paper, we propose MixDemo, a novel GraphRAG framework enhanced with a Mixture-of-Experts (MoE) mechanism for selecting the most informative demonstrations under diverse question contexts. To further reduce noise in the retrieved subgraphs, we introduce a query-specific graph encoder that selectively attends to information most relevant to the query. Extensive experiments across multiple textual graph benchmarks show that MixDemo significantly outperforms existing methods.

This paper proposes a novel alternative to existing sim-to-real methods for training control policies with simulated experiences. Unlike prior methods that typically rely on domain randomization over a fixed finite set of parameters, the proposed approach injects state-dependent perturbations into the input joint torque during forward simulation. These perturbations are designed to simulate a broader spectrum of reality gaps than standard parameter randomization without requiring additional training. By using neural networks as flexible perturbation generators, the proposed method can represent complex, state-dependent uncertainties, such as nonlinear actuator dynamics and contact compliance, that parametric randomization cannot capture. Experimental results demonstrate that the proposed approach enables humanoid locomotion policies to achieve superior robustness against complex, unseen reality gaps in both simulation and real-world deployment.

Accurate and reliable safety metrics are paramount for functional safety verification of ASICs in automotive systems. Traditional FMEDA (Failure Modes, Effects, and Diagnostic Analysis) metrics, such as SPFM (Single Point Fault Metric) and LFM (Latent Fault Metric), depend on the precision of failure mode distribution (FMD) and diagnostic coverage (DC) estimations. This reliance can often leads to significant, unquantified uncertainties and a dependency on expert judgment, compromising the quality of the safety analysis. This paper proposes a novel approach that introduces error propagation theory into the calculation of FMEDA safety metrics. By quantifying the maximum deviation and providing confidence intervals for SPFM and LFM, our method offers a direct measure of analysis quality. Furthermore, we introduce an Error Importance Identifier (EII) to pinpoint the primary sources of uncertainty, guiding targeted improvements. This approach significantly enhances the transparency and trustworthiness of FMEDA, enabling more robust ASIC safety verification for ISO 26262 compliance, addressing a longstanding open question in the functional safety community.

Precipitation nowcasting is critical for disaster mitigation and aviation safety. However, radar-only models frequently suffer from a lack of large-scale atmospheric context, leading to performance degradation at longer lead times. While integrating meteorological variables predicted by weather foundation models offers a potential remedy, existing architectures fail to reconcile the profound representational heterogeneities between radar imagery and meteorological data. To bridge this gap, we propose PW-FouCast, a novel frequency-domain fusion framework that leverages Pangu-Weather forecasts as spectral priors within a Fourier-based backbone. Our architecture introduces three key innovations: (i) Pangu-Weather-guided Frequency Modulation to align spectral magnitudes and phases with meteorological priors; (ii) Frequency Memory to correct phase discrepancies and preserve temporal evolution; and (iii) Inverted Frequency Attention to reconstruct high-frequency details typically lost in spectral filtering. Extensive experiments on the SEVIR and MeteoNet benchmarks demonstrate that PW-FouCast achieves state-of-the-art performance, effectively extending the reliable forecast horizon while maintaining structural fidelity. Our code is available at https://github.com/Onemissed/PW-FouCast.

Hate speech online targets individuals or groups based on identity attributes and spreads rapidly, posing serious social risks. Memes, which combine images and text, have emerged as a nuanced vehicle for disseminating hate speech, often relying on cultural knowledge for interpretation. However, existing multimodal hate speech datasets suffer from coarse-grained labeling and a lack of integration with surrounding discourse, leading to imprecise and incomplete assessments. To bridge this gap, we propose an agentic annotation framework that coordinates seven specialized agents to generate hierarchical labels and rationales. Based on this framework, we construct M^3 (Multi-platform, Multi-lingual, and Multimodal Meme), a dataset of 2,455 memes collected from X, 4chan, and Weibo, featuring fine-grained hate labels and human-verified rationales. Benchmarking state-of-the-art Multimodal Large Language Models reveals that these models struggle to effectively utilize surrounding post context, which often fails to improve or even degrades detection performance. Our finding highlights the challenges these models face in reasoning over memes embedded in real-world discourse and underscores the need for a context-aware multimodal architecture. Our dataset and code are available at https://github.com/mira-ai-lab/M3.

Current language model training commonly applies multi-task Supervised Fine-Tuning (SFT) using a homogeneous compute budget across all sub-datasets. This approach is fundamentally sub-optimal: heterogeneous learning dynamics cause faster-learning tasks to overfit early while slower ones remain under-fitted. To address this, we introduce mSFT, an iterative, overfitting-aware search algorithm for multi-task data mixtures. mSFT trains the model on an active mixture, identifies and excludes the earliest overfitting sub-dataset, and reverts to that specific optimal checkpoint before continuing. Extensive evaluations demonstrate that mSFT consistently outperforms 4 baselines across 10 benchmarks and 6 base models. Further analysis confirms mSFT maintains robust gains across diverse dataset sizes, task granularities, and is insensitive to its single new hyperparameter (compute budget). Notably, at low compute budget, mSFT can improve performance while lowering training FLOPs. Ultimately, mSFT establishes a practical overfitting-aware algorithm for multi-task SFT that maximizes the potential of models across diverse data mixtures.

Long-context LLM inference is bottlenecked not by compute but by the O(n) memory bandwidth cost of scanning the KV cache at every decode step -- a wall that no amount of arithmetic scaling can break. Recent photonic accelerators have demonstrated impressive throughput for dense attention computation; however, these approaches inherit the same O(n) memory scaling as electronic attention when applied to long contexts. We observe that the real leverage point is the coarse block-selection step: a memory-bound similarity search that determines which KV blocks to fetch. We identify, for the first time, that this task is structurally matched to the photonic broadcast-and-weight paradigm -- the query fans out to all candidates via passive splitting, signatures are quasi-static (matching electro-optic MRR programming), and only rank order matters (relaxing precision to 4-6 bits). Crucially, the photonic advantage grows with context length: as N increases, the electronic scan cost rises linearly while the photonic evaluation remains O(1). We instantiate this insight in PRISM (Photonic Ranking via Inner-product Similarity with Microring weights), a thin-film lithium niobate (TFLN) similarity engine. Hardware-impaired needle-in-a-haystack evaluation on Qwen2.5-7B confirms 100% accuracy from 4K through 64K tokens at k=32, with 16x traffic reduction at 64K context. PRISM achieves a four-order-of-magnitude energy advantage over GPU baselines at practical context lengths (n >= 4K).

The Brain Tumor Reporting and Data System (BT-RADS) standardizes post-treatment MRI response assessment in patients with diffuse gliomas but requires complex integration of imaging trends, medication effects, and radiation timing. This study evaluates an end-to-end multi-agent large language model (LLM) and convolutional neural network (CNN) system for automated BT-RADS classification. A multi-agent LLM system combined with automated CNN-based tumor segmentation was retrospectively evaluated on 509 consecutive post-treatment glioma MRI examinations from a single high-volume center. An extractor agent identified clinical variables (steroid status, bevacizumab status, radiation date) from unstructured clinical notes, while a scorer agent applied BT-RADS decision logic integrating extracted variables with volumetric measurements. Expert reference standard classifications were established by an independent board-certified neuroradiologist. Of 509 examinations, 492 met inclusion criteria. The system achieved 374/492 (76.0%; 95% CI, 72.1%-79.6%) accuracy versus 283/492 (57.5%; 95% CI, 53.1%-61.8%) for initial clinical assessments (+18.5 percentage points; P<.001). Context-dependent categories showed high sensitivity (BT-1b 100%, BT-1a 92.7%, BT-3a 87.5%), while threshold-dependent categories showed moderate sensitivity (BT-3c 74.8%, BT-2 69.2%, BT-4 69.3%, BT-3b 57.1%). For BT-4, positive predictive value was 92.9%. The multi-agent LLM system achieved higher BT-RADS classification agreement with expert reference standard compared to initial clinical scoring, with high accuracy for context-dependent scores and high positive predictive value for BT-4 detection.

Multidisciplinary Software Development (MSD) requires domain experts and developers to collaborate across incompatible formalisms and separate artifact sets. Today, even with AI coding assistants like GitHub Copilot, this process remains inefficient; individual coding tasks are semi-automated, but the workflow connecting domain knowledge to implementation is not. Developers and experts still lack a shared view, resulting in repeated coordination, clarification rounds, and error-prone handoffs. We address this gap through a graph-based workflow optimization approach that progressively replaces manual coordination with LLM-powered services, enabling incremental adoption without disrupting established practices. We evaluate our approach on \texttt{spapi}, a production in-vehicle API system at Volvo Group involving 192 endpoints, 420 properties, and 776 CAN signals across six functional domains. The automated workflow achieves 93.7\% F1 score while reducing per-API development time from approximately 5 hours to under 7 minutes, saving an estimated 979 engineering hours. In production, the system received high satisfaction from both domain experts and developers, with all participants reporting full satisfaction with communication efficiency.

Large language models (LLMs) have shown impressive capabilities in code generation. However, because most LLMs are trained on public domain corpora, directly applying them to real-world software development often yields low success rates, as these scenarios frequently require domain-specific knowledge. In particular, domain-specific tasks usually demand highly specialized solutions, which are often underrepresented or entirely absent in the training data of generic LLMs. To address this challenge, we propose DomAgent, an autonomous coding agent that bridges this gap by enabling LLMs to generate domain-adapted code through structured reasoning and targeted retrieval. A core component of DomAgent is DomRetriever, a novel retrieval module that emulates how humans learn domain-specific knowledge, by combining conceptual understanding with experiential examples. It dynamically integrates top-down knowledge-graph reasoning with bottom-up case-based reasoning, enabling iterative retrieval and synthesis of structured knowledge and representative cases to ensure contextual relevance and broad task coverage. DomRetriever can operate as part of DomAgent or independently with any LLM for flexible domain adaptation. We evaluate DomAgent on an open benchmark dataset in the data science domain (DS-1000) and further apply it to real-world truck software development tasks. Experimental results show that DomAgent significantly enhances domain-specific code generation, enabling small open-source models to close much of the performance gap with large proprietary LLMs in complex, real-world applications. The code is available at: https://github.com/Wangshuaiia/DomAgent.

Advanced Persistent Threats (APTs) are stealthy, multi-stage attacks that require adaptive and timely defense. While deep reinforcement learning (DRL) enables autonomous cyber defense, its decisions are often opaque and difficult to trust in operational environments. This paper presents DeepXplain, an explainable DRL framework for stage-aware APT defense. Building on our prior DeepStage model, DeepXplain integrates provenance-based graph learning, temporal stage estimation, and a unified XAI pipeline that provides structural, temporal, and policy-level explanations. Unlike post-hoc methods, explanation signals are incorporated directly into policy optimization through evidence alignment and confidence-aware reward shaping. To the best of our knowledge, DeepXplain is the first framework to integrate explanation signals into reinforcement learning for APT defense. Experiments in a realistic enterprise testbed show improvements in stage-weighted F1-score (0.887 to 0.915) and success rate (84.7% to 89.6%), along with higher explanation confidence (0.86), improved fidelity (0.79), and more compact explanations (0.31). These results demonstrate enhanced effectiveness and trustworthiness of autonomous cyber defense.

Frontier LLM companies have repeatedly assured courts and regulators that their models do not store copies of training data. They further rely on safety alignment strategies via RLHF, system prompts, and output filters to block verbatim regurgitation of copyrighted works, and have cited the efficacy of these measures in their legal defenses against copyright infringement claims. We show that finetuning bypasses these protections: by training models to expand plot summaries into full text, a task naturally suited for commercial writing assistants, we cause GPT-4o, Gemini-2.5-Pro, and DeepSeek-V3.1 to reproduce up to 85-90% of held-out copyrighted books, with single verbatim spans exceeding 460 words, using only semantic descriptions as prompts and no actual book text. This extraction generalizes across authors: finetuning exclusively on Haruki Murakami's novels unlocks verbatim recall of copyrighted books from over 30 unrelated authors. The effect is not specific to any training author or corpus: random author pairs and public-domain finetuning data produce comparable extraction, while finetuning on synthetic text yields near-zero extraction, indicating that finetuning on individual authors' works reactivates latent memorization from pretraining. Three models from different providers memorize the same books in the same regions ($r \ge 0.90$), pointing to an industry-wide vulnerability. Our findings offer compelling evidence that model weights store copies of copyrighted works and that the security failures that manifest after finetuning on individual authors' works undermine a key premise of recent fair use rulings, where courts have conditioned favorable outcomes on the adequacy of measures preventing reproduction of protected expression.

Photovoltaic (PV) power forecasting in edge-enabled grids requires balancing forecasting accuracy, robustness under weather-driven distribution shifts, and strict latency constraints. Existing models work well under normal conditions but often struggle with rare ramp events and unexpected weather changes. Relying solely on cloud-based large models often leads to significant communication delays, which can hinder timely and efficient forecasting in practical grid environments. To address these issues, we propose a condition-adaptive cloud-edge collaborative framework *CAPE* for PV forecasting. *CAPE* consists of three main modules: a site-specific expert model for routine predictions, a lightweight edge-side model for enhanced local inference, and a cloud-based large retrieval model that provides relevant historical cases when needed. These modules are coordinated by a screening module that evaluates uncertainty, out-of-distribution risk, weather mutations, and model disagreement. Furthermore, we employ a Lyapunov-guided routing strategy to dynamically determine when to escalate inference to more powerful models under long-term system constraints. The final forecast is produced through adaptive fusion of the selected model outputs. Experiments on two real-world PV datasets demonstrate that *CAPE* achieves superior performance in terms of forecasting accuracy, robustness, routing quality, and system efficiency.

Multi-turn human-AI collaboration is fundamental to deploying interactive services such as adaptive tutoring, conversational recommendation, and professional consultation. However, optimizing these interactions via reinforcement learning is hindered by the sparsity of verifiable intermediate rewards and the high stochasticity of user responses. To address these challenges, we introduce Implicit Turn-wise Policy Optimization (ITPO). ITPO leverages an implicit process reward model to derive fine-grained, turn-wise process rewards from sparse outcome signals. Unlike volatile token-level rewards, these turn-level signals exhibit superior robustness and may utilize a normalization mechanism to further enhance training stability. We evaluate ITPO across three representative multi-turn collaborative tasks: math tutoring, document writing, and medical recommendation. Empirical results demonstrate that ITPO, when combined with PPO, GRPO, or RLOO, consistently achieves improved convergence than existing baselines. Elaborate trajectory analysis confirms that ITPO infers turn-wise preferences that are semantically aligned with human judgment. Code is publicly available at https://github.com/Graph-COM/ITPO.

Chinchilla Approach 2 is among the most widely used methods for fitting neural scaling laws. Its parabolic approximation introduces systematic biases in compute-optimal allocation estimates, even on noise-free synthetic data. Applied to published Llama 3 IsoFLOP data at open frontier compute scales, these biases imply a parameter underallocation corresponding to 6.5% of the $3.8\times10^{25}$ FLOP training budget and \$1.4M (90% CI: \$412K-\$2.9M) in unnecessary compute at 50% H100 MFU. Simulated multimodal model misallocations show even greater opportunity costs due to higher loss surface asymmetry. Three sources of this error are examined: IsoFLOP sampling grid width (Taylor approximation accuracy), uncentered IsoFLOP sampling, and loss surface asymmetry ($α\neq β$). Chinchilla Approach 3 largely eliminates these biases but is often regarded as less data-efficient, numerically unstable, prone to local minima, and harder to implement. Each concern is shown to be unfounded or addressable, especially when the partially linear structure of the objective is exploited via Variable Projection, enabling unbiased inference on all five loss surface parameters through a two-dimensional optimization that is well-conditioned, analytically differentiable, and amenable to dense, or even exhaustive, grid search. It may serve as a more convenient replacement for Approach 2 or a more scalable alternative for adaptations of Approach 3 to richer scaling law formulations. See https://github.com/Open-Athena/vpnls for details and https://openathena.ai/scaling-law-analysis for other results from this study.

Social biases reflected in language are inherently shaped by cultural norms, which vary significantly across regions and lead to diverse manifestations of stereotypes. Existing evaluations of social bias in large language models (LLMs) for non-English contexts, however, often rely on translations of English benchmarks. Such benchmarks fail to reflect local cultural norms, including those found in Japanese. For instance, Western benchmarks may overlook Japan-specific stereotypes related to hierarchical relationships, regional dialects, or traditional gender roles. To address this limitation, we introduce Japanese cUlture adversarial BiAs benchmarK Under handcrafted creation (JUBAKU), a benchmark tailored to Japanese cultural contexts. JUBAKU uses adversarial construction to expose latent biases across ten distinct cultural categories. Unlike existing benchmarks, JUBAKU features dialogue scenarios hand-crafted by native Japanese annotators, specifically designed to trigger and reveal latent social biases in Japanese LLMs. We evaluated nine Japanese LLMs on JUBAKU and three others adapted from English benchmarks. All models clearly exhibited biases on JUBAKU, performing below the random baseline of 50% with an average accuracy of 23% (ranging from 13% to 33%), despite higher accuracy on the other benchmarks. Human annotators achieved 91% accuracy in identifying unbiased responses, confirming JUBAKU's reliability and its adversarial nature to LLMs.

We obtain the phase diagrams of field theories of intertwined orders in the presence of periodic driving by an external field which preserves all symmetries. We consider both a conventional Landau theory of competing orders, and a fractionalized theory in which the order parameters are distinct composites of an underlying multi-component Higgs field. We work in the large $N$ limit and couple to a Markovian bath. The long time limits are characterized by non-zero average values, oscillations with the drive period and/or half the drive period, quasi-periodic oscillations, or chaotic behavior.

Classical thermodynamics admits a geometric formulation in which work is associated with areas enclosed by cycles in state space. Whether an analogous structure persists in driven, dissipative quantum systems remains an open question. Here we show that quasistatic work in open quantum steady states is governed by an emergent geometric curvature in control-parameter space arising from steady-state coherence. For a driven dissipative two-level system, we construct a work one-form whose curvature determines the work produced in cyclic processes. The work vanishes under strong dephasing, identifying coherence as a necessary condition for nontrivial geometry. However, its magnitude is set not by the coherence itself but by the spatial structure of the curvature: cycles enclosing comparable areas produce different work depending on their location in parameter space. Reversing the cycle orientation reverses the sign of the work, confirming its geometric origin. These results establish a geometric framework for open quantum thermodynamics and identify curvature as the organizing principle of thermodynamic response, with direct implications for driven light--matter systems in cavity quantum electrodynamics.

We investigate Floquet-driven topological phase transitions in an AB-stacked bilayer Haldane lattice with tunable intralayer hopping anisotropy. By combining interlayer hybridization, Haldane flux, and off-resonant circularly polarized light, we obtain controlled transitions among Dirac, semi-Dirac, and higher-Chern insulating phases. As the hopping anisotropy increases, the two inequivalent Dirac points move toward each other and merge at the Brillouin-zone $\mathbf{M}$ point, where a semi-Dirac dispersion emerges with linear and quadratic momentum dependence along orthogonal directions. In this regime, competition between the intrinsic Haldane mass and the Floquet-induced mass drives a sequence of sharp topological transitions with Chern numbers $C=0,\pm1,\pm2$. We further show that interlayer coupling qualitatively reshapes the Floquet band topology by inducing helicity-dependent and valley-selective band inversions at the K and K$'$ points, thereby stabilizing higher-Chern phases in the valence bands. These changes are accompanied by redistribution of the Berry curvature, bulk gap closings, and the collapse or sign reversal of quantized anomalous Hall plateaus. As the system approaches the semi-Dirac limit, the topological phase space narrows and disappears at the critical merger point, beyond which the system becomes topologically trivial even when it remains gapped. Overall, the bilayer geometry broadens the scope of Floquet topological control by enabling dynamically tunable higher-Chern phases and valley-dependent Hall responses governed by interlayer coupling and light helicity.

We investigate the impact of an induced mass term $Δ$ on the current density in graphene subjected to a space- and time-dependent periodic potential $U(x,t)$. By solving the Dirac equation and deriving both the quasi-energy spectrum and the corresponding eigenspinors, we obtain explicit analytical expressions for the current density, which exhibits a clear dependence on $Δ$. We show that $Δ$ acts as a tunable control parameter that governs the amplitude, sign, and resonance structure of Josephson-like current oscillations. For normal incidence and a purely time-periodic potential, our results reveal that the oscillations within the energy gap gradually diminish as the mass term $Δ$ increases. This suppression leads to a weakening of the Josephson-like effect typically observed in such systems. When the potential $U(x,t)$ is periodic in both space and time, the behavior becomes more complex. The current density can take either positive or negative values depending on the magnitude of the induced gap, and it generally decreases over time. As a result, the resonance phenomena--prominent at lower gap values--become progressively less significant as $Δ$ increases. These findings underscore the tunable nature of light-matter interactions and quantum transport in gapped graphene, suggesting potential applications in terahertz (THz) nanoelectronic devices and optically controlled quantum switches.

An important quantity in liquid state theory is the radial distribution function $g(r)$. It can be calculated within the framework of classical density functional theory in two very distinct ways. In the test-particle route, one fixes a single fluid particle, turning it into an external potential in which the inhomogeneous structure of the fluid is calculated by minimising the functional. The second route to $g(r)$ in density functional theory employs the Ornstein-Zernike equation and the pair direct correlation function, that can be obtained from the second functional derivatives of the excess free energy functional. Since typically an approximate excess free energy functional is employed, one generally expects that the test-particle route, which requires only one functional derivative, to be more accurate than the Ornstein-Zernike route. Here we study a two dimensional core-shoulder particle system and present results that challenge this expectation. Our results show that in this system test-particle results for $g(r)$ are not always better than results obtained via the Ornstein-Zernike route.

Advances in hybrid quantum architectures hinge on topological materials that can be synthesized with precise stoichiometric and structural control at the nanoscale. While $Bi_4Te_3$ is a promising candidate due to its dual topological phases, acting as both a strong topological insulator and a topological crystalline insulator, high-quality growth remains challenging due to a narrow stoichiometric window and high sensitivity to surface kinetics. Here, we establish a reproducible molecular beam epitaxy (MBE) process to produce stoichiometric, twin-free $Bi_4Te_3$ thin films with ultra-smooth surfaces and atomically sharp van der Waals stacks. By employing selective area epitaxy (SAE), we realize laterally confined $Bi_4Te_3$ nanostructures that exhibit a feature-dependent stoichiometric deviation. This phenomenon, which we term the selective stoichiometric shift, arises from the unequal lateral diffusion of Bi and Te adatoms, revealing a direct coupling between adatom kinetics and nanoscale compositional stability. Atomic-resolution imaging further uncovers asymmetric van der Waals gaps within the stacking sequence, identifying an intrinsic structural asymmetry between the quintuple and bilayer units. These findings provide fundamental insights into the crystallization of Bi_4Te_3$ and demonstrate a scalable route for integrating functional topological materials into next-generation superconducting hybrid quantum circuits.

Atomically-thin moiré superlattices offer an optically accessible platform for interacting bosons, where strong onsite repulsion $U_{xx}$ suppresses double occupancy and supports excitonic Mott states at unit filling. However, moiré confinement also enhances phonon- and disorder-assisted relaxation, challenging the robustness of these correlated states under dissipation. Here we show that strengthening the intersite exciton repulsion $V_{xx}$ between neighboring moiré cells offers a distinct route to stabilizing unit-filling excitonic Mott states. In H-stacked WSe2/WS2, moiré confinement endows interlayer excitons with an out-of-plane dipole and a pronounced in-plane quadrupolar charge distribution. Helicity-resolved transient photoluminescence, supported by first-principles-informed modelling, reveals that this quadrupolar geometry increases $V_{xx}$ at unit filling by at least a factor of two relative to the dipolar R-stacked excitons. Despite a slight reduction in $U_{xx}$, the enhanced $V_{xx}$ yields a long-lived, valley-polarized excitonic Mott state at unit filling that persists for ~12 ns - more than twice as long as in R-stacks - and remains robust up to ~50 K. Beyond unit filling, the same geometry supports valley-polarized doublons with fourfold longer lifetimes than in R-stacks. These results establish moiré-geometric control of intersite interactions as a route to stabilizing excitonic Mott states and doublons against dissipation in solids.

We introduce the order-separated Grassmann higher-order tensor renormalization group (OS-GHOTRG) method for QCD with staggered quarks in the strong-coupling expansion. The method allows us to determine the expansion coefficients of the partition function, from which we can obtain the strong-coupling expansions of thermodynamical observables. We use the method in two dimensions to compute the free energy, the particle-number density, and the chiral condensate as a function of the chemical potential up to third order in the inverse coupling $β$. Although near the phase transition the expansion is only a good approximation to the full theory at small $β$, we show that the range of applicability can be greatly extended by fits to judiciously chosen transition functions.

We develop a theory of polarized photoluminescence of interface excitons localized at lateral heterojunctions between transition metal dichalcogenide monolayers. We show that the circular selection rules governing interband optical transitions exactly at the band extrema are modified at finite wave vectors. The corresponding wave-vector-dependent corrections to the optical matrix elements result in a net linear polarization of excitonic photoluminescence. We identify two microscopic mechanisms responsible for linear polarization$-$trigonal warping of the electron and hole dispersions and the energy dependence of the effective masses. Their interplay controls both the magnitude and the angle of the emitted light polarization, with distinct dependences on the crystallographic orientation of the interface. Using a microscopic variational approach, we demonstrate that the degree of linear polarization can reach values exceeding 10% in realistic heterostructures. Furthermore, due to the large built-in dipole moment of interface excitons, their optical response can be tuned by an external in-plane electric field, enabling control over the strength and direction of the polarization.

Following the Bethe ansatz we determine the dynamical spectra of the one-dimensional supersymmetric t-J model. A series of fractionalized excitations are identified through two sets of Bethe numbers. Typical patterns in each set are found to yield wavefunctions containing elementary spin and charge carriers, manifested as distinct boundaries of the collective excitations in the spectra of single electron Green functions. In spin channels, gapless excitations fractionalized into two spin and a pair of postive and negative charge carriers, extending to finite energy as multiple continua. These patterns connect to the half-filling limit where only fractionalized spinons survive. In particle density channel, apart from spin-charge fractionalization, excitations involving only charge fluctuations are observed. Furthermore, nontrivial Bethe strings encoding bound state structure appear in channels of reducing or conserving magnetization, where spin and charge constituents can also be identified. These string states contribute significantly even to the low-energy sector in the limit of vanishing magnetization.

The dynamics of glassy materials slows down upon cooling, typically showing either Arrhenius or super-Arrhenius behavior. However, it was recently shown that 2D cell-based models for biological tissues can be continuously tuned between Arrhenius and sub-Arrhenius dynamics. In previous work, using the 2D Voronoi model, we proposed that such atypical dynamical behavior could be a generic feature of the broad class of mechanically under-constrained materials. Our earlier study had left two important points open: (1) many 2D systems are affected by long-wavelength fluctuations and the 2D melting scenario, and (2) the 2D Voronoi model sits exactly at the isostatic point, making it a marginal case rather than a strictly under-constrained one. Both points complicate the interpretation of our 2D Voronoi model results and their generalization to other systems; to remedy this, here we use large-scale simulations to study the glassy behavior of the 3D extension of the Voronoi model. We first show that the structural relaxation time $τ_α$ of the 3D Voronoi model can be tuned between sub-Arrhenius and Arrhenius behavior, like the 2D Voronoi model. We then establish that the four-point susceptibility, the structure factor, and the model's mechanical properties all display trends consistent with the 2D Voronoi model. These results provide strong evidence that sub-Arrhenius glassy dynamics are a generic feature of under-constrained materials across dimensions. Our work thus broadens the class of disordered materials known to have highly unusual glassy phenomenology.

We investigate topological phase transitions for the Haldane and Kane-Mele model in a lattice with $p6$ symmetry, which consists of triangles and hexagons arranged in a two-dimensional geometry. For the Haldane model, which breaks time-reversal symmetry, we calculate the Chern number using a multi-band non-Abelian Wilson loop formalism. By varying the hopping parameters in the triangles and hexagons independently, a large variety of topological phases emerge. In the presence of a next-next-nearest neighbor hopping, the phase diagram becomes even richer, with regions exhibiting high Chern numbers. Then, we consider the Kane-Mele model, for which time-reversal symmetry is preserved, and calculate the number of $π$-crossings in the Concentric Wilson Loop Spectrum (CWLS). This method is appropriate to determine the topological invariant for systems hosting time-reversal and rotational symmetry, but lacking all other symmetries. According to a classification based on $K$-theory, the CWLS invariant reveals topological properties even when more conventional invariants fail to detect them. The formalism was previously successfully applied to systems with 3- and 4-fold symmetry. Here, we surprisingly find that for the 6-fold-symmetry model investigated, the topology identified by this invariant is fragile, therefore questioning the claim that this should be the strong invariant missing in a complete classification of topological insulators.

Hybrid superconductor/semiconductor devices play a crucial role in advancing quantum science and technology by merging the properties of superconductors and semiconductors. To operate these devices at high temperature, Niobium could substitute the widespread aluminum as superconducting element. Niobium devices show the best superconducting properties when shaped by etching, but this technique is often incompatible with semiconductors and two-dimensional materials. Our work investigates the influence of oxygen diffusion on the superconducting transition of Nb nanowires fabricated by lift-off technique. To this scope, we fabricate and measure Nb devices of different width (W) and thickness (t). By using the Berezinskii-Kosterlitz-Thouless (BKT) model for charge transport, we demonstrate that our nanowires behave as two-dimensional superconductors regardless of W and t. While the normal-state transition temperature (TN) remains constant with decreasing W, the temperature of the fully superconducting state (TS) decreases. Thus, the superconducting transition width (δTC) increases as W shrinks, due to oxygen diffusion from the lithography resist occurring during deposition. These insights provide essential knowledge for optimizing Nb-based hybrid quantum devices, paving the way for operating temperatures above 2 K and contributing to the development of next-generation quantum technologies.

This paper introduces soliton_solver, an open-source GPU-accelerated software package for the simulation and real-time visualization of topological solitons in two-dimensional non-linear field theories. The software is structured around a theory-agnostic numerical core implemented using Numba CUDA kernels, while individual physical models are introduced through modular theory components. This separation enables a single computational framework to be applied across a broad class of systems, from nanoscale magnetic spin textures in condensed matter physics to cosmic strings spanning galaxies in high energy physics. The numerical backend provides finite-difference discretization, energy minimization, and GPU-resident evaluation of observables. A CUDA--PyOpenGL rendering pipeline allows direct visualization of evolving field configurations without staging full arrays through host memory. The package is distributed in Python via PyPI and supports both reproducible batch simulations and interactive exploration of metastable configurations, soliton interactions, and model-dependent initial states. We describe the software architecture, numerical workflow, and extensibility model, and we present representative example applications. We also outline how additional theories can be incorporated with minimal modification of the shared numerical infrastructure.

Decoherence in many-body quantum systems can give rise to intrinsically mixed-state phases and phase transitions beyond the pure-state paradigm. Here we study the $(2+1)$D transverse-field Ising model subject to a strongly $\mathbb{Z}_2$-symmetric decoherence channel, with a focus on strong-to-weak spontaneous symmetry breaking (SWSSB). This problem is challenging because the relevant transitions occur in the strong-decoherence regime, beyond the reach of perturbative expansions around the pure-state limit, while conventional quantum Monte Carlo (QMC) methods are hampered by the need to access nonlinear observables and by the sign problem. We overcome these difficulties by developing a QMC algorithm that efficiently evaluates nonlinear Rényi-2 correlators in higher dimensions, complemented by an effective field-theoretic approach. We show that the decohered state realizes a rich mixed-state phase diagram governed by an effective 2D Ashkin-Teller theory. This theory enables analytical predictions for the mixed-state phases and the universality classes of the phase boundaries, all of which are confirmed by large-scale QMC simulations.

Quantum magnets in the $M\mathrm{Nb_2O_6}$ and BaCo$_2$V$_2$O$_8$ families realise frustrated transverse-field Ising models whose competing ferromagnetic and antiferromagnetic couplings generate a sign problem provably intractable for quantum Monte Carlo at any system size, leaving their quantum phase boundaries numerically Inaccessible. Using a D-Wave Advantage2 quantum annealer at $L\leq27$ (729 spins), we obtain the large-$L$ critical points for this model family, measuring quantum-driven transitions at ${g_c^{\mathrm{QPU}}}\in\{0.286,\,0.210,\,0.156,\,0.093\}$ for $α\in\{1.0,\,0.7,\,0.5,\,0.3\}$, where the analytically exact classical threshold is ${g_c^{\mathrm{class}}}(α)=2α/3$. The suppression ratio $r(α)$ exhibits a sharp two-regime structure: the three quasi-1D geometries ($α\leq0.7$) are mutually consistent with a universal plateau $\bar{r}=0.450$ ($χ^2/\mathrm{dof}=1.10$, $p=0.33$), demonstrating that quantum fluctuations destroy approximately $55\%$ of the classical FM stability window independently of coupling anisotropy, while $r$ steps down to the 2D limit above the empirical crossover scale $α^*\approx0.7$. Inner Binder cumulant pairs, which converge fastest to the thermodynamic limit, resolve $r(1.0)\approx0.412$ and a step $Δr=0.038\pm0.015$ from the quasi-1D plateau. A four-point linear fit $r(α)=0.494-0.063\,α$ summarises both regimes; its $α\to0$ intercept recovers the exact 1D result of Pfeuty within 1.7 standard deviations, and its slope is a lower bound on the true crossover amplitude concentrated in $α\in[α^*,1]$. Two sequential blind predictions, confirmed at $0.2σ$ and $0.7σ$ before each measurement, validate the crossover law. All four geometries show a direct ferromagnet-to-paramagnet transition, complete quantum ergodicity ($f_{\rm uniq}=1.000$), and null valence-bond solid order.

The visit probability, quantifying whether a particle has reached a given point for the first time by a specified time, provides access to various extreme value statistics and serves as a fundamental tool for characterising active matter models. However, previous studies have largely neglected how the visit probability depends on the internal degree of freedom driving the active particle. To address this, we calculate the "state-dependent'' visit probability for a Run-and-Tumble particle, that is the probability that the particle first passes through $x$ before time $t$, keeping track of its internal state during first passage. This process may be thought of as the particle "painting'' the positions it passes through for the time in the colour of its self-propulsion state. We perform this calculation in one dimension using Doi-Peliti field theory, by extending the tracer mechanism from previous works to incorporate such "polar deposition'' and demonstrate that state-dependent visit probabilities can be elegantly captured within this field-theoretic framework. We further derive the total volume covered by a right- (or left-) moving Run-and-Tumble particle and compare our results with known expressions for Brownian motion.

Spin-phonon interactions have a dual role in emerging spin-based quantum technologies. While they can be a limitation to device performance through decoherence, they also serve as a critical resource for coherent spin control, detection, and the realization of spin-based quantum networks. However, their direct characterization remains a challenge and is usually material-dependent. Here, we introduce a technique to probe spin-phonon coupling at millikelvin temperatures and gigahertz frequencies, using high-overtone bulk acoustic wave resonators (HBARs) integrated with arbitrary crystals via visco-elastic transfer of thin-film lithium niobate transducers. By tuning the Larmor frequency of dilute spin ensembles into resonance with HBAR modes, we extract the anisotropy and strength of spin-phonon interactions from acoustic dispersion and dissipation measurements. We demonstrate this approach in calcium tungstate (CaWO4) and yttrium orthosilicate (Y2SiO5), achieving cooperativities up to 0.5 for erbium dopant ensembles. Our method enables the study of spin-phonon interactions in complex crystalline materials, with minimal fabrication constraints. These results will facilitate the design of hybrid quantum systems and the quest for ion-matrix combination with enhanced spin-phonon coupling.

Coexistence of different dynamical phases is a hallmark of glassy dynamics. This is well-studied in classical systems where the underlying theoretical framework is that of large deviation theory. The presence of a similar phase coexistence has been suggested in monitored quantum many-body systems, but the lack of suitable methods has yet prevented a systematic large deviation analysis. Here we present a tensor network framework that allows the application of large deviation theory to large quantum systems. Building on this, we locate a series of first-order dynamical phase transitions in a monitored discrete-time many-body quantum dynamics, at the level of the trajectory space. Crucially, our approach provides access not only to large-deviation statistics but also to conditioned quantum many-body states, enabling a microscopic characterization of the dynamical phases and their coexistence.

We present topological insulator (TI)-based single-electron transistors (SETs) as magnetic-field-compatible charge sensing devices that are easily integrable with TI-superconductor hybrid platforms. We observe well-resolved Coulomb diamonds in the charge-stability diagrams of our devices confirming the charge quantization and single-electron transport. In some devices, the Coulomb resonances show persistent shifts corresponding up to $\sim$ e/2 charge. An axial magnetic field further displaces these shifts to higher or lower gate voltages. We find that the axial magnetic-field dependence of the shifts is consistent with the Zeeman shift of a trap state coupled to the SET, and we reproduce the observations using numerical simulations. The resonance shifts are therefore identified as a consequence of the sensitivity of our TI-SET devices to charges in proximity. Establishing this charge sensing capability is a first step toward integrating TI-SETs as charge sensors in more complex TI-based hybrid devices, with the overarching goal of detecting and braiding Majorana zero modes.

We study the nonlinear chaotic dynamics in a system of linear oscillators coupled by social network links with an additional stratification of oscillator energies, or frequencies, and supplementary nonlinear interactions. It is argued that this system can be viewed as a model of social stratification in a society with nonlinear interacting agents with energies playing a role of wealth states of society. The Hamiltonian evolution is characterized by two integrals of motion being energy and probability norm. Above a certain chaos border the chaotic dynamics leads to dynamical thermalization with the Rayleigh-Jeans (RJ) distribution over states with given energy or wealth. At low energies, this distribution has RJ condensation of norm at low energy modes. We point out a similarity of this condensation with the wealth inequality in the world countries where about a half of population owns only a couple of percent of the total wealth. In the presence of energy pumping and absorption, the system reveals features of the Kolmogorov-Zakharov turbulence of nonlinear waves.

We investigate spin-wave modes in confined ferromagnetic resonators with spherical and cylindrical geometries across the exchange-dominated, dipole-exchange, and dipolar interaction regimes. Starting from the linearized Landau-Lifshitz-Gilbert equation, we show that the projection of the total angular momentum and mirror parity are conserved quantities in the problem of axially symmetric resonators. These symmetries provide a natural classification of spin-wave modes and explain the degeneracy of exchange modes, as well as its lifting by dipolar interactions. Numerical analysis shows that the nonlocal dipolar interaction removes the exchange degeneracy and hybridizes modes, leading to avoided crossings between modes that belong to the same symmetry sector. To describe this behavior, we develop a coupled-mode theory formulated directly in terms of dynamical magnetization, which reduces the dipole-exchange problem to a finite system of interacting modes. The resulting framework provides a unified description of spin-wave spectra in confined magnetic particles from the exchange limit to the dipolar regime.

Thermodynamic computing harnesses the relaxation dynamics of physical systems to perform matrix operations. A key limitation of such approaches is the often long thermalization time required for the system to approach equilibrium with sufficient accuracy. Here, we introduce a hybrid digital-thermodynamic algorithm that substantially accelerates relaxation through optimized initializations inspired by the Mpemba effect. In the proposed scheme, a classical digital processor efficiently computes an initialization that suppresses slow relaxation modes, after which the physical system performs the remaining computation through its intrinsic relaxation dynamics. We focus on overdamped Langevin dynamics for quadratic energy landscapes, analyzing the spectral structure of the associated Fokker-Planck operator and identifying the corresponding optimal initial covariances. This yields a predictable reduction in thermalization time, determined by the spectrum of the encoded matrix. We derive analytic expressions for the resulting speedups and numerically analyze thermodynamic implementations of matrix inversion and determinant computation as concrete examples. Our results show that optimized initialization protocols provide a simple and broadly applicable route to accelerating thermodynamic computations.

Generating stable magnetic skyrmions is essential for the practical application of skyrmion-based spintronic devices in thermally agitating environments. Recent advancements have enabled the creation of skyrmions by controlling stripe domain instability through dynamic magnetic-field control. However, deterministic skyrmion creation and effectively managing the thermal stability of skyrmions remain challenges. Here, we present a deep reinforcement learning (DRL) approach to identify advanced dynamic magnetic-field-temperature paths that create skyrmions while controlling stripe domain instability and enhancing their thermal stability. The trained DRL agent discovers an optimized field-temperature path that achieves a higher success rate for skyrmion formation in Fe3GeTe2 monolayers compared to previous fixed-temperature field sweeps. Additionally, the generated skyrmions exhibit longer lifetimes due to their isotropic shape, which tends to suppress internal excitation modes associated with skyrmion annihilation. We demonstrate that these advancements stem from the targeted minimization of the dissipated work, which ensures that the driven skyrmion states remain close to their equilibrium distributions by upper-bounding the Kullback-Leibler divergence. Our findings suggest that a DRL-powered search streamlines the identification of optimized protocols for skyrmion creation and control.

Hierarchical crack patterns that arise during the drying of thin films of colloidal dispersions or polymer solutions on a solid substrate are of interest both from a fundamental standpoint and in the context of the creation of transparent electrodes for optoelectronics. This paper analyzes the morphology of such patterns based on image processing of real-world samples. Graph theory is used to extract chains of edges and analyze the network topology. A method based on the hierarchy of connections is applied to classify cracks by generation. The limitations of existing classification approaches related to the discreteness of the time scale and the use of only a part of the entire pattern are discussed. Three approaches are used to generate artificial hierarchical networks: random uniform partitioning, recursive Voronoi partitioning, and a crack growth simulation model, each modified to reproduce the hierarchical structure. A comparison was made of the geometric characteristics (distribution of crack angles, edge lengths, cell areas, and circularity coefficient) and topological properties (distribution of the number of cell sides) of real and simulated networks. It was shown that the simulation model best reproduces the key features of real cracks, including the characteristic right angles of their connections.

We discuss the order statistics of the particle positions of a gas of $N$ identical independent particles performing Brownian motion in one dimension in a potential that asymptotically behaves like $V(x) \sim x^γ$ for $x\rightarrow+\infty$, with a positive power $γ>0$. We show that in the stationary state, the order statistics that describe how the leaders are reshuffled are universal and independent of $γ$. What depends on $γ$ is the timescale of the leaders' reshuffling, which scales as a power of the logarithm of the population size: $t \sim (\ln N)^\frac{2(1-γ)}γ τ$, where $τ$ is of order one. We derive the probability that the particle which has the $k$th largest value of $x$ at some time $t_1$ will have the $j$th largest value at time $t_2=t_1+t$ in the form of an explicit expression for the generating function for the reshuffling probabilities for all $k\ge 1$ and $j\ge 1$. The generating function, expressed in scaled time $τ$, is independent of $γ$. In particular, we show that the average percentage overlap coefficient of leader lists takes the universal, $γ$-independent form ${\rm erfc}(\sqrtτ)$ for long lists.

We present an exactly solvable model of the Mpemba effect in an overdamped Langevin system confined in a two-dimensional radially symmetric bistable potential. The potential is constructed as a piecewise quadratic-logarithmic function that is continuous and differentiable at the matching radii, enabling an exact mapping of the corresponding Fokker-Planck operator to a Schroedinger-type eigenvalue problem. The relaxation spectrum and eigenmodes are obtained analytically in each region in terms of confluent hypergeometric functions, with eigenvalues determined from matching conditions. Focusing on isotropic equilibrium initial states at inverse temperature $β_{\rm ini}$ quenched to a bath at inverse temperature $β$, we derive explicit expressions for the mode amplitudes governing long-time relaxation. We demonstrate that the coefficient of the slowest mode exhibits non-monotonic dependence on $β_{\rm ini}$ and identify a sufficient crossing condition for the Kullback-Leibler divergence in terms of the two slowest modes, if the global minimum of the potential is located far away from the origin and the second minimum exists near the origin. For corresponding parameters, we demonstrate that the Mpemba effect can be realized. Our results provide a rare example of an analytically tractable two-dimensional model exhibiting anomalous relaxation without any confining walls, extending previous one-dimensional constructions with a hard wall and clarifying the role of radial geometry in nonequilibrium relaxation phenomena.

Why kinetically stable oil droplets in water spontaneously acquire a negative charge remains one of the most vigorously debated questions in interfacial science. Here, we combine neural-network based deep potential molecular dynamics with a data-driven and information theory approach to probe the real-space electron density at an extended decane-water interface. While decane-water clusters show nearly symmetric forward and backward charge transfer (CT) and thus negligible net CT, the extended interface displays a systematic electronic asymmetry, yielding a net CT from water to the hydrocarbon phase producing an average surface charge density of $\sim0.006~e^{-}\,\mathrm{nm}^{-2}$ on the oil phase. This imbalance is accompanied by much larger intra-phase self-polarization, particularly within the hydrocarbon phase, demonstrating that collective many-body polarization dominates the interfacial electronic response. Structural analysis reveals an asymmetry between forward C--H$\cdots$O and backward O--H$\cdots$C motifs, providing a microscopic origin for a net CT from one phase to the other. Curiously, both the water O--H and decane C--H covalent bonds incur subtle contractions which originate from a response to the charge-separation layers at the interface. These features are fully consistent with the weak improper hydrogen-bonds forming at the oil-water interface that results in blue-shifts of the C-H modes.

Surface acoustic waves (SAWs) become a popular driving source in modern condensed matter physics, but most existing theories simplify them as electric fields and ignore the non-uniform Brillouin zone folding effect. We develop a semiclassical framework and reconstruct the electron dynamics by treating SAW as a quasi-periodic potential modulating electronic momentum distribution. This framework naturally explains the experimentally observed DC drag current and predicts acousto-electric Hall effect. The theory further reveals various SAW-driven transport phenomena, emerging anomalous Hall, thermal Hall, and Nernst effects within time-reversal symmetric systems. Illustrated in bilayer graphene and $\mathrm{MX_2}$ (M = Mo, W; X = S, Se, Te), the angular-dependent acousto-electric Hall effect provides an experimental probe for Berry curvature distribution.

Epitaxial heterostructures of two-dimensional van der Waals magnets and topological insulators offer a powerful platform for probing interfacial spin interactions that govern magnetic textures in low-dimensional quantum systems, while simultaneously enabling highly efficient, atomically thin spin-orbit-torque memory and computing architectures. Despite this promise, the fundamental role of these interfacial interactions in determining magnetic domain-phase stability remain largely uncharted. Here, we perform scanning transmission X-ray microscopy to image nanoscale magnetic textures in epitaxial Fe3GeTe2 Bi2Te3 heterostructures, enabled by a thermal-release-tape dry transfer process onto X-ray transparent silicon-rich nitride membranes. Under zero-field-cooled conditions, we observe robust bubble domain phases from 75 to 165 K, and across different number of folds of the multilayer Fe3GeTe2 Bi2Te3 heterostructures. This is in stark contrast with exfoliated single-crystal Fe3GeTe2 flakes, where ZFC stripe domains are observed for flakes thicker than 20 nm and no domains have been reported for thin flakes less than 15 nm. First-principles calculations and micromagnetic simulations reveal that interfacial coupling to Bi2Te3 modifies the magnetic anisotropy and introduces interfacial Dzyaloshinskii-Moriya interaction, shifting the magnetic phase space towards bubble-domain stabilization without field-cooling. Together, our results offer a new strategy for phase-selective control of magnetic domains through interfacial engineering.

Active filaments driven by tangential forces can become trapped in a spinning state when attached to a heavy head, where activity and inertia drive persistent rotation rather than directed transport. Using three-dimensional Langevin dynamics of tangentially driven bead-spring chains anchored to a common heavy head, we demonstrate that increasing the filament number $\Nf$ systematically \emph{rescues} directed transport by sterically preventing the coiled conformations that underlie spinning. The rescue is established through three independent diagnostics: (i)~the mean-square displacement recovers monotonic growth (transport rescue), (ii)~the spatial tangent autocorrelation loses its negative dip signaling helical coiling (conformational rescue), and (iii)~the tangent time autocorrelation ceases crossing zero (orientational rescue). At high bending stiffness ($\kb = 1000$), coiling is fully eliminated at a critical filament number $\Nf^* \approx 3$. At moderate stiffness ($\kb = 100$), residual coiling persists ($\min C_s \approx -0.13$) yet transport is still rescued -- demonstrating that the destruction of spinning \emph{coherence}, not coiling elimination, is the essential mechanism. The multi-filament architecture achieves up to five orders of magnitude transport enhancement. Two physically distinct rescue pathways emerge: at high stiffness, steric constraints force filaments into a coordinated bundle sustaining directed propulsion; at low stiffness, steric interactions destroy orientational coherence, producing enhanced active diffusion. These results demonstrate a purely mechanical, density-independent route to overcome inertia-induced motility arrest, with implications for synthetic microswimmer design, motor-driven filament assays, and multi-filament organization in biological systems.

We investigate the extension of pure-state symmetry protected topological phases to mixed-state regime with a strong U(1) and a weak $\mathbb{Z}_2$ symmetries in one-dimensional spin systems by the concept of quantum channels. We propose a corresponding topological phase order parameter for short-range entangled mixed states by showing that it is quantized and its distinct values can be realized by concrete spin systems with disorders, sharply signaling phase transitions among them. We also give a model-independent way to generate two distinct phases by various types of translation and reflection transformations. These results on the short-range entangled mixed states further enable us to generalize the conventional Lieb-Schultz-Mattis theorem to mixed states, even without the concept of spectral gaps and lattice Hamiltonians.

The noncollinear antiferromagnets Mn$_3$Sn/Ge emerge as promising spin-current sources with both in-plane and out-of-plane spin polarizations, thereby enabling field-free magnetization switching. However, the microscopic origin of the out-of-plane spin polarization remains under debate, specifically whether it arises from the magnetic spin Hall effect (MSHE) or the spin swapping (SSW). Here, we comparatively evaluate the spin torques in single-crystal Mn$_3$Ge/Py bilayers with different crystallographic orientations using the ferromagnetic resonance technique. The distinct angular dependences of the measured spin-torque signals provide clear evidence for the bulk MSHE, which depends on antiferromagnetic order. In addition, we identify the antiferromagnetic-order independent component originating from the interfacial SSW. The coexisting MSHE and SSW, with comparable magnitudes, give rise to the out-of-plane spin polarization. Our study disentangles the origins of spin-torque generation in noncollinear antiferromagnets, providing valuable insights for their spintronic applications.

Nonlinear Hall responses are a direct electrical probe of quantum geometry, but they are symmetry-forbidden in many pristine two-dimensional metals. We show that layer-selective magnetic proximity unlocks intrinsic linear and nonlinear Hall effects in metallic $1H-NbX_2$ ($X=\mathrm{S,Se,Te}$), where native $D_{3h}$ symmetry forces both the anomalous Hall conductivity and the Berry-curvature dipole (BCD) to vanish. Fully relativistic density-functional theory combined with Wannier interpolation reveals that an out-of-plane proximity exchange that preserves $C_3$ generates a sizable sheet anomalous Hall conductivity, $σ^{\mathrm{sheet}}_{xy} \sim 10^{-2}(e^2/h)$, while keeping the BCD exactly zero. Breaking $C_3$ by adding an in-plane exchange component (or an orthogonal two-sided exchange texture) produces a strongly tunable BCD and hence a nonlinear Hall conductivity that is odd and approximately linear in the in-plane exchange scale, reaching $|D_y|$ of order $10^{-2}$ angstrom and maximized in NbTe$_2$. These magnitudes imply a readily measurable second-harmonic Hall voltage in micron-scale Hall bars under mA ac drive. We further propose a dual-interface device in which the signs of the first- and second-harmonic Hall voltages provide two-bit readout using the same contacts.

Dynamic hysteresis, the rate-dependent lagged response of materials to external fields, underpins applications from energy-efficient transformers to gas storage systems. A fundamental yet unresolved question is how the hysteresis loop area $A$ scales with the field sweep rate $R$. Here, we reveal that a competition between the field sweep and thermal fluctuations governs a universal crossover between two scaling regimes: $A - A_0 \propto R^{1/3}$ for $R < R^*$ and $A - A_0 \propto R^{2/3}$ for $R > R^*$, where $A_0$ is the quasi-static area and the crossover rate $R^* \propto T/T_c$ depends on the temperature $T$ and the material's critical temperature $T_c$. We demonstrate these scaling laws universally across experiments of magnetic materials, simulations of Ising and metal-organic framework models, and analytical solutions of a stochastic Langevin equation. This framework not only resolves the long-standing non-universality of reported scaling exponents but also provides a direct design principle for the application of dynamic hysteresis.

The energy spectrum and wave functions of electrons in a single silicon quantum dot provide valuable insights into the capabilities and limitations of such a system in quantum information processing. Here we investigate the low-lying singlet and triplet configurations and spectra in a two-electron silicon quantum dot. To build toward a comprehensive understanding, we first examine the competition between Coulomb interaction and electron kinetic and confinement energy in the absence of valley-orbit coupling, as well as consequences of valley blockade in the presence of an ideal smooth interface. For realistic interfaces the variations in the magnitude and phase of valley-orbit coupling lead to inter-valley leakage, particularly when orbital splittings approach the valley splitting. In our study we particularly focus on the impact on the compositions of low-lying singlets and triplets. We find that for experimentally relevant parameter regimes the ground singlet and triplet states usually contain multiple configurations with significant weights as a result of a complicated competition among valley-orbit coupling, confinement potential, and Coulomb interaction. We further analyze the effects of an out-of-plane magnetic field on these the two-electron spectra. Our findings could have important implications for spin qubits in Si quantum dot in various contexts, such as qubit encoding and spin measurement.

Higher-order symmetric $X$-wave magnets consist of two groups. One includes $d$-wave, $g$-wave and $i$-wave altermagnets, while the other includes $p$-wave and $f$-wave odd-parity magnets. Recently, the possibility of $h$-wave magnets has been discussed. Motivated by this development, we systematically construct an $X$-wave magnet with $\left( N_{X}+1\right) $ nodes in three dimensions from an $X$-wave magnet with $N_{X}$ nodes in two dimensions by means of a dimensional extension, where $N_X=1,2,3,4,6$ for $X=p,d,f,g,i$, respectively. Based on this method, we predict $j$-wave magnets in three dimensions. Then, we argue how to identify each of these $X$-wave magnets experimentally. We show that the $X$-wave magnet is completely identified by measuring the nonlinear spin currents. In particular, we predict that there are no spin currents other than the fourth-order ones such as $σ_{\text{spin}}^{x^{3}y;z}$ in $h$-wave odd-parity magnets in three dimensions and the sixth-order ones such as $σ_{\text{spin}}^{x^{5}y;z}$ in $j$-wave odd-parity magnets in three dimensions. They function as spin-current diodes because the spin current exhibits unidirectional flow independent of the applied electric field.

Measuring the contact angle of a water droplet on a solid surface in air is one of the simplest and most widely used methods for evaluating surface wettability across a wide range of research fields. Wettability can also be evaluated in aqueous environments using the captive bubble method, in which an air bubble is attached to a solid surface. However, it has not yet been experimentally verified whether dynamic contact angles measured by this approach correspond to those obtained in air. In this study, we constructed an experimental system based on the captive bubble method. Dynamic contact angles were measured both in air and in water for smooth polymer surfaces, sandpaper polished surfaces, and hydrophobic surfaces with microstructures. For smooth surfaces, the dynamic contact angles obtained in air and water were nearly identical. Similar agreement was also observed for sandpaper polished surfaces, which exhibited the Wenzel state in air and the reversed gas liquid Wenzel state in water, indicating that comparable dynamic contact angles can be obtained in air and water by the captive bubble method. In contrast, microstructured PMMA surfaces that showed hydrophobic behavior in air exhibited hydrophilic behavior with very small hysteresis in water under degassed conditions. These results provide new insights into wettability in aqueous environments.

In recent years, the application of machine learning to physics has been actively explored. In this paper, we study a method for estimating the ground-state energy of quantum Hamiltonians by applying data-driven Koopman analysis within the framework of variational wave functions. Koopman theory is a framework for analyzing the nonlinear dynamics of vectors, in which the dynamics are linearized by lifting the vectors to functions defined over the original vector space. We focus on the fact that the imaginary-time Schrödinger equation, when restricted to a variational wave function, is described by a nonlinear time evolution of the variational parameter vector. We collect sample points of this nonlinear dynamics at parameter configurations where the discrepancy between the true imaginary-time dynamics and the dynamics on the variational manifold is small, and perform data-driven continuous Koopman analysis. Within our formulation, the ground-state energy is reduced to the leading eigenvalue of a differential operator known as the Koopman generator. As a concrete example, we generate samples for the four-site transverse-field Ising model and estimate the ground-state energy using extended dynamic mode decomposition (EDMD). Furthermore, as an extension of this framework, we formulate the method for the case where the variational wave function is given by a uniform matrix product state on an infinite chain. By employing computational techniques developed within the framework of the time-dependent variational principle, all the quantities required for our analysis, including error estimation, can be computed efficiently in such systems. Since our approach provides predictions for the ground-state energy even when the true ground state lies outside the variational manifold, it is expected to complement conventional variational methods.

The mechanics of single-chain stretching and rupture are central to understanding the resilience of biological polymers and designing strong and tough soft materials such as double-network gels and multi-network elastomers. In this work, we develop a statistical mechanics based model that enables one to determine the distribution of forces along the chain segments. By combining the force distribution with a tilted bond potential that captures the stretch energy stored in these bonds, we calculate the corresponding activation energy required for bond dissociation. This allows us to determine the probability of bond (and consequently chain) failure. The proposed approach is simple, direct, and readily adaptable for constructing higher-level coarse-grained descriptions of damage and fracture in polymer networks. We demonstrated this by applying the theory to three problems of practical interest: (1) toughening via sacrificial bond rupture in polymer chains, (2) toughening of double network hydrogels, and (3) incorporation of the local chain model into a 3-dimensional constitutive relation that captures damage in elastomers. The latter was implemented through the micro-sphere framework, which accounts for different chain orientations, as well as the computationally inexpensive eight chain model. The findings from this work provide a physically-based model to quantify the stretching and failure of a single chain and pave the way to the integration of local damage models into 3-dimensional networks.

We model the spatiotemporal dynamics of cellular protein concentrations near membranes composed of different lipids using a three-variable continuum model for membrane-bound protein, cytosolic protein, and the local composition of a binary lipid membrane. The model contains two globally conserved quantities: the total protein content and the average fractions of the two lipid species. It combines a conserved reaction-diffusion model for protein dynamics with a Cahn-Hilliard equation for lipid demixing. Linear stability analysis of the homogeneous steady state and direct numerical simulations show that the lipid dynamics undergoes classical phase separation, whereas the protein dynamics exhibits oscillatory phase separation for intermediate total protein contents, associated with a long-wavelength instability and traveling domains. In parameter regions where both instabilities are present, we find multiscale patterns with larger-scale traveling and rotating protein domains coexisting with smaller-scale stationary lipid domains. In this regime, traveling protein domains coexist with arrested coarsening of stationary lipid domains above a critical coupling. We further show that the main instabilities and phase diagram are well captured by an extension of a recently proposed conserved FitzHugh-Nagumo model for non-reciprocal pattern formation. The extended model consists of two non-reciprocally coupled Cahn-Hilliard equations with different interface tensions, reflecting the distinct physical properties of lipids and proteins. This also explains the observed asymmetry between static lipid patterns and traveling protein patterns.

The thermal relaxation problem between two parallel plates with the same temperature is investigated, aiming to study the behavior of the free energy of the dense gas described by the Enskog equation. Two types of Enskog equation have been used: one is the Enskog equation with the original Enskog factor, while the other is that with a modified Enskog factor proposed recently in Takata & Takahashi, Phys. Rev. E 111, 065108 (2025). The evaluated free energy is a natural extension of the thermodynamic free energy to the non-equilibrium state. It is observed that this free energy monotonically decreases in time for the modified factor version, while it is not necessarily the case for the original version. Differences are also observed in other quantities in their time evolutions, most typically in the density profile.

In this work, we experimentally investigate spin and orbital torque phenomena using the spin-torque ferromagnetic resonance (ST-FMR) technique in a series of bilayer systems composed of different normal metal (NM) materials. Permalloy (Py) and Ni were employed as ferromagnetic (FM) layers to probe the spin and orbital torque responses, respectively. For the SiO$_2$/FM/NM bilayers, we extracted the damping-like and field-like torque components, as well as the damping-like torque efficiency for each sample, and compared our results with previously reported numerical and experimental data in the literature. Additionally, we experimentally demonstrate the presence of an out-of-plane torque component, which we attribute to interfacial mechanisms and associate with a spin-orbital polarized current along the $z$-direction. This interpretation is supported by the azimuthal angular dependence of the applied magnetic field. Our results provide compelling evidence of orbital torque associated with the orbital Hall effect (OHE) in several materials, thereby broadening the prospects for magnetization switching driven by orbital torque.

Reliable organic ferroelectrics for memory applications require extreme endurance under repeated electrical switching. Here we demonstrate exceptional fatigue resistance in highly crystalline 2-methylbenzimidazole (MBI) films grown by low-temperature deposition followed by restrained crystallization (LDRC) in a simple Pt/MBI/Pt capacitor geometry. Switching kinetics analyzed using the Kolmogorov-Avrami-Ishibashi (KAI) model reveal characteristic millisecond switching times and quasi-one-dimensional domain growth associated with proton transfer along hydrogen-bond chains. Guided by these kinetics, we implemented a stringent fatigue protocol designed to maximize switching stress, involving bipolar switching at approximately 2Ec with 5 ms pulses, well beyond the characteristic switching time, for continuous operation over approximately 2 weeks. The remanent polarization exhibits only a minor wake-up (+10% within the first 10^4 cycles) and ultimately returns to approximately its initial value after 10^8 cycles, with testing limited by experimental duration rather than device failure. This robust endurance is achieved in an unengineered structure and contrasts with polymer ferroelectrics such as P(VDF-TrFE), where comparable performance typically relies on interfacial engineering. The combination of LDRC-enabled high crystallinity and localized proton-transfer switching, which introduces minimal structural perturbation during polarization reversal, enables this outstanding fatigue tolerance and highlights MBI as a simple, fluorine-free platform for durable organic ferroelectric devices.

Mixed quantum-classical models are widely used to reduce the computational cost of fully quantum simulations. However, their general applicability across different classes of problems remains an open question. Here, we address this issue for systems featuring spin-orbit coupling. In particular, we study the interaction dynamics of quantum spin-1/2 and classical orbital momentum in one-dimensional models of Rashba nanowires. We tackle this problem by resorting to a new quantum-classical Hamiltonian model that, unlike conventional approaches, retains the Heisenberg principle and captures correlation effects beyond the common Ehrenfest approach. Based on Koopman wavefunctions in classical mechanics, the new model was recently implemented numerically via a particle scheme -- the koopmon method -- which is extended here to treat spin-orbit coupling. We apply the koopmon method to study the quantum-classical dynamics of nanowire models, with and without the presence of a harmonic potential and in both Rashba-dominated (strong coupling) and Zeeman-dominated (weak coupling) regimes. Considering realistic semiconductor parameters, the results are contrasted with both fully quantum and quantum-classical Ehrenfest dynamics. In the absence of external potential, the koopmon method qualitatively reproduces the features of the fully quantum evolution for all coupling regimes. While it exhibits a slight loss in spin accuracy compared to Ehrenfest simulations, the latter fail to capture the orbital dynamics. In the presence of a harmonic potential, the koopmon scheme reproduces the full quantum results with accuracy levels that are unachievable by the Ehrenfest model in both quantum and classical sectors. We conclude by presenting a test case that exhibits the formation of cat-like states.

Morphogenesis emerges from dynamic feedback among geometry, mechanics, and chemistry; however, disentangling these contributions in living systems remains challenging. Here, we focus on the interplay between geometry and mechanics by developing a minimal in vitro model in which purified microtubules and kinesin motor clusters self-organize into a two-dimensional active nematic cortex at the surface of spherical water-in-oil droplets. The spherical geometry enforces a total topological charge of +2, here realized by four +1/2 defects whose trajectories reveal robust, self-sustained oscillations. Using full-surface reconstructions, we show that the collective dynamics of the defects lead to a periodic switching between planar and tetrahedral arrangements through alternating coiling and hemisphere-crossing phases. By tuning microtubule density, the system spans a continuum from a classic defect-dominated active nematic to a regime resembling an extensile filament confined to a curved surface, where low density is associated with increased trajectory variability and direction reversals. Geometric perturbations introduced through controlled squeezing redistribute curvature and induce the nucleation of additional defects, thereby reorganizing the entire topological landscape while preserving total charge. Together, these results show that periodic morphogenetic-like cycles, defect topology, and material organization can arise solely from the interplay of activity, density, and curvature. This reconstituted system provides a versatile platform for elucidating the coupling between mechanics and geometry underlying shape formation in active biological matter.

The search for carbon-based materials with tailored dimensionality and properties remains an important topic in materials science, particularly for applications in electronics, photonics, and nanomechanics. Among the emerging platforms in this context, graphyne (GY) represents a class of two-dimensional (2D) carbon allotropes composed of benzene rings connected by acetylenic linkages, yielding networks containing both $sp$- and $sp^2$-hybridized carbon atoms. By analogy with the transformation of $sp^2$ carbon networks such as graphene into $sp^3$-bonded diamond through interlayer covalent bonding, we construct three-dimensional (3D) GY-derived frameworks (3DGY) by covalently connecting stacked $α$-, $β$-, and $γ$-GY sheets via out-of-plane acetylene bridges. This approach converts the original $sp^2$ nodes into $sp^3$ centers while preserving the $sp$ character of the acetylenic segments, producing fully $sp$-$sp^3$ carbon networks. Structural relaxation shows that the $α$-derived framework does not converge to a stable configuration within this scheme, whereas the $β$- and $γ$-3DGY phases form stable architectures. Density functional theory (DFT) calculations, combined with ab initio molecular dynamics (AIMD) simulations, confirm the energetic, thermal, and dynamical stability of these two systems and are further used to investigate their structural, mechanical, electronic, and optical properties. Mechanical analysis reveals anisotropic elastic behavior, whereas electronic structure calculations show indirect band gaps of approximately 0.15 eV for $β$-3DGY and 1.65 eV for γ-3DGY. Optical calculations further reveal anisotropic responses, with absorption extending from the infrared to the visible. These results identify β-3DGY and γ-3DGY as new three-dimensional carbon allotropes with distinct mechanical, electronic, and optical properties.

We use nematic Multi-particle Collision Dynamics (N-MPCD) simulations to study confined nematic liquid crystals in square domains, with three distinct mean-field potentials: the classical Maier-Saupe and Marrucci-Greco models, and a more recent model due to Ilg, Karlin, and Öttinger. These potentials incorporate diverse physical features, including spatial gradients and nonlinear dependencies on the order parameter, to describe nematic ordering at mesoscopic scales. We derive coarse-grained equations from a Fokker-Planck description with tensorial closures, and analyse the emergence of order as a function of interaction strength, $U$, in two dimensions. The critical interaction strength depends on the choice of the mean-field potential. We also analytically estimate the nematic coherence length in three dimensions, to establish a rigorous correspondence between the N-MPCD parameters (the system size $R$ and $U$) and the continuum Landau-de Gennes theoretical parameters. We systematically study equilibrium and metastable configurations, including relaxation pathways to stable equilibria, on square domains, for all three mean-field potentials. Our results confirm universal equilibrium and metastable configurations for all three mean-field potentials. Our results also suggest that the N-MPCD predictions are consistent with the continuum Landau-de Gennes predictions, regardless of the choice of the underlying mean-field potential and approximations, for large $R$ and $U$. There are differences for small $R$ and for $U$ near the critical interaction strength, that need to be further explored and quantified for new-age multiscale and multiphysics theories.

Two-dimensional topological insulators feature helical edge states that are remarkably resistant to disorder, making them appeal for energy-efficient electronics and quantum information technologies. In this study, we develop a Te-rod-templated solution growth method to create Bi2Te3 nanoplates with a Corbino geometry. The resulting few-quintuple-layer hexagonal plates are single-crystalline and contain well-defined central pores. Using optimized magnetic force microscopy, we observe clear magnetic contrast at both the inner and outer edges. The signal depends strongly on tip height and oscillation amplitude, allowing us to distinguish genuine magnetic responses from electrostatic and topographic effects. By systematically varying the pore size, we find that edge contrast increases as the distance between edges decreases, suggesting stronger coupling between the inner and outer edge channels. These findings establish a geometry-controlled platform for tuning edge-localized magnetic behavior in Bi2Te3 and open a new path to explore edge interactions in two-dimensional topological insulators.

We establish an exact information-geometric inequality that remains valid regardless of the underlying dynamics, encompassing both Markovian and non-Markovian evolutions within the mixed-state domain. This inequality can be viewed as an extension of thermodynamic speed limits, which are typically formulated as inequalities. For single qubits, we show that this inequality saturates into a strict equality because the density matrix belongs to the quantum exponential family, with the Pauli matrices serving as sufficient statistics. From a practical perspective, this identity enables a non-iterative linear regression approach to continuous-time quantum process tomography, bypassing the local minima issues common in non-linear optimization. We demonstrate the efficiency of this method by estimating the Hamiltonian and dissipation parameters of the Gorini-Kossakowski-Sudarshan-Lindblad (GKSL) master equation. Numerical simulations confirm the validity of this geometric estimator and highlight the necessity of error mitigation near the pure-state boundary where the inverse metric becomes singular.

Fluxonium superconducting qubits have demonstrated long coherence times and high single- and two-qubit gate fidelities, making them a favorable building block for superconducting quantum processors. We investigate the dominant limitations to fluxonium qubit energy relaxation time $T_1$ using a set of eight planar, aluminum-on-silicon qubits. We find that a circuit-based model for capacitive dielectric loss best captures the frequency dependence of $T_1$, which we analyze within both a two-level and a six-level energy relaxation model. We convert the measured $T_1$ into an effective capacitive quality factor $Q_\mathrm{C}^{\mathrm{eff}}$ to compare qubits on equal footing, accounting for independently estimated contributions from $1/f$ flux noise and radiative loss to the control and readout circuitry. We apply this methodology to compare qubits from two fabrication processes: a baseline process and one that applies a fluorine-based wet treatment prior to Josephson junction deposition. We resolve a small improvement of (13.8 $\pm$ 8.4$)\%$ in the process mean $Q_\mathrm{C}^{\mathrm{eff}}$, indicating that the fluorine treatment may have reduced loss from the metal-substrate interface, but did not address the primary source of loss in these fluxonium qubits.

Recent work has proposed fading ergodicity as a mechanism for many-body ergodicity breaking. Here, we show that two paradigmatic random matrix ensembles -- the Rosenzweig-Porter model and the ultrametric model -- fall within the same universality class of ergodicity breaking when embedded in a many-body Hilbert space of spins-1/2. By calibrating the parameters of both models via their Thouless times, we demonstrate that the matrix elements of local observables display similar statistical properties, allowing us to identify the fractal phase of the Rosenzweig-Porter model with the fading-ergodicity regime. This correspondence is further supported through the analyses of quantum-quench dynamics of local observables, their temporal fluctuations and power spectra, and survival probabilities. Our findings reveal that local observables thermalize within the fading-ergodicity regime on timescales shorter than the Heisenberg time, thus providing a unified framework for understanding ergodicity breaking across these distinct models.

We study a simplified model of monoclonal antibodies confined in a patchy random porous medium. Antibodies are represented as Y-shaped particles composed of seven tangential hard spheres with attractive patches on the terminal beads, while the matrix consists of randomly distributed hard-sphere obstacles bearing adhesive sites. The model captures antibody behavior in crowded biological environments with strong short-range antibody-matrix attractions. The theoretical approach combines Wertheim's multidensity thermodynamic perturbation theory, the Flory-Stockmayer theory of polymerization, and scaled particle theory for fluids in porous media. We analyze thermodynamic properties, percolation thresholds, and phase behavior, and compare the selected results with new computer simulations. The interplay between antibody-antibody and antibody-matrix interactions produces a complex phase behavior, including re-entrant phase separation with a closed-loop coexistence region at higher temperatures and conventional liquid-gas separation at lower temperatures.

Although quantum annealing is usually considered as a method for locating the ground states of difficult spin-glass and optimization problems, its use in approximate optimization -- finding low- but not zero-energy states in a reasonably short amount of time -- is no less important. Here we investigate the behavior of quantum annealing at approximate optimization in the canonical mean-field spin-glass models, the spherical $p$-spin models, and find that it performs surprisingly well. Whereas it had long been assumed that infinite-range spin glasses have a unique ``threshold'' energy at which all quench and annealing dynamics become trapped until exponential timescales, recent work has shown that two-stage quenches can in fact reach states below the naive threshold in more generic situations. We demonstrate that quantum annealing is also capable of exploiting this effect to locate sub-threshold states in $O(1)$ time. Not only can it attain energies as far below the threshold as classical annealing algorithms, but it can do so significantly faster: for an annealing schedule taking time $τ$, the residual energy under quantum annealing decays as $τ^{-α}$ with an exponent up to twice as large as that of simulated annealing in the cases considered. Importantly, by deriving and numerically solving closed integro-differential equations that hold in the thermodynamic limit, our results are free from finite-size effects and hold for annealing times that are unambiguously independent of system size.

The central limit theorem provides the theoretical foundation for the universality of the normal distribution: under broad conditions, the asymptotic distribution of a sum of independent random variables approaches a Gaussian. Yet, physical systems described by positive random variable -- from earthquakes to microbial growth to epidemic spreading -- consistently exhibit gamma rather than Gaussian statistics -- what leads to field-specific mechanistic explanations that are non robust to small changes in the model details. We show that gamma distributions emerge naturally from large deviation theory when Padé approximants replace polynomial expansions of the derivative of the scaled cumulant generating function, respecting positivity constraints that the central limit theorem violates. Gamma universality thus emerges as the constrained analog of Gaussian universality, providing a mechanism-free explanation for its pervasive appearance across different disciplines.

The Landau quantization space is based on the respective motion of the electron and hole in a magnetic field and can provide a new route to understand the bound exciton behaviors observed in the experiments. In this paper, we study the two-dimensional exciton properties of monolayer WSe$_2$ in both the real space and Landau quantization space. Focusing on the excitons of zero center-of-mass momentum, we calculate its energy spectrum in both spaces, with the results agreeing well with each other. We then obtain the diamagnetic coefficients and root-mean-square radius, which are consistent with the available $s$ state data in the experiment. More importantly, in the exciton state $nl$, we find that the dominant electron-hole pair component may shift with the magnetic field and the Coulomb interactions, and reveal that the magnetic field will drive the dominant component to be the free electron-hole pair $\{n_e=n+l-1,n_h=n-1\}$, whereas the Coulomb interactions drives it to be the pair of the lower index.

Microswimmers in suspension exhibit collective swimming behaviour, forming various self-organised structures including ordered, aggregated, and turbulent-like structures. When mixed with passive particles phase-separation is known to occur, but due to the difficulty of accurately handling many-body hydrodynamic interactions, the formation of self-organised structures in mixed suspensions has remained unexplored so far. In this study, we investigate the dynamics of mixed dense suspensions of spherical bottom-heavy squirmers and obstacle spheres using Stokesian dynamics in three dimensions, taking hydrodynamic interactions into account. The results show that without an external orientating mechanism the formation of orientational order is in general disturbed by the presence of passive spheres. An initially phase-separated state is metastable for neutral or puller squirmers at high packing densities. When the squirmers are bottom-heavy, phase-separation can occur dynamically in some cases, notably a fibrillar kind of separation for neutral squirmers and pullers at medium densities. We also observed a novel form of lamellar phase-separation for pullers at high densities with strong bottom-heaviness, with a sandwich-like structure consisting of a layer of passive particles pushed by a layer of swimmers, followed by a gap. These results indicate that microstructure and particle transport undergo significant changes depending on the type of swimmer, highlighting the importance of hydrodynamic interactions. These insights allow for a deeper understanding of the behaviour of active particles in complex fluids and to control them using external torques.

Non-equilibrium dynamics of strongly and rapidly driven quantum many-body systems is poorly understood beyond periodic driving, where heating is exponentially slow in the drive frequency (Floquet Prethermalization). In contrast, non-periodic drives were found to exhibit widely different heating scalings with no unifying principle. This work identifies a resonance-suppression principle governing slow heating up to a prethermal lifetime $τ_*$: When the drive's spectral arithmetic structure restricts multiphoton resonances, $τ_*$ is controlled by low-frequency spectral suppression. The principle distinguishes (i) Single-photon suppression, quantified by a low-frequency suppression law $f(Ω)$ for the drive's Fourier Transform weight near $Ω=0$, from (ii) Multi-photon suppression, where nested commutators remain controlled if exceptional arithmetic structure satisfies a subadditive property. Remarkably, if multi-photon suppression holds, $τ_*$ scaling with drive speed $λ$ is governed by $f(Ω)$. This law of $τ_*$ is found through a small-divisor mechanism in this work's iterative rotating frame scheme. Multi-photon suppression breakdown separates $λ$-scaling of $τ_*$ in linear response and non-perturbative theory, shown by a case study of Quasi-Floquet driving. The principle is applied to (i) Resolve inconsistencies in literature on non-periodic driving, and (ii) Provide design principles for engineering prethermal phases of matter in programmable quantum simulators, exemplified by new non-periodic `Factorial' drives with tunable $τ_*$.

We identify a geometric principle governing the location of Hopf and Bogdanov--Takens bifurcations in planar predator--prey systems. The prey coordinate of any coexistence equilibrium undergoing such a bifurcation lies between consecutive critical points of the prey nullcline. The mechanism is algebraic. At critical points of the nullcline, the vanishing of its derivative induces constraints on the Jacobian that prevent the spectral conditions required for bifurcation from being satisfied. We refer to this phenomenon as \emph{spectral rigidity}. The principle is established for three model families and one discrete counterpart with qualitatively different nullcline geometries: a quadratic case (Bazykin model), a cubic case (Holling type~IV with harvesting), and a rational case (Crowley--Martin functional response). In each case, the localization follows from explicit parametric characterizations and symbolic reduction. The analysis extends to discrete-time systems. For a map obtained by forward Euler discretization of the Crowley--Martin model, the Neimark--Sacker bifurcation occurs on the descending branch of the nullcline, providing a continuous--discrete duality governed by the same mechanism. We conjecture that this localization holds for general smooth prey nullclines, with critical points acting as spectral barriers that organise the bifurcation structure.

Vaccination campaigns play a pivotal role in controlling infectious diseases. Their success, however, depends not only on vaccine efficacy and availability but also significantly on public opinion and the willingness of individuals to vaccinate. This paper investigates a coupled opinion-epidemic model on heterogeneous networks, where individual opinions influence vaccination probability, and opinions themselves evolve through a combination of peer interaction and local risk perception derived from observed infection rates. Embedding the coupled dynamics in scale-free networks, particularly barabasi-Albert structures, allows us to examine the role of network heterogeneity beyond homogeneous-mixing assumptions. Using Monte Carlo simulations and a semi-analytical microscopic Markov-chain approach, we derive and numerically validate analytical expressions for the critical infection threshold and stable vaccinated population where risk perception dominated peer influence. Our results show that stronger local risk perception enhances pro-vaccination opinions and suppresses infection, while dominant peer influence can increase long-term infection levels. These findings underscore the importance of accounting for social behavior and network structure when designing effective vaccination and epidemic control strategies.

Robustness of higher-order networks is often quantified by the instantaneous smallest positive eigenvalue of the Hodge $1$-Laplacian under simplex deletion. We show that this observable is generically ill-defined: along a deletion trajectory, eigenvalue branches can switch, so the quantity being monitored may correspond to different nonharmonic modes at different steps. The primary issue is therefore definitional rather than algorithmic. We resolve it by fixing the first nonharmonic branch of the intact complex and following that same branch throughout the damage process, which defines a branch-consistent functional robustness. Triangle sensitivities then follow directly from first-order perturbation theory, making the resulting mode-sensitive deletion protocol a consequence of the observable itself rather than an independent heuristic. Across synthetic and empirical clique complexes, removing only a small fraction of triangles is sufficient to drive the tracked mode to collapse, while graph-level observables remain unchanged because the $1$-skeleton is exactly preserved. The same framework also reveals bridge-like localization of functionally critical simplices and provides a compact predictor of dynamical timescales.

It is shown that nonlocal coupling provides for controlling the collective noise-induced dynamics of non-excitable oscillators in the regime of coherence resonance. This effect is demonstrated by means of numerical simulation on an example of coupled generalized Van der Pol oscillators near the saddle-node bifurcation of limit cycles. In particular, it has been established that increasing the coupling radius allows to enhance the effect of coherence resonance which is reflected in the evolution of the dependence of the correlation time on the noise intensity and the power spectrum transformations. Nonlocal coupling is considered as an intermediate option between local and global coupling topologies which are also discussed as a tool for controlling coherence resonance.

In several natural and engineering systems, changes in control parameters can trigger bifurcations that lead to sustained or growing periodic oscillations, indicating the onset of oscillatory instabilities. Such emergent behaviour often results from positive feedback between interacting subsystems, resulting in large-amplitude oscillations that can be detrimental. Several precursors are available to provide early warning of an impending oscillatory instability. In reality, practical systems may exhibit different sequences of bifurcations, including a primary bifurcation to an oscillatory state that may be either continuous or abrupt, followed by an abrupt secondary bifurcation, and further transitions beyond the secondary bifurcation. Existing precursors for oscillatory instabilities typically forewarn the onset of the primary bifurcation to an oscillatory state and tend to saturate once the system enters the oscillatory regime. Notably, primary bifurcations often involve lower amplitudes compared to the more severe states after secondary bifurcation. In this study, we propose a methodology based on spectral visibility graphs to get forewarning for both primary and secondary bifurcations. The method inherently captures the evolution of the harmonic content of the signal relative to other frequency components. The approach employs tuning of a single sensitivity parameter to detect different sequences of bifurcation. We demonstrate the usefulness of our method for thermo-acoustic and aero-acoustic instabilities in multiple engineering systems involving turbulent flow and reactions. Our methodology can help systems prepare in advance or even avoid undesirable transitions. Tuning the sensitivity parameter allows adaptive, risk-based warnings, ensuring high performance without tipping into undesirable regimes.

Spatial patterning and synchronization are pervasive features of plankton communities, yet the mechanisms that allow such patterns to persist coherently under environmental noise remain unresolved. In vertically structured aquatic ecosystems, plankton populations are often organized into distinct layers, raising the question of how interactions between layers shape both spatial self-organization and robustness. Here, we develop a spatiotemporal ecosystem model of a two-layer plankton community to examine the role of passive diffusive coupling under stochastic environmental fluctuations. We show that interlayer diffusion induces a sharp transition from independent, layer-specific Turing patterns to fully synchronized spatial patterns once the coupling strength exceeds a critical threshold. Importantly, the same coupling mechanism markedly enhances the stability of spatial patterns against environmental noise, extending their persistence far beyond that of non-coupled layers. Moreover, we uncover a trophic hierarchy in noise sensitivity, with zooplankton exhibiting substantially greater vulnerability than phytoplankton. Together, these results identify passive diffusive coupling as a unifying mechanism that simultaneously promotes spatial synchronization and robustness, providing a mechanistic explanation for the persistence of coherent plankton patterns in fluctuating aquatic environments.

Positive and negative flows of the Chen-Lee-Liu model and its various reductions, including Burgers hierarchy, are formulated within the framework of Riemann-Hilbert-Birkhoff decomposition with the constant grade two generator. Two classes of vacua, namely zero vacuum and constant non-zero vacuum can be realized within a centerless Heisenberg algebra. The tau functions for soliton solutions are obtained by a dressing method and vertex operators are constructed for both types of vacua. We are able to select and classify the soliton solutions in terms of the type of vertices involved. A judicious choice of vertices yields in a closed form a particular set of multi soliton solutions for the Burgers hierarchy. We develop and analyze a class of gauge-Bäcklund transformations that generate further multi soliton solutions from those obtained by dressing method by letting them interact with various integrable defects.

We solve the orientation recovery of a tumbling protein in the gas phase from single-event measurements of the spatial positions of its ions after an X-ray laser induced explosion. We simulate diffracted X-ray signal and ion dynamics under experimental conditions and compare our method to conventional orientation recovery in single-particle imaging with X-ray free-electron lasers using only diffraction data. We reconstruct 3D diffraction intensities using orientations recovered from the ion signatures and retrieve the electron density with established phase-retrieval algorithms. We test our orientation recovery procedure on 56 proteins ranging from 14 to 52 kDa (1800 to 6500 atoms), achieving roughly an angular error of around 5°. The resulting 3D electron-density reconstructions are compared to ground-truth volumes simulated at the same nominal resolution, and achieve the resolution at the edge of the detector in conditions similar to current single-particle imaging setups. We investigate the reconstruction quality and demonstrate that ion data can be used for reliable orientation recovery of particles in single-particle imaging, achieving orientation on par or better than currently used recovery techniques. This work shows the potential of ion detection for retrieving additional information from the sample fragmentation, and boost single particle imaging with X-ray lasers in the cases where the diffraction signal is a limiting factor.

The world's power systems are undergoing a rapid transformation, shifting away from carbon-intensive power generation to renewable sources. As a result, electricity is being transported over ever longer distances, while the intrinsic system inertia provided by thermal power plants decreases. Together, these developments raise the probability of cascading line failures and reduce the stability of the system after a system split. In this article, we assess the risk of cascading failures and system splits in the European power grid for different carbon reduction scenarios. We analyze the most likely and most dangerous splits, and identify critical transmission infrastructures and we discuss potential countermeasures that can address the problem of cascades. Our results show that while the risks of splits causing power failures rises with decarbonization, it can be mitigated cost efficiently.

The controlled propagation of spin textures at bifurcations is a critical challenge for racetrack-based logic devices. Here, we investigate the effect of longitudinal and transverse magnetic fields on the propagation of magnetic antivortices at bifurcations within the stripe domain pattern of a reconfigurable NdCo/NiFe racetrack in order to control the preferred antivortex trajectory. Magnetic Transmission X-ray Microscopy experiments were employed to correlate the observed propagation path with the local magnetic configuration. We demonstrate that Zeeman coupling to the magnetization components at the bifurcation core enables switching of the preferred propagation branch using low-amplitude transverse magnetic fields, without modifying the global stripe domain configuration that defines the guiding racetrack landscape. In-plane magnetic anisotropy provides an additional mechanism to break the symmetry between the upper and lower bifurcation branches by tuning the relative orientation between the stripe domain pattern and the longitudinal magnetic fields.

Optical turbulence, driven by fluctuations of the atmospheric refractive index, poses a significant challenge to ground-based optical systems, as it distorts the propagation of light. This degradation affects both astronomical observations and free-space optical communications. While adaptive optics systems correct turbulence effects in real-time, their reactive nature limits their effectiveness under rapidly changing conditions, underscoring the need for predictive solutions. In this study, we address the problem of short-term turbulence forecasting by leveraging machine learning models to predict the atmospheric seeing parameter up to two hours in advance. We compare statistical and deep learning approaches, with a particular focus on probabilistic models that not only produce accurate forecasts but also quantify predictive uncertainty, crucial for robust decision-making in dynamic environments. Our evaluation includes Gaussian processes (GP) for statistical modeling, recurrent neural networks (RNNs) and long short-term memory networks (LSTMs) as deterministic baselines, and our novel implementation of a normalizing flow for time series (FloTS) as a flexible probabilistic deep learning method. All models are trained exclusively on historical seeing data, allowing for a fair performance comparison. We show that FloTS achieves the best overall balance between predictive accuracy and well-calibrated uncertainty.

Raman phonons arise from the inelastic scattering of light and represent quantized molecular motions that mediate a wide range of spectroscopic and nonlinear optical phenomena. In this work, we clarify the physical role of Raman phonons within a previously-developed time-domain framework based on the Raman-induced index modulation, and show that phonons correspond to the oscillatory component of the Raman-induced index modulation. The analysis further reveals a linear phonon-mediated interaction embedded within Raman scattering, in which optical fields couple through wave-vector matching with existing phonons. This mechanism underlies what has long been described as coherent Stokes and anti-Stokes scattering, as well as molecular modulation. Building on this insight, we introduce a phonon-controlled approach that enables efficient conversion into a selected Stokes order by tuning the wave-vector-matching relation between the driven phonons and the targeted Raman process. These results provide a clearer physical interpretation of Raman phonons and its corresponding Raman dynamics and offer new strategies for controlling Raman interactions.

Leadership in social groups is often a dynamic characteristic that emerges from interactions and opinion exchange. Empirical evidence suggests that individuals with strong opinions tend to gain influence, at the same time maintaining alignment with the social context is crucial for sustained leadership. Motivated by the social psychology literature that supports these empirical observations, we propose a novel dynamical system in which opinions and leadership co-evolve within a social network. Our model extends the Friedkin-Johnsen framework by making susceptibility to peer influence time-dependent, turning it into the leadership variable. Leadership strengthens when an agent holds strong yet socially aligned opinions, and declines when such alignment is lost, capturing the trade-off between conviction and social acceptance. After illustrating the emergent behavior of this complex system, we formally analyze the coupled dynamics, establishing sufficient conditions for convergence to a non-trivial equilibrium, and examining two time-scale separation regimes reflecting scenarios where opinion and leadership evolve at different speeds.

Symmetries are ubiquitous in physics and play a pivotal role in light-matter interactions, where they determine the selection rules governing allowed atomic transitions and define the associated conserved quantities. For the up-conversion process of high harmonic generation, the symmetries of the driving field determine the allowed frequencies and the polarization properties of the resulting harmonics. As a consequence, it is possible to establish classical selection rules when the process is driven by coherent radiation. In this work, we show that fluctuation-induced symmetry breaking in the driving field leads to the appearance of otherwise forbidden harmonics. This is achieved by considering bicircular quantum light, and demonstrate that the enhanced quantum fluctuations due to squeezing in the driving field break the classical selection rules. To this end, we develop a quantum optical description of the dynamical symmetries in the process of high harmonic generation, revealing corrections to the classical selection rules. Moreover, we show that the new harmonics show squeezing-like signatures in their photon statistics, allowing them to be clearly distinguished from classical thermal fluctuations.

We present the first systematic comparison of three TiO2 metamaterial solid immersion lens geometries - sub-hemispherical, super-hemispherical, and full-spherical - for label-free super-resolution imaging. Using SEM, we characterised both the cap profiles and the nanoparticle-fluid immersion at the lens-sample interface, revealing that super-hemispherical lenses achieve the deepest immersion and closest contact with sample features. Imaging experiments under wide-field and laser confocal microscopes show that this enhanced immersion drives superior resolution and contrast. In addition, we introduce a deep learning approach based on a SinCUT image translation model to establish a cross-modal mapping between SEM morphology and optical imaging response, enabling virtual optical predictions and providing a first step toward a digital twin representation of mSIL imaging behaviour. Electromagnetic simulations further confirm a direct correlation between immersion depth and far-field main lobe intensity. Our findings demonstrate that careful control of lens shape and nanoparticle-fluid penetration, together with data-driven modelling, is essential to maximise super-resolution performance in TiO2 mSILs.

Solidification governs the microstructure and, therefore, the mechanical response of metal components, yet the atomistic details of nucleation and defect formation are often difficult to determine experimentally. Molecular dynamics can bridge this gap, but only if the interatomic model is both accurate and computationally efficient. Here, we develop a Machine Learning Potential (MLP) for aluminum and demonstrate its near ab initio fidelity when trained with the sequential-refinement workflow that fine-tunes the model on low-energy structures. The favorable scaling of the model enables nanosecond simulations involving millions of atoms, thereby overcoming finite-size effects in simulations of polycrystalline solidification and subsequent mechanical testing. Comparison with classical potentials and recent MLP models, including a general-purpose model, shows that inaccuracies in stacking-fault energetics and diffusion can lead to qualitatively incorrect solidified grain structures and post-solidification mechanical behavior. Since our framework is based on an equivariant graph neural network, it allows for straightforward extensions to multi-component systems, providing valuable guidance for the future design and fine-tuning of both specialized and universal MLPs in computational mechanics simulations.

Spontaneous parametric down-conversion (SPDC) in subwavelength nonlinear nanostructures is emerging as a promising source of quantum light, owing to its intrinsic multifunctionality and ability to generate versatile and complex quantum states. Despite this growing interest, the physical mechanisms governing photon-pair generation in nanostructures remain only partially understood. In particular, experimental investigations of key emission properties in individual resonators, such as spatial directionality and spectral characteristics, are still lacking, and predictive theoretical frameworks with direct experimental validation have not yet been established. Here we measure, for the first time, the spatial and spectral properties of photon pairs generated via SPDC in a lithium-niobate bullseye nanostructured resonator. Both spatial and spectral properties show a resonant behavior, which we describe within an extended quasi-normal-mode theoretical framework. This comparison with the theory is enabled by photon-pair count rates reaching up to 0.45 Hz/mW, to our knowledge, the highest reported to date for a nanostructured resonator. Our results provide new physical insight into SPDC in nanostructures and represent an important step toward predictive design strategies for efficient nanoscale sources of quantum light.

Anisotropic two-dimensional diffraction signals contain more information than the conventional isotropic signals for both gas phase ultrafast electron and X-ray diffraction experiments and are common in typical time-resolved diffraction experiments due to the use of linearly polarized lasers to excite the sample that imprints spatial anisotropy on the molecules. We report an iterative algorithm to restore the missing data at low scattering angles in a two-dimensional diffraction signal, which is essential to obtain real-space representation. The iterative algorithm transforms two-dimensional signals back and forth between the momentum transfer domain and the real space domain through Fourier and Abel transforms and apply real space constraints to retrieve missing signal at low scattering angles. The algorithm only requires an approximate a-priori knowledge of the shortest and longest internuclear distances in the molecule. We demonstrated successful retrieval of the missing signal in simulated patterns and in experimentally measured diffraction patterns from laser-induced alignment of trifluoroiodomethane molecules.

Control over crystallographic orientation in metal-organic framework (MOF) thin films is essential, as many of their functional properties critically depend on exact alignment along a defined crystallographic direction. Spin-assisted layer-by-layer liquid-phase epitaxy (LbL-LPE) offers significant advantages over conventional synthesis approaches, including reduced chemical consumption, shorter processing times, and operation under ambient conditions. In this work, this LbL-LPE spin-coating is established as a robust, high-throughput fabrication protocol suitable for application-ready materials. The flexible pillar-layered framework Zn2BDC2DABCO serves as a proof-of-concept framework for the development of an automated spin-assisted LbL-LPE workflow enabling reproducible fabrication of homogeneous and highly oriented MOF thin films with integrated monitoring of critical processing steps. The protocol incorporates correlative characterization combining grazing-incidence wide-angle X-ray scattering (GIWAXS), infrared and UV-Vis spectroscopy, scanning electron microscopy (SEM), and time-of-flight secondary ion mass spectrometry (ToF-SIMS) to ensure control over surface chemistry, reactant delivery, and film growth. Determination of the degree of orientation and the Hermans orientation parameters provides a key quality metrics for assessing crystal alignment and reproducibility. The automated experimental workflow significantly accelerates the fabrication of thin films whose properties depend on crystal orientation, providing processing optimizations and control that can be readily extended to increasingly complex MOF architectures.

We study two-dimensional periodic metamaterials in which idealised cylindrical inclusions are modelled by boundary conditions. In the scalar time-harmonic setting, the background field satisfies the Helmholtz equation, and high-contrast inclusion limits reduce to Dirichlet or Neumann conditions, with direct analogues in dielectric and acoustic media. By switching the condition assigned to selected inclusions, we break point-group symmetries of the primitive cell and thereby lift symmetry-induced degeneracies in the Floquet--Bloch spectrum of hexagonal and square lattices, opening valley-type band gaps with Berry curvature localised near opposite valleys. To analyse infinite and finite structures within a unified framework, we derive matched-asymptotic point-scatterer approximations for mixed Dirichlet--Neumann arrays. For doubly periodic systems, this yields a finite-dimensional generalised eigenvalue problem for the Floquet--Bloch spectrum; for finite arrays, it yields a generalised Foldy multiple-scattering system. In both hexagonal and square lattices, geometrically identical crystals can realise distinct valley-Hall phases solely through boundary-condition assignment while retaining an overlapping bulk gap. Spatially varying this assignment therefore creates and relocates internal interfaces without altering the underlying geometry, enabling the associated valley-Hall interfacial modes to be repositioned within the same crystal.

Colloidal perovskite nanocrystals (NCs) are a well-proven platform for growing anisotropic structures. Nanowires (NWs) exhibiting a quantum confinement phenomenon and microwires (MWs), which enable lasing, are of particular interest for optoelectronic devices. Synthesis of the latter is challenging. Herein, we report a straightforward access to high-quality CsPbBr3 MW lasers. We utilize a diphenyl ether (DPE) solvent for the hot-injection synthesis. DPE coordinates strongly to Pb2+ and allows to reduce an excess of oleic acid/oleylamine ligand pair well established for PbBr2 dissolution and inhibition of as-formed NCs regrowth. Therefore, a rapid injection of Cs-oleate into the PbBr2-containing solution yields lead-depleted Cs4PbBr6 NCs which slowly release perovskite precursors and produce CsPbBr3 counterparts. The latter transform into NWs through an oriented-attachment mechanism, which in turn evolve into laser MWs. To demonstrate spectrally tunable lasing in MWs we employ YCl3 for ion exchange in perovskite lattice. Resultant CsPb(Cl,Br)3 MWs show high-Q coherent emission in the 485-540 nm range. To highlight the potential of synthesized MWs for integrated optoelectronics, we assemble a device comprising a CsPb(Cl,Br)3 MW laser coupled to MoO3 lossless nanowaveguide, which delivers coherent light to a CsPbBr3 MW photodetector. The device exhibits a nonlinear optoelectronic response applicable for on-chip neuromorphic computing.

Applications for ultra-stable lasers outside controlled laboratory environments require compact and robust optical resonators with reduced sensitivity to temperature fluctuations. The low thermal expansion coefficient (CTE) and the high stiffness make cordierite-based ceramics, such as NEXCERA, attractive for vibration insensitive room-temperature resonators. We revisit the effective CTE of resonators with spacers and mirrors made of different materials and use finite element simulations to analyze the impact of a CTE mismatch in a cordierite-based resonator with mirrors made of ultra-low expansion (ULE) glass or fused silica (FS). This enabled us to determine the CTE of a cordierite spacer from the measured effective CTE of a resonator. We confirm a six-fold larger CTE slope of cordierite around the zero-crossing temperature than in ULE glass. The steep CTE slope, in combination with the large stiffness, makes cordierite-based resonators far less sensitive to CTE mismatch with FS mirrors, thereby eliminating the need for additional compensation rings. We further consider the so far neglected case, where the CTE of the spacer is larger than that of the mirror, and propose resonator designs in which the thermal length change of the spacer is fully or partially compensated by the deflection of the mirrors. This results in a cordierite-based resonator with ULE mirrors whose effective CTE can be close to zero over a temperature range of several tens of Kelvin. We are extending our concept to resonators based on crystalline materials with high stiffness and low isothermal length change, such as silicon, enabling compact and robust room-temperature resonators for terrestrial and space-born applications.

Heatwaves are intensifying as a major climate extreme and have emerged as a growing public health threat in rapidly urbanizing regions such as India. In this study, we integrate long-term heat-related mortality records (1970-2023) with bias-corrected CMIP6 climate projections to quantify future heatwave-related mortality across 67 Indian cities under intermediate (SSP2-4.5) and high-emission (SSP5-8.5) scenarios. A time-series forecasting framework was applied using summer mean temperature as the primary climate driver to project mortality trajectories through the end of the 21st century. Results indicate a strong and sustained increase in heat-related mortality under both scenarios, with multi-fold amplification under SSP5-8.5 relative to SSP2-4.5, reflecting the high sensitivity of health outcomes to emission pathways. Spatial analysis reveals increasing regional divergence under high-emission conditions, with urban regions in the Deccan Plateau, western India, and parts of eastern and northeastern India exhibiting disproportionately higher mortality growth. Multidimensional scaling further highlights emerging clustering of state-level mortality behavior under extreme warming, indicating structurally different regional responses to future heat stress. In contrast, the intermediate mitigation pathway produces more moderate and spatially uniform mortality trends. These findings demonstrate that climate mitigation can substantially reduce both the magnitude and inequality of future urban heat-health burdens. By linking updated climate projections with long-term mortality data at national and sub-national scales, this study provides policy-relevant evidence to support heat adaptation planning and climate-resilient urban development in one of the world's most heat-vulnerable regions.

Multi-Conjugate Adaptive Optics (MCAO) is essential for increasing the corrected Field-of-View (FoV) in astronomical imaging and potentially for free-space optical communications, particularly for small-aperture, transportable systems. We demonstrate the viability and performance of a Refractive-MCAO system utilizing a novel multi-actuator Deformable Lens (DL) as the wavefront correction element. Unlike conventional Deformable Mirrors (DMs), the transmissive nature of the DL simplifies the optical train, making it ideal for compact setups. Using a Shack-Hartmann Wavefront Sensor (SH-WFS) in conjunction with two DLs conjugated to different atmospheric layers, we achieved an extension of the isoplanatic patch up to three times the uncorrected atmospheric isoplanatic angle under moderate turbulence D/r0 = 2. We tested the MCAO system in a setup that emulates a free space optical communication for compact transportable system. In a double-channel, single-mode fiber coupling experiment we demonstrated the efficiency of this method.

Wide-field mid-infrared photothermal (MIP) imaging offers rapid labelfree chemical contrast for biomedical and polymer analysis. However, its field of view (FOV) is limited by the pulse intensity of conventional infrared lasers. Here, we present a wide-field MIP microscope that uses a high-power free-electron laser (FEL) rather than a quantum cascade laser (QCL) as the pump source to achieve a substantially larger FOV. Both implementations use counter-propagating beam paths with a 450 nm LED as the probe source and a CMOS camera that records images using a virtual lock-in detection scheme. QCL nanojoule pulse energies enables FOV of around 45 micrometers for widefield MIP imaging with a sub-micrometer resolution for polystyrene beads, Mycobacterium tuberculosis infected fixed tissues, and laryngeal cancer cryosections. IR spectra in the range of 1000-1800 wavenumbers can be reconstructed by tuning the QCL. FEL pulse energies of up to microjoules expand the FOV by a factor of nearly 20 as demonstrated by wide-field MIP imaging of polystyrene beads, single cells, and murine brain tissue. We discuss current challenges and further improvements to implement high-power IR lasers for wide-field MIP imaging with even larger FOVs in the context of biomedical research and diagnostics.

Quantum-electrodynamical density-functional theory (QEDFT) provides a first-principles framework for describing materials coupled to quantized electromagnetic fields. While QEDFT has successfully captured cavity-induced modifications of electronic structures in atoms and molecules, a fully self-consistent and accurate framework to simulate and predict the structural, phonon-related, polarization and optical response of periodic solids in optical cavities has remained elusive. Here, we introduce a unified QEDFT approach that incorporates collective light-matter coupling in the electronic ground state, density functional perturbation theory for phonons, and real-time time-dependent QEDFT for optical excitations. This framework enables \textit{ab initio} calculations of cavity-modified electronic and phononic dispersions, Born effective charges, dielectric tensors, and both resonant and non-resonant optical absorption spectra. Using wurtzite \ac{GaN} in an optical cavity as a case study, we demonstrate that the quantized vacuum field reshapes electronic, phononic and polarization properties, producing experimentally accessible signatures in the transmission and absorption spectra. These results establish QEDFT as a general first-principles platform for predicting and exploring cavity-modified quantum materials.

Snell law is traditionally regarded as a hallmark of phase-propagating phenomena such as optical, acoustic, elastic, electromagnetic, and quantum waves. In contrast, purely diffusive processes, such as Fourier heat conduction and chemical diffusion, are generally considered incapable of exhibiting refractive/reflective behavior. In this letter, we demonstrate that although diffusion waves including thermal diffusion, mass diffusion, Lindblad quantum diffusion, and electromagnetic diffusion do not follow Snell law in either time or frequency-domain, nevertheless they obey a spectral form of Snell law which reveals a hidden analog of wave refraction/reflection within the mathematical structure of diffusion dynamics. Remarkably, the spectral refraction ratio is governed not by the diffusion coefficient itself but by the constitutive relations of the media across the interface, establishing a new physical paradigm for diffusion-wave fields. Importantly, while each spectral eigenmode satisfies a rigorous Snell-type refraction relation, the inverse Fourier-Laplace transformation mixes these modes and suppresses any persistent real-space refraction angle, thereby reconciling the modal-level directionality with the long-standing absence of geometric refraction in diffusive systems.

We present the Ising noise filter, a highly portable, graph-based pre-filtering algorithm for early-stage background suppression in particle accelerators and astrophysical detectors. Standard noise rejection methods relying on track fitting suffer from severe combinatorial explosion. Our method bypasses this by mapping individual detector hits to a network of binary spins and minimizing an energy functional. The interaction kernels are physics-informed, tailored to the underlying physics and geometry of the experiment. We demonstrate the efficacy of this approach in two distinct experimental regimes. Applied to the Baikal-GVD neutrino telescope the filter yields fast, standard-quality noise rejection with 96.8\% recall for astrophysical neutrinos. For the SPD detector at the NICA collider the filter attains recall of 97\% on a toy Monte Carlo sample. Furthermore, when combined with a Peterson--Hopfield network for track finding, our physics-informed coupling improves the TrackML score from 0.5 to 0.95.

We investigate the quantum evolution of the pump field in second-harmonic generation under strong pump depletion. Starting from a coherent state, the pump develops a nonclassical phase-space structure resembling a Schrödinger cat state. This behavior originates from phase instability induced by vacuum fluctuations of the harmonic mode. A rigorous quantum analysis has been performed for mean photon numbers up to $\langle \hat n \rangle = 100$ in pump mode. For larger photon numbers, up to $\langle \hat n \rangle = 10^{7}$, the dynamics have been analyzed using a classical trajectory method with sampled initial conditions that reproduces the main features of the quantum evolution. The results indicate that nonlinear frequency conversion can generate macroscopic superposition-like states of the pump field. Although the resulting state is not pure due to correlations with the second-harmonic wave, it remains non-classical with negative zones of Wigner function. These results indicate that strongly nonlinear frequency conversion can provide a scalable route toward macroscopic nonclassical states of light.

We derive the conserved energy-like quantity and ensemble measure for Martyna--Tobias--Klein (MTK) barostats in which only a restricted subset of the cell degrees of freedom are active. In the standard fully anisotropic MTK formulation, the number of barostat degrees of freedom is $d^{2}$, where $d$ is the spatial dimension. When only $n_c$ axes of the cell matrix are allowed to fluctuate, the conserved energy-like quantity retains the same functional form but with $d^{2}$ replaced by $n_c$ in every term that counts barostat degrees of freedom. The derivation builds on the generalized Liouville framework for non-Hamiltonian systems and the existing MTK integration machinery. We verify that this quantity is exactly conserved, show that the resulting dynamics samples the isothermal--isobaric ensemble restricted to the submanifold of cell shapes in which inactive components are held fixed, and provide a complete Liouville-operator-based integration scheme for the masked MTK variant.

We present a system for generating arbitrary, triaxial magnetic waveforms with a spectral content spanning from DC to tens of kHz, a critical capability for quantum control and spin manipulation. To compensate for amplifier-coil dynamics, we implement a data-driven approach to identify a numerical compensation model. The method parametrizes the system response using a Finite Impulse Response (FIR) filter calibrated on the specific waveform to be generated. The application of a driving signal designed via frequency-domain inversion of the identified model enables the synthesis of complex field sequences with sharp transitions between static and single- or multi-frequency temporal segments. The work is validated with experimental results demonstrating waveform fidelity and transient performance, thereby showcasing the precision and feasibility of the method.

Quantum probes that enable enhanced exploration and characterization of complex systems are central to modern science, spanning applications from biology to astrophysics and chemical design. In large many-body quantum systems, interactions delocalize phase information across many degrees of freedom, dispersing it away from accessible measurements and limiting the scalability of quantum metrology. Here we show that experimentally accessible Clifford operations acting jointly on quantum states and observables can refocus this distributed information. These operations implement what we term {\it Clifford lensing}--transformations that coherently localize phase information onto a reduced set of degrees of freedom, mapping optimal measurements onto observables of reduced Pauli weight. We establish a correspondence between quantum error-correcting codes and interferometric constructions that enforce deterministic phase kickback, and generalize this to circuits that concentrate many-body phase information onto a controllable subset of qubits. We further develop partial shadow tomography protocols for estimating subsystem-supported phases. We experimentally demonstrate these principles in liquid-state nuclear magnetic resonance systems of up to fifteen qubits, achieving optimal sensing with constrained resources. Our results establish a scalable route to coherent control of information flow in interacting quantum systems, enabling many-body quantum sensing and multimode interferometry across complex architectures.

Finite-difference time-domain (FDTD) simulations often involve physical quantities spanning multiple orders of magnitude, such as the speed of light or electromagnetic field amplitudes. The standard practice for maintaining numerical accuracy in many FDTD implementations is to use 32-bit or 64-bit floating-point values to represent the electric and magnetic fields. However, this approach is not always optimal when recording field values, particularly during time-reversible gradient computation where electric and magnetic field values need to be saved at the boundary of the simulation domain. Since this memory bottleneck is often the limiting factor in time-reversible inverse design for nanophotonics, we present two field optimizations for enhancing memory efficiency in FDTD simulations. Using a smaller bit-width representation of field values as well as interpolation, we achieve similar accuracy at lower memory cost. This approach is particularly beneficial for GPU-accelerated computing, where reduced-precision data types are increasingly preferred due to their computational efficiency and prevalence in machine learning frameworks. We integrate our approach into FDTDX, an open-source, differentiable FDTD solver that natively supports time-reversible gradient computation. Our approach is especially important for future developments towards large-scale open-source simulations, which are critical for advancing computational nanophotonic applications.

Voltage breakdown in high-voltage pulsed vacuum systems arises from nonlinear multiscale interactions among circuit dynamics, kinetic plasma evolution, and ion-induced secondary electron emission (SEE) at electrode surfaces. Although circuit-plasma co-simulation frameworks couple lumped circuits with particle-in-cell (PIC) solvers, most neglect energy-resolved SEE and its feedback to both plasma and circuit, limiting predictive capability. We present a self-consistent framework for multiscale circuit-plasma coupling that incorporates ion-energy-dependent SEE into the electrode boundary of an electrostatic PIC solver. The emitted electron flux is included in the surface charge update, leading to a modified Poisson boundary condition that couples plasma and circuit within a unified formulation. Two integration strategies are developed: (i) a fully implicit strict coupling scheme solving the plasma-circuit system monolithically, and (ii) a weak coupling scheme based on operator splitting, compatible with SPICE solvers and enabling partitioned time integration with one-step lag. The framework is applied to a Tesla-transformer-driven vacuum capacitor with ion injection. Results show that SEE alters surface charge evolution, triggering rapid voltage collapse and sustaining a near-zero-voltage plateau, while SEE-free models fail. Agreement between strict and weak coupling confirms robustness. The method provides a unified framework for predictive simulation of multiscale circuit-plasma interactions.

Low-frequency magnetic fields carry vital information for neuroscience, navigation, and Earth science. However, they are generally weak, making it challenging to measure them with compact, room-temperature magnetometers. To overcome this challenge, we combine an on-chip optomechanical magnetometer with a high-permeability flux concentrator. Beyond boosting sensitivity and bandwidth, exploiting the concentrator's nonlinear response converts low-frequency magnetic fluctuations into higher-frequency signals where the sensor is intrinsically most responsive. This sidesteps the technical noise that has long constrained the application of optomechanical magnetometry at low frequencies. Our measurements show order-of-magnitude improvements in sensitivity and extend performance into the sub-hertz regime, achieving below 20 nT Hz$^{-1/2}$ down to 3 Hz and less than 100 nT Hz$^{-1/2}$ at 0.1 Hz. Because this approach requires no redesign of the underlying architecture, it can be readily applied across magnetometer technologies, opening the way to practical low-frequency sensing for applications from brain activity mapping to undersea navigation and biomedical diagnostics.

Optical skyrmions are structured vector fields with nontrivial polarization topology and subwavelength-scale features. One common approach to generating optical skyrmions is the superposition of a zeroth-order Bessel beam and a higher-order Bessel beam carrying orbital angular momentum, with each beam possessing an orthogonal circular polarization state. However, creating such complex beams typically requires bulky free-space optical setups; therefore, recent efforts have focused on compact optical skyrmion generators based on metasurfaces. Nevertheless, achieving the degrees of freedom required for simultaneous phase and polarization control remains challenging because of the limited design flexibility of conventional meta-atoms. Here, we address this challenge by employing an inverse-design approach and demonstrate a single-layer metasurface that generates high-fidelity optical skyrmions. We employ an adjoint-based topology-optimization method to design a silicon metasurface that converts an incident beam into an optical skyrmion without the need for additional optical components. The optimized metasurface generates an optical skyrmion with skyrmion number $(N_\mathrm{sk}) = 0.970$. This work demonstrates that inverse design can be a promising route to compact skyrmion generators, and our approach provides a basis for near-field particle manipulation and the generation of independent topological bits in dense photonic integration.

Visible photonic integrated circuits underpin applications ranging from AR/VR to quantum control, yet lack a high-resolution, nondestructive diagnostic comparable to the optical frequency-domain reflectometry used in infrared silicon photonics. Here we adapt spectral-domain optical coherence tomography to measure guided-mode back-reflections in visible PICs. Broadband visible light injected into a circuit generates back-reflections that interfere with a depth-referencing local oscillator, and the resulting spectral fringes are recorded on a spectrometer. We validate the approach by resolving multiple round-trip echoes in a waveguide-coupled ring resonator using only single-port access. We then extend it to circuits integrated with diamond quantum micro-chiplets, clearly resolving input and output facets as well as PIC--QMC transition regions. The system achieves shot-noise-limited sensitivity, 50 dB dynamic range, 8 um axial resolution in silicon nitride, and a 2 mm imaging depth at 6 dB roll-off. SD-OCT therefore provides a practical, high-resolution diagnostic for visible PICs that uses a broadband probe source and requires only single-port optical access, enabling rapid characterization of propagation loss, backscattering, and dispersion.

Wavefront-free behavior and the resulting quantification difficulties are intrinsic limitations of vibrothermography due to the diffusive nature of thermal fields. This work proposes a generalized virtual wave reconstruction framework to address the absence of propagation features in thermal diffusion-wave fields and its impact on quantitative defect characterization. Unlike conventional virtual wave formulations restricted to Dirac-type excitations, the proposed approach establishes a rigorous spatiotemporal mapping between diffusion-wave fields and virtual wave fields under arbitrary heat-generation conditions without simplifying assumptions on thermo-mechanical coupling. The resulting ill-posed inversion problem is solved using truncated singular value decomposition (T-SVD) and the alternating direction method of multipliers (ADMM) to enhance numerical stability and suppress noise amplification. Numerical simulations demonstrate that the reconstructed virtual wave fields recover propagation characteristics absent in temperature distributions, leading to improved defect boundary definition, contrast enhancement, and depth-resolved analysis. Experiments on CFRP laminates validate the robustness of the approach and show significantly improved signal-to-noise ratio, spatial clarity, and defect size estimation compared with conventional thermographic processing methods.

We present a widefield two-dimensional electronic spectroscopy microscope (2DESM) that integrates multidimensional coherent spectroscopy with optical imaging, enabling femtosecond temporal and micrometer spatial resolution. The broadband coverage (1.4-1.8 eV) allows the direct acquisition of spatially resolved two-dimensional electronic spectroscopy (2DES) maps of relevant near infrared excitations without the need for spatial scanning. By capturing both spectral and spatial domains simultaneously, 2DESM overcomes limitations of pump-probe microscopy and scanning 2DES, providing access to decoherence dynamics, inhomogeneous broadening, and coherent coupling in heterogeneous systems. As a proof-of-concept we performed 2DESM measurements on bilayer WSe2 encapsulated in hBN, revealing distinct spatial variations in excitonic dynamics. These results validate the ability of 2DESM to link local environments with ultrafast coherent processes and establish 2DESM as a versatile platform for probing quantum coherence, many-body interactions, and non-local energy transfer in two-dimensional materials, heterostructures, and micrometer-scale optoelectronic devices.

We report several technical approaches that significantly improve the performance of a vapor-cell atomic electrometer operating in the quasi-DC frequency domain ($\ll$ 1 kHz). With a very small active volume of approximately 11 mm$^3$ inside the vapor cell, we demonstrated a noise floor for electric field (E-field) sensitivity ranging from 0.2 to 7.7 mV/m$\sqrt{\rm Hz}$ for a frequency band of 1--100 Hz. Our work utilizes only a bare vapor cell for electrometry, without any metal parts or electrodes, to ensure minimal distortion of the measured E-field and to minimize the effective sensing volume for high spatial resolution. The E-field-sensitive atomic state (Rydberg state) is excited and read out optically, maximizing the simplicity of the system design and enabling the miniaturization of quasi-DC E-field sensors for potential applications, such as diagnostics of electronics without physical contact, communications in and below the super-low frequency (SLF) band, proximity detection, remote activity surveillance, tracing charge signatures, and research in bioscience and geoscience.

Using a closed quantum optical coupled-dipole model, we investigate why sub-sevenfold symmetries are likely absent in the stacked-ring scaffolds of light-harvesting 2 (LH2) complexes in purple photosynthetic bacteria.

Underwater observatories have recently emerged as an efficient mean of marine biodiversity monitoring. In order to conduct data muling from the underwater sensors in an efficient and cost-effective way, we consider the use of optical wireless communications to transmit the data from the underwater sensors to an aerial node close to the water surface, such as an unmanned aerial vehicle (UAV). More specifically, we utilize a direct water-to-air (W2A) optical communication link between the sensor node equipped with an LED emitter and the UAV equipped with an ultra-sensitive receiver, i.e., a silicon photomultiplier (SiPM). To characterize this particularly complex communication channel, we introduce a ray-tracing algorithm based on the Monte Carlo method, incorporating the impact of bubbles modeled through the Mie scattering theory and a realistic sea surface representation derived from the JONSWAP spectrum. Additionally, we incorporate into this model the channel losses resulting from UAV instability under windy weather conditions. Furthermore, we conduct a comprehensive analysis of the wireless channel, examining the influence of key parameters such as wind speed, transmitter configurations, and receiver characteristics. Finally, we evaluate the end-to-end performance of the system by analyzing the average bit-error rate at varying depths and data rates, providing valuable insights into the feasibility and efficiency of the proposed approach.

Origami-inspired robots offer rapid, accessible design and manufacture with diverse functionalities. In particular, origami robots without conventional electronics have the unique advantage of functioning in extreme environments such as ones with high radiation or large magnetic fields. However, the absence of sophisticated control systems limits these robots to simple autonomous behaviors. In our previous studies, we developed a printable, electronics-free, and self-sustained oscillator that generates simple complementary square-wave signals. Our study presents a quadrature oscillation system capable of generating four square-wave signals a quarter-cycle out of phase, enabling four distinct states. Such control signals are important in various engineering and robotics applications, such as orchestrating limb movements in bio-inspired robots. We demonstrate the practicality and value of this oscillation system by designing and constructing an origami crawling robot that utilizes the quadrature oscillator to achieve coordinated locomotion. Together, the oscillator and robot illustrate the potential for more complex control and functions in origami robotics, paving the way for more electronics-free, rapid-design origami robots with advanced autonomous behaviors.

Foundation models have the potential to extend the discovery reach for anomaly detection searches. When studying the large OmniLearned foundation model on data from the CMS experiment, unexpected behavior was observed in a mass sideband. The purpose of this paper is to perform a full analysis, including a complete background estimate, on the phase space picked out by the large model. We find that the background estimation describes the data well in validation regions, but is unable to accurately model the signal region. We invite further scrutiny of these events and our methods.

This work provides a brief review of the numerical methods for modelling the optical propagation of an ultrashort, intense laser pulse in plasmas and ionizable gases. These methods are implemented in an open-source simulation toolkit Axiprop, which is now actively used for design studies in the context of laser plasma acceleration of electrons (LPA). We present two examples of such studies: optical plasma waveguide generation, and phase-locked flying-focus LPA. These tools enable the identification of complex effects impacting both energy deposition during waveguide formation, and the pulse propagation dynamics that governs wakefield structure, ultimately determining the properties of the accelerated electron bunches. The developed tools and presented simulation designs are relevant to experiments on laser plasma interactions at modern high power laser facilities.

Wafer-to-wafer (WxW) bonding is a key enabler for three-dimensional integration, including hybrid bonding for fine-pitch Cu-Cu interconnects. During bonding, wafer deformation and the air entrapped between the wafers interact through a strongly coupled, time-dependent fluid-structure interaction (FSI) that can produce non-intuitive bonding dynamics and process sensitivities. This paper develops a mathematically consistent reduced-order model for WxW bonding by deriving a Kirchhoff-Love plate equation for wafer bending from three-dimensional linear elasticity and coupling it to a Reynolds lubrication equation for the inter-wafer air film. The resulting nonlinear plate-Reynolds system is discretized and solved monolithically in the high-performance FEniCSx framework using a $C^0$ interior-penalty formulation for the fourth-order plate operator, standard continuous Galerkin discretization for the pressure field, implicit time integration, and a Newton solver with automatic differentiation. Simulations reproduce experimentally reported probe-displacement histories for multiple initial gaps and verify force equilibrium at the bond front, where the Reynolds pressure acts as an effective contact reaction. Parametric studies reveal nonlinear, and in some cases non-monotonic, sensitivities of bonding-front kinetics to the initial gap, air viscosity, and interfacial energy, providing actionable trends for process optimization.

Trapped ions are one of the most advanced platforms for quantum technologies, with applications ranging from quantum computing to precision timekeeping. A crucial step towards more compact and scalable systems involves integrating photonic integrated circuits (PICs) into surface ion traps to enable on-chip light delivery and optical addressing of individual ions. Currently, most implementations rely solely on transverse-electric (TE) mode grating couplers, where the emitted light is polarized in the plane of the chip. In this work, we design, fabricate and characterize silicon nitride (Si\(_3\)N\(_4\)) PIC components, including incoupling structures, splitters, and grating couplers that support both TE and transverse-magnetic (TM) modes with comparable optical losses. We benchmark the PIC at 760\,nm, which is a typical wavelength for Yb$^{+}$-applications. The fabricated grating couplers enable the outcoupling of collimated free-space beams for both polarizations, exhibiting distinct emission angles. This dual-polarization capability gives more flexibility in polarization control and expands the accessible optical design space for trapped-ion quantum technologies.

Marathons are now common ways of seeing cities, yet little is known about how representative their routes are. Using 311 marathon routes across five continents, we compare landmarks and amenities along the course with those elsewhere in the same city, finding that museums are 15.7 times denser near the route and that the median city has about 8.5 times more luxury brands near the route than elsewhere in the city. These patterns persist under perturbed routes with the same start and finish lines: monuments and landmarks, in particular, are more prevalent on the race course than on similar alternative routes, suggesting that marathons function as intentionally selective urban portraits.

We derive a Riemannian metric on three-dimensional color space from the Fisher information of neural population codes in the visual pathway. Photoreceptor adaptation, retinal opponent channels, and cortical population encoding each map onto a geometric construction, producing a metric tensor whose components correspond to measurable neural quantities. The resulting 17-parameter model is fitted jointly to four independent threshold datasets: MacAdam's (1942) chromaticity ellipses, the Koenderink et al. (2026) three-dimensional ellipsoids, Wright's (1941) wavelength discrimination function, and the Huang et al. (2012) threshold color difference ellipses, covering 96 independently measured discrimination conditions across varied chromaticities and luminances. The joint fit achieves STRESS of 23.9 on MacAdam, 20.8 on Koenderink et al., 30.1 on Wright, and 30.8 on Huang et al.

Our subjective experience of color is typically described by abstract properties such as hue, saturation, and brightness that do not directly correspond to sensory signals arising from cones in the retina. Along the hue dimension, certain colors -- red, green, blue, and yellow -- appear unique in that they are not perceived as a combination of other colors, and the pairs red-green and blue-yellow appear opposites. However, the anatomical and physiological correlates of these 'unique hues' within the brain and the reason for their existence remain a mystery. Here, we demonstrate a direct connection between these hues and the statistics of the natural visual environment. Analysis of simulated cone responses on a dataset of 503 calibrated natural images reveals a strongly non-Gaussian distribution in 3D color space, with heavy tails in distinct, asymmetrically arranged directions. A sparse coding model is then adapted to this data so as to minimize the total sum of coefficients on the basis vectors for representing the data. A six basis-vector model converges to the four unique hues in addition to black and white. Moreover, we find that the nonlinear nature of inference in the sparse coding model yields both excitatory and inhibitory interactions among latent variables; the former facilitates combining adjacent pairs of unique hues to encode intermediate hues situated between them, while the latter enforces mutual exclusivity between opposite unique hues. Together, these findings shed new light on the distribution of color in the natural environment and provide a linking principle between this structure and the phenomenology of color appearance.

Capturing dynamic spatiotemporal neural activity is essential for understanding large-scale brain mechanisms. Functional magnetic resonance imaging (fMRI) provides high-resolution cortical representations that form a strong basis for characterizing fine-grained brain activity patterns. The high acquisition cost of fMRI limits large-scale applications, therefore making high-quality fMRI reconstruction a crucial task. Electroencephalography (EEG) offers millisecond-level temporal cues that complement fMRI. Leveraging this complementarity, we present an EEG-conditioned framework for reconstructing dynamic fMRI as continuous neural sequences with high spatial fidelity and strong temporal coherence at the cortical-vertex level. To address sampling irregularities common in real fMRI acquisitions, we incorporate a null-space intermediate-frame reconstruction, enabling measurement-consistent completion of arbitrary intermediate frames and improving sequence continuity and practical applicability. Experiments on the CineBrain dataset demonstrate superior voxel-wise reconstruction quality and robust temporal consistency across whole-brain and functionally specific regions. The reconstructed fMRI also preserves essential functional information, supporting downstream visual decoding tasks. This work provides a new pathway for estimating high-resolution fMRI dynamics from EEG and advances multimodal neuroimaging toward more dynamic brain activity modeling.

Understanding how animals move through heterogeneous landscapes is central to ecology and conservation. In this context, step selection functions (SSFs) have emerged as the main statistical framework to analyze how biotic and abiotic predictors influence movement paths observed by radio tracking, GPS tags, or similar sensors. A traditional SSF consists of a generalized linear model (GLM) that infers the animal's habitat preferences (selection coefficients) by comparing each observed movement step to random steps. Such GLM-SSFs, however, cannot flexibly consider non-linear or interacting effects, unless those have been specified a priori. To address this problem, generalized additive models have been integrated in the SSF framework, but those GAM-SSFs are still limited in their ability to represent complex habitat preferences and inter-individual variability. Here we explore the utility of deep neural networks (DNNs) to overcome these limitations. We find that DNN-SSFs, coupled with explainable AI to extract selection coefficients, offer many advantages for analyzing movement data. In the case of linear effects, they effectively retrieve the same effect sizes and p-values as conventional GLMs. At the same time, however, they can automatically detect complex interaction effects, nonlinear responses, and inter-individual variability if those are present in the data. We conclude that DNN-SSFs are a promising extension of traditional SSF. Our analysis extends previous research on DNN-SSF by exploring differences and similarities of GLM, GAM and DNN-based SSF models in more depth, in particular regarding the validity of statistical indicators that are derived from the DNN. We also propose new DNN structures to capture inter-individual effects that can be viewed as a nonlinear random effect. All methods used in this paper are available via the 'citoMove' R package.

Background: Autism spectrum disorder (ASD) is characterized by significant clinical and biological heterogeneity. Conventional group-mean analyses of eye movements often mask individual atypicalities, potentially overlooking critical pathological signatures. This study aimed to identify idiosyncratic oculomotor patterns in ASD using an "outlier analysis" of smooth pursuit eye movement (SPEM). Methods: We recorded SPEM during a slow Lissajous pursuit task in 18 adults with ASD and 39 typically developed (TD) individuals. To quantify individual deviations, we derived an "outlier score" based on the Mahalanobis distance. This score was calculated from a feature vector, optimized via Principal Component Analysis (PCA), comprising the temporal lag ($Δ$t) and the spatial deviation ($Δ$s). An outlier was statistically defined as a score exceeding $\sqrt{10}$ (approximately 3.16$σ$) relative to the TD normative distribution. Results: While the TD group exhibited a low outlier rate of 5.1%, the ASD group demonstrated a significantly higher prevalence of 38.9% (7/18) (binomial P = 0.0034). Furthermore, the mean outlier score was significantly elevated in the ASD group (3.00 $\pm$ 2.62) compared to the TD group (1.52 $\pm$ 0.80; P = 0.002). Notably, these extreme deviations were captured even when conventional mean-based comparisons showed limited sensitivity. Conclusions: Our outlier analysis successfully visualized the high degree of idiosyncratic atypicality in ASD oculomotor control. By shifting the focus from group averages to individual deviations, this approach provides a sensitive metric for capturing the inherent heterogeneity of ASD, offering a potential baseline for identifying clinical subtypes.

The deployment of extremely large-scale antenna array (ELAA) in sixth-generation (6G) communication systems introduces unique challenges for efficient near-field channel estimation. To tackle these issues, this paper presents a theory-guided approach that incorporates angular information into an attention-based estimation framework. A piecewise Fourier representation is proposed to implicitly encode the near-field channel's inherent nonlinearity, enabling the entire channel to be segmented into multiple subchannels, each mapped to the angular domain via the discrete Fourier transform (DFT). Then, we develop a joint subchannel-spatial-attention network (JSSAnet) to extract the spatial features of both intra- and inter-subchannels. To guide theoretically the design of the joint attention mechanism, we derive upper and lower bounds based on approximation criteria and DFT quantization loss mitigation, respectively. Following by both bounds, a JSSA layer of an attention block is constructed to assign independent and adaptive spatial attention weights to each subchannel in parallel. Subsequently, a feed-forward network (FFN) of an attention block further captures and refines the residual nonlinear dependencies across subchannels. Moreover, the proposed JSSA map is linearly computed via element-wise product combining large-kernel convolutions (DLKC), maintaining strong contextual learning capability. Numerical results verify the effectiveness of embedding sparsity information into the attention network and demonstrate JSSAnet achieves superior estimation performance compared with existing methods.

The International Telecommunication Union (ITU) identifies "Artificial Intelligence (AI) and Communication" as one of six key usage scenarios for 6G. Agentic AI, characterized by its ca-pabilities in multi-modal environmental sensing, complex task coordination, and continuous self-optimization, is anticipated to drive the evolution toward agent-based communication net-works. Semantic communication (SemCom), in turn, has emerged as a transformative paradigm that offers task-oriented efficiency, enhanced reliability in complex environments, and dynamic adaptation in resource allocation. However, comprehensive reviews that trace their technologi-cal evolution in the contexts of agent communications remain scarce. Addressing this gap, this paper systematically explores the role of semantics in agent communication networks. We first propose a novel architecture for semantic-based agent communication networks, structured into three layers, four entities, and four stages. Three wireless agent network layers define the logical structure and organization of entity interactions: the intention extraction and understanding layer, the semantic encoding and processing layer, and the distributed autonomy and collabora-tion layer. Across these layers, four AI agent entities, namely embodied agents, communication agents, network agents, and application agents, coexist and perform distinct tasks. Furthermore, four operational stages of semantic-enhanced agentic AI systems, namely perception, memory, reasoning, and action, form a cognitive cycle guiding agent behavior. Based on the proposed architecture, we provide a comprehensive review of the state-of-the-art on how semantics en-hance agent communication networks. Finally, we identify key challenges and present potential solutions to offer directional guidance for future research in this emerging field.

The proliferation of civilian and commercial unmanned aerial vehicles (UAVs) has heightened the demand for reliable radio frequency (RF)-based drone identification systems that can operate under dynamic and uncertain airspace conditions. Most existing RF-based recognition methods adopt a closed-set assumption, where all UAV types are known during training. Such an assumption becomes unrealistic in practical deployments, as new or unknown UAVs frequently emerge, leading to overconfident misclassifications and inefficient retraining cycles. To address these challenges, this paper proposes a unified incremental open-set learning framework for RF-based UAV recognition that enables both novel class discovery and incremental adaptation. The framework first performs open-set recognition to separate unknown signals from known classes in the semantic feature space, followed by an unsupervised clustering module that discovers new UAV categories by selecting between K-Means and Gaussian Mixture Models (GMM) based on composite validity scores. Subsequently, a lightweight incremental learning module integrates the newly discovered classes through a memory-bounded replay mechanism that mitigates catastrophic forgetting. Experiments on a real-world UAV RF dataset comprising 24 classes (18 known and 6 unknown) show effective open-set detection, promising clustering performance under the evaluated noise settings, and stable incremental adaptation with minimal storage cost, supporting the potential of the proposed framework for open-world UAV recognition.

We propose an alternating optimization framework for maximizing energy efficiency (EE) in reconfigurable intelligent surface (RIS) assisted distributed MIMO (D-MIMO) systems under both coherent and non-coherent reception modes. The framework jointly optimizes access point (AP) power allocation and RIS phase configurations to improve EE under per-AP power and signal-to-interference-plus-noise ratio (SINR) constraints. Using majorization-minimization for power allocation together with per-element RIS adaptation, the framework achieves tractable optimization of this non-convex problem. Simulation results for indoor deployments with realistic power-consumption models show that the proposed scheme outperforms equal-power and random-scatterer baselines, with clear EE gains. We evaluate the performance of both reception modes and quantify the impact of RIS phase-shift optimization, RIS controller architectures (centralized vs. per-RIS control), and RIS size, providing design insights for practical RIS-assisted D-MIMO deployments in future 6G networks.

Most algorithms for hyperspectral image unmixing produce point estimates of fractional abundances of the materials to be separated. However, in the absence of reliable ground truth, the ability to perform abundance uncertainty quantification (UQ) should be an important feature of algorithms, e.g. to evaluate how hard the unmixing problem is and how much the results should be trusted. The usual modeling assumptions in Bayesian models for unmixing rely heavily on the Euclidean geometry of the simplex and typically disregard spatial information. In addition, to our knowledge, abundance UQ is close to nonexistent. In this paper, we propose to leverage Aitchinson geometry from the compositional data analysis literature to provide practitioners with alternative tools for modeling prior abundance distributions. In particular we show how to design simplex-valued Gaussian Process priors using this geometry. Then we link Aitchinson geometry to constrained sampling algorithms in the literature, and propose UQ diagnostics that comply with the constraints on abundance vectors. We illustrate these concepts on real and simulated data.

The sub-THz spectrum offers numerous advantages, including massive multiple-input multiple-output (MIMO) technology with large antenna arrays that enhance spectral efficiency (SE) of future systems. Hybrid precoding (HP) thus emerges as a cost-effective alternative to fully digital precoding regarding complexity and energy consumption. However, sub-THz frequencies introduce hardware challenges, particularly phase noise (PN) from local oscillators (LOs). We analyze PN impact on MIMO systems using HP, leveraging singular value decomposition and common LO architecture. We adopt the Gaussian PN (GPN) model, recognized as accurate for describing PN behavior in sub-THz transmissions. We derive a lower bound on achievable SE and provide closed-form bit error rate expressions for quadrature amplitude modulation (QAM), specifically 4-QAM and 16-QAM, under high-SNR and strong GPN conditions. These analytical results are validated through Monte Carlo simulations. We show that GPN can be effectively counteracted with a single pilot symbol in single-user MIMO systems, unlike single-input single-output systems where mitigation proves infeasible. Simulation results compare conventional QAM against polar-QAM tailored for GPN-impaired systems. Finally, we introduce perspectives for further improvements in performance and energy efficiency.

Near-field beamfocusing with extremely large aperture arrays can effectively enhance physical layer security. Nevertheless, even small estimation errors of the eavesdropper's location may cause a pronounced focal shift, resulting in a severe degradation of the secrecy rate. In this letter, we propose a physics-informed robust beamforming strategy that leverages the electromagnetic (EM) caustic effect for near-field physical layer security provisioning, which can be implemented via phase shifts only. Specifically, we partition the transmit array into caustic and focusing subarrays to simultaneously bypass the potential eavesdropping region and illuminate the legitimate user, thereby significantly improving the robustness against the localization error of eavesdroppers. Moreover, by leveraging the connection between the phase gradient and the EM wave departing angle, we derive the corresponding piece-wise closed-form array phase profile for the subarrays. Simulation results demonstrate that the proposed scheme achieves up to an 80% reduction of the worst-case eavesdropping rate for a localization error of 0.25 m, highlighting its superiority for providing robust and secure communication.

Highly directional mmWave/THz links require rapid beam alignment, yet exhaustive codebook sweeps incur prohibitive training overhead. This letter proposes a sensing-assisted adaptive probing policy that maps multimodal sensing (radar/LiDAR/camera) to a calibrated prior over beams, predicts per-beam reward with a deep Q-ensemble whose disagreement serves as a practical epistemic-uncertainty proxy, and schedules a small probe set using a Prior-Q upper-confidence score. The probing budget is adapted from prior entropy, explicitly coupling sensing confidence to communication overhead, while a margin-based safety rule prevents low signal-to-noise ratio (SNR) locks. Experiments on DeepSense-6G (train: scenarios 42 and 44; test:43) with a 21-beam discrete Fourier transform (DFT) codebook achieve Top-1/Top-3 of 0.81/0.99 with expected beam probe of 2 per sweep and zero observed outages at θ = 0 dB with margin Δ = 3 dB. The results show that multimodal priors with ensemble uncertainty match link quality and improve reliability compared to ablations while cutting overhead with better predictive model.

Deploying six-dimensional movable antenna (6DMA) systems in Internet-of-Vehicles (IoV) scenarios can greatly enhance spectral efficiency. However, the high mobility of vehicles causes rapid spatio-temporal channel variations, posing a significant challenge to real-time 6DMA optimization. In this work, we pioneer the application of 6DMA in IoV and propose a low-complexity, instantaneous channel state information (CSI)-free dynamic configuration method. By integrating vehicle motion prediction with offline directional response priors, the proposed approach optimizes antenna positions and orientations at each reconfiguration epoch to maximize the average sum rate over a future time window. Simulation results in a typical urban intersection scenario demonstrate that the proposed 6DMA scheme significantly outperforms conventional fixed antenna arrays and simplified 6DMA baseline schemes in terms of total sum rate.

This paper addresses the modeling gap between complex dispersive-medium characterization and clutter statistical analysis in single-snapshot frequency diverse array multiple-input multiple-output ground-penetrating radar (FDA-MIMO-GPR). Existing FDA-MIMO clutter studies have rarely incorporated subsurface dispersion, dissipation, and random inhomogeneity in an explicit statistical framework. To bridge this gap, a continuous relaxation spectrum is adopted to describe complex media, and a statistical propagation chain is established from random relaxation-spectrum perturbations to complex permittivity, complex wavenumber, steering-vector perturbation, medium-induced additional clutter covariance, and total clutter covariance. On this basis, the effects of medium randomness on covariance spectral spreading, effective rank, effective clutter-subspace dimension, and target-clutter separability are further characterized. Numerical results show close agreement between the derived theory and Monte Carlo sample statistics across multiple stages of the propagation chain. The results further indicate that medium uncertainty not only changes clutter-covariance entries, but also reshapes its eigenspectrum and effective subspace, thereby influencing the geometric separation between target and clutter. The study provides an explicit and interpretable theoretical interface for embedding complex-medium uncertainty into FDA-MIMO-GPR clutter statistical analysis.

Analog beamforming holds great potential for future terahertz (THz) communications due to its ability to generate high-gain directional beams with low-cost phase shifters. However, conventional analog beamforming may suffer substantial performance degradation in wideband systems due to the beam squint effect. Instead of relying on high-cost true-time delayers, we propose an efficient six-dimensional movable antenna (6DMA) architecture to mitigate the beam-squint effect. In particular, we study a wideband wide-beam coverage problem in this paper, aiming to maximize the minimum beamforming gain over a given range of azimuth/elevation angles and frequencies by jointly optimizing the analog beamforming vector, the MA positions within a two-dimensional (2D) region, and the three-dimensional (3D) rotation angles of the antenna array. However, this problem is non-convex and intractable to solve optimally due to the coupling of the spatial and frequency domains and that of the antenna weights, positions and rotation. To tackle this problem, we first derive an optimal solution to it in a special case with azimuth or elevation angle coverage only. It is shown that rotating a uniform linear array (ULA) is sufficient to achieve global optimality and eliminate beam-squint effects. While for other general cases, an alternating optimization (AO) algorithm is proposed to obtain a high-quality suboptimal solution, where the antennas' beamforming weights, positions, and rotation angles are alternately optimized by combining successive convex approximation (SCA), sequential update with Gibbs sampling (GS), and hybrid coarse- and fine-grained search. Simulation results demonstrate that our proposed scheme can significantly outperform conventional antenna arrays without antenna movement or rotation, thus offering a cost-effective solution for wideband transmission over THz bands.

Active incoherent millimeter-wave (AIM) imaging is a recently developed technique that has been shown to generate fast millimeter-wave imaging using sparse apertures and Fourier domain sampling. In these systems, spatial frequency sampling is determined by cross-correlation between antenna pairs, making array geometry an important aspect that dictates the field of view (FOV) and image quality. This work investigates the impact of array redundancy and spatial sampling diversity on AIM image reconstruction performance. We present a comparative study of three receive array configurations, including one simple circular design and two arrays obtained through optimization strategies designed to maximize unique spatial samples while preserving system resolution and FOV. Performance is evaluated using the image-domain metrics of structural similarity index (SSIM) and peak sidelobe level (PSL), enabling a quantitative assessment of reconstruction fidelity and artifact suppression. We perform experimental validation using a 38-GHz AIM imaging system, implementing a 24-element receive array within a 48-position reconfigurable aperture. Results demonstrate that optimized array configurations improve spatial sampling efficiency and yield measurable gains in reconstruction quality compared to a conventional circular array, highlighting the importance of array design for AIM imaging systems.

Existing cell-free integrated sensing and communication (CF-ISAC) beamforming algorithms predominantly rely on classical optimization techniques, which often entail high computational complexity and limited scalability. Meanwhile, recent learning-based approaches have difficulty capturing the global interactions and long-range dependencies among distributed access points (APs), communication users, and sensing targets. To address these limitations, we propose the first Set Transformer-based CF-ISAC beamforming framework (STCIB). By exploiting attention mechanisms, STCIB explicitly models global relationships among network entities, naturally handles unordered input sets, and preserves permutation invariance across APs, users, and targets. The proposed framework operates in an unsupervised manner, eliminating the need for labeled training data, and supports three design regimes: (i) sensing-centric, (ii) communication-centric, and (iii) joint ISAC optimization. We benchmark STCIB against a convolutional neural network (CNN) baseline and two state-of-the-art optimization algorithms: the convex-concave procedure algorithm (CCPA) and augmented Lagrangian manifold optimization (ALM-MO). Numerical results demonstrate that STCIB consistently outperforms the CNN, achieving substantially higher ISAC performance with only a negligible increase in runtime. For instance, in regime (iii), at $η$=0.4, STCIB improves the sensing and communication sum rates by 14.8 % and 31.6 %, respectively, relative to the CNN, while increasing runtime by only 0.26 %. Compared with CCPA and ALM-MO, STCIB offers significantly lower computational cost while maintaining modest performance gains. In regime (i), for a 3.0 bps/Hz communication threshold, the runtime of STCIB is only 0.1 % and 0.3 % of that required by CCPA and ALM-MO, respectively, while improving the sensing sum rate by 4.45 % and 5.9 %.

Quantum field theory (QFT) on fractal spacetimes is a program aiming at quantizing the gravitational interaction consistently at all energy scales thanks to an intrinsically or dynamically induced multiscale or multifractal-like spacetime geometry that regularizes the infinities of standard QFT. We reach the goal of this program and formulate a field theory of quantum gravity which is shown to be super-renormalizable and unitary at all perturbative orders. Viable and unviable ways to test this proposal through black holes and gravitational waves are discussed.

We present a theoretical investigation on the electronic structure and properties of radium monochalcogenides, with chalcogens O, S, and Se, as well as the ionic species RaO +/-. Our approach combines fully relativistic and partially relativistic quantum-chemistry methods. Electronic properties are obtained using the exact two-component Hamiltonian-based coupled-cluster approach with single, double, and perturbative triple excitations [CCSD(T)+ X2C], while potential energy curves are computed using an internally contracted multireference configuration interaction method, including relativistic effects through small-core pseudopotentials and Pauli-Breit operator diagonalization (MRCI+Q+ECP+SO). The dimers exhibit very large permanent dipole moments and sizable dipolar polarizabilities, while the Franck-Condon factors among the lowest electronic states are highly non-diagonal. These features are discussed in terms of the divalent character of the chemical bonding in the neutral species.

We construct nested Calderbank-Shor-Steane code pairs with non-vanishing coding rate from Hsu-Anastasopoulos codes and MacKay-Neal codes. In the fixed-degree regime, we prove relative linear distance with high probability. Moreover, for several finite degree settings, we prove Gilbert-Varshamov distance by a rigorous computer-assisted proof.

We provide a recursively defined sequence of flag circuits which will detect logical errors induced by non-fault-tolerant $R_{\overline{Z}}(\fracπ{2^l})$ gates on CSS codes with a fault distance of two. As applications, we give a family of circuits with $O(l)$ gates and ancillae which implement fault-tolerant logical $R_{Z}(\fracπ{2^l})$ or $R_{ZZ}(\fracπ{2^l})$ gates on any $[[k + 2, k, 2]]$ iceberg code and fault-tolerant circuits of size $O(l)$ for preparing $|\fracπ{2^l}\rangle$ resource states in the $[[7,1,3]]$ code, which can be used to perform fault-tolerant $R_{\overline{Z}}(\fracπ{2^l})$ rotations via gate teleportation, allowing for implementations of these gates that bypass the high overheads of gate synthesis when $l$ is small relative to the precision required. We show how the circuits above can be generalized to $π( x_0.x_{1}x_{2}\ldots x_{l}) = \sum_{j}^{l} π\frac{x_j}{2^j}$ rotations with identical overheads in $l$, which could be useful in quantum simulations where time is digitized in binary. Finally, we illustrate two approaches to increase the fault-distance of our construction. We show how to increase the fault distance of a Cliffordized version of the T gate circuit to $3$ in the Steane code and how to increase the fault-distance of the $\fracπ{2}$ iceberg circuit to $4$ through concatenation in two-level iceberg codes. This yields a targeted logical $R_{\overline{Z}}(\fracπ{2})$ gate with fault distance $4$ on any row of logical qubits in an $[[(k_2+2)(k_1+2), k_1k_2, 4]]$ code.

Dark energy is one of the deepest puzzles in modern cosmology, and mounting evidence suggests that it is not just a cosmological constant but a genuinely dynamical component. Although cosmology and dynamical systems theory emerged from different disciplines, dynamical systems methods have become essential tools to uncover the qualitative evolution of the universe. The equations governing homogeneous and isotropic cosmologies can be naturally written as systems of ordinary differential equations, making them an ideal arena for dynamical system analysis. This review begins with a sharp, streamlined introduction to the standard dynamical systems toolkit widely used in cosmology. We then move on to alternative formulations based on polar and hyperbolic variable transformations. These approaches unlock powerful new ways to probe a broad spectrum of scalar field dark energy models, to set and constrain initial conditions, and to analyze tracking behavior across wide classes of potentials. The review is self-contained, but consistently directs the reader to more specialized and in-depth treatments where needed.

We investigate the constraining power of future post-reionization and galaxy surveys on possible interactions between dynamical dark energy and dark matter. The analysis focuses on the interaction strength and the dark energy equation of state parameters, in addition to the six standard cosmological parameters. Using fiducial values obtained from the current observational bounds (Planck 2018 + DESI DR2 + Pantheon+), mock datasets for upcoming 21-cm intensity mapping, galaxy clustering and cosmic shear observations from the SKA-mid, and for the upcoming large-scale survey from the Euclid mission, were generated. Subsequently, Markov chain Monte Carlo analyses combining current cosmological data with these mock datasets were performed to forecast parameter constraints. The results indicate that both SKA-mid and Euclid observations can significantly improve constraints on interacting dark sector parameters. In particular, the interaction strength and dark energy equation of state parameters can be constrained considerably tighter than current combined constraints from Planck 2018, DESI DR2 and Pantheon+. Comparing different probe combinations and survey configurations, it is found that SKA2 provides the tightest projected constraints, particularly on the interaction strength, while Euclid achieves a precision broadly comparable to that of SKA1. The results highlight the potential of these upcoming surveys to probe interactions within the dark sector.

The quantum Mpemba effect is a phenomenon characterized by an exponential relaxation from a non-equililbrium state to a steady state. This effect was predicted with an analysis of the Liouvillian superoperator and experimentally demonstrated in a three-level system. In this work, the system dynamics of the Mpemba effect is predicted within the steepest-entropy-ascent quantum thermodynamics framework considering a single constituent three-level isolated system. The system is projected from a four-dimensional Hilbert space onto a three-dimensional one using the Feshbach projection in order to compare the theoretical results with experimental data. Since the quantum Mpemba effect is characterized by a dissipative acceleration, the relaxation parameter, $τ_D$, plays a fundamental rol in the dissipative dynamics predicted by the model and is determined using machine learning methods, resulting in a model that thermodynamically describes this phenomenon at the quantum level.

Quantum devices featuring mid-circuit measurement and reset capabilities, such as quantum computers and dual-species Rydberg quantum simulators, enable the realization of quantum cellular automata. These systems evolve in discrete time following local updates implemented by unitary gates, and allow for the realization of both closed and synthetic open dynamics. Here, we focus on quantum cellular automata that implement minimal models of classical and quantum transport. To illustrate our ideas, we focus on a discrete-time totally asymmetric simple exclusion process and investigate how coherent dynamical contributions allow for the emergence of quantum effects and correlations. We find that bipartite entanglement dominates the transient evolution, while stationary states can retain quantum correlations beyond entanglement. Our results suggest viable routes for realizing transport models on quantum devices and characterizing collective quantum correlations in strongly driven systems.

This work presents the computation of nuclear magnetic shielding and magnetizability tensors for paramagnetic molecules, using a magnetically induced current density framework to account for orbital and spin contributions. We demonstrate that the methodology proposed by Soncini[1] is physically equivalent to the formalisms of Pennanen and Vaara[2] and Franzke et al.[3], provided that scalar and spin-orbit relativistic effects are included within the ground-state spin density. In our model, these corrections are implemented through a Zeroth-Order Regular Approximation (ZORA) formulation of the current density. The resulting magnetizability tensor is fully consistent with the general Van Vleck formulation, recovering the temperature-dependent Curie contribution through the explicit integration of the magnetically induced spin current density. This methodology offers a straightforward computational route that bypasses the complex evaluation of g-tensors and Zero-Field Splitting (ZFS) Hamiltonians, requiring only a ground-state spin density incorporating relativistic effects. Notably, scalar relativistic effects are shown to be essential for capturing the Heavy-Atom Light-Atom (HALA) effect in 1H and 13C shieldings. To maintain efficiency, relativistic effects on the orbital contribution are neglected as they are negligible for light atoms. This approach represents an optimal compromise for paramagnetic complexes involving transition metals up to the second row, where the HALA effect is primarily driven by scalar relativistic corrections within the ground-state spin density. Neglecting spin-orbit terms in the orbital contribution significantly streamlines the calculation without loss of accuracy, providing the pNMR community with a robust tool for characterizing open-shell systems.

We introduce a model of quantum walkers interacting with a magnetic impurity localized at the origin. First, we study a model of a single quantum walker interacting with a localized magnetic impurity. For a simple case of parameter values, we analytically obtain the eigenvalues and the eigenvectors of bound states, in which the quantum walker is bound to the magnetic impurity. Second, we study a model with two quantum walkers and one magnetic impurity, in which the two quantum walkers indirectly interact with each other via the magnetic impurity, as in the Kondo model. We numerically simulate the collision dynamics when the spin-spin interaction at the origin is of the XX type and the SU(2) Heisenberg type. In the case of the XX interaction, we calculate the entanglement negativity to quantify how much the two quantum walkers are entangled with each other, and find that the negativity increases drastically upon the collision of the two walkers. We compare the time dependence for different statistics, namely, fermionic, bosonic, and distinguishable walkers. In the case of the SU(2) interaction, we simulate the dynamics starting from the initial state in which one fermionic walker is in a bound eigenstate around the origin and the other fermionic walker is a delta function colliding with the first walker. We find that a bound eigenstate closest to the singlet state of the first walker and the magnetic impurity is least perturbed by the collision of the second walker. We speculate that this is a manifestation of Kondo physics at the lowest level of the real-space renormalization-group procedure.

We develop a new benchmarking scheme for the Decoded Quantum Interferometry (DQI) algorithm quantifying the number of quantum gates required to obtain an optimal solution to a problem amenable to DQI. We apply the benchmarking scheme to the Binary Paint Shop Problem (BPSP) in order to benchmark the performance of DQI against a state of the art classical solver. To do so, we provide an explicit construction of a quantum circuit implementation of a greedy decoder for low-density parity check codes arising from max-2-XORSAT problems.

One of the most intriguing proposals for wavefunction collapse is the Diosi Penrose model, in which collapse is driven by stochastic fluctuations of the Newtonian potential. We argue that a closely related effective structure can emerge in string theory if, as recently suggested, the present cosmic acceleration is sourced by instant folded strings and their decay products. A key difference, however, is that in this stringy setting the noise is naturally colored in time rather than white. As a result, the scenario is significantly less constrained by existing experiments than the standard Diosi Penrose model.

We analyze quasinormal modes, grey-body factors, and absorption cross-sections of a massive scalar field in four-dimensional Einstein--Gauss--Bonnet black-hole spacetimes within a stability-constrained coupling window. High-order WKB-Padé spectra show that increasing field mass typically reduces damping and drives the system toward long-lived, quasi-resonant behavior. The scattering sector follows the same potential-barrier physics: larger effective barriers suppress transmission and low-frequency absorption, while the Gauss--Bonnet coupling has a comparatively mild impact over the stable range. These results provide a compact baseline for massive-field spectroscopy in higher-curvature black-hole backgrounds.

This article provides a gentle, visual introduction to the basic concepts of differential geometry appropriate for students familiar with special relativity. Visual methods are used to explain basics of differential geometry and build intuition for all types of Riemannian and Lorentzian manifolds of constant curvature. A visual derivation of the Thomas precession is given, showcasing the utility of differential geometry while also pointing a spotlight at certain intricacies of Minkowski space crucial from a pedagogical perspective. In addition, a straightforward method to generate some Carter-Penrose diagrams -- suitable for students with no differential geometry knowledge -- is presented, and a new method of indicating distortion on spacetime diagrams is shown.

Our electronic structure theory for crystalline solids is commonly built on the periodic potential assumption $V(\mathbf r)=V(\mathbf r+\mathbf R)$ for every lattice translation $\mathbf R$, enabling Bloch eigenstates, crystal momentum as a good quantum number, and the standard quasiparticle-based description of the behavior of metals. Because the zero-point motion of the ions, however, in correlated electron systems the electronic environment experienced by an itinerant electron is neither static nor self-averaging at the single-particle level, even in perfectly stoichiometric crystals, leading to a distribution of local Kondo scales that spans two orders of magnitude in temperature. We discuss, through a comparison between uniform scenarios and one that breaks with perfect lattice translational symmetry, how incorporating this distribution yields a unified description for all heavy-fermion systems at the quantum critical point.

A recently introduced recurrence-relation ansatz applied to the Bose-Hubbard model is here used in the generalized Aubry-Andre model. The resulting modified Aubry-Andre model allows for a simple parametrization of the solutions in terms of three parameters, viz., the system energy when the quasiperiodicity amplitude Delta = 0, the site mu where the particle is initially localized, and the tuning parameter -1 < alpha < 1 that determines the regions of localized or extended states. The standard Aubry-Andre form corresponds to alpha = 0.

The performance of quantum processors in the noisy intermediate-scale quantum (NISQ) era is severely constrained by environmental noise and other uncertainties. While the recently proposed quantum control robustness landscape (QCRL) offers a powerful framework for generating robust control pulses for parametric gate families, its application has been practically restricted to quasi-static noise. To address the spectrally complex, time-varying noise prevalent in reality, we propose filter function-enhanced QCRL (FF-QCRL), which integrates filter function formalism into the QCRL framework. The resulting FF-QCRL algorithm minimizes a generalized robustness metric that faithfully encodes the impact of stochastic processes, enabling robust pulse-family generation for parametric gates under realistic time-varying noise. Numerical validation in a representative single-qubit setting confirms the effectiveness of the proposed method.

The partial trace is commonly introduced in quantum mechanics as an algebraic operation used to define reduced states of composite systems. However, the probabilistic origin of this operation goes systematically unnoticed in the literature. Here, we show that the partial trace emerges naturally from the requirement of consistency between the Born rule for measurement probabilities and the classical marginalization of probability mass functions. Starting from the classical marginalization rule relating joint and marginal probability distributions, we impose that the reduced density operator of a subsystem must reproduce the local measurement statistics derived from the global state. We show that this requirement directly leads to the standard expression of the partial trace. From this perspective, the reduced density operator appears not as an ad hoc algebraic construction, but as a natural consequence of the probabilistic structure of quantum mechanics.

The Talbot effect -- a near-field diffraction phenomenon in which a periodic wavefront self-images at regular distances -- can be transposed to the time--frequency domain via the space--time duality between diffraction and dispersive broadening. We exploit this analogy to define the time--frequency (TF) Talbot effect and show that it implements different Clifford operations on TF Gottesman-Kitaev-Preskill (TF-GKP) qubits (Phys. Rev. 102, 012607), a class of qubit states encoded in the discretised frequency and time-of-arrival degrees of freedom of entangled photon pairs, whose logical basis corresponds to even and odd components of an entangled frequency combs. These states are intrinsically robust against small frequency and temporal displacements, which can be further corrected by linear or nonlinear quantum error-correction schemes. We analyse the role of the comb envelope and peak width relative to the free spectral range, and show that a compromise must be made between the gate fidelity of the Clifford gates induced by TF-Talbot operation and the error-correction capacity of the code. We then demonstrate that the signature of the TF-Talbot effect is directly accessible via the generalised Hong-Ou-Mandel interferometer: all six logical GKP states can be unambiguously distinguished by introducing a frequency shift of half the comb periodicity in one interferometer arm. We conclude with a feasibility analysis based on current experimental technology, identifying the comb finesse as the key figure of merit for both gate performance and correctability. This conclusion extends naturally to quadrature GKP states, where a shear in quadrature phase space is precisely a Talbot effect.

We provide a set of general tools to study the problem of stellar equilibrium in any gravitational theory in which spherically symmetric spacetimes satisfy master field equations taking the form of an equality between an identically conserved tensor, with derivatives of up to second order in the metric, and an identically conserved matter tensor. We derive the most general expression for the Tolman--Oppenheimer--Volkoff equation of stellar equilibrium that is compatible with these minimal requirements. A general discussion of the conditions that guarantee geodesic completeness at the center of symmetry is also presented. The equations of stellar equilibrium are integrated in a subset of the space of allowed deformations of general relativity proposed by Ziprick and Kunstatter, allowing us to illustrate universal aspects associated with the weakening of the strength of gravity, such as the mitigation of the Buchdahl limit obtained in general relativity or the existence of static solutions describing regular black holes with perfect fluid cores.

In projected descriptions of quantum dynamics, the importance of an eliminated degree of freedom is routinely assessed by deleting it and measuring the system's response. This conflates two effects: the channel's intrinsic contribution and the reorganization of the surviving model space. Here we disentangle them in continuum-discretized coupled-channels (CDCC) scattering, decomposing the Feshbach dynamic polarization potential (DPP) channel by channel while keeping the full Green's function intact, and comparing with conventional bin-deletion from the coupled equations. For $d$+$^{58}$Ni the two approaches reproduce the same elastic $S$-matrix to 0.45\%, yet a channel ranked first by one diagnostic is ranked fifth by the other. A frozen-basis protocol, zeroing couplings without reducing the basis, yields rankings that track the DPP closely ($ρ_{\rm DPP,frozen} = 0.94$) and are uncorrelated with standard deletion ($ρ_{\rm frozen,del} = -0.37$), establishing that the discrepancy is dominated by model-space reorganization. Pairwise analysis reveals quantum anti-synergy: adjacent channels partially cancel through off-diagonal Green's-function coherence, in all 10 tested pairs by the DPP and 8 of 10 by deletion. The asymmetry between excluding a degree of freedom from the effective interaction and deleting it from the model space is algebraic and general; basis-preserving decoupling, implementable in any coupled-channel code, isolates the reorganization component.

The detection of gravitational waves from a core-collapse supernova in the Milky Way or its vicinity represents a unique opportunity to probe the inner workings of these explosions. In this review, I briefly summarize our current understanding of the supernova explosion mechanism and then outline the physical processes that shape the supernova gravitational wave signal. The review highlights how the various components of the signal have the potential to constrain the progenitor rotation, the proto-neutron star structure, the nuclear equation of state, the nature of hydrodynamic instabilities, and the violence of turbulent motions in the supernova core. I also highlight some open questions and uncertainties in the theory of supernova gravitational wave astronomy as well as challenges for further progress. Specifically, there is a need to develop large model databases, systematic uncertainty quantification and methods for evidence assessment to prepare for multi-messenger observations from a Galactic supernova.

Atom loss is a dominant error source in neutral-atom quantum processors, yet its correlated structure remains largely unexploited by existing quantum error correction decoders. We analyze the performance of the surface code equipped with teleportation-based loss-detection units for neutral-atom quantum processors subject to circuit-level, partially correlated atom loss and depolarizing noise. We introduce and implement a decoding strategy that exploits loss correlations, effectively converting the \textit{delayed} erasure channels stemming from atom loss to erasure channels. The decoder constructs a loss graph and dynamically updates loss probabilities, a procedure that is highly parallelizable and compatible with real-time operation. Compared to a decoder that assumes independent loss events, our approach achieves up to an order-of-magnitude reduction in logical error probability and increases the loss threshold from $3.2\%$ to $4\%$. Our approach extends to experimentally relevant regimes with partially correlated loss, demonstrating robust gains beyond the idealized fully correlated setting.

We study the consistency of several early-Universe scenarios within a framework of non-minimal effective sca\-lar--ten\-sor gravity. We show that bounce, inflation, and genesis stages are supported within the aforementioned theory. Consequently, this framework can serve as a viable model of the early Universe, where accelerated expansion is driven by the theory's own intrinsic degrees of freedom. Notably, the theory also provides two different values of the Hubble parameter, potentially explaining the different values of the Hubble constant measured from galaxy clusters and relic radiation, respectively.

We investigate how Restricted Boltzmann Machines (RBMs) encode antiferromagnetic order when trained as variational ansätze for one-dimensional Heisenberg spin rings with periodic boundary conditions. Through systematic hidden unit analysis and ablation studies on $N=4$ and $N=8$ spin systems, we show that individual hidden units spontaneously specialize to capture staggered magnetization patterns characteristic of antiferromagnetic ground states. Hidden units naturally segregate into two classes: those essential for ground-state energy and correlation structure, and supplementary units providing smaller corrections. Removing important units induces clear energy penalties and disrupts the staggered correlation pattern in $C_{zz}(r)$, whereas removing supplementary units has modest effects. Single-unit analysis confirms that no individual hidden unit reproduces the full antiferromagnetic correlations, indicating that quantum order emerges through collective encoding across the hidden layer. Extending this analysis to $N=8$ through $20$ with hidden unit densities $α= 2$ to $5$ and ten independent seeds per configuration, we find that the fraction of important hidden units decreases with system size, consistent with sublinear growth $m' \sim N^k$ ($k \approx 0.4$). The energy-correlation impact relationship persists for small to moderate system sizes, though it weakens for the largest systems studied. These results provide a quantitative framework for RBM interpretability in quantum many-body systems.

We derive expressions for the first and second derivatives of the quintessence potential $V(φ)$, in terms of $λ= -V^{\prime}/V$ and $Γ= (V^{\prime \prime}/V)/(V^\prime/V)^2$, as functions of the quintessence density fraction $Ω_φ$ and the cosmographic parameters $q$, $j$, and $s$. Our mapping is not explicitly a function of the equation of state parameter $w$. We use these results, along with recent observational data, to derive expansions of $V(φ)$ about the present-day value of the scalar field, $φ_0$.

This study provides an analytic and numerical characterization of a class of regular, asymptotically flat black holes described by a deformed static spherical metric. The model is grounded in a four-dimensional non-polynomial quasi-topological framework in which higher-curvature corrections remain dynamically nontrivial while the static spherical sector retains a reduced-order structure, enabling tractable black-hole solutions with regular cores. Starting from the existence conditions of horizons and regularity, the allowed parameter domain and the extremal bound are derived. Hawking temperature, shadow radius, photon-ring Lyapunov exponent, and ISCO binding efficiency are then analyzed across the physically allowed parameter space. We further extend the analysis to Novikov--Thorne thin-disk accretion by deriving the flux kernel, effective-temperature profile, and bolometric luminosity scaling, and by providing representative numerical datasets for these quantities. A coherent trend emerges: increasing the deformation parameter drives the solution away from Schwarzschild behavior, reducing temperature, shadow size, and photon-orbit instability rate while enhancing orbital binding efficiency and accretion luminosity; increasing the exponent $ν$ suppresses deformation effects and restores Schwarzschild-like observables. These results provide a compact phenomenological map linking horizon structure, thermodynamics, optical signatures, dynamical instability, and thin-disk accretion diagnostics in this regular black-hole family.

The prevalence of quantum crosstalk is an important barrier to scaling frequency-addressable qubit architectures, with dynamic crosstalk being particularly difficult to detect and suppress. This form of crosstalk refers to unintended interactions driven by the gate control fields themselves. Here, we minimize dynamic crosstalk using quantum optimal control based on the perfect entangler spectrum, where spectral peaks signal unwanted entanglement with spectator qubits. Focusing on parametric gates in tunable coupler systems, we derive pulse shapes that eliminate dynamic crosstalk. Remarkably, only minimal pulse modifications are required to mitigate the form of crosstalk that is otherwise most difficult to predict. The ability to suppress dynamic crosstalk via the perfect entangler spectrum establishes a generalizable control principle for eliminating unwanted interactions in quantum hardware.

We prove a globally hyperbolic spacetime with locally Lipschitz continuous metric and timelike distributional Ricci curvature bounded from below obeys the timelike measure contraction property. The remarkable class of examples of spacetimes that are covered by this result includes impulsive gravity waves, thin shells, and matched spacetimes. As applications, we get new comparison theorems for Lipschitz spacetimes in sharp form: d'Alembert, timelike Brunn-Minkowski, and timelike Bishop-Gromov. Under appropriate nonbranching assumptions (conjectured to hold in even lower regularity), our results also yield the timelike curvature-dimension condition, a volume incompleteness theorem, as well as exact representation formulas and sharp comparison estimates for d'Alembertians of Lorentz distance functions from general spacelike submanifolds. Moreover, we establish the sharp timelike Bonnet--Myers inequality ad hoc using the localization technique from convex geometry. Alongside, we prove a timelike diameter estimate for spacetimes whose timelike Ricci curvature is positive up to a "small" deviation (in an $L^p$-sense). This adapts prior theorems for Riemannian manifolds by Petersen-Sprouse and Aubry to Lorentzian geometry, a transition the former two anticipated almost 30 years ago.

The standard weak lensing formalism assumes that the lensing map relating the observed image of a source to its intrinsic shape depends only on the deflection angle. We show that this description is incomplete beyond linear perturbation theory, even when only scalar perturbations are present at first order. Using the Jacobi map formalism, we derive expressions for the rotation field, shear B-modes, and their angular power spectra at second order in relativistic perturbation theory. In the standard formalism, rotation and shear B-modes share the same spectrum, however, this degeneracy is broken once the parallel transport of the Sachs basis is consistently taken into account. We quantify this correction numerically, finding a difference of about $5\%$ on large angular scales $\ell \sim 5$ for sources at redshift $z_\mathrm{s} = 0.5$. We also investigate frame-dragging effects, which are usually neglected in weak lensing. We present the first analytical derivation of the corresponding impact on the angular power spectrum of shear B-modes and show that it becomes the dominant contribution on scales $\ell \lesssim 10$. While both Sachs-basis rotation and frame dragging significantly affect shear B-modes on large scales, their effect on the observed galaxy ellipticity is of order $1\%$, making these nonlinear relativistic corrections challenging to detect in practice. Our results are supported by relativistic simulations of weak lensing observables, including the first numerical study of frame dragging in the power spectra of the lensing convergence and cosmic shear.

Previous analyses of shocks generated by shell-crossing singularities are affected by inappropriate gauge choices, and no systematic method is available for selecting a consistent gauge. In this work, we develop such a method for effective quantum gravity coupled to a dust shell. We illustrate it in classical general relativity and verify it numerically, with results in agreement with the Israel junction condition. We also show that gauges such as the Painlevé-Gullstrand and Schwarzschild ones are incompatible with the presence of a dust shell when imposed on the whole spatial slice. This explains the difficulties in previous treatments. The framework developed here provides a basis for studying shell-crossing singularities and shock dynamics in generally covariant effective black-hole models.

We revisit scalarized black holes in Einstein-scalar-Gauss-Bonnet gravity and analyze the thermodynamic phase transition between the Schwarzschild solution of general relativity and scalarized black holes. Restricting to spherically symmetric configurations, we investigate several classes of scalar-Gauss-Bonnet coupling functions. For the simplest quadratic coupling that triggers spontaneous scalarization, the scalarized solutions are thermodynamically disfavored and no phase transition occurs. For an exponential coupling, the phase structure depends strongly on the coupling parameter, allowing for the absence of a transition, a continuous second-order transition, or a discontinuous first-order transition. For couplings leading to purely nonlinear scalarization, we find either a first-order transition or no transition. These results reveal a rich phase structure of scalarized black holes controlled by the scalar-Gauss-Bonnet coupling.

This study explores the bouncing solutions within the framework of modified $f(Q, L_m)$ gravity. We examine four prominent bouncing models, the symmetric bounce, super bounce, oscillatory bounce, and matter bounce, each of which has been extensively analyzed in the context of modified gravity theories. Our investigation focuses on the behavior of the Hubble parameter, the evolution of the scale factor, and the equation of state (EoS) parameters. Notably, the dynamics of the scale factor and Hubble parameter effectively support the bouncing scenario. During the bouncing epoch, the EoS parameters fall within the phantom region, reinforcing the viability of the bounce. To further validate the bouncing scenario, we assess the energy conditions associated with each model. Our findings reveal a violation of the null energy condition at the bouncing epoch, which successfully characterizes the model's bouncing behavior.

We investigate the entanglement entropy of a massive scalar field using the spherical shell lattice model introduced by Das and Shankaranarayanan. A systematic numerical analysis is performed to study the dependence of the entropy on the field mass and on the size of the entangling region for both ground and excited states. For the ground state, we find that the entanglement entropy is exponentially suppressed by the field mass, reflecting the presence of a finite correlation length, while the geometric area-law scaling remains robust for all masses. For localized excited states, however, we uncover a qualitatively different behavior. The excess entropy does not exhibit universal scaling in the dimensionless variable mR. Instead, numerical results show that data points with identical mR but different (m,R) pairs do not collapse onto a single curve, demonstrating a clear violation of simple scaling. This breakdown is traced to the presence of an additional length scale associated with the finite width of the wave-packet excitation. This result identifies the coexistence of multiple infrared scales as a key feature of excited-state entanglement in massive quantum field theories. Mutual information provides an additional finite diagnostic of correlations in the chosen nested geometry. The numerical results show a strong dependence on the field mass, although the detailed behavior is sensitive to the geometric setup used in the calculation. These findings clarify how particle mass and excitation structure jointly determine entanglement properties, and suggest that the matter contribution to the generalized entropy in semiclassical gravity may depend on independent infrared parameters rather than on a single correlation scale. Implications for black hole entropy and the island formula are briefly discussed.

We investigate the entanglement dynamics of two giant atoms coupled to a common waveguide. By introducing additional phase modulation at each coupling point, every photon propagation path is jointly controlled by two distinct coupling phases, enabling precise and flexible manipulation of the entanglement evolution. This phase engineering induces destructive interference among different paths, leading to entanglement dynamics in nested giant atoms that become equivalent to those of small atoms, as well as dynamical equivalence between separated and braided configurations. Furthermore, the proposed scheme significantly enhances the robustness of entanglement against variations in the phase shift, offering a practical route to generate stable entanglement and enabling quantum devices with programmable propagation and controllable memory effects.

Miniaturized quantum light sources that operate directly in optical fibers are an attractive platform for optical quantum technologies. However, most miniaturized spontaneous parametric down- conversion (SPDC) sources still rely on objective-lens-based free-space pumping and collection, which limits compactness, robustness, and direct compatibility with fiber-based systems. Here we demonstrate a lens-free in-line SPDC photon-pair source by integrating a van der Waals NbOI2 flake directly onto the end facet of an optical fiber. In this configuration, the generated photon- pairs are efficiently collected into optical fibers, eliminating the need for bulk free-space collection optics. Despite the limited numerical aperture of the single-mode fiber, efficient photon-pair collection with high purity, characterized by a coincidence-to-accidental ratio of up to ~4600, is achieved in an ultracompact configuration. These results establish van der Waals ferroelectric materials as a promising platform for fiber-integrated quantum light sources and provide a pathway toward compact, alignment-free quantum photonic devices.

Quantum circuits that generate coherent superpositions of stochastic processes are key to many downstream quantum-accelerated tasks, such as risk analysis, importance sampling, and DNA sequencing. However, traditional methods for designing such circuits from data face immense challenges, given the exponential growth in the size of the associated probability vectors as the desired simulation time horizon increases. Here, we introduce quantum sequence models that leverage a recurrent quantum circuit structure to generate coherent superpositions with circuit complexity that grows linearly with the desired time horizon; together with a recurrent variant of the parameter-shift rule, we train these models from observational data. When benchmarked against baseline quantum Born machines, our constructions exhibit orders-of-magnitude improvements in model accuracy in data-sparse regimes.

Exceptional points (EPs) in non-Hermitian systems give rise to enhanced sensitivity and chiral state transfer, which are important for quantum technologies. Although parameter trajectories encircling EPs can control symmetric and chiral state transfer, their robustness against practical perturbations and their role in quantum sensing remain largely unexplored. Here, we study three time-modulated parameter loops in a non-Hermitian two-level system to show how trajectory design governs state-transfer symmetry, robustness, and sensing performance. Trajectories avoiding the EP support robust symmetric transfer, while those encircling the EP yield chiral transfer governed by the topological winding number, whose robustness depends on the distance to the EP and the encircling direction. For quantum sensing, trajectory engineering enables tuning of sensitivity amplitude, time window, and parameter selectivity in both eigenvalue-based and eigenstate-based sensors. Notably, eigenstate-based sensing achieves full parameter selectivity that is unattainable with eigenvalue-based methods. Our results establish a quantitative connection between trajectory topology and system dynamics, providing a unified framework for robust state-transfer protocols and high-performance quantum sensors.

We suggest a tight inequality for norms in $d$-dimensional space $l_p $ which has simple formulation but appears hard to prove. We give a proof for $d=3$ and provide a detailed numerical check for $d\leq 200$ confirming the conjecture. We conclude with a brief survey of solutions for kin problems which anyhow concern minimization of the output entropy of certain quantum channel and rely upon the symmetry properties of the problem. Key words and phrases: $l_p $-norm, Rényi entropy, tight inequality, maximization of a convex function.

Nonadiabatic holonomic quantum computation (NHQC) offers intrinsic resilience to certain control imperfections. However, conventional nonadiabatic holonomic protocols are constrained by the fixed-pulse-area condition, which limits flexibility and prolongs duration of small-angle gates. Here we experimentally demonstrate a universal brachistochrone nonadiabatic holonomic quantum gate scheme in a trapped 40Ca+ ion, and realized the construction of pX gate under the conventional NHQC, brachistochrone NHQC (BNHQC) and composite BNHQC (CBNHQC) protocols. By characterizing the performance of gate performance in the presence of dissipation, Rabi-frequency errors and detuning errors, we show that BNHQC and CBNHQC outperform conventional NHQC, and BNHQC can offer a favorable balance between operation speed and robustness. It further shows that keeping high fidelity and strong robustness need decrease the accumulated population of excited state in the evolution process. These results highlight nonadiabatic holonomic computation as a practical route toward fast and robust quantum gates in trapped-ion platforms.

We simulate the thermalization dynamics for minimally truncated SU(2) pure gauge theory on linear plaquette chains with up to 151 plaquettes using IBM quantum computers. We study the time dependence of the entanglement spectrum, Rényi-2 entropy and anti-flatness on small subsystems. The quantum hardware results obtained after error mitigation agree with extrapolated classical simulator results for chains consisting of up to 101 plaquettes. Our results demonstrate the feasibility of local thermalization studies for chaotic quantum systems, such as nonabelian lattice gauge theories, on current noisy quantum computing platforms.

Quantum computers, currently in the noisy intermediate-scale quantum (NISQ) era, have started to provide scientists with a novel tool to explore quantum physics and chemistry. While several electronic systems have been extensively studied, Frenkel excitons, as prototypical optical excitations, remain among the less-explored applications. Here, we first use variational quantum deflation to calculate the eigenstates of the Frenkel Hamiltonian and evaluate the observables based on the oscillator strength for each eigenstate. Furthermore, using NISQ quantum computers requires performing error mitigation techniques alongside simulations. To deal with noisy qubits, we developed a deep-learning-based framework combined with a post-selection technique to learn the noise pattern and mitigate the error. Our mitigation methods work well and outperform the conventional post-selection and remain valid on real hardware.

We develop a comprehensive thermodynamic description for a zero-temperature boson gas in curved spacetime, integrating energy conservation with information-theoretic principles. Using the hydrodynamic Madelung representation within the ADM formalism, we establish two fundamental relationships: an energy balance equation representing the first law of thermodynamics from a spacetime perspective, and an information-theoretic constraint connecting Fisher entropy to the dynamical evolution of the boson density. This dual formulation clearly separates energy transport from information conservation while revealing how quantum information is preserved in curved backgrounds. The introduction of a stochastic velocity provides a bridge between quantum potential effects and underlying spacetime fluctuations, suggesting a gravitational basis for quantum stochastic behavior. We demonstrate the consistency of our framework through detailed analyses of quantum systems in both Minkowski and Schwarzschild spacetimes. This work provides a unified foundation for studying relativistic bosonic systems, with direct relevance to boson stars and scalar field dark matter models.

Graph states are a key resource for measurement-based quantum computation and quantum networking, but state-preparation costs limit their practical use. Graph states related by local complement (LC) operations are equivalent up to single-qubit Clifford gates; one may reduce entangling resources by preparing a favorable LC-equivalent representative. However, exhaustive optimization over the LC orbit is not scalable. We address this problem using the split decomposition and its quotient-augmented strong split tree (QASST). For several families of distance-hereditary (DH) graphs, we use the QASST to characterize LC orbits and identify representatives with reduced controlled-Z count or preparation circuit depth. We also introduce a split-fuse construction for arbitrary DH graph states, achieving linear scaling with respect to entangling gates, time steps, and auxiliary qubits. Beyond the DH setting, we discuss a generalized divide-and-conquer split-fuse strategy and a simple greedy heuristic for generic graphs based on triangle enumeration. Together, these methods outperform direct implementations on sufficiently large graphs, providing a scalable alternative to brute-force optimization.

Flavor oscillations in curved space-time provide a novel channel to explore the unknown parameters of neutrinos. In this work, the gravity-modulated CP violations (CPVs) in neutrino oscillations were investigated under the Reissner-Nordstrom, Hayward, and Simpson-Visser metric. The interplay among the CPV, the properties of neutrinos, and the space-time is illustrated with analytical and numerical methods. The morphologies of the flavor-oscillation curves show that the information on the mass-ordering, the absolute mass, and the gravitational parameters could be encoded into the amplitudes and periods of the CPVs. Hence, the characteristic of the space-time background may be identified through its modulation effects on the CPV, such as amplification, damping on the amplitudes.

The mathematical rules used to handle systems of identical quantum particles bring into question whether the elementary constituents of matter, such as electrons, have the fundamental characteristics of persistence and reidentifiability that are usually attributed to classical particles. However, despite considerable philosophical debate, the metaphysical profile of these entities remains elusive. Previous debates have taken the mathematical rules, and the language in which these are usually couched, as a starting point. Here, we argue that this methodology is inherently limited, and develop a new conception of identical particles based on a recent mathematical reconstruction of these rules. Using this reconstruction, we demonstrate that the special behaviour of identical particles originates in the confluence of identicality and the active nature of the quantum measurements. We propose that identical particles are appropriately viewed as potential parts of a whole, and show how this leads to striking consequences such as restricted transtemporal identity.

Dephasing noise is a ubiquitous source of decoherence in current atomic sensors. We address the problem of entanglement-assisted frequency estimation subject to classical dephasing noise with full spatial correlations (collective) and arbitrary temporal correlations. Our contributions are threefold. (i) We derive rigorous, state-independent bounds on the achievable estimation precision, showing how they are entirely determined by the short-time behavior of the decoherence function. For temporally uncorrelated (Markovian) dephasing, precision is limited by a probe-independent constant. For temporally correlated stationary noise, the bound approaches the noiseless limit for classical states, precluding any asymptotic quantum advantage. (ii) We show that these scaling bounds are tight, by constructing generalized Ramsey protocols that saturate them. These optimal protocols use squeezing at the input and before readout, both of which are available in state-of-the-art atomic interferometers. Implementing a perfect-echo protocol, which reaches Heisenberg scaling in the absence of noise, remains optimal in this noisy setting, irrespective of the noise temporal correlations. (iii) We prove that arbitrary collective open-loop control cannot lift the no-go for super-classical precision scaling under either Markovian or colored stationary noise, highlighting the detrimental nature of full spatial correlations. In the latter case, temporal correlations may nonetheless enable constant-factor improvements over the standard quantum limit, which may still be important in practical metrological scenarios.

We introduce the architecture and timing algorithm to realize a time-bin-encoded quantum photonic neural network (QPNN): a reconfigurable nonlinear photonic circuit inspired by the brain and trained to process quantum information. Unlike the typical spatially-encoded QPNN, time-encoded networks require the same number of photonic elements (e.g. phase shifters or switches) regardless of their size or depth. Here, we present a model of such a network and show how to include imperfections such as losses, routing errors and most notably distinguishable photons. As an example, we train the QPNN to realize a controlled-NOT gate, based on a hypothetical ideal Kerr nonlinearity. We then extend our model to a realistic two-photon nonlinearity due to scattering from a single, semiconductor quantum dot coupled to a photonic waveguide. We show that, using this realistic nonlinearity, the QPNN can be trained to act as a Bell-state analyzer which operates with a fidelity of 0.96 and at a rate only limited by losses. We further show that time gating can raise this fidelity to over 0.99, while still maintaining an efficiency exceeding 0.9. Overall, this work lays a framework for the first QPNN encoded in time, and provides a clear path to the scaling of these networks.

A quantum particle on a circle in a quadratic potential exhibits a spectrum that is not harmonic, despite having all algebraic properties of the quantum harmonic oscillator. This raises the question where the usual algebraic argument -- implying integer gaps -- fails. The answer is illuminating and covers a surprisingly rich range of physical phenomena for such a simple model.

We study non-perturbative quantum gravitational corrections to the thermodynamics and quantum work distribution of the $n$-dimensional Schwarzschild--Tangherlini--Anti-de Sitter black hole. Starting from the corrected entropy $S = S_0 + η\, e^{-S_0}$, where $S_0$ is the Bekenstein--Hawking entropy, we derive the modified specific heat, internal energy, Helmholtz free energy, and Gibbs free energy in closed form. The specific heat retains the classical divergence at $r_h^{*}=l\sqrt{(n-3)/(n-1)}$ for $n\geq 4$, but the quantum correction suppresses its magnitude by up to $78\%$ at small horizon radii. In the extended phase space, the uncharged black hole admits no van der Waals critical point; however, the non-perturbative correction induces a Hawking--Page transition for $n\geq 4$ that is absent in the semi-classical limit. The corrected Gibbs free energy turns negative at small $r_h$, opening a thermodynamic channel with no classical counterpart. Using the Jarzynski equality and Jensen inequality, we obtain the quantum work distribution during evaporation. The free energy difference $ΔF$ between two black hole states undergoes a sign reversal at small horizon radii for $n\geq 4$ when $η=1$, flipping the average quantum work from negative to positive. This sign reversal grows with the spacetime dimension, reaching $\langle W\rangle \approx +4.31$ for $n=10$. These findings demonstrate that non-perturbative quantum gravitational effects qualitatively alter the phase structure and evaporation energetics of AdS black holes, and they cannot be captured by perturbative corrections alone.

Recent comparisons of quantum repeater protocols have highlighted the strong near-term potential of multiplexed two-way architectures for long-distance quantum communication. At the same time, advances in hollow-core fiber (HCF) technology motivate a re-examination of the physical transmission medium as an architectural lever in quantum network design. In this work, we compare emerging anti-resonant HCFs against conventional silica single-mode fibers (SMFs) in multiplexed two-way quantum repeater networks. We evaluate their performance under both telecom and memory-native transmission, accounting for frequency-conversion overheads, coupling efficiencies, memory decoherence, and operational noise. We find that HCF significantly outperforms SMF across a wide range of regimes. With memory-native transmission, HCF yields up to an order of magnitude improvement in secret-key rate per channel use under realistic conversion efficiencies. Even at telecom wavelengths, HCF enables larger optimal repeater spacing, improving rate--cost tradeoffs and reducing repeater requirements. We further quantify the role of memory quality, hardware efficiency, detector and conversion losses, and two-qubit gate noise in shaping these gains. These results show that recent advances in HCF materially expand the design space of practical terrestrial quantum repeater networks.

We analyze various gravity theories involving de-Sitter, quadratic $\mathcal{R}^2$ and non-minimally coupled scalar in the light of application of the Dyson-Schwinger technique involving exact background solution of the Green's function. We denote specific set of solutions for the metric to move towards a quantum analysis of the theory. This kind of solutions is identified as conformally flat metric. Such a conclusion naturally arises in the use of the Dyson-Schwinger equations in the study of the Yang-Mills theory through the mapping theorem. We show a sequence of cosmological phase transitions starting from the breaking of such conformal invariance that can be hindered by the presence of the non-minimal coupling.

Recently, for the first time, a metric of a static spherically symmetric generalized uncertainty inspired quantum black hole was derived. We apply the modified Newman-Janis algorithm to this metric and derive its rotating counterpart. We show that this metric has all the correct limits, while due to Newman-Janis side effects, the singularity which was resolved in the static case, is introduced back into the model. However, the slowly-rotating limit of this black hole is singularity-free. Furthermore, we show that the presence of quantum parameters modifies the location of the horizons, temperature, and entropy of the black hole, and allows the existence of naked singularities even if the ratio of the spin parameter to mass of the black hole is less than unity. Finally, by computing the shadow parameters of this black hole and comparing them with data from the Event Horizon Telescope for both M87* and Sgr A*, we set bounds on one of the quantum parameters of the model, and show that there is a limit on the angular momentum of M87* if this model is valid.

We present a comprehensive and self-contained framework for the efficient classical simulation of Clifford circuits acting on $d$-dimensional qudits, including realistic Pauli/Weyl noise via stochastic simulation. Our approach uses the stabilizer tableau formalism for qudits of arbitrary dimension and tracks both stabilizer and destabilizer generators under Clifford updates. The classical simulation remains efficient with simple algebraic Clifford update rules over $\mathbb{Z}_d$. Computational basis measurements in prime dimensions are handled by a generalized Aaronson-Gottesman (CHP) procedure. In composite dimensions, $\mathbb{Z}_d$ is not a field and the standard tableau reduction fails, so we employ an exact Smith normal form decomposition to enable efficient sampling. Noise is modelled as probabilistic mixtures of Weyl operators that act only on the tableau's phase column. For fast simulation of noisy circuits, we support Pauli frames, respectively generalized Weyl frames, and introduce a noise-pushing technique that allows all noise processes to be consolidated into a single phase update at the end of the circuit. Using this representation, circuit fidelity can be determined entirely by the single accumulated phase-shift parameter $Δτ$, reducing fidelity estimation to a simple phase check per shot. Our codebase supports tableau simulation and conventional state-vector and density-matrix backends for qudits, and also includes circuit and tableau visualisations. Additionally, we provide tests and Jupyter notebooks for validation and illustration. This framework forms the basis for a scalable, open-source strong+weak stabilizer simulator including noise and can be found publicly available at https://github.com/QUICK-JKU/QuickQudits.

We show that dynamical dark energy theories can imprint $O(1)$ modifications on the quasi-normal mode (QNM) spectrum characterising black hole ringdown. The time dependence of dynamical dark energy naturally gives rise to cosmological 'hair' around a black hole. Taking the cubic Galileon as a concrete example, which admits the only known stable solution of this kind, we parametrically connect the cosmological and black hole regimes, derive the induced QNM shifts and forecast the resulting dark energy constraints. We find that the dark energy field profile can be constrained with an accuracy of up to $10^{-2}$ for LVK and $10^{-4}$ for LISA.

State cloning and state transposition are fundamental transformations which, despite being desirable, cannot be perfectly implemented in quantum theory. In this work, we determine the optimal approximation for transforming $N$ qudits into $K$ copies of their transposition and prove the optimal fidelity for pure input states. We further show that the optimal qudit \(N\!\to\!K\) transposition map is the complementary channel of the optimal universal symmetric \(N\!\to\!N\!+\!K\) quantum cloning machine on pure states, implying that both tasks can be simultaneously realised by the same quantum operation and attain the same optimal fidelity. We then present an explicit quantum circuit that simultaneously implements optimal \(N\!\to\!K\) transposition and \(N\!\to\!N\!+\!K\) cloning and discuss its gate efficiency. Lastly, we investigate mixed-state \(N\!\to\!1\) qudit transposition and find the optimal performance in terms of white noise visibility, yielding the optimal structural physical approximation of state transposition in the multicopy regime.

Oscillation modes of neutron stars, a key target for third-generation gravitational wave detectors, encode key information about their constituent nuclear matter. In this work, we study the effect of viscosity on oscillations of cold, polytropic, spherically symmetric neutron stars. We focus on purely radial oscillations and work perturbatively to linear order within two hydrodynamic frameworks: the acausal covariant generalization of the Navier-Stokes equations proposed by Eckart, and the causal generalization formulated by Bemfica, Disconzi, Noronha, and Kovtun (BDNK). We find that viscosity damps the radial modes on millisecond timescales and induces fractional shifts in the oscillation frequency which increase both with the compactness and viscosity of the star, reaching up to the percent level for the fundamental mode with bulk viscosities $ζ\sim10^{30}\mathrm{g}/\mathrm{cm}/\mathrm{s}$. For more viscous stars, the oscillation frequency decreases, becoming zero (i.e., an overdamped mode) for $ζ\gtrsim10^{31}\mathrm{g}/\mathrm{cm}/\mathrm{s}$. We also study the linear threshold of gravitational collapse. Consistent with recent analytic results in the small viscosity regime, we find that viscosity in Eckart theory cannot stabilize an unstable inviscid star. We provide numerical evidence that viscosity in BDNK theory is similarly unable to prevent gravitational collapse, but it slightly modifies the threshold of collapse. Overall, our results advance our understanding of the impact of viscosity on the oscillation modes of neutron stars, a key component of viscous asteroseismology with next-generation gravitational wave detectors.

We present the first joint gravitational-wave cosmological inference with LISA extreme mass-ratio inspirals at $z\lesssim1$ (galaxy redshifts) and massive black hole binaries at $z\gtrsim1$ (electromagnetic counterparts). Combining these standard sirens reduces cosmological degeneracies and yields competitive constraints on the Hubble constant $H_0$ and the dark-energy equation-of-state parameter $w_0$. This highlights LISA's potential for late-time cosmology across a broad redshift range with systematics distinct from electromagnetic distance indicators.

A key aspect in quantum information is to understand the advantage offered by quantum systems over classical ones in communication tasks. In recent years, a fundamental approach to this problem has been developed, focusing on quantum correlations in prepare-and-measure scenarios. Inspired by the developments in Bell nonlocality and device-independent information processing, this line of research aims to characterize the possibilities and limits of quantum systems for communication, in particular to precisely capture the advantage they offer over classical systems. In addition to fundamental insights, these ideas also underpin the concept of semi-device-independent quantum information processing. Exploring trade-offs between security, performance and ease-of-implementation, this approach opens promising directions for novel quantum information processing technologies and devices. A number of protocols and proof-of-principle demonstrations have been reported in recent years, in particular for quantum randomness certification and key distribution. Here, we provide a comprehensive introduction to quantum prepare-and-measure correlations and semi-device independent applications.

Solid-state spin defects are promising qubits for quantum network nodes. A key challenge towards larger networks is creating defects with high yield into nanophotonic devices, while maintaining good optical and spin properties. Here, we demonstrate the creation of V2 centers in nanopillars fabricated from commercial bulk-grown 4H-silicon carbide using a pulsed above-bandgap (UV) laser. We observe an eleven-fold increase in the V2 center occurrence after UV laser illumination. These laser-induced V2 centers exhibit narrow optical linewidths and spectral diffusion rates comparable to naturally occurring V2 centers in nanopillars of the same material. Furthermore, we measure a spin coherence time of $T_{2}^{\mathrm{DD}} = 3.6 \pm 0.3~\text{ms}$ under dynamical decoupling, consistent with dephasing by the nuclear-spin bath. This demonstration of the in-situ, post-fabrication generation of coherent V2 centers in nanostructures in widely available bulk-grown 4H-SiC, shows the potential for above-bandgap laser illumination for scalable defect creation in integrated photonic devices.

Given the recent interest in perspectival quantum reference frames (QRFs), we ask how quantum properties in the perspectival picture relate to their global, non-perspectival counterparts. It is instructive to establish this link, as most known results in quantum information theory are derived in the latter context. Specifically, we find sufficient conditions under which global entanglement decomposes into a combination of perspectival entanglement and coherence -- a phenomenon that we call entanglement transference. We apply this result to non-inertial QRFs, in particular, revisiting the problem of entanglement degradation. We find that entanglement degradation in the perspectival picture can be offset by an increase in coherence resources. The non-inertial problem may also provide clues to understanding perspectival QRFs in curved spacetime.

We propose a quantum sensing protocol for coupled qubit-oscillator systems that surpasses the standard quantum limit by exploiting a geometric phase for dark matter searches. Instead of letting the qubit evolve freely under a weak dark matter background, we combine large coherent displacements and squeezing operations within the evolution protocol, thereby mapping the signal onto an enhanced geometric phase. This new protocol increases the quantum Fisher information to surpass standard quantum limit and leads to a substantial improvement in dark photon and axion detection sensitivity, opening a new paradigm for cavity-based dark matter detection.

Quantum reference frames provide a relational description of multipartite quantum systems in which physical states and observables are defined relative to quantum observers. Yet different observers can assign different entropies to the same system, raising the question of how such observer-dependence is constrained. We identify a family of frame-independent diagonal Rényi entropies for arbitrary subsystems, yielding a generalized multipartite coherence-entanglement tradeoff. For ideal frames, the observer-dependence of subsystem entropy admits an exact decomposition into a sum of single-frame coherences and inter-frame correlations; for non-ideal frames, it is instead bounded by the dimension of an effective relational Hilbert space determined by the representation structure of the frames. Our results place quantitative limits on how much quantum observers can disagree about subsystem entropy, with potential implications for observer-dependent entropy assignments in gravitational settings.

Agreement theorems are no-go results about rational disagreement: if two agents start from a common prior and their posterior beliefs are common knowledge, they cannot assign different probabilities to the same event. Standard treatments of the result have the agents reason about an underlying state of the world, which has lead some to ask whether the result can extend to quantum or postquantum phenomena, where such a description may no longer be appropriate. We derive an operational version of Aumann's agreement theorem without assuming an objective state of the world and instead focusing only on what is observed. This allows us to establish the theorem's validity in quantum theory and even in situations with indefinite causal order or involving hypothetical postquantum phenomena. We comment on seemingly contradictory results in the literature and point to the one place where the theorem might fail: Wigner's friend-type situations.

In this paper, we prove energy and Morawetz estimates for solutions to Teukolsky equations in spacetimes with metrics that are perturbations, compatible with nonlinear applications, of Kerr metrics in the full subextremal range. The Teukolsky equations are written in tensorial form using the non-integrable formalism in \cite{GKS22}, and we follow the approach in \cite{Ma} of relying on a Teukolsky wave/transport system. The estimates are proved by extending the ideas from our earlier result \cite{MaSz24} on the corresponding problem for the scalar wave, notably the use of $r$-foliation-adapted microlocal multipliers for the wave part, and by incorporating techniques from \cite{Ma} to control the linear coupling terms between the components of the Teukolsky wave/transport system. Additionally, in order to adapt the methodology of \cite{MaSz24} to tensorial waves, we introduce a well-suited regular scalarization procedure which is of independent interest. This result, alongside our companion paper \cite{MaSz24}, is an essential step towards extending the current proof of Kerr stability in \cite{GCM1} \cite{GCM2} \cite{KS:Kerr} \cite{GKS22} \cite{Shen}, valid in the slowly rotating case, to a complete resolution of the Kerr stability conjecture, i.e., the statement that the Kerr family of spacetimes is nonlinearly stable for all subextremal angular momenta.

We investigate the potential to detect low-frequency gravitational waves (GWs) through their imprints on the proper motions of distant quasars observed by the Gaia mission. Using astrometric data from Gaia DR3, we simulate the effect of GWs on the proper motions of quasars, incorporating their actual sky positions and measurement uncertainties. We investigate two data analysis techniques for the extraction and characterization of GW signals from quasar proper motions: Vector Spherical Harmonics (VSH) and angular correlation functions, commonly referred to as Hellings-Downs curves (HDC). Using realistic simulated data, we forecast their sensitivity and accuracy to GWs, and evaluate the impact of systematic errors. From these simulations, we derive an upper limit on the amplitude of a stochastic GW background, constrained by the observational timespan, astrometric precision, and the sky distribution of quasars. Compared to HDC, VSH appears more statistically robust, less prone to selection effects, and with a significantly smaller computational cost, scaling as N. The HDC method is more sensitive for detecting gravitational waves, but its complexity scales as N^2. We find that, with Gaia DR3 proper motion errors, the lower limit for a detectable GW strain is of 10^{-11}, with possible improvements to about 3 x 10^{-12} for the next Gaia Data Release 4 (for the same number of quasars). This limit holds for a stochastic GW spectrum integrated over all frequencies less than half the inverse of the 34-month observational timespan of Gaia DR3, corresponding to approximately 5.6 nHz. We also investigate how different data-restriction and weighting schemes influence the final estimate of the gravitational wave strain.

We examine the validity of classical energy conditions in nonsingular bouncing cosmological solutions arising in quadratic curvature gravity minimally coupled to a scalar field. Focusing on the null, weak, strong, and dominant energy conditions, we perform a systematic analysis under two distinct formulations of the energy-momentum tensor. In the first approach, the energy-momentum tensor is assumed to be sourced solely by the scalar field, whereas in the second, an effective energy-momentum tensor is constructed that incorporates the higher-curvature corrections characterizing deviations from general relativity. Our results reveal that, in the scalar-field description, the null, weak, and dominant energy conditions remain satisfied throughout the cosmological evolution, while the strong energy condition is necessarily violated during the bounce phase, enabling the avoidance of the initial singularity. In contrast, when the effective energy-momentum tensor is considered, all four energy conditions are violated near the bounce, highlighting the intrinsically non-Einsteinian nature of the underlying gravitational dynamics. These findings clarify the role of higher-order curvature terms in facilitating nonsingular cosmological bounces, providing important insights into the energy condition violations required in modified theories of gravity.

In this work, we investigate the coupling of General Relativity with Nonlinear Electrodynamics (NED), governed by a general Lagrangian $\mathcal{L}(\mathcal{F})$, to address the axial singularity of four-dimensional black strings. Through a model-independent analysis, we scrutinize the viability of regular configurations by extending no-go theorems, originally formulated for spherical spacetimes, to cylindrical symmetries. We provide a comprehensive mathematical proof that regular, purely electric black strings cannot be generated by any NED Lagrangian that recovers the Maxwell limit in the weak-field regime, establishing a fundamental constraint for cylindrical topologies. Despite these limitations, we employ specific mathematical frameworks to construct new exact solutions for black strings, including cylindrical analogues of the well-known Bardeen and Hayward regular black hole classes. Each solution is analytically derived, and we demonstrate that their curvature invariants remain finite everywhere, effectively replacing the axial singularity with a regular core. Furthermore, we evaluate the physical consistency of these new metrics by subjecting them to stringent causality and unitarity constraints. Our results provide a comprehensive classification of the conditions under which NED can regularize cylindrical spacetimes and offer new insights into how topological differences between spherical and axial symmetries influence the global structure and the physical viability of non-singular gravitational objects in nonlinear gauge theories.

According to the GEVAG (Generalized Entropy Varying-G) framework, any modification to the Bekenstein-Hawking area law would also lead to a varying-$G$ gravity theory in which the effective gravitational constant $G_\text{eff}$ becomes area-dependent. Among a myriad of generalized entropy functions explored in the literature, of special interest is the logarithmic correction of quantum gravity. In this work, we apply GEVAG to investigate the effect of logarithmic correction on very early-time cosmology, including the conditions for inflation. We found that if the coefficient of the logarithmic correction term is negative, $G_\text{eff}$ becomes twice that of the current value; whereas, a positive coefficient leads to a very small value of $G_\text{eff}$, which may ameliorate the "arrow of time" problem. In fact, slow-roll inflation becomes more natural in the latter case. We make some comparisons with the constant-$G$ approach and reveal the advantages of the GEVAG approach. For example, it can evade the sudden singularity that could otherwise arise when the coefficient of the logarithmic correction term is negative. We also check the validity of the generalized second law and comment on the range of the various parameters.

Efficiently sampling from high-dimensional, multi-modal posteriors is a central challenge in Bayesian inference for astrophysics, especially gravitational-wave astronomy. Popular families of methods like Markov-chain Monte Carlo, nested sampling, and importance sampling all rely on proposal distributions to guide exploration. Because prior knowledge of the target is often limited, practitioners can adopt adaptive proposals that iteratively refine themselves using information gained from previously drawn samples. Traditional adaptive strategies, however, struggle in high-dimensional multi-modal settings: complex, non-linear correlations are hard to capture, and hyperparameters typically require tedious, problem-specific tuning. To address these issues, we introduce Parallel Adaptive Reweighting Importance Sampling (PARIS). PARIS models its proposal as a Gaussian mixture whose component centers are the existing samples and whose component weights match the current importance weights. New draws from the proposal therefore concentrate around high-weight regions, while candidate points in unexplored areas receive intentionally inflated weights. As the algorithm continuously reweights all samples up to the latest proposal, any initial over-weighting self-corrects once additional neighbor samples are collected. To enable rapid reweighting, we present an efficient update scheme and evaluate PARIS on illustrative toy problems and more realistic gravitational-wave parameter estimation tasks. PARIS achieves accurate posterior reconstruction and evidence estimation with substantially fewer function evaluations than competing approaches, highlighting its promise for widespread use in astrophysical data analysis.

We present a unified description of first-order cosmological phase transition dynamics that links the phenomenological friction model employed in hydrodynamic simulations to the microscopic treatment based on Boltzmann equations. We derive an approximate analytical expression for the chemical potential and demonstrate that the resulting friction parameter $\tildeη$ follows a simple power-law dependence on the transition strength ($\propto v_n^4/T_n^4$). Incorporating this scaling into a phenomenological framework accurately reproduces the terminal wall velocities obtained from the full microscopic analysis performed using \texttt{WallGo}. This approach offers an efficient method to quantify out-of-equilibrium contributions to friction and reliably estimate bubble-wall velocities.

Given any compact connected matrix Lie group $G$ and any lattice dimension $d\ge 2$, we construct a massive Gaussian scaling limit for the $G$-valued lattice Yang-Mills-Higgs theory in the "complete breakdown of symmetry" regime. This limit arises as the lattice spacing tends to zero and the (inverse) gauge coupling constant tends to infinity sufficiently fast, causing the theory to "abelianize" and yield a Gaussian limit. This complements a recent work by Chatterjee (arXiv:2401.10507), which obtained a similar scaling limit in the special case $G= SU(2)$.

In this note we show how the consistent truncations on half-supersymmetric branes of Leung and Stelle and Lin, Skrzypek and Stelle fit into the general exceptional generalised geometry analysis of Cassani \emph{et al.}. Each solution defines a torsion-free $Spin(n)$ structure in the $Spin(n,n)\times \mathbb{R}^+$ generalised geometry introduced by Strickland--Constable, where $n$ is the dimension of the space transverse to the brane. Embedding this into the appropriate exceptional generalised geometry then defines the truncation. As a by-product we derive a new consistent truncation on the IIA NS5-brane to six-dimensional $\mathcal{N}=(2,0)$ supergravity coupled to a tensor mutliplet, and new consistent truncations on the D6- and D7-branes to seven- and eight-dimensional pure half-maximal supergravity respectively.

We present an automated, GPU-accelerated framework for many-body perturbation theory (MBPT) calculations of the zero-temperature nuclear equation of state (EOS) based on chiral nucleon-nucleon (NN) and three-nucleon (3N) interactions. Automated diagram generation and evaluation enable the computation of all diagrams up to fifth order in the MBPT expansion at the normal-ordered two-body level in infinite matter, with residual three-body contributions explicitly included up to third order. Multi-GPU acceleration of 3N normal ordering, a novel Monte Carlo integrator (called PVegas), and further advances in high-performance computing enable us to evaluate all 840 fifth-order diagrams with controlled numerical uncertainties. We investigate the MBPT convergence up to fifth order in pure neutron matter (PNM) and symmetric nuclear matter (SNM) for two sets of chiral interactions, study neutron star matter, and present fourth-order results for asymmetric matter including normal-ordered 3N forces. The framework enables systematic MBPT studies with harder interactions and benchmarks against nonperturbative methods. It can be further extended to finite-temperature EOS calculations and to improved uncertainty quantification using emulation and resummation techniques.

In the case of a D-dimensional non-Abelian 1-form gauge theory (without any interaction with the matter fields), we show that the application of the Noether theorem does not lead to the derivations of the Becchi-Rouet-Stora-Tyutin (BRST) and anti-BRST charges that obey (i) the (anti-)BRST invariance, and (ii) the nilpotency property (despite the fact that these charges are derived from the infinitesimal, continuous and nilpotent (anti-)BRST symmetry transformations). This happens because of the presence of the {\it non-trivial} Curci-Ferrari (CF) condition on our non-Abelian theory (whose limiting case is the Abelian gauge theory where the CF-type restriction is trivial and the corresponding Noether (anti-)BRST charges turn out to be nilpotent as well as (anti-)BRST invariant together). We exploit the theoretical strength of the basic canonical approach to prove (i) the non-nilpotency of the conserved Noether (anti-)BRST charges, and (ii) the (anti-)BRST invariance of the consistently modified versions of the Noether conserved (anti-)BRST charges. We very briefly comment on the nilpotency property of the consistently modified versions of the conserved (anti-)BRST charges, too. Our present observations are very sacrosanct because they have been proven by (i) using the beauty of the symmetry properties, and (ii) exploiting the theoretical strength of the basic canonical (anti)commutators.

There is a well understood way of generating random coverings of a fixed manifold by sampling homomorphisms from the fundamental group of this manifold into the symmetric group. We prove a central limit theorem for the number of connected components of these random coverings when the fundamental group is nilpotent. This provides a nonabelian generalization of an earlier result by the author and Shannon Starr in the case of the torus where the fundamental group is a free abelian group of rank at least two. Our result relies on the work of du Sautoy and Grunewald on the subgroup growth zeta functions of nilpotent groups, and on Delange's generalization of the Wiener-Ikehara Tauberian theorem.

The NUCLEUS experiment aims to measure coherent elastic neutrino-nucleus scattering (CE$ν$NS) at unprecedentedly low nuclear recoil energies using gram-scale cryogenic calorimeters operated at the Chooz nuclear power plant in France. Access to recoil energies at the $\mathcal{O}(10~\mathrm{eV})$ scale enables CE$ν$NS studies at extremely low momentum transfer and provides enhanced sensitivity to new physics. In this work, we present sensitivity projections for the upcoming NUCLEUS technical and physics runs, incorporating a data-driven treatment of the low-energy excess (LEE) observed during commissioning. We develop a likelihood framework that exploits reactor-power variation to disentangle signal and background in a low signal-to-background regime and to assess the impact of the dominant systematic uncertainties. For the Technical Run with a 7 g CaWO$_4$ target, we find competitive sensitivity to several scenarios beyond the Standard Model, which do not require a CE$ν$NS observation. For the Physics Run, assuming complete suppression of the LEE, we project a 4.7 $σ$ observation of CE$ν$NS with a statistical precision of about 20 % in 1 year, enabling a determination of the weak mixing angle at the lowest momentum transfer probed to date with CE$ν$NS and leading CE$ν$NS-based constraints on the neutrino charge radius and new mediator models.

We report a search for the rare decay $B^{+} \rightarrow K^{+} τ^{+} τ^{-}$ using $1.2 \times 10^9$ $Υ(4S)$ mesons produced near threshold in electron-positron collisions and collected by the Belle and Belle~II experiments. We fully reconstruct the hadronic decay of one $B$ meson produced in the $Υ(4S)\rightarrow B^{+} B^{-}$ decay, and search for $B^{\pm}\rightarrow K^{\pm} τ^{+}τ^{-}$ candidates among the remaining collision products, reconstructing a charged kaon and leptonic decays of the $τ$ leptons. We optimize the selection for best sensitivity and look for an excess over background at low values of the residual energy detected in the calorimeter after full event reconstruction. We observe no significant excess and set the limit $\mathcal{B}(B^{+}\rightarrow K^{+}τ^{+}τ^{-})< 0.56\times 10^{-3}$ at the 90% confidence level, improving on the only previous result by a factor of four.

Taub-NUT black holes are somewhat exotic solutions to the vacuum Einstein equations, which have received limited attention in gravitational phenomenology. We use the soft behaviour of scattering amplitudes to compute the memory effect of the waveform resulting from the scattering of Kerr-Taub-NUT black holes. Due to the non-linear nature of gravity, NUT charges introduce intriguing features in the soft dynamics, which have no counterpart in the closely related setting of monopole charges in electromagnetism. In addition to this potentially realistic problem, we also comment on the purely academic problem in complexified gravity of the scattering of self-dual Taub-NUT black holes, which have been discussed recently in the context of celestial holography.

Left-Right Symmetric theory has proved to be one of the most successful models in explaining the origin of neutrino mass and mixings, and the phenomenological origin of neutrinoless double beta decay, lepton flavor violation as well as leptogenesis within its regime. In the current work, left-right symmetric model has been extended with a sterile neutrino per generation, which acts as a dark matter candidate within the model. The model has been realized using $A_{4}$ modular symmetry and the primary focus of the work rests on investigating the impact of assigning different modular weights to the particle content of the model and hence study its effects on the results pertaining to relic abundance of dark matter and also leptogenesis within its framework.

In this work, we continue the investigation of correlation numbers in $\mathcal{N}=1$ super Minimal Liouville Gravity (SMLG), with physical fields in the Ramond sector. Building upon our previous construction of physical operators and the evaluation of three-point correlation functions involving Ramond and Neveu-Schwarz (NS) insertions, we now turn to the analytic computation of four-point correlation numbers. This development is motivated by the framework established for the bosonic Minimal Liouville Gravity and its supersymmetric NS analog, where the integration over moduli space in correlation functions can be performed explicitly using the higher equations of motion (HEM) in Liouville theory. In particular, if one of the insertions corresponds to a degenerate field, the four-point amplitude can be expressed in terms of boundary contributions obtained from the OPE structure of logarithmic counterparts of ground ring elements. We aim to adapt and generalize this approach to the Ramond sector.Our result is a closed-form analytic expression for four-point correlation numbers involving Ramond fields.

In this paper, we investigate leptogenesis under the neutrino mass anarchy hypothesis in both type-I and type-II seesaw models. We first revisit the corresponding study in the type-I seesaw framework with two improvements: in contrast to Ref.[25], where an approximate $U(1)$ flavor symmetry was imposed to ensure sizable hierarchies among the right-handed neutrino masses and Yukawa couplings, we adopt a fully general anarchy scenario with completely random and structureless neutrino mass and Yukawa matrices; moreover, given the crucial role of lepton flavors in both the generation and washout of the lepton asymmetry, flavor effects are consistently incorporated throughout our analysis. We then extend our investigation to the type-II seesaw framework, in which leptogenesis proceeds via the out-of-equilibrium decays of a scalar triplet.

Going from lower to higher multiplicity events in proton-proton and proton-lead collisions, the data show a stronger than linear growth of the D-meson normalized yields. In this contribution we try to understand this behavior using a Monte Carlo event generator which implements the $k_T$-factorization formalism. We use different spatial distribution of matter in the proton and in the lead nucleus at the initial stage of these collisions. We find that, with all the tested spatial distributions, the model reproduces well the behavior seen in the data. We conclude that this observable is not appropriate for a detailed study of the spatial distribution of matter in the proton.

By exploiting small-scale structure formation probed by Lyman-$α$ forest observations, we study constraints on a model of dark matter from dark sector decay. We compute the phase space distribution of the dark matter and the linear matter power spectrum. We map the non-thermal dark matter distribution in this dark matter model to an approximate thermal warm dark matter distribution, and use this approximation to obtain a constraint from the Lyman-$α$ forest observation. We combine the latest Lyman-$α$ forest bounds with the constraint from the Big Bang Nucleosynthesis. As these two probes offer highly complementary constraints, we impose strong limits on sub-GeV dark matter. Consequently, masses lighter than $\sim 10^{-1}$ GeV are excluded, thereby significantly limiting the allowed parameter space. More broadly, our findings demonstrate the utility of small-scale structure observations in testing non-thermal dark matter paradigms, offering valuable insights for exploring a wider class of late-time decay models.

We formulate a partonic description of hadronic gravitational form factors within QCD, focusing on the three-point function of the energy-momentum tensor and two gluon currents. Despite the lack of exact conformal symmetry in QCD, the correlator may be organized around the conformal limit through momentum-space CFT methods, suitably adjusted for gauge-fixing effects. This yields a tensor decomposition into spin-2, spin-1, and spin-0 sectors, with the spin-0 contribution governed by the conformal anomaly. The corresponding anomaly form factor satisfies a mass-independent dispersive sum rule and allows a dilaton-like interpretation. In the light-cone limit, this term and an additional traceless structure become dominant, indicating an effective anomaly-mediated description relevant to hadronic gravitational form factors.

Based on a novel method for producing antineutrons via $J/ψ$ decays, we report a study of $\bar{n}p$ inelastic scattering into final states containing kaons. The analysis uses $(10087\pm44)\times 10^6$ $J/ψ$ events collected at the BESIII detector operating at the BEPCII storage ring. Antineutrons are produced via $J/ψ\to p π^{-} \bar{n}$ decays and tagged by the detected protons and pions, resulting in antineutron momenta ranging from 0 to 1174~MeV/$c$, while target protons are provided by the hydrogen in the beam-pipe material. The cross sections of the reactions $\bar{n}p \rightarrow K^{+}K^{-}π^{+}$ and $\bar{n}p \rightarrow K^{+}K^{-}π^{+}π^{0}$ are measured to be $0.53^{+0.15}_{-0.12} \pm 0.08$~mb and $1.09^{+0.36}_{-0.30} \pm 0.31$~mb respectively, where the first uncertainties are statistical and the second systematic. Due to limited statistics, the intermediate states in these processes are not investigated. The observation of clean antineutron-proton scattering events indicates the potential of this approach for future investigations of antineutron-proton interactions.

We investigate the dynamical generation of charmonium-like ($I=0$) with spin-parity $J^{PC}=0^{++}, 1^{++}, 2^{++}$, and $3^{--}$ in the mass range of $3.6$ to $4.3$ GeV. We employ the off-shell coupled-channel formalism, constructing kernel amplitudes from effective Lagrangians that respect heavy-quark spin-flavor and chiral symmetries. To focus solely on dynamically generated states, we explicitly exclude $s$-channel pole diagrams and include only $t$- and $u$-channel meson exchanges. Solving the integral equations, we identify six poles in the complex energy plane. In the scalar ($0^{++}$) sector, we find a bound state below the $D\bar{D}$ threshold and a resonance at $\sqrt{s_R}=(3861-i\,23)\,\mathrm{MeV}$. For the axial-vector ($1^{++}$) sector, the experimentally observed $χ_{c1}(3872)$ is reproduced as a bound state near the $D\bar{D}^*$ threshold, alongside a broader resonance at $(3961-i\,32)\,\mathrm{MeV}$, which is a plausible candidate for the $X(3940)$. Furthermore, we find a narrow tensor ($2^{++}$) state at $4005\,\mathrm{MeV}$ and a vector ($3^{--}$) state at $4030\,\mathrm{MeV}$. The present results demonstrate that coupled-channel dynamics, particularly involving the $D^*\bar{D}^*$ channel, play a crucial role in the formation of these charmonium-like exotic states.

We demonstrate that a vector-like fourth family of quarks induces a genuine chiral enhancement in $b\to sγ$, which is absent in the Standard Model (SM). The coexistence of doublet and singlet states allows the chirality flip to occur inside the loop, leading to contributions proportional to the heavy vector-like mass. The resulting amplitude is enhanced by a factor $\overlineλ_d v_H/m_b$, which can be as large as $\mathcal{O}(40)$ for moderate Yukawa couplings. This leads to sizable deviation from the SM prediction even for $\mathcal{O}(\mathrm{TeV})$ vector-like quark masses and small mixing angles. We find that $\mathrm{Br}(\overline{B}\to X_sγ)$ provides the most stringent constraint on this scenario among a wide range of precision observables.

The STAR collaboration at RHIC (BNL) and the ALICE collaboration at LHC (CERN) published recently dielectron ($e^+e^-$ pair) spectra in the intermediate mass region (IMR), $M_{e^+e^-}$ = (1-3) GeV, which show a constant, i.e. energy-independent, emission temperature $T_{IMR}\simeq$ 287+-27 MeV, at all bombarding energies $\sqrt{s_{NN}}$ from 27 to 200 GeV. What causes this strange 'Thermostat' behaviour? Why the temperature is so small and constant, although the bombarding energy is increased by orders of magnitude, so the early temperatures of the created parton plasma ought to rise?

A one-loop resummation scheme, inspired by Dyson-Schwinger (DS) formalism of strongly coupled quantum field theories, is applied to spin-1/2 magnetic monopoles (MMs), in the context of an effective field theory (EFT), invariant under the gauge group U(1)_em x U(1)', where U(1)' is a dual strongly coupled Abelian interaction, associated with a "dark photon". An ultra-violet fixed point structure is found in the resummed theory, which is purely non-perturbative, due to different boundary conditions of the resummation equations, compared to the weak coupling (perturbative) case. A self-consistent identification of the renormalized coupling of the MM to the electromagnetic photon in the fixed-point theory with the magnetic charge, compatible with the Dirac quantization condition, is made. This provides for the first time a formal justification of the use of tree-level Drell-Yan and photon-fusion MM production processes in collider searches, and of the corresponding cross sections and MM mass bounds thereof. The latter provide a means to constrain the resummed-EFT parameters experimentally. The DS resummation applies here primarily to the case of elementary (structureless) MMs. However, this approach may also be applied to the last stage (collapse) of the formation of composite MM pairs at colliders, in case they behave as quantum excitations, with their core radius comparable to the Compton wavelength, thereby avoiding the extreme suppression of their production.

Driven by the growing agreement between the experimentally measured muon anomalous magnetic moment and its SM prediction, we reexamine phenomenological consequences of the MSSM, which is embedded in the supersymmetric $SU(4)_C \times SU(2)_L \times SU(2)_R$ Pati-Salam model. In contrast to earlier studies that predominantly favored a specific sign for the Higgsino mass parameter, our analysis systematically explores both $μ> 0$, and $μ< 0$ scenarios in light of current collider, cosmological, and DM constraints. Within this framework, we identify viable parameter space regions where the observed DM relic density is reproduced through multiple mechanisms: co-annihilations involving sbottom-neutralino, gluino-neutralino, stop-neutralino, stau-neutralino, and chargino-neutralino coannihilation, as well as resonant s-annihilation channel via the pseudoscalar Higgs boson. We demonstrate that all such scenarios are consistent with present bounds from LHC supersymmetry searches, the Planck~2018 DM relic density bound, and current limits from DD DM searches. Our results reveal characteristic mass spectra associated with these mechanisms. In particular, sbottom-neutralino coannihilation typically requires sbottom masses near $2.8~\text{TeV}$, while gluino-neutralino and stop-neutralino coannihilation scenarios allow gluino masses in the range $1$--$3~\text{TeV}$ and stop masses between $1$ and $3.5~\text{TeV}$. In coannihilation-dominated regions, the stau and chargino masses may reach values as high as $3.8~\text{TeV}$, whereas viable $A$ resonance solutions are realized for pseudoscalar Higgs masses spanning approximately $1.6$--$3.8~\text{TeV}$. We anticipate that a portion of the parameter space will be accessible to supersymmetry searches in LHC Run-3 and future runs.

In this paper we show how the Baxter Q-operator and the product formula for eigenfunctions of two-particle hyperbolic Ruijsenaars system can be derived from the genus one Moore-Seiberg duality identity in two-dimensional Liouville conformal field theory. We expect that this relation would reveal genuine role of the Moore-Seiberg identity in integrable systems.

We study Knizhnik-Zamolodchikov (KZ) connection in the presence of irregular singularities, that is, poles of higher order. We consider both the case of a universal connection and the case when it is associated with a specific simple Lie algebra, such as $\mathfrak{su}(2)$. We give some general results about the monodromies of such flat connections in the configuration spaces of points, and provide explicit examples of topological invariants of links (more generally, tangles) realized by the monodromy.

In this paper we study the ultraviolet behavior of the invariant charge in QED. We show that for complex momenta the invariant charge does not have Landau pole singularity. We can define new invariant charge as real part of standard invariant charge. New invariant charge is limited from above and does not have Landau pole singularity. Also we use the $1/N$ perturbation theory for the investigation of the ultraviolet behavior of the invariant charge. To this aim we consider QED with imaginary charge which is asymptotically free but nonphysical model. In QED with nonphysical imaginary charge we can reliably calculate the ultraviolet asymptotics for the $(1/N)^k$ correction to the invariant charge, namely: $α_k(\frac{p^2}{μ^2}, α) \sim (\ln(\frac{p^2}{μ^2}))^{-k-1}$ at $k > 1$ and $α_1(\frac{p^2}{μ^2}, α) \sim (\frac{\ln(\ln(\frac{p^2}{μ^2})}{\ln^2(\frac{p^2}{μ^2})})$ at $k =1$. The $1/N$ perturbation theory coincides for QED with imaginary charge and standard QED with real charge. It means in particular that ultraviolet behavior of the $(1/N)^k$ correction $α_k(\frac{p^2}{μ^2}, α)$ in real QED coincides with the corresponding asymptotics for QED with imaginary charge. We propose also to use the modified $1/N$ expansion which is ultraviolet finite. The comparison of the standard QED and nonphysical QED with imaginary charge gives hint that in other non asymptotically free models like supersymmetric QED, scalar QED or Wess-Zumino model ultraviolet asymptotics of the invariant charge coincides with leading log approximtion.

We revisit the dark matter phenomenology of a flavor-dependent $U(1)_X$ gauge extension of the Standard Model, where anomaly cancellation predicts the existence of exactly three fermion generations and requires the presence of three right-handed neutrinos. In Ref.~\cite{VanLoi:2023utt}, a strong hierarchy between the vacuum expectation values of two singlet scalars, $\La_2 \gg \La_1$, renders all $\mathbb{Z}_2$-odd scalar states heavy, resulting in a two-component dark matter scenario composed exclusively of fermions. In the present work, we relax this simplifying assumption and consider a more general mass spectrum. In particular, scalar mixing can naturally lead to a situation in which the lightest $\mathbb{Z}_2$-odd particle is a scalar rather than a fermion. As a consequence, the model admits a qualitatively new realization of two-component dark matter consisting of one fermionic and one scalar component, in addition to the purely fermionic scenario studied previously. We perform a dedicated phenomenological analysis of these two-component dark matter realizations, focusing on the coupled thermal freeze-out dynamics and the resulting relic abundance. Constraints from the observed relic density and current direct-detection limits are taken into account, and viable regions of parameter space are identified.

Reactor CEvNS experiments provide a powerful probe of the physics beyond the Standard Model (BSM) with the intense flux of neutrinos, photons, and other particles produced in nuclear reactors. In this work, we investigate the sensitivity of reactor CEvNS experiments to the dark axion portal, which connects the axion or axion-like particle to the dark photon. We focus in particular on the RECODE experiment, while also considering other reactor-based experiments such as CONUS and MINER. We find that reactor CEvNS experiments offer enhanced sensitivity in the sub-MeV region compared with the existing constraints from B factories, and can probe the portal coupling down to $\mathcal{O}(10^{-3})$ for $G_{aγγ'}$.

In this work, we systematically investigate the evolution of spin polarization and quantum entanglement in cascade decays of baryon-antibaryon pairs, which are produced in electron-positron annihilation. We derive a fully analytical spin density matrix explicitly expressed in terms of spin polarization observables, extend this formalism to multi-step cascade decay scenarios, and establish compact recursive relations to facilitate density matrix calculations for such processes. It is found that when maximal parity violation occurs during a decay, the resulting final-state particles are fully polarized and exist in a non-entangled state. Furthermore, we demonstrate that quantum entanglement amplification is a generic characteristic of charge-conjugate decays under $CP$ conservation when the initially produced baryon-antibaryon pair is polarized.

An attempt to evade the strict uniqueness of consistent interactions involving spin-2 particles is made by modifying the Noether procedure from the outset. A vector field is introduced, coupled to a graviton already at the level of quadratic mixing. The byproduct is a gauge-invariant mass for the vector and novel consistent interactions, here derived and tested up to quartic order. A simple geometric interpretation of the vector field appears possible.

We study the neutral massive Maxwell (Proca) equation on subextremal Reissner--Nordström exteriors. After spherical-harmonic decomposition, the odd sector is scalar, while the even sector remains a genuinely coupled $2\times2$ system. Our starting point is that this even system admits an exact asymptotic polarization splitting at spatial infinity. The three resulting channels carry effective angular momenta $\ell-1$, $\ell$, and $\ell+1$, and these are precisely the indices that govern the late-time thresholds. % For each fixed angular momentum we develop a threshold spectral theory for the cut-off resolvent. We prove meromorphic continuation across the massive branch cut, rule out upper-half-plane modes and threshold resonances, and obtain explicit small- and large-Coulomb expansions for the branch-cut jump. Inverting this jump yields polarization-resolved intermediate tails together with the universal very-late $t^{-5/6}$ branch-cut law. % At the full-field level, high-order angular regularity allows us to sum the modewise leading terms on compact radial sets and obtain a two-regime asymptotic expansion for the radiative branch-cut component of the Proca field, with explicit coefficient fields and quantitative remainders. We also analyze the quasibound resonance branches created by stable timelike trapping, prove residue and reconstruction bounds, and derive a fully self-contained dyadic packet estimate. As a result, the unsplit full Proca field obeys logarithmic compact-region decay, while the radiative branch-cut contribution retains explicit polynomial asymptotics and explicit leading coefficients.

Recent studies invoke a unified description of different neutron star observables using metamodels, which parametrize the Equation of State (EoS) of neutron star matter close to nuclear saturation density in terms of few nuclear parameters. In this light, the bulk viscosity in the neutrino-transparent regime of dense nuclear matter composed of neutrons, protons and electrons has been recently shown to be mostly sensitive to the value of the nuclear symmetry energy. As muons are also present at densities around nuclear saturation, we further analyse in this manuscript their impact on this transport coefficient as a function of the slope $L$ of the symmetry energy. We find that muons introduce both relevant qualitative and quantitative effects in the bulk viscous dissipation. Increasing $L$ by a factor two has an effect of several orders of magnitude on the (frequency-independent) bulk viscosity. We also find that for all values of $L$ the frequency-dependent bulk viscosity presents a double peak structure for some values of the density, absent without muons. This also represents changes in orders of magnitude of the viscosity in narrow windows of densities that could be attainable in a neutron star for enough high values of $L$. We present a systematic numerical analysis of both second-order transport coefficients, frequency-dependent bulk viscosity, and damping times of density oscillations as a function of the density and the slope, and find when these could be relevant for the dynamics of the merger of neutrons stars.

We introduce the FLASH haloscope experiment and present its electronic read-out system, currently under development. FLASH searches for Dark Matter (DM) particles and High-Frequency Gravitational Waves (HFGWs) using two cryogenic resonant cavities to scan the radio frequency spectrum between 117 and 360 MHz, looking for signals as weak as $10^{-22}$ W. The signal readout uses Microstrip Superconducting Quantum Interference Amplifiers (MSAs) as low-noise amplifiers and Software-Defined Radio (SDR) techniques to acquire, preprocess and reduce the physics signal into a format suitable for permanent storage and offline analysis.

We show that the tower of hard exclusive amplitudes in holographic light-front QCD leads to a spectral generator $G(α,λ)$ which encodes the full Regge spectrum. The construction assumes a Poisson distribution of the Fock expansion, where $λ$ represents the average parton multiplicity above the minimal valence configuration. The resulting generator yields a Regge spectrum invariant under continuous $λ$-deformations and provides analytic representation of physical form factors, including their interference structure.

We explore isospin imbalanced strongly interacting matter within the two-flavor Linear Sigma Model with quarks, an effective model for low-energy QCD. At one loop order, including quark, pion, and sigma fluctuations while respecting chiral symmetry, we find that the formation of an isospin condensate necessarily gives rise to a Goldstone mode. This mode enforces a nontrivial relation between the chiral and isospin condensates through the mixing of charged pions and the sigma field in the condensed phase. From the resulting thermodynamic potential, we compute the speed of sound and observe a pronounced peak as a function of the isospin chemical potential. Although the peak of the speed of sound may be described at tree-level and including only quarks in the analysis, meson dynamics introduces further constraints that influence the position and width of the peak which making it to align well with lattice QCD simulations. Therefore we identify that the shape and position of the peak is a consequence of the Goldstone mode dynamics and of the associated charged pion sigma mixing.

M-theory geometric engineering on manifolds of special holonomy yields a rich class of novel field theories. In this paper, we construct new 3d $\mathcal{N}=2^{\ast}$ and $\mathcal{N}=4^{\ast}$ gauge theories, realized as mass-deformations of theories with 16 supercharges, within this framework. These arise from non-compact 8d geometries given by fibrations of $\mathbb{R}^{4}/Γ_{ADE}$ over Biberbach 4-manifolds. The existence of consistent $Spin(7)$-structures on the 8d spaces requires the rotational holonomy of the Biberbach spaces to act on the $Sp(1)$-structure of the fibers. Furthermore, we analyze Higgsing the 7d $\mathcal{N}=1$ $ADE$ gauge theories induced by the action of a permutation group on the centres of the corresponding $\mathbb{R}^{4}/Γ_{ADE}$ spaces. We show that this operation admits a natural interpretation in terms of nilpotent, upper-triangular, Higgsing, although it breaks supersymmetry. Supersymmetry is restored by fibering the singular geometry over a compact internal space, whose structure group is chosen to coincide with the permutation group to implement the nilpotent Higgsing. We refer to such backgrounds as T-geometries, where ``T'' denotes the triangular nature of the nilpotent Higgsing. Within this framework, we investigate the nilpotent Higgsing of the 3d $\mathcal{N}=2^{\ast}$ and 4d $\mathcal{N}=1^{\ast}$ theories, where the rotational holonomy groups of the Bieberbach spaces realize the permutation groups. We demonstrate that the Higgs branch moduli are encoded by specific elements of the Slodowy slices associated with nilpotent elements. Moreover, we demonstrate that additional elements of the same slice give rise to non-chiral charged matter under the unbroken gauge algebra. We establish that both the Higgs branch moduli and the charged matter are massless and admit a natural interpretation as localized matter.

We develop the theoretical framework for semi-inclusive deep-inelastic scattering on a polarized spin-1 target and apply it to scattering on the polarized deuteron with spectator nucleon tagging. In Part I (previous article) we present the general form of the semi-inclusive cross section and polarization observables for the spin-1 target. In Part II (this article) we consider deep-inelastic scattering on the polarized deuteron with spectator nucleon tagging as a special case of target fragmentation. Methods of light-front quantization are employed to separate nuclear and hadronic structure in the high-energy process and achieve a composite description. The light-front wave function of the polarized deuteron is obtained from a rotationally covariant 3-dimensional wave function in the center-of-mass frame of the proton-neutron system. The tagged structure functions are computed in the impulse approximation. The momentum and spin distribution of the active nucleon are controlled by the deuteron polarization and the detected spectator momentum ($D/S$ wave ratio). The cross section and spin asymmetries are evaluated for general deuteron polarization (vector and tensor, longitudinal and transverse) as functions of the spectator momentum. Tensor-polarized spin asymmetries of order unity are achieved for spectator momenta $\sim$ 300 MeV, which select configurations with large $D$-wave. Sum rules for the tagged spin structure functions are derived. The results can be used for simulations of spectator tagging in future polarized fixed-target experiments (Jefferson Lab) or at the Electron-Ion Collider.

We develop the theoretical framework for semi-inclusive deep-inelastic scattering on a polarized spin-1 target and apply it to scattering on the polarized deuteron with spectator nucleon tagging. In Part I (this article) we present the general form of the semi-inclusive cross section and polarization observables for the spin-1 target. A relativistically covariant formulation in terms of 4-vectors and invariant polarization parameters is employed. The target polarization is described by a spin density matrix with vector and tensor polarization. The spin and azimuthal angle dependence of the semi-inclusive cross section is derived and parametrized in terms of invariant structure functions. To validate the result, the structure functions are expressed as photon-target helicity amplitudes with known symmetry properties. The expressions presented here are kinematic (no assumptions about particle production dynamics) and valid in all regions of the deep-inelastic final state (current and target fragmentation regions). In Part II (following article), we consider deep-inelastic scattering on the polarized deuteron with spectator nucleon tagging as a special case of target fragmentation. The semi-inclusive structure functions are computed by separating nuclear and hadronic structure, and the polarization observables are explored as functions of the tagged nucleon momentum.

Scintillation light, produced alongside ionisation charge from particle interactions, plays a critical role in liquid argon time projection chamber (LArTPC) detectors. A detailed understanding of its production and detection mechanisms is essential for robust calibration, systematic uncertainty evaluation, and physics analysis. This article describes the MicroBooNE light simulation, light-based triggering schemes, photomultiplier tube gain calibration, light response stability, and light-based systematic uncertainties over the course of five years of data collection. In addition, we present a measurement of scintillation light triggering efficiency, focusing on the lowest-light regime relevant to rare-event searches and low-energy neutrino interactions. Finally, we discuss two notable observations in MicroBooNE's data, both reported here for the first time: an approximately 50% decline in MicroBooNE's light yield over time, concentrated in the first two years of running; and a higher than expected O(200 kHz) rate of single photoelectron noise. The results presented provide an important benchmark of long-term light detection performance in LArTPC neutrino detectors.

The collinear and transverse-momentum-dependent jet fragmentation function and the radial profile for $B^{\pm}$ mesons in jets are measured. The $B^{\pm}$ mesons are reconstructed through the $J/ψ(\to μ^{+} μ^{-}) K^{\pm}$ decay channel using proton-proton collision data collected during 2016-2018 with the LHCb detector at a center-of-mass energy of $\sqrt{s}=13$ TeV, corresponding to an integrated luminosity of $5.4$ fb$^{-1}$. The results complement recent measurements of jet fragmentation functions for heavy-flavor hadrons and suggest a growing contribution of gluon fragmentation to $B^{\pm}$ mesons as the jet transverse momentum increases.

We develop a semi-analytical model to describe the cosmological evolution of networks of cosmic strings with small-scale structure, by extending the velocity-dependent one-scale model to include an additional lengthscale describing the typical interkink density. We study the impact of the different physical processes involved in the production and removal of small-scale structure from cosmic strings on the attainment of a full linear scaling regime, in which the characteristic lengths of the network and of small-scale structure evolve proportionally to physical time and the root-mean-squared velocity of the network remains constant. We find, using this novel velocity-dependent two-scale model, that quite generally small-scale structure does not prevent the attainment of a linear scaling regime since, even if not enough kinks are carried away when loops are chopped from the network, gravitational backreaction is generally enough to ensure that the interkink density scales. We find, however, that this regime is characterized by a smaller energy density and root-mean-squared velocity when compared to strings without small-scale structure and that this reduction may be significant when scaling is maintained by gravitational backreaction. In this case, we also find that, before reaching full scaling, the network should evolve in a transient quasi-scaling regime, in which its evolution is very similar to that of cosmic strings without small-scale structure.

Ultra-relativistic oxygen-oxygen (OO) collisions are a promising arena in which to probe rare, large-angle, high momentum-transfer $2\rightarrow2$ Molière scatterings between energetic jet partons and quasiparticles in quark-gluon plasma (QGP). As a jet propagates through the droplet of QGP formed in the same collision, its constituents lose energy to and excite wakes in the medium, and may scatter off quark- and gluon-like quasiparticles in QGP. Using the hybrid strong/weak coupling model, we show that including Molière scatterings between jet partons and medium quasiparticles is essential to reproduce recent CMS measurements of charged-particle suppression in OO collisions with this model. We then present the first theoretical study of how jet-medium interactions modify the internal structure of jets in OO collisions. We find that Molière scatterings broaden the Soft Drop splitting angle $R_g$, enhancing the population of $R=0.4$ and $R=0.8$ jets with $R_g\gtrsim0.2$ in OO collisions relative to pp collisions. Energy-energy correlators (EECs) provide a complementary probe, exhibiting enhanced large-angle correlations within jets due to jet-induced wakes and Molière scattering. In both cases, we propose an experimental measurement where the relevant OO/pp ratio can, if enhanced above unity in future data as in our calculations, be a distinctive, model-independent, detection of hard scattering off QGP quasiparticles. We furthermore use our calculations of EECs to show how the angular scale corresponding to the deflection of jet or medium partons by Molière scattering is imprinted in the EEC for jets with radius $R_{\rm jet}\sim0.8$ in OO collisions. These results demonstrate that jet substructure measurements in OO collisions are promising avenues to probe the quasiparticles that emerge at short distances within an otherwise strongly coupled medium.

The scalar potential of any particle-physics model must be bounded from below (BFB). We consider the extension of the Standard electroweak Model with three $SU(2)$ doublets of scalars and a symmetry under which each of those doublets changes sign. In the absence of necessary and sufficient conditions for boundedness from below (BnessFB) for this specific model, we argue that one may use ever more necessary conditions. We introduce a Mathematica code, StableWein, that implements this idea. The user is allowed to choose the level of accuracy that they want in the determination of BnessFB; more precision means the use of more necessary conditions, and usually entails a longer running time for the code. Our investigation suggests that our procedure and code can be extremely precise in the determination of the potentials that are BFB. In addition, we introduce a machine-learning code that identifies, with more than 99% accuracy, which potentials are BFB.

We use heavy-nucleus effective field theory to compute radiative corrections to two-neutrino double-$β$ decay ($2νββ$). Our main result is the first derivation of a universal radiative-correction factor for double-weak decays -- the analogue of the Sirlin function in single-$β$ decay -- independent of nuclear matrix elements and excitation energies. This "double-weak Sirlin function" depends on the individual electron energies as well as their relative angle and differs significantly from the approximation obtained by summing two single-$β$ decay Sirlin functions. In addition, we calculate the nuclear-structure-dependent component of the radiative corrections and find that they can still be neglected at current experimental sensitivities. On the other hand, the double-weak Sirlin function induces distortions of the electron energies and angular spectra that are comparable in size to the leading nuclear-structure corrections parametrized by the ratio of nuclear matrix elements, $ξ_{31}$. Our results indicate that extractions of nuclear structure information and tests of the Standard Model from high-precision $2νββ$ measurements must include double-weak radiative corrections, implying that recent extractions of $ξ_{31}$ should be revisited.

Understanding the novel QCD phenomenon under high external magnetic fields of hot and dense medium help us to develop a better understanding of the underlying quark dynamics of baryons. Using a hybrid approach based on the effective field theory that treats quarks as the fundamental degrees of freedom and calculating the individual contribution of valence, sea and orbital angular moment of sea quark, the magnetic moment of a given baryon of the decuplet family is calculated. The incorporation of Landau quantization in the vector and scalar densities of baryons help us to obtain the impact of external magnetic field on the properties of baryons within the chiral SU(3) quark mean field model (CQMF). In the present study, effective masses of the baryons are calculated using CQMF while the framework of chiral constituent quark model ($χ$CQM), extended to SU(4) sector, is used to obtain the effective magnetic moments of decuplet baryons under the influence of magnetic field.

The repeated scattering of photons by thermal electrons at low temperatures is described by the Kompaneets equation and its generalized forms that include anisotropies and higher order temperature corrections. In this work, we use the boost operator approach to derive the related expressions in a transparent way that showcases the generality of the formalism and its application to radiative transfer problems. We consider the simplest form of the Kompaneets equation for the scattering in isotropic media at the leading order in the electron temperature and then include anisotropies in the photon field, reproducing previously obtained expressions for the evolution equations. For this we use expressions for the scattering operator in the electron rest frame up to first order in the electron recoil, O(h nu/m_e c^2), but then work at all orders in the electron momentum, p, as easily obtained with the boost operator approach. This shows how specific transformation rules can be formulated that allow simplification of the otherwise cumbersome and repetitive calculations. We also confirm the expressions for higher order temperature corrections in isotropic media, highlighting the validity of the approach presented here. As part of the derivation, we find expressions for the boost operator in general boost directions which we believe will also be useful in other applications of the formalism.

Data unfolding -- the removal of noise or artifacts from measurements -- is a fundamental task across the experimental sciences. Of particular interest are applications in physics, where the dominant approach is Richardson-Lucy (RL) deconvolution. The classical RL approach aims to find denoised data that, once passed through the noise model, is as close as possible to the measured data in terms of Kullback-Leibler (KL) divergence. This requires that the support of the measured data overlaps with the output of the noise model, a hypothesis typically enforced by binning, which introduces numerical error. As a counterpoint, the present work studies an alternative formulation using a Wasserstein loss. We establish sharp conditions for existence and uniqueness of optimizers, answering open questions of Li, et al., regarding necessary conditions for existence and uniqueness in the case of transport map noise models. We then develop a provably convergent generalized Sinkhorn algorithm to compute approximate optimizers. Our algorithm requires only empirical observations of the noise model and measured data and scales with the size of the data, rather than the ambient dimension. Numerical experiments on one- and two-dimensional problems inspired by jet mass unfolding in particle physics demonstrate that the optimal transport approach offers robust, accurate performance compared to classical RL deconvolution, particularly when binning artifacts are significant.

Superorbital modulations has been detected in the supergiant High-Mass X-ray binary 4U 1538-52 using long-term monitoring with the Neil Gehrels Swift Observatory Burst Alert Telescope (BAT). The source also exhibits a long-term pulse period evolution as seen with Rossi X-ray Timing Explorer (RXTE), INTEGRAL, and Fermi Gamma-Ray Burst Monitor (GBM) that appears uncorrelated with changes in its X-ray flux. To investigate the mechanisms causing these superorbital modulations and its possible dependence on pulse period changes, we analyzed long-term monitoring with Swift-BAT and Monitor of All Sky X-ray Image Gas Slit Camera (MAXI-GSC) to construct dynamic power spectra and superorbital intensity profiles. In addition, we used pointed X-ray observations from Nuclear Spectroscopic Telescope Array (NuSTAR) and Neutron Star Interior Composition Explorer mission (NICER) to investigate the pulsation and spectral properties across different superorbital and orbital phase intervals. We find the presence of superorbital modulations in the MAXI-GSC 2-20 keV lightcurves, consistent with the periodicity observed with the Swift-BAT lightcurves. However, no significant changes are detected in the pulse profiles or spectral parameters across different superorbital, orbital, or pulse-change intervals. This lack of spectral or timing variations with orbital and superorbital phases suggests that the mechanisms driving the observed superorbital modulation and pulse period changes are likely associated with large-scale stellar wind structures, such as Co-Rotating Interaction regions, within the stellar wind of the supergiant companion.

The James Webb Space Telescope (JWST) has detected, through gravitational lensing, several young massive star clusters (YMCs), which are considered as relevant building blocks of high redshift galaxies. In this work, we show how a significant fraction of these YMCs could act as relevant birth places for intermediate-mass black holes. We first consider the formation of massive clusters and show that the population of YMCs is consistent with a steep mass-radius relation, which includes a relevant spread of roughly an order of magnitude. We pursue a comparison of this population with young star clusters in the local Universe and Milky Way globular clusters, including an analysis of the characteristic timescales. The YMCs show a wide spread over these properties, but include systems with both short relaxation times as well as relatively short collision timescales, implying they could go through efficient core collapse, which would lead to runaway collisions. We provide quantitative estimates of the sizes of the clusters that could efficiently form intermediate-mass black holes through a runaway collision-based channel, suggesting that these roughly correspond to the systems beyond the $1σ$ scatter in the mass-radius relation. This implies a fraction of ~16% of YMCs as candidates to form intermediate-mass black holes. We show that above a mass limit of ~6x10^6 M_sun, compact star clusters are likely to retain gas even in the presence of strong supernova feedback, altering the dynamics in the central core and providing the possibility to rapidly grow the central object both via gas dynamical friction and Bondi accretion. Finally, we consider the possibility of a gas-dominated regime, in which strong gravitational torques may inhibit star cluster formation and instead directly form a high-mass black holes, as suggested to have occurred in the infinity galaxy.

Distinguishing active galaxies from star-forming galaxies is essential for understanding galaxy evolution. Diagnostic methods like the BPT (Baldwin, Phillips, and Terlevich) diagram use optical emission-line ratios to separate galaxies. However, with growing availability of large surveys and high-resolution instruments, manually identifying galaxy types has become increasingly challenging. In this study, we investigate machine learning to classify active and star-forming galaxies using properties like stellar mass, stellar velocity dispersion, colour, redshift, and [O III] luminosity. These new approaches enable faster AGN/star-forming galaxy classification than the BPT diagram and provide a flexible, scalable alternative that can complement traditional diagnostics, particularly for large surveys or low-quality data. We employ four classification algorithms -- Decision Tree, Random Forest, Support Vector Classifier (SVC), and k-Nearest Neighbours (KNN) -- using the Galaxy Zoo 1 dataset derived from the SDSS sample. The dataset contains 47,675 galaxies within the redshift range 0.02--0.05, including 17,002 pure star-forming and 2,254 active galaxies, labeled using the BPT diagram. These labels train and evaluate our models through confusion matrices, learning curves, and receiver operating characteristic (ROC) curves. Among the four algorithms, the SVC and Random Forest models achieve the highest accuracy of approximately 93\%, while KNN shows the lowest at 88\%. Furthermore, we estimate supermassive black hole masses using stellar velocity dispersion ($σ$) and the $M_{\rm BH}-σ$ relation. We apply four regression models -- Random Forest Regressor, Support Vector Regressor (SVR), KNN Regressor, and Polynomial Regression. All four models produce similar results, with $R^2$ values from 0.75 to 0.77, indicating consistent performance.

Extended $γ$-ray emission has been observed around several nearby pulsars and interpreted as inverse Compton radiation from relativistic electrons and positrons diffusing in the surrounding interstellar medium. In this work, we present a unified analysis of the TeV halos associated with the Geminga and Monogem pulsars, combining GeV--TeV $γ$-ray observations within a common physical framework. Assuming continuous injection of $e^\pm$ pairs from pulsar wind nebulae, we model the resulting $γ$-ray emission by accounting for particle diffusion and radiative energy losses. We find that the observed spectra of both Geminga and Monogem are well reproduced by the model, provided that particle transport in the vicinity of the pulsars is significantly suppressed with respect to the average Galactic diffusion. The injection spectra require cutoff energies of several tens of TeV, consistent with efficient acceleration in pulsar wind nebulae. Using the best-fit models inferred from the $γ$-ray data, we also evaluate the contribution of Geminga and Monogem to the local cosmic-ray positron flux measured by AMS-02. We find that the suppressed diffusion around the sources strongly limits the positron flux reaching the Earth, resulting in a subdominant contribution over the entire energy range probed by the experiment excect for the highest energy at around 1 TeV. Our results support an interpretation in which TeV halos trace regions of inhibited particle diffusion around pulsars, without implying a significant impact on the local cosmic-ray positron spectrum. This combined analysis highlights the importance of extended $γ$-ray observations for constraining particle transport in the vicinity of Galactic cosmic-ray sources.

The redshifted 21 cm absorption trough from cosmic atomic hydrogen is one of the most promising probes of the early Universe, but its detection is challenged by bright foregrounds and instrumental systematics. In this work we quantify the impact of antenna mismodelling on signal recovery within a fully Bayesian, forward-modelled data analysis pipeline. We show that discrepancies between simulated and modelled antenna beams lead to frequency dependent errors in antenna temperature that can bias parameter inference. In particular, we demonstrate that orientation mismatches at the level of 0.25 degrees can significantly bias recovered signal parameters in typical observing scenarios. However, we also show that Bayesian evidence can be used to infer antenna orientation within this precision by scanning over model realisations. For structural mismodelling, we find that broadband recovery of all signal parameters requires accurate beam knowledge, but that partial recovery remains possible. Signal frequency and width can be robustly recovered under restricted frequency bands even when the antenna structure is imperfectly modelled, but signal depth is highly sensitive to beam errors. These results quantify the level of beam knowledge required for forward-modelled global 21 cm experiments and highlight the importance of observing strategy and antenna design in mitigating beam-sky coupling systematics.

We present homogeneous, near-infrared ($JHK_s$ bands) time-series observations of the classical Carina dwarf Spheroidal (dSph) galaxy to determine accurate and precise distances using the pulsating stars as standard candles. These observations cover two Carina dSph fields ($\sim10.8'\times10.8'$) obtained with the FourStar infrared camera mounted on the 6.5-m Magellan Telescope. We collected precise photometric measurements of 43 RR Lyrae, 11 anomalous Cepheids (ACep), and 102 dwarf Cepheids (DCep) in Carina dSph. Using RR Lyrae, we obtained a distance modulus of $20.079\pm0.028\mathrm{(statistical)}\pm0.045\mathrm{(systematic)}$~mag, or a distance to Carina of $103.7\pm1.3\mathrm{(statistical)}\pm2.2\mathrm{(systematic)}$~kpc. The literature calibrations based on SX Phoenicis or delta-Scuti stars were used to anchor the $JHK_s$ period-luminosity relations for DCep. This resulted in a distance modulus that is in excellent agreement with RR Lyrae based determination. Finally, the distance moduli estimates using the ACep were found to be systematically smaller than the RR Lyrae-based distance modulus, suggesting a metallicity dependence on the ACep period-luminosity relation.

On $2024$ October $5$, BL Lacertae ($2200+420$) experienced one of its brightest gamma-ray flares. We conducted simultaneous follow-up observations in the $u$, $v$, $g$, $r$, $i$, and $z$ bands from $2024$ October $17$ to November $21$ using the Mephisto telescope and its two $50$ cm twin auxiliary photometric telescopes of Yunnan University. Intraday variability (IDV) was detected in the $g$, $r$, $i$, and $z$ bands. The IDV duty cycle increased with observing frequency across these bands. The shortest variability time-scale, derived from auto-correlation analysis, constrains the upper limit of the black hole mass to be $M_{\bullet} \lesssim 10^{8.29} M_{\odot}$ assuming a Kerr black hole, and $M_{\bullet} \lesssim 10^{8.77} M_{\odot}$ assuming a Schwarzschild black hole. The emission region responsible for the observed variability has a size of $R \le 3.51 \times 10^{14}$ cm and is located at a distance of $R_H \le 2.83 \times 10^{15}$ cm from the central supermassive black hole. This distance is approximately three orders of magnitude smaller than the typical radius of the broad-line region, indicating that the emission region lies well within it. A general bluer-when-brighter (BWB) trend was detected on intraday time-scales, suggesting that shock-accelerated relativistic electrons enhance the high-energy particle population, leading to spectral hardening. A potential quasi-periodic oscillation (QPO) with a period of $\sim 100.77$ minutes was detected with $>99.99$ per cent confidence, consistent with predictions from the magnetic reconnection model. These observed optical intraday variabilities and colour variations of BL Lacertae can be well explained by the turbulent jet model.

The fact that luminous X-ray coronae and Ultra-Fast Outflows (UFOs) are both inferred to originate from the innermost regions of active galactic nuclei (AGNs) suggests a deep physical connection between them. However, standard magnetic buoyancy models struggle to transport sufficient energy through the radiation-pressure-dominated inner disc to sustain both the phenomena, creating a theoretical energy deficit. In this work, we propose an outflow-based model with energy transport enhanced by the Photon Bubble Instability (PBI) in the inner region. By coupling this enhanced energy supply with the MHD turbulence-driven mass-loading mechanism appropriate for weakly magnetized standard discs, we solve the dynamical and thermodynamic structure of the corona. We find that the model can successfully launch high speed winds matching observed UFO kinematics provided the mechanical acceleration efficiency is high ($f_{\rm acc}\gtrsim 0.5$). Furthermore, the model naturally reproduces the observed spectral evolution found in AGN coronae: as the accretion rate increases, the corona becomes optically thicker and cooler and produces a softer spectrum. Our results support an extended slab-like coronal geometry and suggest that UFOs and X-ray coronae in the inner discs are manifestations of the same magnetic activity.

A more realistic description of the magnetosphere is crucial for understanding the radiation emitted by pulsars. In this paper, we revisit the annular gap model by employing a rotating dipole field, which is more realistic than the static dipole field, as an approximation of the magnetic structure of the pulsar magnetosphere. Compared with the static dipole field approximation, the open field-line region, including both the core and annular gaps, is significantly enlarged, and the two regions become asymmetric with respect to the fiducial plane. We apply this model to three young gamma-ray pulsars with distinct light-curve morphologies, PSRs J0631$+$1036 (single peak), J1709$-$4429 (double peaks), and J1048$-$5832 (three peaks). Using viewing geometries constrained by radio polarization measurements, the annular gap model within the rotating dipole field successfully reproduces the main morphological features of their gamma-ray light curves above 0.1 GeV. Our model provides a framework for interpreting pulsar high-energy emission, which can be used to analyze the emission properties of high-energy pulsars.

Metal-poor (MP) stars are important targets for investigating the chemical evolution of the early universe. Among them, Carbon-Enhanced Metal-Poor (CEMP) stars have attracted extensive attention due to their rarity and astrophysical significance. Owing to their low occurrence rate, the identification of MP stars and CEMP stars remains a task of considerable scientific value. In this study, we investigate the search for CEMP stars based on the low-resolution stellar spectra from LAMOST (Large Sky Area Multi-Object Fiber Spectroscopic Telescope) DR11 and propose a deep-learning-based approach for this purpose. By analyzing the LAMOST DR11 spectral library, we identify 1,408 CEMP star candidates. For ease of reference and further use, we provide the estimated stellar parameters for these objects, including $T_\texttt{eff}$, $\log~g$, [Fe/H], and [C/H].

Interactions with the host galaxy strip stars and dark matter from the outer regions of satellite galaxies. Meanwhile, some stars from the central regions can migrate outward due to dynamical heating, producing an excess in the outer surface brightness relative to the extrapolation of the inner Sérsic profile. Recently discovered dark matter deficient galaxies (DMDGs) appear to be representative examples of such tidally disturbed systems. In this work, we investigate how the break radius, defined as the radius beyond which this surface brightness excess emerges, forms and evolves, by performing $N$-body simulations of a satellite galaxy interacting with a host, where the satellite serves as a plausible progenitor of a DMDG. Our simulations naturally reproduce a break radius consistent with that observed in DMDGs. We find that the break radius grows over time and exhibits a characteristic evolutionary behaviour: during each pericentric passage it briefly contracts due to tidal compression, and then rapidly and strongly expands as the satellite undergoes dynamical relaxation. After the satellite reaches a quasi-equilibrium configuration, the break radius shows only mild variations until the next pericentre. Across our suite of simulations, the ratio of the break radius to the effective radius remains approximately constant, even when we change the orbital parameters and internal structure of the satellite. Based on these findings, we develop a prescription for predicting the time evolution of the break radius, which can be used to constrain the tidal interaction history of satellite galaxies, including DMDGs and splashback galaxies.

Rotating Radio Transients (RRATs) are a class of pulsar-like objects characterized by intermittent radio emissions. Among them, RRATs that exhibit both RRAT and normal pulsar (NP) states may represent a key evolutionary stage from nulling pulsars to RRATs. We performed a detailed analysis of RRAT J1574+4703 using the Five-hundred-meter Aperture Spherical Radio Telescope (FAST) at a frequency of 1250 MHz. Our findings indicate that this RRAT spends approximately 98% of its time in the RRAT state, with the remainder spent in an NP state exhibiting nulling behavior. Additionally, we observed distinct integral pulse profiles and polarization properties between the two states, suggesting that they originate from different emission heights and magnetospheric structures. Furthermore, it was observed that the NP states of this RRAT exhibit mode switching, with ~44% of the time spent in the normal mode and ~39% in the abnormal mode. Notably, abnormal modes are predominantly detected at the onset and termination of the NP states. This discrepancy between the modes indicates potential instability in the magnetospheric processes that govern the NP states.

We present GalSyn (Galaxy Synthesizer), a modular and flexible Python package for generating synthetic spectrophotometric observations from hydrodynamical galaxy simulations. GalSyn employs a particle-by-particle spectral modeling approach that enables the rapid production of large synthetic datasets required for statistical population studies, offering a computationally efficient alternative to full radiative transfer codes. Users have full control over the spectral modeling choices, including the choice of stellar population synthesis engine, stellar isochrones, spectral libraries, and initial mass functions. Dust attenuation is modeled at the spatially resolved level via a line-of-sight column density method, with a comprehensive suite of fixed and adaptive attenuation laws. A decoupled kinematics model independently Doppler-shifts the stellar and nebular components, enabling realistic synthetic IFU data cubes. It also provides features to add observational realism, including PSF convolution and multi-component noise simulation. Beyond imaging and spectroscopic data cubes, GalSyn reconstructs spatially resolved physical property maps and star formation histories. Alongside this paper, we present the first public data release of synthetic imaging observations and spatially resolved star formation histories generated from the IllustrisTNG simulation suites, comprising four mock extragalactic survey fields (with areas of $5$, $8$, $137$, $365$ arcmin$^{2}$), progenitor histories of 290 local massive galaxies ($\log(M_{*,z=0}/M_{\odot}) > 10.5$) tracked across $0<z<5$, and 259 major-merger systems. Each galaxy data cube contains imaging in 47 filters spanning HST, JWST, Euclid, Rubin/LSST, and the Roman Space Telescope. GalSyn is publicly available at https://github.com/aabdurrouf/GalSyn.

We present the confirmation and characterization of a long stream (S-stream) in the southern part of M83. This feature is revealed using deep wide-field photometric data obtained by the Hyper Suprime-Cam (HSC) mounted on the Subaru Telescope. Using individual red giant branch (RGB) stars, we successfully trace the stream over a large length of $\sim 81$~kpc and a considerable width of $\sim 9$ kpc. With a mean surface brightness of ${\langle μ_{\it V} \rangle} \sim 31.8_{-1.9}^{+1.3}$ mag arcsec$^{-2}$, it is one of the most diffuse extragalactic streams currently known. The mean photometric metallicity of the stream is $\langle[{\rm M/H}]\rangle = -1.23\pm0.02$ dex with a standard deviation of $0.28\pm0.01$ dex, and we estimate the stellar mass to be $(8.5_{-2.8}^{+4.2}) \times 10^6~{\rm M_\odot}$ from the luminosity of RGB stars. Compared to its well-known northern counterpart, the S-stream is slightly more metal-poor, but our large-area RGB map shows compelling evidence that these two features are related, originating from a single low-mass merger event. We identify density variations along the S-stream, which more likely reflect intrinsic density structure within the progenitor rather than the interaction with dark matter subhalos. Similarities between the morphology of the S-stream and some features in the \HI distribution suggest that a minor merger event may have disturbed and redistributed M83's outer \HI gas, leading to triggered star formation and the formation of the XUV disk.

We present optical and near-infrared (NIR) observations of the Type Ib supernova (SN) 2024rbc. Emission from the first CO overtone, resting on a dust continuum at $2.3-2.4$ $μ$m, was observed at 62 days post-explosion. The CO band heads are not seen; the emission is broad and devoid of sharp spectral structure. This is the first observation of CO in the ejecta of a Type Ib SN reported in literature. Fitting a LTE model to the CO overtone derives a mass of $(5.2 \pm 1.2)$ $\times$ 10$^{-4}$ $M_{\odot}$, a temperature of $4040 \pm 435$ K, and a velocity of $5905 \pm 1960$ km s$^{-1}$. We also fitted a modified blackbody model to the dust continuum, deriving a dust temperature of $910 \pm 10$ K and a mass of $(1.3 \pm 0.1)$ $\times$ $10^{-3}$ $M_{\odot}$. Furthermore, the spectra of SN 2024rbc exhibit strong He I lines and numerous neutral and ionized metal lines. Comparing the spectral evolution of SN 2024rbc to other Type Ib, Ic, and IIb SNe indicates it is a Type Ib SN. Additionally, fitting SN light curve models of helium star progenitors computed with the STELLA code to photometric observations indicates a $^{56}$Ni mass of $0.07$ $M_\odot$ and an ejecta mass of $1.7$ $M_\odot$. We also compare the velocities of key optical lines to examine the evolution of the ejecta. Lastly, we discuss the observed CO and dust emission and its implications for early-Universe dust formation.

Galaxy surveys provide finite catalogs of objects observed within bounded volumes, yet clustering statistics are often interpreted using theoretical frameworks developed for infinite point processes. In this work, we formulate key statistical quantities directly for finite point processes and examine the structural consequences of finite-number and finite-window constraints. We show that several well-known features of galaxy survey analysis arise naturally from finiteness alone. In particular, non-vanishing higher-order connected correlations can occur even in statistically independent samples when the total number of points is fixed, and the integral constraint in two-point statistics appears as an exact identity implied by the finite-number condition rather than as an estimator artifact. We further demonstrate that counts-in-cells and point-centered environmental measures correspond to distinct statistical ensembles. Using Palm conditioning, we derive an exact relation between random-cell and point-centered statistics, showing that the latter probe a tilted version of the underlying distribution. These results provide a probabilistic framework for separating structural effects imposed by finite sampling from correlations reflecting genuine astrophysical processes. The formulation presented here remains valid for realistic survey geometries and finite data sets and clarifies the interpretation of commonly used clustering statistics in galaxy surveys.

The Crab pulsar experienced two relatively small glitches separated by only 20 days in September and October 2025. IXPE observed the source twice, with delay times since the glitch epoch ranging between 35 and 75 days, depending on the observation. We carried out a multi-method analysis to investigate whether there is evidence for significant changes in the polarization properties of the pulsar, underlying possible variations in the pulsar magnetosphere itself following the glitches. Specifically, we performed: (1) phase-averaged polarimetry of the Crab pulsar before and after the glitches, following an approach similar to that adopted in 2019 by PolarLight, a non-imaging CubeSat-class photoelectric polarimeter which observed a change in the X-ray polarization within 100 days after a stronger glitch in July 2019; (2) a comparison, before and after the glitch, of phase-resolved X-ray polarimetry with IXPE, not possible with PolarLight. Furthermore, we investigated, by means of phase-resolved optical (OPTIMA) polarimetry, whether a significant change in the X-to-optical lag was present in the data before and after the glitch. We find no evidence of a change in the polarization for the pulsar emission before and after the glitch, We use the upper limits obtained to estimate the maximum change in magnetic obliquity allowed by the data, using the standard rotating vector model and assuming that the glitch is due to a neutron-star quake. We constrain this maximum change to be no greater than $\pm4^{o}$ at the 95\% confidence level.

A small but growing class of binaries containing hot ($T_{\rm eff}\sim10^5\rm~K$) white dwarfs (WDs) and rapidly rotating, apparently subgiant companions -- including the prototype, Abell 35 -- show companions that are too large and luminous to be ordinary main-sequence stars yet too numerous to be explained as finely tuned near-twin binaries. We argue that these stars are instead main-sequence accretors temporarily inflated out of thermal equilibrium by recent mass transfer. For the subgiant of Abell 35, a new Gaia DR3 astrometric orbit ($P_{\rm orb} = 790$ d) combined with updated photometric and spectroscopic constraints yield $T_{\rm eff} \approx 4900~\rm K$, $R \approx 3~R_{\odot}$, near-solar metallicity, and rapid rotation aligned with the orbit ($v_{\rm rot} \approx 195~\rm km~s^{-1}$), indicating substantial recent accretion and spin-up. Dynamical mass limits disfavor a coeval twin-binary origin, supporting the inflated-accretor interpretation. We test this scenario using self-consistent MESA binary evolution calculations with a new accretion prescription in which accreted material retains a fraction of its infall energy. The accretor expands to giant-like radii when $\dot{M}$ is high yet remains within its Roche lobe, allowing stable mass transfer even for mass ratios traditionally considered unstable. After mass transfer ceases, the star contracts on Myr timescales through a bloated, rapidly rotating phase whose temperatures, radii, and spins match those observed in Abell 35-type systems. This framework explains the population without fine tuning and unifies Abell 35-type binaries with post-AGB binaries, blue lurkers, and wide WD$+$main-sequence systems as successive stages of the same post-mass-transfer evolutionary pathway.

AGNs are known to be variable in almost all wavelengths and timescales. The shortest variability timescale of AGNs can be used to probe the smallest scale structures within AGNs. We aim to measure the shortest detected variability timescale, $t_{min,ul}$, of type 1 radio-quiet Seyfert galaxies and analyse their characteristics. We extracted TESS light curves of 47 Seyfert 1 galaxies. We measured the PSDs of the sample, modelled by a power law model plus a constant noise, and constrained the shortest detected AGN variability timescale as the power law component exceeds the constant noise and systematic uncertainties indicated by the upper limits of non-variable quiescent galaxies' PSDs. We measured the upper limits of the shortest variability timescale to be $\log(t_{min,ul}/hrs)=0.85\pm0.55$. We compared these upper limits to a range of theoretical AGN variability timescales, and the natural interpretation of our measured $t_{min,ul}$ is the light crossing scale from a coherently varying region, where the measured $t_{min,ul}$ corresponds to the range from a few to thousands of gravitational radii. A significant fraction of these light crossing scales is smaller than the accretion disk emission sizes measured by quasar microlensing, reverberation mapping, or theoretical accretion disk models. Since we only measure the upper limits, the true physical shortest variability timescales are even shorter. We measure the power law index to be $2.0\pm0.2$, and find weak anticorrelations with the black hole mass and luminosity. Our analysis suggests that the shortest optical variability is driven by a compact region smaller than the accretion disk size, potentially by X-ray reprocessing. Alternatively, this shortest timescale variability suggests that the accretion disk can be inhomogeneous potentially caused by turbulence from magnetorotational instability or magnetic reconnections. (abridged)

Magnetic fields and turbulence may play a crucial role in the evolution of molecular clouds and ultimately in the formation of dense cores and stars. Despite being studied in many molecular clouds, the exact role of magnetic fields and turbulence in star formation is still poorly understood. Here, we report the high resolution plane of sky magnetic field (B_pos) morphology toward the high mass star forming region G327.29, obtained with the 12-meter of the Atacama Large Millimeter/sub-millimeter Array (ALMA) telescope. From our analysis, we obtain a complex B_pos morphology where the magnetic field orientation is uniformly distributed across the entire range from -90 to +90 deg. The observed area is composed of one filament and one dense central clump, which harbor multiple dense cores. The total magnetic field strengths (B_tot) in these regions are 1.4 \pm 0.7 mG and 2.0 \pm 0.8 mG at a number density (n) of 6.8 \pm 1.5 x 10^5 and 1.1 \pm 0.3 x 10^6 cm^-3 , derived from the angular dispersion function (ADF) method. The virial parameters (α vir )in these regions are 7.7 \pm 7.1 and 0.7 \pm 0.6, suggesting that the regions may be gravitationally bound or unbound after accounting for the errors. Moreover, the ratio of turbulent to magnetic energy (~ 0.25) indicates that the magnetic field is dynamically more important than turbulence. The relative influence of turbulence and magnetic fields on core dynamics appears to depend on how the B_tot scales with gas density (\r{ho}) in the densest regions. In summary, this work presents a comprehensive analysis of the relative roles of magnetic fields, turbulence, and gravity in regulating high-mass star formation in G327.29, enabled by high-resolution ALMA observations.

The detectability of $\rm{C_6H_5CN^+}$ (benzonitrile cation) in the diffuse ISM is re-evaluated. A holistic evidentiary framework suggests $\rm{C_6H_5CN^+}$ is relatively absent in the diffuse ISM owing to the following concurrently: a marginal intramolecular vibrational energy redistribution (IVR) favoring fragmentation, recurrent fluorescence being an improbable mechanism in this case to prevent dissociation, mismatches between observed DIBs and experimental results, and the hitherto absence of DIBs matching any similarly sized cations. The putative gap in bottom-up synthesis is reaffirmed (diffuse ISM), and although DIB sources are largely unknown, within a broader approach the lines can help benchmark astrochemistry paradigms. The results relied on new advantageous Daly et al. experimental spectra, an expanded observational DIB analysis (APO catalog), and complementary $ω$B97X-D/cc-pVTZ computations.

The orientation of triaxial galaxy clusters with respect to the line-of-sight is expected to be one of the prime sources of scatter and potential bias in optical observables (e.g., richness and weak-lensing signal) of galaxy clusters. In this work, we use the observed shape of the central Brightest Cluster Galaxy (BCG) as proxy for the orientation along the line-of-sight for clusters selected via the Sunyaev-Zel'dovich (SZ) effect from the South Pole Telescope (SPT) and Atacama Cosmology Telescope (ACT) surveys, matched to optically selected clusters from the Dark Energy Survey Year 3 (DES). We construct two samples of clusters that are designed to be identical in SZ mass estimate and redshift but with the roundest vs. the most elliptical BCGs, which we expect to correspond to BCGs (and clusters) with major axes aligned along the line-of-sight vs. in the plane of the sky, respectively. We find that the optical richness of round-BCG clusters is $\sim 10$\% larger than that of elliptical-BCG clusters, in agreement with the expectation from projection effects and presenting the first such detection in data. The density profiles, however, are not in agreement with the expectation from projection effects: the 1-halo term (below $6~h^{-1}\rm{Mpc}$) of both the weak-lensing and galaxy density profiles are the same for the subsamples, contrary to previous studies based on X-ray selected clusters. In the 2-halo regime (above $6~h^{-1}\rm{Mpc}$), we find a significant excess of the elliptical-BCG cluster profiles compared to the round-BCG cluster profiles, which is the opposite of the expectation from numerical simulations. We hypothesize that the intrinsic shape of the BCG reflects not just the orientation angle, but also intrinsic properties of the cluster which can affect both the SZ signal and the amplitude of the 2-halo term.

Polycyclic aromatic hydrocarbons (PAHs) are key dust components in galaxies and play a fundamental role in the physics of the interstellar medium (ISM), yet their response to AGN feedback remains debated. We present a spatially resolved analysis of PAHs in the central $7^{\prime\prime}\times12^{\prime\prime}$ ($\sim100\times200$ pc$^2$) of Centaurus A. We use JWST/MIRI-MRS observations at 5-28 $μ$m from the MIRI European consortium GTO program MICONIC, with angular resolution of $0.35^{\prime\prime}-1^{\prime\prime}$ (about 6-17 pc). We derive PAH moment-0 maps via local continuum subtraction and extract one-dimensional spectra from five regions of interest, including the nucleus, the circumnuclear disc, and a PAH-deficient region. The spectra are decomposed into continuum, emission lines, and PAHs to measure feature intensities and equivalent widths (EWs). PAH emission is primarily distributed in a ring-like structure with localized enhancements at $\sim40$ pc from the nucleus. A distinct PAH-deficient region is observed to the north-west, roughly perpendicular to the jet axis, and coincident with enhanced ionized-gas velocity dispersion and inflowing molecular streamers. The 11.3/7.7 $μ$m and 6.2/7.7 $μ$m ratios exceed model predictions for pericondensed PAHs, indicating processed populations with more open structures. The 11.3/12.7 $μ$m ratio suggests a dominance of solo hydrogen sites and partial dehydrogenation, particularly in the PAH-deficient region, where shocks likely drive erosion. The largest EWs are found in the ring, while reduced values in the deficient region point to partial destruction; in the nucleus, low EWs are mainly due to continuum dilution.

Dust extinction curves provide one of the main avenues to understanding the detailed nature of dust grains and accounting for the effects of dust on observations of many astrophysical objects. For the first time, spectroscopic ultraviolet (UV) extinction curves are measured in M33 expanding the sample of Local Group galaxies with such measurements to five. These curves are based on Hubble Space Telescope/Space Telescope Imaging Spectrograph spectra and literature photometry from the UV to the near-infrared. The four measured curves show large variations in their UV shapes including their 2175 A bump and UV slope strengths. The average extinction of these four sightlines is lower than the averages for other Local Group Galaxies and does not follow the Milky Way R(V) dependent relationship. The variations between UV extinction shape parameters and gas-to-dust ratios for the M33 sightlines fall within the variations seen in the combined sample of UV extinction curves in the Milky Way, Large and Small Magellanic Clouds, and M31. The correlation with gas-to-dust ratio is much stronger than the correlation with global metallicity. This strengthens the picture that local conditions like radiation field density and shocks dominate over global galaxy properties like metallicity in determining the wavelength dependence of dust extinction.

Disc winds from active galactic nuclei (AGN) can be launched by radiation pressure acting on spectral lines. However, launching a line-driven wind in the X-ray rich environment of AGN is challenging, as the wind easily gets over-ionized. Previous simulations suggested that X-ray self-shielding could enable line driving, though it remained unclear whether this relied on simplified treatments of radiation and ionization. Here, we revisit the X-ray shielding scenario using the first multi-frequency, multi-directional Monte-Carlo radiative photo-ionization hydrodynamical simulations of AGN line-driven winds. We find that sustaining a steady wind with mass-loss rates of $\approx20\%$ of the accretion rate requires an unrealistically weak X-ray flux ($α_{\rm OX}<-3$). For stronger X-ray emission ($-3<α_{\rm OX}<-1$), self-shielding is only transient, leading to episodic ejections with mass-loss rates approaching the accretion rate. Our steady winds naturally produce FeLoBAL, HiBAL, and broad emission line signatures, depending on the disc spectral energy distribution and the observer's inclination. At moderate X-ray luminosities ($α_{\rm OX}\sim-3$), transient winds can generate short-lived BAL and ultra-fast outflow (UFO) features. At the highest X-ray luminosities ($α_{\rm OX}\sim-1$), the winds are too ionized to form BALs, but still produce UFOs. These results imply that additional physics is required to explain BAL outflows at realistic X-ray levels and to drive winds strong enough for AGN feedback. Nonetheless, our simulations provide a new framework for interpreting the observed diversity of AGN outflow signatures with fully coupled radiation and dynamics.

We present preliminary SPHEREx maps of diffuse Galactic emission tracing polycyclic aromatic hydrocarbons (PAHs) and ionized hydrogen gas, and we study their relationship across the Galactic plane. Since its launch in early 2025, the SPHEREx space telescope has been conducting an all-sky near-infrared spectral survey from 0.75 to 5.0 microns. We produce a large-scale map of the 3.3-micron PAH emission feature, which is bright and detectable throughout the Galactic plane, and find a strong correlation with the thermal dust radiance measured by Planck. We also trace ionized hydrogen gas by producing a map of Brackett-alpha emission at 4.05 microns. By combining the two maps, we identify extended shells of PAH emission associated with photodissociation regions surrounding ionized gas. We construct a PAH abundance map and find a significant anticorrelation between PAH abundance and ionized hydrogen, indicating systematic PAH depletion within ionized gas regions across the Galactic plane and demonstrating that ionizing radiation is a dominant driver of PAH abundance variations. These early SPHEREx results provide a large-scale view of PAHs and ionized hydrogen and preview the capability of the mission to map diffuse emission in the interstellar medium.

Today's most massive ellipticals are proposed to originate from starbursting galaxies in $z\gtrsim2$ overdensities. To discern what triggers these starbursts, and their $z=0$ descendants, we performed a detailed case study of five gas-rich galaxies in the $z=2.41$ overdensity, HATLAS J084933.4+021443. Using 0.15" resolution CO(4-3), [C I] 1-0, and dust-continuum observations, we characterised their cold gas morphology and kinematics. We find two rotating discs, W and C, both exhibiting non-axisymmetric radial gas motions (consistent with bars). Of the two extreme starbursts, W is a lopsided, rotation-dominated disc with a rotation velocity of $\sim520$ km s$^{-1}$, whereas T is most likely a late-stage merger. Combined with recent studies, we find that $\gtrsim42\%$ of gas-rich, massive starbursts in overdensities are rotation-dominated discs, a fraction not yet systematically reproduced by galaxy evolution models. Beyond $z=1$, disc galaxies with rotation velocities of $>400$ km s$^{-1}$ reside almost exclusively in overdensities, consistent with early mass assembly in dense environments. By comparing to local early-type galaxies with cold gas discs, we confirm that these systems already reside in halos comparable to the most massive $z\sim0$ ellipticals at the centres of groups and clusters. Despite their extreme star-formation rates, these discs lie on the same $σ-$SFR locus as lower-SFR field galaxies, implying that stellar feedback remains the dominant turbulence driver. We postulate that this is because inflowing gas is effectively transported through ordered streaming, such that only a small fraction of kinetic energy feeds disc-wide turbulence.

Galaxy clusters offer powerful laboratories for studying galaxy evolution in dense environments. In this context, the CLONE, Constrained LOcal and Nesting Environment, project provides a zoom-in hydrodynamical simulation of the Virgo cluster, including AGN and supernovae feedback, with a resolution down to 350 pc, designed to mirror Virgo's observed properties. Previous work showed that this replica and Virgo share the same history, mass and luminosity distributions including the central M87. This study examines several observational relations extending to lower stellar masses than previous synthetic-population studies: star formation density, (specific) star formation rate, metallicity and quenched fraction of galaxies as a function of stellar mass and cluster-centric distance. The aim is to assess how simulated and observed trends compare. Despite slightly low metallicity and high, but then enough, quenched fraction, simulated galaxies reproduce key observational trends even without averaging or accounting for observational uncertainties, aside from the consideration of projection effects: At fixed stellar mass, cluster galaxies form fewer stars than field counterparts. Most galaxies are quenched but for those of intermediate mass or isolated. Low-mass galaxies are highly quenched implying a sharp metallicity drop, and low metallicity does not imply youth. Quenching occurs earlier for the most massive and the smallest galaxies than for those of intermediate mass at least until they enter the cluster. Quenched galaxies have undergone dark matter stripping. Gas depletion drives quenching, especially in low-mass galaxies and the farther from the cluster center they are. Overall, the synthetic population reproduces jointly multiple observational trends, making it a valuable tool to probe processes from jellyfish galaxies to cluster-core gas dynamics. [Shorten]

Estimates for the total number of Milky Way (MW) satellites are often generated from a combination of the observed number of satellites in surveys, adjustments for the completeness of those surveys, and theoretical expectations from halo assembly modelling. One of the features of this modelling is disruption by the MW stellar disc. We examine the effect of degrees of disc disruption on inferred satellite counts, by means of an N-body simulation of a MW-mass halo plus a toy model for this disruption. We use a fictional all-sky survey to show that high resilience to disc disruption predicts small populations of satellites that are radially very concentrated around the central galaxy and are hosted by massive subhaloes, while low resilience predicts many more satellites with a less concentrated radial distribution and hosted within less massive subhaloes. We show that the most massive subhaloes are particularly susceptible to disruption due to their radial orbits, and in their putative absence galaxy formation must occur in lower mass haloes that have a shallower radial number density profile. We then demonstrate this phenomenon for a combination of the Pan-STARRS and DES surveys. It is therefore necessary to account for uncertainty in the disc disruption radius when making predictions for MW satellite distributions.

We present a comprehensive analysis of the host galaxies of 11 luminous fast blue optical transients (LFBOTs). We model new and archival host photometry and spectroscopy with Prospector. We determine that all LFBOT hosts are actively star-forming with recent bursts of star formation and have a median stellar mass of $\log(M_*/M_\odot)=9.61^{+0.74}_{-1.61}$, present-day star formation rate SFR=$0.95^{+18.37}_{-0.91} M_\odot$yr$^{-1}$, and gas-phase oxygen abundance metallicity 12+log(O/H)=$8.71^{+0.17}_{-0.40}$. To contextualize these results, we compare them to the host properties of Hydrogen-poor superluminous supernovae (SLSNe-I), several core-collapse supernova subtypes (CCSN; SNe Ibc, II, and Ibn) and long gamma-ray bursts (LGRBs). We find that LFBOT hosts are more star-forming than CCSN hosts, but less star-forming than SLSN-I hosts. We further show that LFBOT hosts are more metal-poor than SN Ibc and II hosts, but more metal-rich than SLSN-I and LGRB hosts. Finally, we find that, similar to SLSNe-I and unlike CCSNe and LGRBs, a large fraction (>30%) of LFBOTs occur in their hosts' faintest pixel or outside their host galaxy's light. Our results indicate that LFBOTs have massive stellar origin that do not trace active star-forming regions within their hosts and have a weaker metallicity-dependence than other extreme transients. For these reasons, we favor a compact-object and Wolf-Rayet star merger progenitor scenario. Future discoveries of LFBOTs with the Rubin observatory will help to increase their sample size and place firmer constraints on their environments and progenitors.

Most galaxies closely follow the radial acceleration relation (RAR), which tightly links the observed accelerations to those predicted by Newtonian gravity from visible baryonic matter. Galaxy clusters, however, deviate from this relation. Several explanations have been proposed. Some of them predict that even some galaxies in clusters should deviate, but this hypothesis remains largely untested. We test it here by analyzing acceleration profiles for 17 early-type galaxies, derived from Jeans modeling of their globular cluster systems in our older work. Our sample spans central galaxies in clusters and groups, non-central galaxies, isolated ones, and-uniquelly for this paper-centrals in galactic subclusters, which are smaller clusters being accreted by larger ones. We compare these profiles to the standard RAR for non-cluster galaxies and its counterpart for clusters. We find that isolated and non-central galaxies adhere to the standard RAR. In contrast, central galaxies of clusters, subclusters, and groups exhibit enhanced accelerations in most cases, tracing instead the cluster acceleration behavior either partly or fully. The radius at which divergence from the standard RAR begins tends to decrease with increasing group mass. These findings imply that if cluster fields depart from the standard RAR due to undetected material, it must be dynamically cold and collisionless, such as non-baryonic cold dark matter, but also compact clouds of cold gas.

The vertical phase-space spirals in the Milky Way are clear evidence of disequilibrium. However, they are challenging to study because phase mixing signals evolve under the influence of many different dynamical processes and can be driven by many sources of disequilibrium. We characterize phase spirals in two simulations -- one test particle and one N-body -- with basis function expansions, using these to derive winding times ($T_{\rm fit}$). We find that phase spirals in the test particle simulation wind up as expected from pure phase mixing theory while those in the self-consistent simulation do not. Specifically, in the N-body simulation we find that (i) the onset of winding is delayed, (ii) the winding rate is slowed, and (iii) the rate of winding oscillates with time. The extent of these effects depends on the azimuthal action $J_φ$ of the phase spiral region. We build some physical intuition for these effects through 1-D toy models which follow a group of co-moving stars traveling through several different evolving potentials. We find that phase spiral winding can be delayed until the group no longer moves coherently with the midplane of the (perturbed) potential and oscillates with time as the group experiences (e.g.) a breathing mode traveling through the disk. Rates of winding change as the vertical structure of the disk evolves. The modifications to winding are strongest in the inner galaxy where the disk potential dominates. We conclude that in the Milky Way, all calculations of the winding time should be interpreted as lower limits and that the most trustworthy winding times are likely in the outer disk.

The Perseus cluster is a nearby cool-core galaxy cluster that hosts an archetypal radio mini-halo. Recent Low Frequency Array (LOFAR) High Band Antenna (HBA) observations at 120 - 168 MHz have revealed the presence of a giant radio halo within the cluster with a size of 1.1 Mpc enveloping the mini-halo. By exploring the spectral properties of the radio emission at low frequencies, we can gain deeper insights into the nature of this emission and improve our understanding of its origin. Here we present LOFAR Low Band Antenna (LBA) images of the cluster between 30.0 - 57.7 MHz, with a resolution of 19.2'' x 15.0'' and a r.m.s. noise of 3.7 mJy/beam . In our images, we detect both the mini-halo and giant radio halo. We measured the spectral indices between 44 and 144 MHz of the mini-halo and giant radio halo to be -1.34 +- 0.10, and -1.01 +- 0.11, respectively. An alternative and more direct measurement of the spectrum of the giant radio halo results in a spectral index of -1.28 +- 0.15. The discrepancy between both values is caused by the poor ionospheric conditions. In addition, we study two X-ray 'ghost cavities' in the cluster. These cavities are thought to have been produced by an older outburst from the central AGN 3C 84. We measure a spectral index between 44 and 144 MHz for the radio plasma in these cavities of -1.86 +- 0.12 and -1.90 +- 0.12 for the northwest and southern ghost cavities, respectively. Furthermore, by including VLA 352 MHz data, we find that the spectrum steepens at higher frequencies. These results are consistent with the ghost cavities being filled with old and aged radio plasma. We also detect the tailed radio galaxies NGC 1265 and IC 310. In our analysis, these sources show signs of spectral steepening along their tails.

Because direct measurements require resolved stellar populations including low-mass stars, determining the stellar initial mass function (IMF) has been a historically difficult problem even within our own Galaxy and impossible everywhere else. As a result, even though it is predicted that the IMF should vary depending upon the properties of each individual star-forming molecular cloud, it is standard to assume a Universal IMF. Using recent observations from {\em Gaia}, it is now possible to test for IMF variation using resolved stellar populations in open clusters and a parameterization that separates properties of the IMF from subsequent dynamical evolution. Here, we show that the IMF is not Universal but instead varies across individual Galactic stellar populations, reflecting evolution in the average conditions of molecular clouds over cosmic time. This evolution is consistent with the predictions of a simple astrophysical model in which the IMF is environmentally-dependent and the Milky Way reflects typical galactic behavior in recent cosmic history. Thus, observational evidence now agrees with long-standing theoretical and numerical predictions.

Omega Centauri ($ω\,$Cen) is the most massive and chemically complex star cluster in the Milky Way and is widely regarded as the surviving nuclear star cluster of an accreted dwarf galaxy. However, its parent host remains uncertain. Here, we investigate a scenario in which Sequoia, Thamnos, and Gaia--Enceladus (GE) are debris from a single disrupted progenitor, the $ω\,$Dwarf, whose nucleus survives today as $ω\,$Cen. Using APOGEE and GALAH abundances together with Gaia astrometry, we reconstruct the chemical structure across this progenitor adopting orbital energy as a proxy for pre-merger radius. We find that the chemically evolved (younger Al-N-He-rich) population is strongly concentrated toward the inner regions, representing a population formed after/during the merger, while the primordial population represents a dwarf-galaxy-like population, supporting a common dwarf-galaxy origin for its components. The metallicity profile shows an inverted U-shaped gradient similar to those observed in present-day nucleated dwarf galaxies. At the same time, the inner regions ($ω\,$Cen+Thamnos) are more $α$-enhanced than the outskirts, pointing to shorter and more efficient star formation and indicating that the nucleus may have assembled through the merger of inspiraling globular clusters. Neutron-capture abundances reveal a Eu-rich, r-process-dominated outskirts and inner regions enhanced in [Ba/Eu] and [La/Eu], requiring delayed enrichment and more complex chemical evolution. Finally, our analysis shows that Sequoia and Thamnos naturally fit an outside-in stripping sequence around $ω\,$Cen, whereas the connection with GE remains unsure.

In most galaxies, the central black hole accounts for no more than a percent of the total mass in stars. Recently, however, extremely over-massive black holes with ratios of 10% have been reported in dwarf galaxies at z<1 and at cosmic dawn (z>5.5) by JWST. Both findings have been interpreted as signatures of the still mysterious origins of super-massive black holes, such that most of the black hole mass was built at birth rather than through black hole accretion. Here we show that among evolved galaxies over-massive black holes are also present, indicating that overmassive BHs are not a signature unique to black hole formation channels. The first large-area sky survey of the eROSITA X-ray telescope on board SpectrRG identified 200 quasars by their luminous hard X-ray radiation. These signpost rapidly growing black holes. Complementary optical spectroscopy from the Sloan Digital Sky Survey and archival UV to IR photometric data combined with galaxy-quasar decomposition techniques allow us unbiased estimates of cosmological distances, black hole masses and host galaxy stellar masses. We securely identify a sample of over-massive black holes with BH-to-host ratios of more than 5%, which may have undergone exponential accretion spurts lasting about a billion years. Our survey identified a high space density of at least 4/Gpc^3 of overmassive black holes near cosmic noon. This indicates an accretion channel disconnected from the stellar population that cause strong deviations from galaxy scaling relations. This channel is currently not part of galaxy evolution models. The identified channel, if applicable also for the first billion years of cosmic time, can explain JWST AGN without requiring them to signify imprints of black hole seeding mechanism.

We examine solutions of the hydrodynamic equations for dark matter (DM) modeled as a Bose-Einstein condensate (BEC) with axionlike interaction, forming a spherically symmetric halo in dwarf galaxies. Small perturbations and decoherence of the BEC DM arise from changes in the gravitational background induced by subgalactic baryonic processes. Focusing on the events in the central region of a galaxy, overlapping with the stable DM core, we consider three scenarios: (i) expansion of a gaseous shell mimicking stellar explosions, (ii) collapse of a shell modeling star formation, and (iii) contraction of a stellar cluster toward the galactic center, driven by dynamical friction within a gaseous shell. Numerical parameters are extracted from observational data for NGC 2366. Our results show central DM density increases of 0.01 percent and DM wind velocities of up to several meters per second. A greater increase in density is observed at lower wind speeds and vice versa. These results raise the question of whether minor DM variations significantly affect star formation. In analyzing the fate of the cumulative impact of baryonic processes, we turn to the quantum excitation model with a discrete spectrum in finite volume. In the inhomogeneous DM halo, including unstable phase, metastable excitations associated with false vacuum states decay over 32 million years. This induces the decay of the system's evolutionary operator. Meanwhile, the Beliaev damping, originating from the decay of stable quasiparticles, emerges in the next order of perturbation.